<%BANNER%>

Functional Ability Profiles and Young Children's Social Competence

Permanent Link: http://ufdc.ufl.edu/UFE0043247/00001

Material Information

Title: Functional Ability Profiles and Young Children's Social Competence Exploring Relationships in the Pre-elementary Education Longitudinal Study Data Set
Physical Description: 1 online resource (365 p.)
Language: english
Creator: Mclaughlin, Tara W
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: ability -- children -- competence -- functional -- person-oriented -- social -- young
Special Education, School Psychology and Early Childhood Studies -- Dissertations, Academic -- UF
Genre: Special Education thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Children with social competence have the skills to achieve social goals, they know when to use appropriate behaviors in social contexts, and they refrain from inappropriate behavior in social contexts. Social competence has been identified as a priority outcome for young children with disabilities. Research has been conducted to examine relationships between various child and contextual factors and social competence. The International Classification of Functioning, Disability, and Health for Children and Youth (ICF-CY) provides a framework to describe and quantify children's functional abilities and to explore associations among functioning, contextual factors, and child outcomes. The present study was a correlational study conducted through secondary analyses of the Pre-Elementary Education Longitudinal Study (PEELS) data set. The purpose of the present study was to combine an ICF-CY approach to quantify children's functional abilities with person-oriented analytic techniques (i.e., latent class analysis) to derive empirically subgroups of children with similar profiles of functional abilities. Regression analyses were conducted to examine relationships between subgroup membership and children's social competence, variance in social competence explained by subgroup membership and disability category, and if contextual factors moderated relationships between subgroup membership and social competence. Results from the latent class analyses showed that five subgroups emerged. Each subgroup had a distinct and interpretable functional ability profile. Functional ability profile subgroup membership was associated with children's social competence outcomes (social skills R2 = .20 and problem behaviors R2 = .115), and notable differences in these outcomes were identified between subgroups. Subgroup membership accounted for more variance in social competence outcomes than disability category. Four contextual factors moderated the relationship between children's functional ability profile subgroup membership and social skills or problem behaviors. The present study provides information about the prevalence and nature of different functional ability profiles found in a nationally representative sample of young children with disabilities. Findings demonstrate the use of person-oriented analyses combined with a functional approach shows promise for identifying subgroups of children with similar characteristics and for examining associations among functional abilities, contextual factors, and important outcomes. Results help inform policy and practice related to characterizing and quantifying children's functional abilities.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Tara W Mclaughlin.
Thesis: Thesis (Ph.D.)--University of Florida, 2011.
Local: Adviser: Snyder, Patricia Ann.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2013-06-30

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0043247:00001

Permanent Link: http://ufdc.ufl.edu/UFE0043247/00001

Material Information

Title: Functional Ability Profiles and Young Children's Social Competence Exploring Relationships in the Pre-elementary Education Longitudinal Study Data Set
Physical Description: 1 online resource (365 p.)
Language: english
Creator: Mclaughlin, Tara W
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: ability -- children -- competence -- functional -- person-oriented -- social -- young
Special Education, School Psychology and Early Childhood Studies -- Dissertations, Academic -- UF
Genre: Special Education thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Children with social competence have the skills to achieve social goals, they know when to use appropriate behaviors in social contexts, and they refrain from inappropriate behavior in social contexts. Social competence has been identified as a priority outcome for young children with disabilities. Research has been conducted to examine relationships between various child and contextual factors and social competence. The International Classification of Functioning, Disability, and Health for Children and Youth (ICF-CY) provides a framework to describe and quantify children's functional abilities and to explore associations among functioning, contextual factors, and child outcomes. The present study was a correlational study conducted through secondary analyses of the Pre-Elementary Education Longitudinal Study (PEELS) data set. The purpose of the present study was to combine an ICF-CY approach to quantify children's functional abilities with person-oriented analytic techniques (i.e., latent class analysis) to derive empirically subgroups of children with similar profiles of functional abilities. Regression analyses were conducted to examine relationships between subgroup membership and children's social competence, variance in social competence explained by subgroup membership and disability category, and if contextual factors moderated relationships between subgroup membership and social competence. Results from the latent class analyses showed that five subgroups emerged. Each subgroup had a distinct and interpretable functional ability profile. Functional ability profile subgroup membership was associated with children's social competence outcomes (social skills R2 = .20 and problem behaviors R2 = .115), and notable differences in these outcomes were identified between subgroups. Subgroup membership accounted for more variance in social competence outcomes than disability category. Four contextual factors moderated the relationship between children's functional ability profile subgroup membership and social skills or problem behaviors. The present study provides information about the prevalence and nature of different functional ability profiles found in a nationally representative sample of young children with disabilities. Findings demonstrate the use of person-oriented analyses combined with a functional approach shows promise for identifying subgroups of children with similar characteristics and for examining associations among functional abilities, contextual factors, and important outcomes. Results help inform policy and practice related to characterizing and quantifying children's functional abilities.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Tara W Mclaughlin.
Thesis: Thesis (Ph.D.)--University of Florida, 2011.
Local: Adviser: Snyder, Patricia Ann.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2013-06-30

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0043247:00001


This item has the following downloads:


Full Text

PAGE 1

1 F UNCTIONAL ABILITY P ROFILES COMPETENCE : EXPLORING RE L A TIONSHIPS IN THE PRE ELEMENTARY EDUCATION LONGITUDINAL STUDY DATA SET By TARA WAY MCLAUGHLIN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2011

PAGE 2

2 2011 Tara Way McLaughlin

PAGE 3

3 I n mem ory of Robert Patten McLaughlin

PAGE 4

4 ACKNOWLEDGMENTS There are many people who have shaped my education and career, and each of them have contributed to my knowledge and learning throughout the doctoral journey I have no doubt that these individuals will conti nue to shape my education and career throughout my life For their guidance, support, encouragement, tough love, and kindness I am very thankful T he gratitude that I can express on this page however, will never come close to the appreciation, admirat ion, and respect that I have for those who have shaped my path First, I want to give special thanks to my husband, mother sister, and brother for their unconditional support, love, and sacrifice that has allowed me to pursue my passi on and work I learn from them each and every day and I am lucky to have them as my family I have also be family that has been a tremendous support The opportunities to connect and work with researchers, faculty, and do ctoral students both here at UF and from around the county has enriched my education and my life My deepest thanks and appreciation goes to Pat ricia Snyder my chair, for her unconditional support of my interests, my career, and me Pat is a mentor, a colleague Pat knows how to challenge me, support me, and help me be successful all at the same time She is a patient, supportive, and thoughtful mentor, and t he opportunity to work with Pat can only be described in a million Pat has shown me how to think carefully and deeply about the intersections among research, practice, and policy. She instills in her students a sense of curios i ty and questioning of ideas that she supports with her knowledge and he r own careful examination of ideas through research, implementation in practice, and evaluation of

PAGE 5

5 policy. I am truly grateful for her guidance, wisdom, and support to complete this dissertation research and many endeavors during my doctoral program. I am thankful to all my committee members, each who ha s contributed to my education in unique and important ways I thank James Algina for his tremendous support and encouragement through his dedication of time, mentoring, teaching, and, when needed, repeated explanations I have greatly enjoyed work ing with Dr Algina and consider myself lucky to have been one of his students I thank Maureen Conroy for her willingness to give support as she was still getting to know me and for providing thoughtful and hel pful feedback I thank Diane Ryndak for sharing her passion of education and advocacy for children and adults with disabilities And I thank Maria Denney for her encouragement and support I thank Bob Crow for being a life long teacher and life long learner I have benefitted from his knowledge and advice I thank Dave Lawrence for his support of early childhood studies and the University of Florida I am thankful for the opportunities created because of his passion and dedication to education I thank Salih Rakap, Crystal Crowe, Cathy Pasia, Cinda Clark, Dana Kasian, and Katrina Moore for making our team a great team I want to give special thanks to Katrina Moore for being a great friend and colleague Her willingness to listen and her support have helped me every step of the way I also thank Elaine Carlson and the PEELS staff at Westat for their help with technical and substantive questions about the data set Thank you to Tamara Daley and Elaine Carlson for sharing the syntax to create the PEELS Severity Index In addition, thank you to the American Educational Research Association and their

PAGE 6

6 Governing Board for their support to conduct this research This research was supported by a grant from the American Educational Research Association which receives funds for its "AERA Grants Program" from the National Science Foundation under Grant #DRL 0941014 Opinions reflect those of the author and do not necessarily reflect those of the granting agencies.

PAGE 7

7 TABLE OF CONTENTS p age ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ .......... 11 LIST OF FIGURES ................................ ................................ ................................ ........ 14 LIST OF ABBREVIATIONS ................................ ................................ ........................... 15 LIST OF CONCEPTUAL DEFINITIONS ................................ ................................ ........ 16 LIST OF OPERATIONAL DEFINITIONS ................................ ................................ ....... 19 ABSTRACT ................................ ................................ ................................ ................... 21 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 23 Backgr ound ................................ ................................ ................................ ............. 23 Problem Statement ................................ ................................ ................................ 26 Purpose of the Study ................................ ................................ .............................. 30 Conceptual Framework ................................ ................................ ........................... 32 Context for the Study ................................ ................................ .............................. 34 Research Questions ................................ ................................ ............................... 36 Rationale for the Study ................................ ................................ ........................... 36 Social Competence as Key Child Outcome ................................ ...................... 37 Social Competence Framework ................................ ................................ ....... 38 ................................ ... 39 Measuring and Examining Social Competence ................................ ................ 41 Concerns with Categorical Descriptions of Disability ................................ ........ 42 Functional Approaches as Alternate Method for Describing Child Characteristics ................................ ................................ .............................. 43 Methods to Create Subgroups of Children with Similar Profiles of Abilities ...... 45 Ne ed for Additional Studies Using the PEELS Data Set ................................ .. 4 6 Need for Empirical Studies of Social Competence and Functional Characteristics ................................ ................................ .............................. 47 Methodological Rationale ................................ ................................ ................. 49 Summary of Rationale ................................ ................................ ...................... 50 Importan ce of the Study ................................ ................................ .......................... 51 Definitions of Terms ................................ ................................ ................................ 52 Delimitations ................................ ................................ ................................ ........... 52 Limitations ................................ ................................ ................................ ............... 54 Summary ................................ ................................ ................................ ................ 55

PAGE 8

8 2 REVIEW OF THE LITERATURE ................................ ................................ ............ 58 Search Procedures ................................ ................................ ................................ 59 Social Competence ................................ ................................ ................................ 60 Defining S ocial Competence ................................ ................................ ............ 61 Challenges Related to the Measurement of Social Competence ..................... 62 Outcomes Associated with Social Competence ................................ ............... 63 Child Factors Associated with Social Competence ................................ .......... 65 Contextual Factors Associated with Social Competence ................................ 67 Summary Related to Social Competence ................................ ......................... 70 Use of Disability Categories to Characterize Children ................................ ............ 71 Concerns with Categorical Approaches ................................ ........................... 72 Current Variations in Eligibility Determination Systems ................................ .... 74 Illustration of Concerns and Implications for Research ................................ .... 77 Summary Related to Categorical Descrip tions of Disability .............................. 79 The ICF CY Framework ................................ ................................ .......................... 80 History of the ICF CY ................................ ................................ ....................... 81 ICF CY Applications ................................ ................................ ......................... 85 ICF CY Usability ................................ ................................ ............................... 86 ICF CY in Educational Contexts ................................ ................................ ....... 87 Summary Related to ICF CY ................................ ................................ ............ 90 PEELS Data Set ................................ ................................ ................................ ..... 90 Public Information from the PEELS Reports ................................ ..................... 91 Studies Conducted with the PEELS Data Set ................................ .................. 95 Summary Related to Peels Data Set ................................ ................................ 98 Empirical Studies with Direct Relevance for the Present Study .............................. 99 Studies that Used Functional Ability Composite Scores ................................ 100 Person Oriented Analytic Approaches to Examine Child Outcomes .............. 103 Summary of Empirical Studies Related to Present Study ............................... 114 Social competence ................................ ................................ ................... 115 Functional ability compared to disability category ................................ .... 115 Person oriented approaches to identify subgroups ................................ .. 116 Examination of contextual factors ................................ ............................ 118 Summary ................................ ................................ ................................ .............. 119 3 METHODOLOGY ................................ ................................ ................................ 131 Research Quest ions ................................ ................................ ............................. 132 Research Design ................................ ................................ ................................ .. 132 Hypothesized Relationships ................................ ................................ .................. 133 PEELS Study and Data Set ................................ ................................ .................. 136 PE ELS Sampling Strategy ................................ ................................ .............. 137 PEELS Instrumentation ................................ ................................ .................. 139 PEELS Response Rates and Imputation for Missing Data ............................. 141 PEELS Variables Selected for Analysis in the Present Study ............................... 141 Criterion Variables ................................ ................................ .......................... 142 Explanatory Variables ................................ ................................ .................... 144

PAGE 9

9 ICF related functional profiles ................................ ................................ .. 144 Disability category ................................ ................................ .................... 149 Descriptive Variables ................................ ................................ ...................... 150 Child factors ................................ ................................ ............................. 151 Family factors ................................ ................................ ........................... 152 Environmental factors ................................ ................................ .............. 154 Procedures ................................ ................................ ................................ ........... 155 Data File Preparation ................................ ................................ ..................... 155 Primar y Sampling Unit, Stratification, and Sampling Weights ........................ 156 Missing Data ................................ ................................ ................................ ... 157 Analyses ................................ ................................ ................................ ......... 158 Research Question 1 ................................ ................................ ............... 158 Research Question 2 ................................ ................................ ............... 159 Research Question 3 ................................ ................................ ............... 160 Research Question 4 ................................ ................................ ............... 161 Statistical Software ................................ ................................ ......................... 161 Summary ................................ ................................ ................................ .............. 161 4 RESULTS ................................ ................................ ................................ ............. 169 Context for Reporting and Interpreting Findings ................................ ................... 169 Research Question 1 ................................ ................................ ............................ 170 Data Analyses to Conduct Latent Class Analyses ................................ .......... 171 Generating and Evaluating Latent Class Models ................................ ............ 173 Interpreting the Selected Latent Class Model ................................ ................. 176 Severity and number of limitations on functional ability variables ............ 176 Types of limitations on functional ability variables ................................ .... 177 Probability of being assigned to a profile ................................ ................. 179 Functional ability profile summary ................................ ............................ 180 Demographic and Descriptive Information for Members of Each Subgroup ... 180 Research Question 2 ................................ ................................ ............................ 185 Using Most Probable Class Membership to Examine Social Competence ..... 186 Relationships Between Subgroup Membership and Social Skills and Problem Behaviors ................................ ................................ ...................... 187 Resear ch Question 3 ................................ ................................ ............................ 190 Associations Between Explanatory Variables ................................ ................. 191 Estimating Individual and Combined Contributions of Explanatory Variables 192 Individual and Combined Contributions of Explanatory Variables .................. 193 Research Question 4 ................................ ................................ ............................ 194 Examination of Non Malleable Child Factors and Contextual Factors ............ 195 Moderating Influence of Contextual Factors on the Relationship Between Subgroup Membership and Social Skills ................................ ..................... 196 Moderating Influence of Contextual Factors on the Relationship Between Subgroup Membership and Problem Behaviors ................................ .......... 198 Summary ................................ ................................ ................................ .............. 201 5 DISCUSSION ................................ ................................ ................................ ....... 232

PAGE 10

10 Interpretation of Study Findings ................................ ................................ ............ 233 Characteristics of Latent Class Subgroups Based on Functional Ability Profiles ................................ ................................ ................................ ........ 233 Profile patterns in relation to functional ability variables ........................... 236 Profile patt erns in relation to child, family, and school variables .............. 240 Comparison of the selected latent class model with existing studies ....... 242 Social Skills and Problem Behaviors ................................ ........................... 247 Social Skills and Problem Behaviors ................................ ........................... 249 Disability Category Compared to Subgroup Membership ............................... 253 Non Malleable Child Factors and Contextual Factors ................................ .... 257 Implications of the Present Study ................................ ................................ ......... 261 Conceptual Implications of the Findings ................................ ......................... 262 Methodological Implications of the Findings ................................ ................... 263 Practical Implications of the Findings ................................ ............................. 265 Policy and Research Implications of the Findings ................................ .......... 267 Recommendations for Future Research ................................ ............................... 268 Summary ................................ ................................ ................................ .............. 270 APPENDIX A ................................ .......... 274 B PEELS DISABILITY SEVERITY INDEX VARIABLES ................................ .......... 277 C CHILD, FAMILY, AND ENVIRONMENTAL VARIABLES ................................ ...... 281 D VARIABLE CODING AND ANALYTIC SYNTAX ................................ ................... 291 E STATISTICAL MODELS ................................ ................................ ....................... 332 F LATENT CLASS MODELS NOT SELECTED ................................ ....................... 335 REFERENCES ................................ ................................ ................................ ............ 346 BIOGRAPHICAL SKE TCH ................................ ................................ .......................... 365

PAGE 11

11 LIST OF TABLES Table p age 1 1 Studies related to aspects of present study ................................ ........................ 56 2 1 ............................... 120 2 2 Population estimates across disability categories by state: Minnesota, Washington, and Wisconsin ................................ ................................ ............. 121 2 3 International Classification of Functioning, Disability, and Health (ICF) chapters ................................ ................................ ................................ ............ 122 2 4 Selected codes of ICF classification system: Body functions component ......... 123 2 5 Studies related to aspects of present study ................................ ...................... 124 2 6 Predictor, criterion, and contextual variables and related analysis included in studies ................................ ................................ ................................ .............. 126 3 1 ................................ 163 3 2 Response rates in the PEELS data set ................................ ............................ 164 3 3 PEELS Disability Severity Index and related ICF CY codes ............................. 165 4 1 Proportion of sample for each response category for PEELS Disability Severity Index ................................ ................................ ................................ ... 204 4 2 Proportion of sample for each response category used in latent class analysis ................................ ................................ ................................ ............ 205 4 3 Model fit statistics for two through seven class models ................................ .... 206 4 4 Model implied means (standard deviations) for 5 class models ....................... 207 4 5 Average latent class posterior probabilities for most likely class membership .. 208 4 6 Demographic and descriptive information for children within each profile ........ 209 4 7 Demographic and descriptive information for families within each profile ......... 210 4 8 Descriptive information about child and family activities within each profile ..... 211 4 9 Descriptive information about programs or schools for children and families within each profile ................................ ................................ ............................. 212 4 10 Comparative demographic information for samples used in analyses. ............. 213

PAGE 12

12 4 11 PKBS 2 composite scores for each profile ................................ ....................... 214 4 12 PKBS 2 sub scale scores for each profile ................................ ........................ 215 4 13 PKBS 2 composite scores standardized mean difference effect sizes ............. 216 4 14 Percentage of total sample by p rofile and disability category ........................... 217 4 15 Percentage of profile sample by disability category ................................ .......... 218 4 16 Percentage of total sample for disability categories within the low incidence disability category by disability category and profile ................................ ......... 219 4 17 Variance accounted with the addition of explanatory variables: R 2 .................. 220 4 18 Hold out analyses for R 2 ................................ ................................ ................... 221 4 19 Modified categorical coding for moderation analyses ................................ ....... 222 4 20 Moderation of differences among profiles on social skills by non malleable child factors and contextu al factors ................................ ................................ .. 223 4 21 Moderation of differences among profiles on problem behaviors by non malleable child factors and contextual f actors ................................ .................. 224 4 22 Mean problem behaviors standard scores by race/ethnicity and profile ........... 225 4 23 Mean problem behaviors standard scores by parent rating of neighborhood safety and profile ................................ ................................ .............................. 226 A 1 Functional profile: Child A ................................ ................................ ................. 275 A 2 Functional profile: Ch ild B ................................ ................................ ................. 275 A 3 Functional profile: Child C ................................ ................................ ................ 276 A 4 Functional profile: Child D ................................ ................................ ................ 276 B 1 PEELS Disability Severity Index variables ................................ ........................ 277 C 1 Child factor variables from PEELS data set ................................ ...................... 281 C 2 Family characteristics variables from PEELS data set ................................ ..... 282 C 3 Parent child interaction variables from PEELS data set ................................ ... 284 C 4 Environmental factor variables from the PEELS data set ................................ 288 F 1 Model implied means (standard deviations) for 2 class models ....................... 342

PAGE 13

13 F 2 Model implied means (standard deviations) for 3 class models ....................... 343 F 3 Model implied means (standard deviations) for 4 class models ....................... 344 F 4 Model implied means (standard deviations) for 6 class models ....................... 345

PAGE 14

14 LIST OF FIGURES Figure p age 1 1 International Classification of Functioning, Disability, and Health (WHO 2007).. ................................ ................................ ................................ ................ 57 2 1 International Classification of Impairments, Disability, and Handicaps (ICIDH) framework. ................................ ................................ ................................ ........ 129 2 2 International Classification of Functioning, Disability, and Health framework (WHO, 2007). ................................ ................................ ................................ ... 130 3 1 Hypothesized relationship betw een functional ability variables, functional .......................... 166 3 2 Hypothesized rel ationships between functional profile membership, disability category membership, and social competence ................................ ................. 167 3 3 Hypothesized moder ator relationships involving functional ability profile competence ................................ ................................ ................................ ...... 168 4 1 Profile means across functional ability variables.. ................................ ............ 227 4 2 Mean PKBS 2 score for each profile. ................................ ............................. 228 4 3 Moderation of differences on social skills standard scores between profiles 4 and 5 by child activities. ................................ ................................ .................... 229 4 4 Moderation of differences on problem behaviors standard scores betwe en profiles 1 and 3 and between profiles 1 and 5 by parent child activities. .......... 230 4 5 Moderation of differences on problem behaviors standard scores between profiles 1 and 5 by regular child activities. ................................ ........................ 231

PAGE 15

15 LIST OF ABBREVIATION S ICD 10 International Classification of Disease, Tenth Edition ICF International Classification of Functioning, Disability, and Health ICF CY International Classificati on of Functioning, Disability, and Health for Children and Youth ICIDH I nternational Classification of Impairments Disability, and Handicaps IDEA Individuals with Disabilities Education Improvement Act IES Institute of Education Sciences PEELS Pre Elementary Education Longitudinal Study WHO Word Health Organization WHO FIC World Health Organizations Family of International Classifications

PAGE 16

16 LIST OF CONCEPTUAL DEFINITIONS Activity The execution of a task or action by a child or individual (WHO, 2007). Activity limitations D ifficulties a child or individual might have completing an activity (WHO, 2007). Body structures A natomical parts of the body and their components (WHO, 2007). Developmental delay Disability catego which [sic] (McLean et al., 1991, p.1). Developmental domains B road physiological and sociological processes related to human development t hat might include motor development, communication and language development, cognitive development, social development, emotional development, or physical development (Bailey & Wolery, 1992). Eligibility categorie s or disability categories Fourteen categ ories described in the Individuals with Disabilities Education Improvement Act that can be used as part of eligibility determination or for administrative reporting purposes Disability categories for children age 3 t hrough 5 include autism, deaf blind, d eafness, developmental delay, emotional disturbance, hearing impairments, mental retardation, multiple disabilities, orthopedic impairments, other health impairments, specific learning disability, speech or language impairments, traumatic brain injury, vis ual impairments including blindness (NICHCY, 2009) Because definitions for each disability category might vary by state definitions are not provided for each disability category Environmental factors A spects of physical, social, and attitudinal conte xts in which people live and conduct their lives (WHO, 2007). Functional ability profile I performance or ability of same aged peers across a range of developmental or performance domains ( Simeonss on & Bailey, 1991) Impairments P roblems in body function or structure including a deviation or loss (WHO, 2007)

PAGE 17

17 Malleable characteristics A ttributes of a child that might change as process of development or intervention or might be altered by different settings and contexts (e.g., cognitive abilities, language and communication skills; IES, 2011). N on malleable characteristics A ttributes of a child that are not altered or changed by interventions or varying contexts (e.g., age, sex, and race/ ethnicity; IES, 2011) Parent P arent, primary familial caregiver, or legal guardian used as the primary respondent for the PEELS parent interview ( Carlson, Posner, & Lee, 2008) Participa tion A c (WHO, 2007). Participation restrictions P roblems a child or individual might experience in involvement in life situations (WHO, 2007). Preschool children with disabilities Children ages 3 through 5 years of age who are eligible for early childhood spec ial education services under Section 619 of the Individuals with Disabilities Education Improvement Act. Preschool eligibility determination criteria P rocedures and definitions used to determine if a child meets the established requirements set by each s tate to receive special education and related services under Part B (Section 619) of the Individuals with Disabilities Education Improvement Act (IDEA; Dunst & Trivette, 2004) Preschool population Children ages 3 through 5 years of age. Severity of disability T he cumulative influence of a functioning in daily activities, with recognition of the influence Simeonsson & Scarborough, 2001). Social competence A term used to refer to a multi dimensional construct of to achieve social goals and using appro priate behaviors for a given social context (i.e., social skills) and (b) the absence of or refraining from inappropriate use of behaviors (i.e., problem behavior) in a social context (Odom, McConnell, & Brown, 2008; Odom, McConnell, & McEvoy, 1992)

PAGE 18

18 Yo ung child with a disability C hild 3 4, or 5 years of age who has been identified as eligible for special education and related services under Part B Section 619 of Individuals with Disabilities Education Improvement Act.

PAGE 19

19 LIST OF OPERATIONAL DEFINITIONS Child factors Seven variables related to non malleable child language, whether the child had an individualized family service plan (IFSP) before 3 years of age, number of weeks child w as born premature, and child birth weight. Variables measured in the PEELS study and investigator developed coding of PEELS data elements were used to define these variables (Appendix C provides complete descriptions of these variables). Disability category One of seven investigator developed disability categories used in this study. These categories are speech or language impairments developmental disability, autism, emotional behavioral disturbance, mental retardation, learning disability, and lo w incidence disability. was identified as the primary disability category identified in the PEELS data set. Primary disability category is provided in the child demographic file and represents the disability category used t o identify the child for special education services. Environmental factors environment: Neighborhood safety, community income level, school/program quality, parent satisfaction with special educat ion services, program support of social interaction, number of children with and without individualized IEP goals. Variables measured in the PEELS study and investigator developed coding of PE ELS data elements were used to define these variables (Appendix C provides complete descriptions of thes e variables). Family factors Includes two types of variables related to families. The first type is described as family circumstances, which were liv ing environment, respondent role, martial status, parent education, and family income. The second type is described as parent child interactions, which were child activities, family school activities, child participation in regular activities, regular chi ld activities, and parent child activities. Variables measured in the PEELS study and investigator developed coding of PEELS data elements were used to define these variables (Appendix C provides complete descriptions of these variables).

PAGE 20

20 Functional ab ility profile subgroup membership The probability that a child in the PEELS sample will be associated with a subgroup based on similar functional characteristics measured by the 15 items from the PEELS based Disability Severity Index (Appendix B provides a complete description of the variables associated with the 15 items). Child assignment to a subgroup was based on results from latent class statistical analyses conducted using Mplus (Muthen & Muthen, 2007). Low incidence disability An investigator deve loped category that included eight disability categories with small sample sizes in the PEELS data set: hearing impairment, deaf/blind, deafness, multiple disabilities, orthopedic impairments, other health impairments, traumatic brain injury, and visual im pairment. Mental retardation An investigator developed disability category that included mild mental retardation and moderate mental retardation as identified in the PEELS data set. Social competence The use of social skills and the absence of problem behavior as measured by teacher ratings on items included on the Preschool and Kindergarten Behavior Scales, Second Edition (Merrell, 2002). Social skills evaluative judgments of 34 items that describe adaptive or positive behaviors that are likely to lead to positive personal and social outcomes. Problem behavior is measured by problem behaviors commonly seen in the presch ool population.

PAGE 21

21 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy F UNCTIONAL ABILITY P ROFILES AND YOUNG COMPETENCE : EXPLORING REL A TIONSHIPS IN THE PRE ELEMENTARY EDUCATION LONGITUDINAL STUDY DATA SET By Tara Way McLaughlin December 2011 Chair: Patricia Snyder Major: Special Education Children with social competence have the skills to achieve social goals, they know when to use appropriate behaviors in social contexts, and they refrain from inappropriate behavior in social contexts S ocial competen ce has been identified as a priority outcome for young children with disabilities Research has been conducted to examine relationships between various child and contextual factors and social competence The International Classification of Functioning, Disability, and Health for Children an d Youth (ICF CY) provides a framework to describe and quantify functional abilities and to explore associations among functioning contextual factors, and child outcomes The present study was a correlational study conducted through secondary a nalys e s of the Pre Elementary Education L ongitudinal Study (PEELS) data set. The purpose of the present study was to combine an ICF CY approach to quantify function al abilities with person oriented analytic techniques (i.e., latent class analys is) to derive empirically subgroups of children with similar profiles of functional abilities R egression analyses were conducted to examine relationships between subgroup

PAGE 22

22 membership and explained by subgroup membership and disability category, and if contextual factors moderated relationships between subgroup membership and social competence. Results from the latent class analyses showed that five subgroups emerged Each subgroup had a distinct and interpretable functional ability profile F unctional ability profile subgroup membership was associated with outcomes (social skills R 2 = .20 and problem behaviors R 2 = .115) and notable differences in these outcomes were identified between subgroups Subgroup membership accounted for more variance in social competence outcomes than disability category. Four contextual factor s moderated the relationship between functional ability profile subgroup membership and social skills or problem behaviors The present study provide s information about the prev a l e nce and nature of different functional ability profiles found in a nationally representative sample of young children with disabilities. Findings d e monstrate the use of person oriented analyses combined with a functional approach show s promise for identify ing subgroups of children with similar characteristics and for examin ing associations among functional abilities, contextual factors, and important outcomes Results help inform policy and

PAGE 23

23 CHAPTER 1 INTRODUCTION In the present study, secondary analyses were conducted using an extant, large scale, nationally represent ative data set involving young children with disabilities during their preschool and early elementary school years. This study was designed to examine relationships among child functio nal ability profiles, disability categories and contextual variab les and to examine the associations between these variables and In this chapter, background information and the problem statement are described to situate the need for the present study. T he purpose, the conceptual framework and the context of the study are discussed and research questions are stated. The r ationale includes a review of relevant literature and further describes the need for the present study. The importance of the study and a description of delimitations an d limitations are included in this chapter. Background Preschool children with disabilities are eligible for special education and related services through the Individuals with Disabilities Education Improvement Act of 2004 20 U.S.C.§ 1418 (i.e., Part B Section 619 Preschool grants). Over 700,000 children ages 3 through 5 years receive special education and related services (IDEA Data, 2008). IDEA guarantees these children a free, appropriate public education and related services in the least restrictiv e environments, access to the general preschool education curriculum, and individualized education programs (IEPs) to promote school readiness. Fourteen disability categories are specified in the federal law and can be used to identify or describe presch ool children with disabilities. The categories are autism, deaf blindness, deafness developmental delay emotional disturbance, hearing impairment s,

PAGE 24

24 mental retardation, multiple disabilities, orthopedic impairments, other health impairments, speech or la nguage impairment s specific learning disability, traumatic brain injury, and visual impairments including blindness (NICHCY, 2009). Disability categories are used for different purposes under IDEA For example, disability categories are an integral part of preschool eligibility determination Preschool eligibility determination refers to the procedures and definitions used to determine if a child meets the established requirements set by each state to receive special education services The disability categories outline d in federal law might be used to dete rmine eligibility for services. States choose which categories to adopt, how to define them, and determine the criteria for the categories (Muller & Markowitz, 2004) For a child to receive special education services under Section 619, they must meet the criteria for at least one disability category Children are identified and a primary disabi lity category as a result of eligibility determination for special education services. Disabil ity categories also are used as part of reporting and accountability requirements specified in the law (Muller & Markowitz, 2004; Smith & Schakel, 1986) Under IDEA 2004, 20 U.S.C.§ 1412 (i.e., Part B S ection 61 2 State Eligibility), states are required to submit s tate p erformance p lans (SPP) and a nnual p erformance r eports (APR) to report status and progress toward meeting established targets or benchmarks on various progress monitoring indicators Within the context of services for children 3 through 5 year s of age states monitor and report data on a variety of indicators, including descriptive information on the number of children being served (i.e., child count data), educational placement, and the percent of preschool children with individualized education programs who demonstrate improve ment in relation to three

PAGE 25

25 functional outcome s. IDEA 2004, 20 U.S.C.§ 1418 (i.e., Part B S ection 61 8 Program Information) accountability requirements specify that performance data be provided in relation to target populations identified by gender, disability category, race/ethnicity, and limited English proficiency (IDEA, 2004) IDEA mandated data that are collected annually from states are stored in a repository referred to as the Data Analysis System (DANS) whi ch is managed by the Office of Special Education Programs (OSEP) In addition to accountability requirements under IDEA Section 618, OSEP, in partnershi p with the National Center for Special Education Research (NCSER) in the Institute of Education Science s (IES) has funded seven national studies focused on implementation of IDEA 2004, 20 U.S.C.§ 1464 (i.e., Part D Section 664 Studies and Evaluations). National studies are designed to (a) assess the i mplementation and effectiveness of IDEA (b) provide inf ormation on how to implement IDEA more effectively, and (c) provide information to inform legislation and policy related to IDEA (IDEA, 2004) These studies include the National Early Intervention Longitudinal Study (NEILS); the Pre Elementary Education L ongitudinal Study (PEELS); the Special Education Elementary Longitudinal Study (SEELS); the National Longitudinal Transition Study 2 (NLTS2); the Special Education Expenditure Project (SEEP); and the Study of State and Local Implementation and Impact of th e Individuals with Disabilities Education Act (SLIIDEA) Secondary analyses of national studies are used to examine many facets of educational services and child outcomes for children with disabilities ( OSEP, 2006 ). Taken together, the DANS data related to IDEA Section 619 and the federally funded IDEA national studies that focus on young children with disabilities provide a n

PAGE 26

26 opportunity to examine educational services and outcomes for young children with disabilities. Analyses and secondary ana lyses of national data can help guide policy recommendations and future legislation, inform the development of prevention and intervention programs, improve service provision, and advance research related to childhood disability. The present study involve d secondary analyses of the Pre Elementary Education Longitudinal Study (PEELS) data set with a focus on examining Problem Statement Children with social competence have the necessary skills to achieve social goa ls (i.e., social skills), know when to use appropriate behaviors in a given social context (i.e., social skills) and refrain from inappropriate behavior (i.e., problem behaviors ) in a social context (Brown, Odom, McConnell, & Rathel, 2008; Con roy, Brown, & Olive, 2008; Odom, McConnell, & Brown, 2008) Social competence is a desired early childhood outcome for young children, including young children with disabilities The relation to overall development al status and school success is reflected in the selection of two of the three IDEA S ection 618 indicators for which states must demonstrate child progress Under accountability provisions associated with the IDEA s tates are required to report child outc omes related to (a) positive social emotional skills (including social relationships) and (b) use of appropriate behaviors to meet their needs (Early Childhood Outcome s Center, 20 09 ). Research related to young development of social competence is influenced by child factors (e.g., age, gender, skills, and abilities) and contextual factors (e.g., family factors and environmental factors; Guralnick, 1999; McCollum & Ostro sky, 2008; Odom, McConnell, & Brown,

PAGE 27

27 2008; Qi & Kaiser, 2003 ; Raver, Gershoff, & Aber, 2007 ). C hild factors and c ontextual factors might facilitate or impede attainment of skill s related to social competence (Huffman, Mehlinger, & Kerivan, 2001) For example, research has shown that children with poor communication skills often have more difficulties with social competence (i.e., child factor impedes social competence; Herbert Myers, Guttentag, Swank, Smith, & Landry, 2006). Research has also s hown that high quality programs and responsive teacher c ontextual factors might facilitate social competence; Burchinal, Vandergrift, Pianta, & Mashburn, 2010). Thus, the identificati competence is important for the development and delivery of interventions and services Fox, Dunlap, & Powel l, 2002). For children with disabilities, large scale analyses that examine factors associated Blackorby & Cameto, 2004; Wei & Marder, 2011) or focus on children identif ied by a specific disability category (e.g., Sanford, Levine, & Blackorby, 2008; Wagner et al., 2006). Moreover, studies that examine correlates of social competence that involve children with disabilities have not consistently included other child factor s or contextual factors that might be related to Concerns have been raised abou t using IDEA based disability categories alone to examine child outcomes such as social competence (Blackorby et al., 2002). Variations in state eligibility determination systems have resulted in disability categories that are idiosyncratic to individual states (Danaher, 2007; Muller & Markowitz, 2004) This variability presents

PAGE 28

28 several limitations when aggregating data in nationally representativ e data sets and using disability category as a correlate of child outcomes (M a cMillian & Resch ly, 1998) Of primary relevance to the present study is the ability to (a) compare children by disability category from different states, and (b) distinguish dif ferences between children within a disability category (Florian et al., 2006; Simeonsson, Simeonsson, & Hollenweger, 2008) Disability category functional abilities or levels of functioning and subsequent suppo rt needs and offers limited, if any, information about secondary conditions or contextual factors that might ing (Florian et al 2006 ; Forhan, 2009; Marder, 2009 ) Given these concerns, the exclusive use of disabil ity category as a correlate of social competence outcomes might hamper federal efforts to explain variations in child outcomes and interpret research related to childhood disability when examining social competence for children with varying abilities Information is needed beyond children Beyond disability category, t he identification of other individual variables potentially related to desir ed outcomes might not be a more effective approach Referred to as a variab le oriented approach, analytic methods that examine the influence of individual variables on outcomes such as social competence in a he terogeneous group of children, might help inform associations among key variables, but not account for complex associati Boulerice, & Vitaro, 2000). Analytic methods to identify subgroups of children with similar patterns of abilities, referred to as person oriented approach es, are increasingly being used in early education and early childhood special education research

PAGE 29

29 ( Campbell, Shaw, & Gilliom, 2000; Konold & Pianta, 2005) A person oriented approach us es analytic methods that cluster or group children who display similar patterns of strengths and needs across identified domains. These profiles are then examined in relation to desired outcomes. Th e rationale for a person orien ted approach is to focus on outcomes for children with similar characteristics by identifying common patterns or profiles of chi For children with disabilities, child characteristics might be examined using a functional ability profile approach (Hobbs, 1975; Simeonsson et al., 2008) A functional approach focuses on specifying charac teristics or abilities of the child across a variety of developmental or performance domains (Simeonsson, 2003) within the context of his or her environment, which might promote or constrain participation (Snyder, 2006) Functio nal approaches have been used to create an abilities across developmental or performance domains functional abilities have been associated with their social competence (e.g., communication skills, play skills; Odom, McConnell, & Brown, 2008). The exploration of patterns or profiles of functional abilities related to social competence might help identify associations that have important impli cations for the development of interventions and services for young children with disabilities (Haapasalo et al., 2000). To enhance understandings about factors associated with desired outcomes for young children with disabilities beyond disability categor ies functional approaches to might be combined with person oriented analytic techniques to identify sub groups of children with similar functional ability profiles and to examine

PAGE 30

30 relationships with social competence. To date, these methods ha ve not been widely used with nationally representative data sets focused on young children with disabilities. Purpose of the Study The primary purpose of the present study was to explore relationships between empirically derived subgroups of children with distinct and interpretable functional ability profiles and their social competence The present study evaluated whether use of subgroups formed on the basis of functional ability profiles were viable correlates of social competence and examined the explanatory power of both functional ability profile subgroup membership and disability category in relation to social competence. The extent to which c ontextual factors moderated relationships between functional ability profile subgroup memb examine these relationships, secondary analys e s were completed using cross sectional data from the Pre Elementary Education Longitudinal Study (PEELS) national ly representative data set The p resent study offered an alternative methodological approach and extend ed substantive analyses to include the influences of contextual factors as moderators of the onal ability profile subgroup membership and their social competence In the present study, a functional approach was combined with p erson oriented analytic techniques to identify subgroups of children with similar characteristics (i.e., referred to as s imilar functional ability profiles) that were distinct from other sub groups identified through the analytic approach Functional ability profiles represented common patterns of functional abilities that exist across subgroups of children For example, wi thin the PEELS sample of children with disabilities, there might have been a subgroup

PAGE 31

31 of children who had no limitations related to communication skills, social skills, cognitive skills, and motor skills. Concurrently, there might have been a different su bgroup of children with substantial limitations and need for extensive supports in relation to these skills. The functional ability profile subgroups represented children with different combinations of strengths and needs across select functional ability variables The number and profile descriptions of the subgroups were not determined a priori, but rather were derived from the data based on the selection of functional ability variables represented in the PEELS data set. Thus, person oriented analyses related to identifying subgroups of children with similar functional ability profiles were exploratory in nature. The intent of these exploratory analyses was to examine if the data supported the hypothesis that distinct and interpretable subgroups exist in the PEELS sample of children with disabilities. Following confirmation of distinct and interpretable subgroups with similar functional ability profiles, analyses focused on examining relationships between these functional ability profile subgroups and which functional ability profile subgroup membership was related to their social competence and added explanatory power beyond disability category was examined. To understand further associations be tween functional ability profile membership and social competence, the influence of contextual factors as moderators and social competence w as examined. In the present study, functional ability variables, used to create subgroups with similar functional ability profiles, were considered malleable child factors. Malleable

PAGE 32

32 factors refer to the characteristics and conditions that might be alte red by context or intervention (e.g., language skills, cognitive skills; IES, 2011). In contrast, non malleable factors refer to attributes or conditions that can not be alte red by context or intervention (e.g., age, gender; IES, 2011). Current requests for applications from the federal Institute of Education Sciences emphasize the importance of conducting research to examine associations among desired outcomes, such as social competence, and malleable factors. This type of research should help inform the development an d evaluation of interventions and policy related to malleable factors that can be affected by early intervention and early childhood special education services to improve outcomes for young children with disabilities (IES, 2011). Conceptual Framework The present study used the International Classification of Functioning, Disability, and Health for Children and Youth (ICF CY ) framework described by the World Health Organization (WHO, 2007) to guide the conceptualization of the study research questions, the creation of the functional ability profiles, and the identification of contextual factors hypothesized to be related to functional ability profiles and their social competence This framework offers a way to conceptualize and describe a s developing characteristics within his or her surrounding environment, while and adaptation. The ICF CY highlights the unique nature of child development and s intensity, and consequence over time (Lollar & Simeonsson, 2005). The current ICF CY framework illustrates the multi dimensional interactions among disability/heal th conditions, functioning and disability, and contextual factors that

PAGE 33

33 affect all individuals (WHO, 2007). Figure 1 1 illustrates the primary components of the framework. Within the ICF CY framework, a disability/health condition (i.e., disability from a diagnostic perspective) interfaces with key components related to functioning and disability (i.e., disability used as a global term). These components can be viewed dy functions (i.e., p hysiological and psychological), body str ucture (i.e., anatomical parts), activities (i.e., tasks a child completes) and participation (i.e., the int egration of activities in life). Alternatively, these same components can be viewed from a negative impairment of body structure, activity limitation, and participation restrict ion (De Kleijn De Vrankri jker, 2003; WHO, 2007). The framework highlights the inf luence of contextual factors, including both environmental factors (e.g., physical, social, and attitudinal environment) and personal factors ( e.g., age, gender, ethnicity) on an individuals overall well being and adaptation with regard to human functionin g and restrictions on functioning (WHO, 2007). The framework provides a practical approach to classification of disability and a series of categories, codes, and rati ngs related to dimensional features of function across body functions and structures, activities and participation, and environmental functional characteristics are classified by using a pre specified taxonomy and by applying a standard numeric rating or qualifier. A rating of 0 is applied if the child has normal function related to the code. Normal function is defined by what is typical of the

PAGE 34

3 4 average same aged peer. A ratin g of 1 is applied if the child has mild functional impairment, limitation, or restriction related to the code. A rating of 2, 3, or 4, is applied if a child has moderate, severe, or complete functional impairment, limitation, or restriction, respectively, related to the code. This coding system provides direct in relation to same aged peers. The ICF CY taxonomy and related numeric ratings were not directly employ ed as a measurement system in the present study because this is not one of the intended uses for the ICF CY. Rather, the ICF CY framework was the basis for identifying variables related to functional ability and environmental context (i.e., family, school and community) that were available in the PEELS data set. Context for the Study D ata from the restricted version of the Pre Elementary Education Longitudinal Study (PEELS) funded by the U.S Department of Education, National Center for Special Education Research (NCSER) were used in the present study. The PEELS data set was obtained through a restricted use license agreement with the Institute of Education Sciences (IES) The PEELS data set was selected for this study because it offers descriptive information on the characteristics, functioning, and school experiences of children with disab ilities during early childhood and information related to family, school, and community e nvironments The PEELS data set is based on a nationally representative sample of 3,10 0 children with disabilities Children in the PEELS sample were 3, 4, or 5 years of age and had an individualized education program (IEP) or individualized family serv ice plan (IFSP) at the time they were recruited into the study. Data for the PEELS data set were

PAGE 35

35 collected in four waves from the 2003 2004 through 2006 2007 school years, and follow up data were collected during the 2009 2010 school year (Markowitz et al., 2006) Wave 1 cross sectional data were selected for this study because they provide information on children in preschool or kindergarten settings PEELS data collection focused on the characteristics of children receiving early childhood speci al education, the programs and services they receive, their transitions from preschool to school age settings, and how children with disabilities function and learn in preschool and school age settings The PEELS data set contains information about variab characteristics (malleable and non malleable) educational services, academic performance and environmental conditions In addition, the data set contains information on parent s per spectives of their child ren development and fu nctioning and educational services Th e secondary analyses conducted in th e present study focus ed on variables related to child characteristics, disability category social competen ce In addition to the variables available in the data set, the PEELS data set was selected because Daley, Simeonsson, and Carlson (2009) conducted a previous study with the PEELS data set that used variables from the p arent interview to identify or deriv e 15 variables related to In the Daley et al study, functional ability variables were used to create a composite score of functional ability and to examine relationships between this composite score and a variety of child outcome variables The present study extends this work by using the same variables from the parent interview to identify subgroups of children with similar functional ability profiles rather than a summated composite score and to explore potential modera tors of

PAGE 36

36 Detailed descriptions of the PEELS data set and variables from the data set that were used to address the research questions in the present study are provided in Chapter 3. Research Questions The following research questions guided the secondary analyses conducted in the present study: 1. What distinct and interpretable functional ability profile subgroups emerge when using person oriented analytic techniques to examine functional abilit y variables contained in the PEELS data set for young children with disabilities? 2. What is the strength of the relationship between functional ability profile subgroup membership and social competence? 3. What are the individual and combined contributions of f unctional ability profile subgroup membership and disability category membership to the explanation of social competence? 4. To what extent do non malleable child factors and contextual factors moderate the relationship between functional ability profile subgroup membership and social competence? Rationale for the Study readiness, success in school, and emotional well being Studies that examine malleable factors related to so cial competence might help inform important prevention and intervention efforts for children with or at risk of poor social competence outcomes The creation of subgroups with similar functional ability profiles was proposed in contrast to the traditional use of disability categories as a primary correlate of child outcomes for children with disabilities in national data sets Functional ability profiles were used to address concerns about the validity of categorical descriptions of disability In additi on, the influences of non malleable child factors and contextual

PAGE 37

37 factors on the relationships between functional ability profiles and competence were also examined in the present study To justify the rationale for the present study, th e following topics are discussed (a) social competence as a key child outcome, (b) the social competence framework used competence, (d) issues related to measuring and examining social c ompetence, ( e ) concerns with categorical characterizations of disability, (f) functional approaches as an alternate method to describe child characteristics, and (g) methods to create subgroups of children with similar profiles of functional abilities In addition, the rationale for the present study was informed by previous investigations conducted using the PEELS data set and other empirical studies that used large scale data sets to examine the child functioning related to child outcomes or to explore person oriented statistical techniques To conclude this section, a methodological rationale and summary of the substantive rationale for the secondary analyses conducted in the present study are presented Social Competence as K ey Child Outcome In recent years, the complex emotions and essential social skills that develop favorable outcomes for children through systematic and planned earl y interventions that target these skills have received national attention Shonkoff and Phillips ( 2000 ) highlighted the critical being, social competence, and cognitive skills in the landmark book From Neurons to Neighborhoods As noted by the National Council on the Developing Child (2004a),

PAGE 38

38 will influence school achievement and emotional well being later in life F or chi ldren with disabilities, promoting social competence can be particularly difficult due to poor communication skills, difficulty establishing peer relationships, and the increased prevalence of challenging behaviors when compared to peers without identified disabilities or those at risk for disabilities (Dunlap et al., 2006; Guralnick, 2006; Odom, McConnell, & Brown, 2008) Evidence has shown that persistent maladaptive social and challenging behavior in young children might lead to continued problem behavi ors peer rejection, school failure, and other social and academic challenges through out adolescence ( Missall & Hojnoski, 2008; National Council on the Developing Child, 2004 b ) Given the potential for poor outcomes, measuring monitoring and promoting t young children with disabilities, is a national priority (e.g., IDEA preschool outcomes) Social Competence Framework For the purpose of the present study, social competence was viewed from a social behavioral perspective This perspective emphasizes performance based assessment of social competence Within a performance based perspective, social performan ce in social settings (Odom, McConnell, & Brown, 2008) The present study was guided by the definition of social competence presented by Odom, McConnell, and McEvoy (1992), which was updated by Odom, McConnell, and Brown (2008) These authors conceptu alized social competence based on two

PAGE 39

39 for the social context From this p erspective, social competence is observed when children use social skills to achieve social goals (i.e., social skills) use appropriate behaviors in a given social context (i.e., social skills), and refrain from inappropriate use of behaviors (i.e., probl em behaviors ) in a social context ( Brown, Odom, McConnell, & Rathel, 2008; Conroy, Brown, & Olive, 2008; Odom et al., 2008 ) In addition to the performance based view of social competence, Odom and colleagues (1992; 2008) highlighted the influential role of child factors (both malleable and non social competence The hypothesized contributions of child and contextual factors to performance based social competence a re consistent with findings reported the empirical literature (cf Clements, Reynolds, & Hickey, 2004; Guralnick, 1999; McCrae & Barth, 2008; Rouse & Fantuzzo, 2009) ompetence, previous studies have suggest ed that associations among child factors and contextual factors competence should be considered For child factors, b oth malleable and non malleable child factors have been associated with soci al competence Malleable child factors associated with social competence inclu and characteristics related to language and communication skills, cognitive abilities, level of assertiveness, activity level, ability to sust ain attention, and regulation of emotions (e.g., Herbert Meyers, Guttentag, Swank, Smith, & Landry, 2006; Qi & Kaiser, 2003) Non malleable characteristics, such as age, sex, race/ethnicity have also been ce (e.g., Campbell et al., 2000; Mendez, McDermott, & Fantuzzo, 2002; Raver, Gershoff, & Aber, 2007)

PAGE 40

40 For contextual factors, both family factors and environmental factors have been Family factors associated with social competence include family characteristics such as family structure and size, socio economic circumstances, marital status, parent education, and mental health status of caregivers (Krishnakumar & Black, 2002; Loeber & Hay, 1997; NICHD Research Network, 2003; Raver et al., 2007; Schmidt, Demulder, & Denham, 2002) and parent child interactions such as parental attitudes and behaviors and parental involvement (Guralnick, 1999; Oravecz, Koblinsky, & Randolph, 2008; Raver et al., 2007; Schmidt et a l., 2002) competence include community factors such as the socio economic status of the community or safety of the neighborhood, and school factors such as the quality of childcare or school programs or the provision of interventions or supports to promote social competence (Domitrovich, Cortes, & Greenberg, 2007; Loeber & Hay, 1 997; Odom et al., 1999; Oravecz, Koblinsky, & Randolph, 2008; Romano, Kohen, & Findlay, 2010) R esearchers have examined the effect of contextual factors in terms of risk and resiliency Research on risk has focused on the identification of factors associated with undesirable outcomes. Research on resiliency has focused on of risk factors; the se positive factors are referred to as promotive or protective factors ( cf. Werner & Smith, 2001) A series of seminal research studies: the Isl e of Wright study (Rutter, 1979), Rochester Longitudinal study (Sameroff, Seifer, & Zax, 1982), and examined the influence of risk

PAGE 41

41 factors promotive factors, and protective factors nt, including their social competence Researchers have noted the individual contribution of any one factor does not consistently predict or explain related domains (Campbell et al. 2000; Krishnakumar & B lack, 2002; Sameroff & Seifer, 1983) When multiple contextual variables are included in an analysis, the contribution of variables to predict or explain outcomes change s across samples and are affected by which variables are included in the analysis the measures used to quantify the constructs, and the analytic techniques used (Gutman, Sameroff, & Cole, 2003; Raver et al., 2007; Sameroff & Seifer, 1983) Measuring and Examining Social Competence Odom et al (2008) described the importance of an assessment method that integrates multiple sources and methods to generate meaningful information about The recommendation for multi component assessment of social competen ce, however, is tempered by the constraints of resources and the purpose for examining social competence (Odom et al., 2008) complex Previous studies have shown a range factors associated with the development of social competence As noted previously, the individual contribution of any one factor does not consistently relate to and the contributions of different variables change s across samp les (Campbell et al. 2000; Gutman, Sameroff, & Cole, 2003; Krishnakumar & Black, 2002; Raver et al., 2007; Sameroff & Seifer, 1983) Current national efforts to examine factors associated with

PAGE 42

42 child outcomes emphasize the importance of exploring the relations hips between outcomes and malleable factors as well as mediators or moderators of identified associations (IES, 201 1 ) The exploration of malleable factors is intended to guide policy decisions and info rm the development of interventions and services that can improve outcomes for children (IES, 2011) For children with disabilities, a common variable included as a correlate of child Given the focus o n examining relationships between child outcomes and malleable child factors that can guide national policy and inform interventions static disability categories might not be the most meaningful variable Concerns with Categorical Descriptions of Disabi lity Fourteen disability categories can be used to determine if preschool children are eligible for special education and related service under IDEA To receive services a child must meet the established criteria for a disability category and subsequentl y receive a primary disability classification or label The federal law and associated implementing regulations provide only general criteria for determining eligibility for special education and related services by specifying the federal disability categ ories (Muller & Markowitz, 2004) S pecific recommendations for eligibility determination are provided for only one disability category (i.e., learning disability) The absence of guidance in the law and in federal regulations has allowed states to set th eir own terms, definitions, and criteri a for eligibility determination, although state systems must meet the requirements for federal reporting (i.e., Section 618 requirements) Analysis of state system s have highlighted variations across states in the a doption of disability categories and the criteria for each category used to determine eligibility for

PAGE 43

43 special education and related services for preschool aged children (Danaher, 2007) and school aged children (Muller & Markowitz, 2004) Florian and colleagues (2006) noted there is as much within category variation as there is between category variation for the IDEA disability categories IDEA disability categories provide limited, if any, information about disability severity, limitations in functio ning, or potential secondary or additional disabilities (Chambers Perez, et al., 2004). These between and within category variations have resulted in disability categories that are unique to individual states and provide limited information about the fun ctional characteristics of the child identified with a disability (Simeonsson, Bailey, Smith, & Buysse, 1995) Because of the idiosyncratic and limited nature of disability categories, concerns about comparing data for children by disability category from different states and concerns about distinguish ing differences in characteristics between children within a disability category when data are aggregated across states have been raised (Florian et al., 2006; Simeonsson, Simeonsson, & Hollenweger, 2008) G iven abilities, alternative frameworks to describe and quantify disability from a functional approach have been proposed (Hobbs, 1975; Simeonsson et al., 2008). Funct ional Approaches as Alternate Method for Describing Child Characteristics Researchers and organizations have proposed functional approaches as alternatives for characterizing and describing children with disabilities in relation to their abilities (i.e., s trengths and weakness), level of functioning (sometimes referred to as severity of disability or impairment), and resulting support needs (American Association of Intellectual Developmental Disabilities Definition Manual, AAIDD, 2010; ICF CY, WHO, 2007; Sn yder, Bailey, & Auer, 1994) Functional approaches provide information

PAGE 44

44 development or performance (e.g., cognitive, motor, language, vision ; Simeonsson, 2003) From an ICF C Y approach, function is viewed in terms of interactions among characteristics of the person his or her activities and his or her environment (Snyder, 2006) Functional descriptions of children have been recommend to help guide decisions for individualized services and to encourage consideration of disability within social and environmental context s (Lollar & Simeonsson, 2005) Functional descriptions of children have also been recommended for use in research to examine relationships between child characteristics and desired outcomes (Simeonsson, 2003) The ICF CY provides one framework and associated taxonomy to describe and characterize children using a functional approach The use of the ICF CY has been promoted as a framework that provide s a common language to facilitate communication and knowledge related to childhood disability ( Lollar & Simeonsson, 2005; Simeonsson, 2009) Functional measurement approaches, based on the ICF CY (or previous versions), have been used to create an overall composite of functional ability (or developmental or performance domains For example, t he ABILITIES Index (Simeonsson & Bailey, 199 1 ) is a judgment based rating scale designed to profile the functional abilities of children across nine developmental and performance domains : audition (hearing), behavior and social skills, intellectual functioning, use of limbs, intentional communication, tonicity, integrity of health, eyes (vision), and structural status The ABILIITIES Index has been used to calculate an overall score to indicate the severity of disability and the index has been

PAGE 45

45 cluded on the index (Buysse, Smith, B ailey, & Simeonsson, 199 3 ) Simeonsson, Bailey, Smith, and Buysse (1995) noted that the ability to cluster or group children with similar profiles might support the examin ation of relationships between child characteri stics and Methods to Create Subgroups of Children with Similar Profiles of Abilities Researchers in education, early education, special education, and early intervention are increasingly using met hods to identify subgroups of children with similar characteristics and to describe these subgroups by profiles of abilities Bergman and Magnusson ( 1997 ) r efer to this as a person orien ted approach A person oriented approach contrasts with a variable o riented approach A variable oriented approach examin es the influence of individual variables on outcome s in a heterogeneous group of children and it is assumed that identified relationships apply across all children in the population (Collins & Lanza, 20 10) A person oriented approach examines individuals and individual patterns of characteristics and their relationships (Collins & Lanza, 2010) In order to study groups of individuals based on patterns of characteristics, person oriented approaches can be used to identify subtypes or typologies in a population (i.e., homogenous groups from a heterogeneous population ; Collins & Lanza, 2010; McCut cheon, 1987) Researchers in early childhood have used these person oriented analytic methods that cluster or sub group children who display similar patterns of strengths and needs across identified domains ( Campbell et al. 2000; Konold & Pianta, 2005) Each cluster or sub group shares a similar profile that is distinct from other sub groups identified through the analytic approach After subgroups are identified, additional analytic

PAGE 46

46 methods can be used to examine relationships between subgroup membership and other outcomes In addition to using person oriented techniques to create homogenous groups from a hetero geneous population, this appro ach has been used to create sub groups of children with different patterns of functioning or ability within a homogenous population on defined attributes (e.g., race/ethnicity, gender; Mendez, Fantuzzo, & Cicchetti, 2002) Ne ed for Additiona l Studies Using the PEELS Data S et The Pre Elementary Education Longitudinal Study (PEELS) followed a nationally representative sample of children receiving special education services from 2003 2004 through 2008 2009 Reports about finding s from analyses conducted with the PEELS sample have been published by PEELS investigators and are publically available on the PEELS website The findings from PEELS reports highlight trends and patterns in educational services, transition experiences, an d characteristics of young children with disabilities In addition, PEELS and other investigators have published studies using the PEELS data set in peer reviewed journals Studies published to date have focused on a range of topics (e.g., information ab out data set preparation, parent satisfaction with services, the influence of home literacy environments on literacy skills) To date, only one published study using the PEELS data set is directly related to the present study Daley, Simeonsson, and Car lson (2009) used data in the PEELS data set to create a Disability Severity Index Index items were based on responses given by parents of children enrolled in the PEELS study to structured interview ties To construct the index, the authors identified abilities and were related to domains associated with the ABILITIES Index (Simeonsson

PAGE 47

47 & Bailey, 1991) and corresponded with codes o n the ICF CY They examined a composite score from both a 15 item and a 6 item version of the index to scores on a range of child outcome measures and determined the patterns of associations were similar across both indices For parsimony, the authors se lected the 6 item version to create a composite score that represented a severity of disability index The authors examined relationships between the composite score from the Disability Severity Index and select outcome variables, including aspects of soc ial competence They examined the contribution of the composite score from the Disability Severity Index to variance explained in the outcome variables beyond variance accounted for by the use of disability categor y Daley and colleagues reported disability category alone accounted for 17% of the score variance skills scores When disabili ty category and functional ability score were used together, 26% of the variance in social skills scores was accounted for by these two variables The authors noted th e increase in variance explained by the two variables gory and functional ability] are different constructs, with less overlap than might be predicted given traditional ideas about certain categories p 548 ). Need for Empirical Studies of Social Competence a nd Functional Characteristics Eight empirical studies directly relevant to the present study were identified in the extant literature All identified studies included a large sample of children (i.e., children ages 3 through 5 included in study) Studie s were analyzed to examine the extent to which investigations (a) were conducted with U.S based nationally representative samples, (b) were conducted with children with disabilities, (c) were based on ICF CY

PAGE 48

48 haracteristics, (e) used person oriented analytic techniques to identify subgroups with similar profiles, (f) compared functional ability to disability category, (g) examined outcomes related to social competence, and (e) considered contextual factors as p art of the analyses Table 1 1 shows the relevant studies identified for review and the components of each study that relate to specific aspects of the present study Four of these studies were conducted with nationally representative samples Two studi es were conducted with children with disabilities All studies included children ages 3 through 5, however, three of these studies were conducted exclusively with young children (i.e., ages 3 through 5) Two studies were based on the ICF CY framework T characteristics Six studies used person oriented techniques to create subgroups of children with similar profiles disability categories Six studies ex amined the relationship between descriptions of es Six studies considered contextual factors as part of the examination of relationships between descriptions of utcomes Although no single study included each aspect included in the present study, studies reviewed were helpful for informing the design of the present study and building support for the substantive and methodological rationale for the present study The extent to which each study reviewed contributed to the present study is described in Chapter 2 Nonetheless, given the limited research that integrates all aspects of the present study into one investigation, the secondary analyses conducted in the present

PAGE 49

49 study offered a unique contribution to the research base in early childhood special education and related fields. Methodological Rationale The present study employed statistical analysis techniques to create subgroups of children with similar funct ional ability profiles The possible techniques to create these subgroups aim to identify groups with internal cohesion (i.e., homogeneity within group) and external isolation (i.e., separation form other groups; Everitt, Landau, & Leese, 2001) These te p 1) An important aspect of the present study was examining a potential technique to create subgroups based on profiles of functional abilities for young children with disabilities categories, the ability to describe children by subgroups with shared characteristics and level of functioning offers a n altern ative way to examine relationships between child characteristics and important outcomes Although previous studies have used person oriented techniques to create subgroups of children with disabilities with similar functional ability profiles (Granlund, E rickson, & Ylven, 2004 ; Simeonsson et al., 1995 ), these studies were conducted with small samples of convenience The present study explores using a person oriented an alytic approach in a nationally representative sample of young children with disabilitie s In addition, previous investigations have predominately used cluster analysis and the present study used latent class analysis to generate subgroups of children with shared profiles (Muthen & Muthen, 2007) Magidson and Vermunt (2006) indicated that interest in latent class models has increased with the development of statistical

PAGE 50

50 software programs that can perform this analysis with more than just a few variables They also noted latent class models have gained popularity over other methods because they use model based approaches to estimate membership probabilities in order to classify cases into the appropriate subgroup Following classification into a subgroup, relationships among variables, including subgroup membership, can be explored. In the present study, t he combination of using a functional approach to 15 variables and a person oriented analytic approach to examine outcomes for subgroups of children with similar functional ability profile s off ers a preliminary investigation of a methodological approach that might be particularly useful to examine outcomes for children with disabilities The present study builds on the complementary nature of a functional approach and a person oriented analytic approach to e xplore relationships with social competence for young children with disabilities. Summary of Rationale A need exists in the current literature base to examine the associations among child functioning, disability/health condition, and contextu al factors, in relation to school readiness, school achievement, and later in life well being, further examining variables that have been associated with social competen ce is important to help guide future research and policy and inform interventions and practice The present study addressed concerns identified in the literature related to using categorical descriptions of disability to examine relationships in research by exploring

PAGE 51

51 functioning The use of empi rically derived subgroups of children with similar profiles of functioning related to malleable child abilities offered a person oriented approach to examine the relationship between functional ability profile subgroup membership and key outcomes such as s ocial competence Exploring the influence of non malleable child factors and contextual factors was consistent with previous research on social competence and consistent with factors described in the ICF CY framework Importance of the Study The seconda ry analyses conducted in this study were important for several reasons First, this study extended secondary analyses conducted with the PEELS data set (i.e., Daley et al., 2009) by exploring the identification of subgroups with similar functional ability profiles to characterize children with disabilities instead of a composite score of functional ability Second, f indings from the pr esent study might inform policy research, and practice related to characterizing disability and function when examining a ssociations between children with disabilities and socially important outcomes such as social competence Previous research has suggested that functional ability profiles might account for more variance in criterion variables than disability category Th ird, e functional abilities will contribute to understanding relationship s between social competence and malleable child factors such as functional abilities This profile approach might offer additional information beyond c disability category t o help inform the provision of targeted supports or services for children whose profile of functional abilities is associated with social competence challenges Fourth, the exploration of select contextual factors as potential moderators of functional profiles might inform policies related to the provision of family services or additional preschool intervention supports or services for children Finally, the ICF CY

PAGE 52

52 framework, which guided this study, offers a holistic system to describe functional abilities that provide s information about children beyond disability category while considering contextual factors that influence child functioning functional abilities might be useful for informing t he type and intensity of intervention supports to achieve desired child outcomes. Definitions of Terms Conceptual definitions were presented to provide a definition of terms used in the present study and operational definitions were presented to provide a description of how variables were operationalized for analys e s (List of Definitions) Throughout the present study, terms and definitions used to refer to disability categories in previous research studies might not be consistent with recommended anti bia s language in the Publication Manual of the American Psychological Association (APA, 2010) These terms and definitions, however, reflect those used by authors at the time the study was conducted and are used to provide an accurate representation of previ ous research studies Delimitations The present study involved secondary analyses of an extant data set; the study did not involve an original research design with primary data collection The focus of social competence; other areas of child development or performance were not examined The present study focused on the examination of social competence status when ch ildren were 3, 4, or 5 years of age The study did not examine social competence traj ectories (i.e., change over time) or prediction of social competence at a later age (i.e., status at an older age) The study employed measures of social competence Ratings

PAGE 53

53 f rom parents or peer nominations were not available in the PEELS data set and, therefore, not included in the present study Variables used to create subgroups with similar functional ability profiles were variables included in the PEELS data set based on parent interviews The variables selected represent a sample of functional abilities related to domains that were previously identified in published research The selection of child and contextual factors were based on a review of the literature and sub sequent identification of these variables available in the PEELS data set T hese variables are not exhaustive of all child and contextual factors Moreover, selected variables were based on response formats available in the PEELS data set For example, some interview items provided respondents with a restricted response format (e.g., 1 = not safe 2 = safe = very safe ; related to the extent a parent feels the neighborhood is safe for their child to play). Alternatively, some items were summed to create a continuous variable (e.g., number of different activities that a child participates in on a monthly basis). Some of these variables might not be the most robust measure or proxy for the construct of interest Additional empirical examination of these variables (e.g., the extent to which a interview or questionnaire item reflects a reliable or valid score related to the construct of interest) was not conducted as part of the present study. L atent classes in the present study represent subgroups of children with a similar pattern of functioning on 15 functional ability variables included in the PEELS data set Children were assigned to a profile based on their most likely class membership. It is acknowledged that children within a subgroup will have individual differences related to

PAGE 54

54 their pattern of functioning on the 15 functional ability indicators. Fu nc t ional ability p rofile subgroups are intended to represent salient pattern s of functioning in the PEELS sample of children with disabilities. a functional ability profile subgroup is not intended to replace disability category or i ndividualized information about children that might be used to inform eligibility for special education services or development of IEP s Limitations Several limitations of the present study are noted First, Odom and colleagues (2008) recommend the use o social competence from a performance based perspective The present study employed only one measure and one method (i.e., teacher ratings) to evaluate Teacher ratings ar The extent to could not be determined Second, the items used to abilities were derived from a parent interview abilities were not conducted nor were judgments about functional abilities obtained from other informants As analyses were completed, analytic decisions were made in the conduct of the secondary analyses related to data structure and analytic methods These decisions and resulting analytic procedures are described in Chapter 4 and should be considered when interpreting findings in the present study

PAGE 55

55 Summary Social competence has been identified as an important outcome for young children with disabilities The examination of factors associa ted with social competence of young children with disabilities might help guide policy recommendations; inform service provision, including prevention and intervention programs for children with disabilities and their families; and advance future research on examining correlates of social competence In the present study, secondary analyses of the PEELS data set were conducted to explore associations between empirically derived subgroups of children with similar functional ability profiles and their socia l competence profile subgroup membership was examined as a correlate of competence beyond the us e disability categor y In addition, the extent to which non malleable child factors and contextual factors m oderate d relationships between functional ability profile subgroup membership and was examined The ICF CY conceptual framework informed the selection of variables and research questions examined in the present study.

PAGE 56

56 Table 1 1 Studies r elated to a spects of p resent s tudy Citation National s ample Children with d isabilities Ages 3 5 ICF Functional c haracteristic s Person o riented t echnique s Compared to d isability c ategories Relationship to s ocial c ompetence Contextual f a ctors c onsidered Chambers Perez, et al., 2004 X X 0 a X X X X Daley, Simeonsson, and Carlson, 2009 X X X X X X X Hair, Halle Terry Humen Lavelle and Calkins, 2006 X X 0 b X X X Haapasalo, Tremblay, Boulerice, and Vitaro, 2000 0 a X X X Janson and Mathiesen, 2008 0 a X X Konold and Pianta, 2005 X 0 b X X X Sanson, Letcher, Smart, Prior, Toumbourou, and Oberklaid, 2009 0 a X X X Stephen s Petr a s, Fabian, and Walrath, 2009 X 0 c 0 a X X X Note X refers to aspect of present study reflected in the cited study, 0 refers to aspect of present study somewhat reflected in the cited study, and blank cell refers to aspect of present study not reflected in the cited study a Children age s 3, 4, or 5 years included in larger sample of school age children b Referred to as school readiness skills c All children identified as youth receiving mental health services.

PAGE 57

57 Figure 1 1 International Classification of Functioning, Disability, and Health (WHO 2007) Reprinted with permission from the International Classification of Functioning, Disability and Health: Children and Youth Version (pp. 17), by the World Health Organization, 2007, Geneva, Switzerland: WHO Press.

PAGE 58

58 CHAPTER 2 REVIEW OF THE LITERA TURE In the p resent chapter, a review of the literature is conducted The review provides the rationale and background for the conceptual framework for the study, the research questions addressed, the identification of variables of interest from the Pre Elementary Edu cation Longitudinal Study (PEELS) data set, and the analytic techniques used in the present study The review of the literature covers five major topics: (a) (b) concer ns about using IDEA disability categor ies to characterize young children (c) the International Classification of Functioning, Disability, and Health for Children and Youth (ICF CY) framework, (d) findings from PEELS studies, and (e) empirical research rel ated to the present study. The first four topics contribute to the present study in unique and complementary ways First, social competence has been identified as a desired early childhood outcome (Early Childhood Outcomes Center, 2009 ; Hebb e ler & Kahn, 2008 ) and has been associated with school readiness and later achievement (Shonkoff & Phillips, 2000) competence during the early childhood years, however, requires further ex amination Second, concerns with use of IDEA disability categor y labels to describe young children for research purposes motivated the decision to examine a functional approach to describe children with disabilities in the present study Third, the ICF C Y guided the functional approach used in the present study Fourth, the PEELS data set provide d a unique opportunity to describe and examine the experiences of young children with disabilities during the formative years of preschool and early elementary s chool

PAGE 59

59 Previous investigations using the PEELS data set offer a rationale for the significance of the present study Finally, the fifth topic addressed in the present review of the literature is empirical research related to the present study Empirical studies included in this part of the literature review were those conducted with large samples (i.e., more than 500 participants) that included young children ages 3 through 5 and investigated (a) the contribution of child functioning beyond disability ca tegories related to child outcomes or (b) the use of person oriented analytic techniques ( cf. Magidson & Vermunt, 2006 ) to identify subgroups of children with similar profiles of abilities and to examine relationships between these subgroups and child outc omes. Search Procedures Research studies, articles, and reports reviewed in this chapter were identified in a number of ways First, the following electronic databases were used to identify peer reviewed articles: Education Full Text and EBSCO Host Platfo rm for Academic Search Premier, CINAHL, Psychology and Behavioral Sciences Collection PsycINFO, and Teacher Reference Center For the five topics previously described, searches were conducted using combinations of keywords For social competence, keywor ds included social competence, social skills, problem behavior, young children, preschool, children, disability, predictor, and outcome For disability categories, keywords included disability category, developmental delay, and categorical classification For the ICF CY framework, keywords included ICF, ICF CY, young children, preschool, children, and functional profile For the PEELS data set, keywords included PEELS, disability, and children For the empirical research related to the present study ke ywords included social competence, behavior, social skills, functional profile, functional ability, functioning, predictor, person oriented, and children Second, searches of relevant

PAGE 60

60 websites were conducted to identify national reports and data (e.g., In dividuals with Disabilities Education and Improvement Act Section 618 Child Count Data) including: PEELS, Institute of Education Science (IES), Office of Special Education Programs (OSEP), Educational Resource Information Center (ERIC), IDEA data, National Research Council, and World Health Organization (WHO) Third, the reference lists of identified studies, articles, and reports were reviewed Articles, reports, and studies obtained through the electronic search procedures were reviewed by title and ab stract to determine if the source should be included as part of the literature review Sources were included in the review of literature related to (a) social competence and contextual factors, (b) concerns about us ing IDEA disability categor ies to charac terize young children and (c) the ICF CY framework, if the source provided background to the topics or was illustrative of key issues All articles, reports, and studies that were related to the PEELS data set were included in the review Empirical stud ies reviewed in the final section of this chapter were selected based on the following inclusion criteria: the study was conducted with a large sample that included young children ages 3 through 5 that examined (a) the contribution of child functioning bey ond disability categor y related to child outcomes, or (b) the use of person oriented analytic techniques to identify subgroups of children with similar profiles of abilities and to examine relationships between these subgroups and child outcomes All sources reviewed in this chapter were available in English Social Competence competence development, school readiness and achievement, and later in life well being ( National

PAGE 61

61 C ouncil on the Developing Child, 2004a ; Shonkoff & Phillips, 2000 ) The relevance of this review for the present study is to describe issues related to examining young cial competence, including definitions and measurement, outcomes associated with social competence, and factors that promote or hinder the development of Defining Social Competence The present study defined social competence from a performance based perspective as described by Odom, McConnell, and McEvoy (1992) and Odom, McConnell, and Brown (2008) These authors described social competence by two key eving social goals for the social context This view of social competence emphasizes the necessary skills se of appropriate behaviors for a given social context (i.e., social skills), and the absence of or refraining from inappropriate use of behaviors (i.e., problem behavior s ) in a social context ( Brown, Odom, McConnell, & Rathel, 2008; Conroy, Brown, & Olive 2008; Odom et al., 2008 ) Using a performance based approach, social competence is characterized by making evaluative judgments about the social performance of children in social contexts ( Odom & McConnell, 1985; Odom et al., 1992; Odom et al., 2008) Table 2 1 shows the made For example, a child might be judged to have good social skills and use socially appropriate behaviors (i.e., no problem behavior s ) Alternatively a child might have good social skills but display socially inappropriate behaviors

PAGE 62

62 Within this framework, a variety of social behaviors might be observed to evaluate For example, initiating and maintaining int eractions, engaging in play, attending and listening to others, maintaining friendships, responding to adult or peer requests, playing or sharing with others, regulating emotions and feelings, or showing empathy for others are all skills that are positivel y related to social competence Given the range of social behaviors related to social competence, Odom and colleagues (2008) emphasized the importance of conducting multi component assessments of social competence to inform a more comprehensive profile of Challenges Related to the Measurement of Social Competence development, the measur ement of social competence has received attention in the empirical literature Odom and colleagues (2008) outlined various methods that settings and to inform judgments about social competence These include observational strategies to count the number or rate the effectiveness of social behaviors children display ( cf. Brown, Odom, & Holcombe, 1996), parent or teacher ratings of child behavior ( cf. Merrell, 200 2 ), social prob lem solving tasks ( cf. Webster Stratton & Lindsey, 1999), sociometric approaches using peer ratings or peer nominations ( cf. Wu, Hart, Draper, & Olsen, 2001), and measures of friendships ( cf. Gifford Smith & Brownell, 200 3 ) Odom and colleagues noted that each approach, competence Shonkoff and Phillips (2000) described concerns with each measurement approach For example, observational methods focus on specific skills in limited

PAGE 63

63 contexts and do not capture enduring interactions Teacher and parent ratings provide Socio metric approaches using peer ratings or peer nominations provide insight into The stability and applicability of these ratings beyond defined peer groups requires further examination. Raver and Zigler (1997) noted conc erns about using individual measures to inform These authors also competence within the complex social demands of div erse environments that children experience with peers and adults They acknowledged, however, the importance of cost effectiveness and feasibility when making assessment decisions for research or evaluation purposes Current recommendations for measuring competence emphasize the selection of measures that reflect the research or practical questions to be addressed (Odom et al., 2008) Outcomes Associated with Social Competence Despite measurement challenges, empirical evidence has demo nstrated the years of life and important outcomes such as school readiness, later school success, and overall well being (Odom et al., 2008; Raver & Zigler, 1997; Shonkoff & Phillips, 2000) These outcomes have been identified from both positive and negative frames In a positive frame, children with social competence engage in more peer social interactions and maintain more friendships From a negative frame, children wit h social

PAGE 64

64 competence challenges experience lower rates of peer acceptance and engage in fewer social interactions Research often reflects the negative frame with respect to poor outcomes associated with social competence challenges For example, childre competence might result in conduct problems, which have been linked to peer rejection and isolation, substance abuse, school drop out, juvenile delinquency and incarceration, and depression and other mental health concerns in adult years (Campbell, Breaux, Ewing, & Szumowski, 1986; Campbell et al., 2006) Many researchers have cautioned, however, these later in life outcomes are not definitively associated with early conduct problems and have suggested further examinations are needed of the complex processes and associated factors that result in undesirable outcomes (Campbell, Shaw, & Gilliom, 2000; Loeber & Hay, 1997) For example, Campbell and colleagues identified that concerns related to problem behaviors and poor social skills were either (a) time limited to the period of early childhood, (b) were not identified in early childhood but emerged during adolescence, or (c) were present in early childhood and adolescence These authors noted the latter group of children w as identified wi th problem behaviors that were considered seve re and persistent across settings during early childhood Campbell et al emphasized that, for some children, identified concerns about social competence might be transient or reflect the process of developmen t Although concerns related to social competence might not persist over time, researchers have also identified deleterious outcomes young children with social competence challenges might experience more immediately These outcomes relate to child

PAGE 65

65 social interactions, poor adaptation to school and new environments, lowered teacher expectations and academic opportunities, and, in some cases, preschool or early school sus pension or expulsion (Gilliam, 2005; Ladd & Price, 1987, McIntyre, Blacher, & Baker, 2006; Perry, Dunne, McF adden, & Campbell, 2008) The extent to which these later in life and more immediate outcomes affect children with disabilities continues to be ex amined empirically (Odom et al., 2008) Some research has shown that social competence challenges experienced by young children with disabilities have been associated with peer rejection, fewer friendships and social interaction opportunities, poor adaptation to school, and increases in challenging behaviors (Baker et al., 2003; Guralnick, Hammond, Con nor, & Neville, 2006; McIntyre et al., 2006) Increases in challenging behaviors for children with disabilities have been associated with parental str ess and exclusion from general education settings (Baker et al., 2003; Lecavalier, Leone, & Witt 2006) Sigman and Ruskin (1999) noted the extent to which children with intellectual disabilities experience independence in later in life settings (e.g., em ployment, independent living) is as much a function of their social competence as their cognitive competenc i es Child Factors Associated with Social Competence Researchers have conducted investigations to explicate associations between attributes of chil dren and their social competence Odom et al (2008) referred to child inside factors Inside out factors refer red In the present study, c hild f actors identified in the literature that were related to social competence were characterized as malleable or non malleable characteristics Non malleable factors refer to aspects of the child that are not altered or changed by interventions or varying

PAGE 66

66 co ntexts (IES, 2011) malleable child factors Malleable factors refer to child characteristics that are amenable to change due to developmental progression or intervention or that might be altere d by different settings and contexts (IES, 2011) abilities or language and communication skills change over time based on developmental progression or intervention would be considered malleable factors Previous resear ch has demonstrated the associations between the non malleable For example, some studies have demonstrated that boys are more likely than girls to demonstrate physical aggres sion or externalizing problem behaviors (Mendez, McDermott, & Fantuzzo, 2002; Olson & Hoza, 1993) Children of different race/ethnicity groups have shown different social competence trajectories and vary in the extent to which certain skills are demonstra ted (Raver, Gershoff, & Aber, 2007) demonstration of social competence becomes more complex and sophisticated as they age (Mendez et al., 2002). Because of the developmental nature of social competence, many researchers highlight the importan ce of distinguishing between age expected social skills and behaviors and other indicators that might be perceived as social tantrums, grabbing of toys, and crying are n ormal expressions of young children learning to navigate a social play setting at certain ages. The persistence of these behaviors at older ages might be indicative of social competence challenges. Malleable child characteristics and skills have also be en shown to be associated with including c

PAGE 67

67 communication skills, executive functioning, and temperament ( Center on the Developing Child at Harvard University, 2011; Herbert Meyers, Guttentag, Swank, Smith, & Landry, 2006; Qi & Kaiser, 2003) For the purposes of this review, executive function is defined as working memory, inhibitory control, and cognitive flexibility (Center on the Developing Child at Harvard University, 2011); and temperament is defined as regulation of emotions (Herbert Meyers et al., 2006; Shonkoff & Phillips, 2000). Cognitive, language, or communication delays might be evidenced by young children with disabilities For these children, social interactions and demonstration of social competence in social contexts might be more difficult (Guralnick, 1999, Schneider & Goldstein, 2008; Strain, Schwartz, & Bovey, 2008; Drasgo w, Lowery, Turan, Halle, & Meadan, 2008) Odom and colleagues (2008) noted, however, that many children with identifi ed disabilities are socially competent and benefit from strong peer relations and social interactions The extent to which children with disabilities experience additional difficulties with social competence might be more associated with their functional abilities, as reflected in various malleable factors (e.g., use of arms and legs, cognitive and language abilities, hearing, vision, healt h status) than their disability status Contextual Factors Associated with Social Competence In addition to child factors, researchers have also emphasized the role of & Keane, 2010; Odom et al., 2008; Qi & Kaiser, 2003) Odom et al (2008) referred to outside in factors Outside in t have

PAGE 68

68 been associated with social competence In the present study, these contextual factors are referred to as family factors and environmental factors and family characte ristics, including socio economic status, parent education, martial status and family size, family stress and disruption, domestic violence, substance abuse, and history of poor mental health or depression (Krishnakumar & Black, 2002; Loeber & Hay, 1997; N ICHD Research Network, 2003; Raver et al., 2007; Schmidt, Demulder, & Denham, 2002) Factors related to parental attitudes and behaviors that influence parent competence Specific parent al attitudes and behaviors that have been examined in belief that they can affect change in their child ren ren ation of social opportunities for children, and parent child interactions (Guralnick, 1999; Oravecz, Koblinsky, & Randolph, 2008; Raver et al., 2007; Schmidt et al., 2002). Researchers have noted the transactional nature of d family factors, specifically the transactional nature of parent child interactions (Olson & Lunkenheimer, 2009; Rutter, 1979; Sameroff, 2009; Sameroff, Seifer, Baldwin, & Baldwin, 1993). defined as including select community and school factors Community factors that have been economic status of the r & Hay, 1997; Oravecz et al., 2008; Romano, Kohen, & Findlay, 2010) School factors

PAGE 69

69 programs and the provision of interventions or supports to promote social competence (e.g., social skills training, positive behavior supports; Domitrovich, Cortes, & Greenberg, 2007; Elias, Gara, Schuyler, Branden Muller, & Sayette 1991; Odom et al., 199 9) Research has examined the individual contribution of select contextua l factors to The explanatory patterns and consistency of competence has varied across studies, population samples, and analytic methods us ed to examine these relationships (Krishnakumar & Black, 2002; Raver et al., 2007; Sameroff & Seifer, 1983). Shonkoff and Phillips (2000) noted that individual factors in isolation are unlikely to influence social development Contemporary perspectives ab out social competence and their social competence Researchers have often examined these factors in terms of risk and have focused on the identification of factors associ ated with undesirable outcomes. For example, in a series of studies about hard to manage preschool boys, Campbell and colleagues used a cluster analysis technique to examine subgroups based on different patterns of risk factors. Five groups were identifi ed: boys with child and family risk factors, boys with child risk factors only, boys with family risk factors only, and two groups of boys with low or no risk factors (Campbell, 1994; Campbell, March, Peirce, Ewing, & Szumowski, 1991). These studies showe d boys in the group with family and child risk factors

PAGE 70

70 consistently displayed lower social skills and more problem behaviors, with high rates of externalizing problem behaviors when compared to other boys in the sample. Another key aspect of studies on contextual factors has been the identification of cf. Werner & Smith, 2001) These factors have been referred to as promotive or protective factors Gutman and colleagues (2003) defined promotive factors as variables positively related to positive outcomes and protective factors as variables positively related to positive outcomes for children in a high risk group but not for children in a low risk group From an analytic perspective, promotive factors address main effects and protective factors address interaction effects Summary Related to Social Competence Social competence has emerged as a national priority outcome for young children and young children with disabiliti es given the role of social competence in many aspects success, and later in life well being Non malleable and malleable child factors and select contextual factors have been examined in relation to childre Researchers have noted the interacting and accumulating influence s of contextual factors in relation to social competence Researchers continue to identify a range of risk factors, promotive factors, and protective factors that mig ht be associated with or influence development and their social competence The review of the literature related present study in several ways For example, the performance based view of social competence described by Brown and colleagues (2008) was consistent with the assessment of social competence available in the PEELS data set (i.e., Preschool and

PAGE 71

71 Kindergarten Behavior Rating Scale, Second Ed ition [PKBS 2]; Merrell, 2002), which define d how social competence was operationalized in the present study The PKBS 2, however, was the only standardized measure available in the PEELS data set, which limited the use of multiple measures The child an d contextual factors identified in the selection of child and contextual variables from the array of variables available in the PEELS data set Selecte d variables included potential risk factors or protective factors and were used to examine the moderating role of these factors on the relationship between functional ability profiles and social competence Use of Disabilit y Categories to Characterize Children This sectio n of the present literature review focuses on the use of disability categories under the Individuals with Disabilities Education Improvement Act (IDEA, 2004) The relevance of this review to the present study is to illustrate concerns with using disabilit y categ ories to examine relationships with outcomes for young children, including social competence Thirteen disability categories are specified in the IDEA for children age 3 through 21 The categories are autism, deaf blindness, deafness emotional d isturbance, hearing impairment s, mental retardation, multiple disabilities, orthopedic impairments, other health impairments, speech or language impairment s specific learning disability, traumatic brain injury, and visual impairments including blindness For children ages 3 through 9, the add itional disability category of developmental delay is also speci fied in the law (NICHCY, 2009) To be eligible for services under IDEA, a child must meet the established criteria for a disability category and subsequ ently receive a primary Many unintended consequences and

PAGE 72

72 concerns about a categorical approach to characterize disability have been described in the literature (e.g., Florian et al., 2006; Hobbs, 1975; Sim eonsson & Scarborough, 2001) The following sections review key issues related to the use of disability categor y and variations in state eligibility systems, particularly with respect to the use of disability categor y as predictor or explanatory variables in research Concerns with Categorical Approaches In 1975, Nicolas Hobbs authored The Future of Children: Categories, Labels, and Their Consequences for the federally funded project focused on the c lassification of e xceptional c hildren In this seminal report, Hobbs described issues associated with using a categorical approach to describe children with disabilities or children with social and economic risk factors Some of the issues identi fied by Hobbs, and subseq uently by others over the years, include (a) the stigmatizing effects of categorical labels (Smith & Schakel, 1986 ; Simeonsson & Scarborough, 2001); (b) the poor reliability and validity of diagnostic procedures that lead to classification ( Haring, Lovett et al., 1992; McLean, S mith, McCormick, Schakel, & McE voy, 1991 ); ( c ) the imprecision of categorical labels to inform interventions or support services (Florian et al., 2006; Reschly, 1996; Simeonsson & Scarborough, 2001); and ( d ) the variation of disabili ty categories and eligibility determination criteria across states and agencies (Linehan, 2001; Snyder, Bailey, & Auer, 1994) Others have noted that the categorical approach locates the disability as a fixed characteristic of the child or person (Burke & Ruedel, 2008; Florian et al., 2006) and provides limited or no information a bout weaknesses, support needs, or secondary conditions (Forhan, 2009; Haring Farron Davis et al., 1992)

PAGE 73

73 For young children with disabilities, the use of developmental delay was intended related services (McLean et al., 1991) The use of this non categorical option was at the discretion of the state States could chose to adopt the category, determine the age range in which the category might be applied, and provide discretion to local education agencies (LEAs) about whether to adopt the use of developmental delay within a district (Linehan, 2001) Advocates of the non categorical o ption encouraged a broad definition in which [sic] represents significant delay in 1) Proponents of developmental for young children asserted this approach would remediate some of the problems identified with the categorical approach, such as encourage service individualization and improve continuity of service between Part C and Part B Despite these potential benef its, researchers cautioned that developmental categorical option as intended (Haring, Lovett et al., 1992, p 155) In 2000, the National Association of State Directors of Special Educat ion (NASDE) in collaboration with Frank Porter Graham Child Development Center at the University of North Carolina (UNC) convened a meeting of key stakeholders to identify trends in state practices related to the use of developmental delay as an eligibilit y category (Linehan, 2001) As part of this summit, researchers from UNC reported findings from a survey of state directors of special education This study revealed that state directors viewed the use of developmental delay option in four different ways First, some directors considered the category as one that was different than all other Part B

PAGE 74

74 categories (i.e., children not eligible for other categories could be identified for services with developmental delay) Second, some directors considered the developmental delay category to encompass all possible categories represented in Part B for preschool aged children (i.e., all preschool aged children with disabilities should be classified as developmental delay) Third, some believed the category was a ppropriate to use to identify the early manifestation of disabilities that would later be identified by other Part B categories Finally, some state directors felt the developmental delay category provided an option for a functional approach The differe nces in the special s provided insight into the variation of practices across states and the processes by which developmental delay often became another category in some states. Current Variations in Eligibility Determination Systems In 2004, the National Association of State Directors of Special Education (NASDE) commissioned a report on state terminology, definitions, and eligibility criteria for disability categories In this report, Muller and Markowitz (2004) described the vague criteria that the federal regulations provided with regard to determining eligibility across all disability categories The authors noted that federal regulations only provide specific recommendation s for eligibility determination for one disability category (i.e., learning disability) The absence of guidance in the federal regulations allowed states to set their own terms, definitions, and criteria for eligibility determination as long as the state creates a structure that satisfies the requirements for federal reporting (i.e., Section 618 requirements) The report described the extensive variation across the 50 states and 3 non state U.S jurisdictions with regards to the use of federal terms, sta te definitions,

PAGE 75

75 and state criteria for determining eligibility for each of the 13 disability categories and the optional category of development delay Their report highlighted that most states adopt the terms used in the federal legislation to describe a disability category One notable exception is the term mental retardation ; only about half the states use the term mental retardation while other states use other comparable terms such as cognitive delay, cognitive impairment, or intellectual disabilit y Muller and Markowitz also reported that states did not use all categories or some states combined categories They reported that 15 states included deafness under the category of hearing impairment, seven states did not use the category of multiple dis abilities, and three states included autism, other health impairment, or traumatic brain injury under the category of orthopedic impairment A major focus of the report was the analysis of eligibility criteria, including specificity of criteria, types of assessments used to determine eligibility categories, use of an outside professional as part of the eligibility determination process, incorporation of qualitative data to inform decision making, and membership requirements for the eligibility determinati on team specified by each state Across states, the specificity of the criteria and required assessments was variable by disability category More than half the states outlined specific criteria for each disability category with the exception of deaf bli ndness, deafness, and multiple disabilities On average, most states did not require the use of an outside professional to contribute to eligibility determinations across disability categories Very few states required the use of qualitative data or spec ified requirements for the composition of the eligibility determination team.

PAGE 76

76 In 2005, Muller, Markowitz, and Srivastava conducted follow up investigations from their previous report to determine whether use of specific terms or the presence or absence of an eligibility criteria resulted in differences in proportions of children served by a specific disability category The researchers selected five disability categories (i.e., autism, emotional disturbance, mental retardation, specific learning disability and speech or language impairments ) and used data from the 50 states for children age 6 through 21 served under IDEA Independent t tests were used to assess differences between groups (i.e., term/criteria used group and term/criteria not used group) On average, from the selection of terms and criteria the authors examined, use of a specific term or criteria did not result in statistically significant differences in the proportion of children served by different disability categories across the 50 stat es with two notable exceptions First, the use an outside professional as part of the requirements for the diagnosis of autism was associated with a lower proportion of children being identified for services in the category of autism Second, the use of the term mental retardation was associated with a lower p roportion of students being identified for services in this category compared to states that used other comparable terms (e.g., cognitive delay, cognitive impairment, or intellectual disability) Danaher (2007) compiled an analysis of eligibility policies and practices for preschool aged children with disabilities variability among states over the years in relation to eligibility determination and eligi bility categories The report outlined variation in (a) the use of developmental delay across states, (b) the age range to which developmental delay is applied, (c) the relationship between the use developmental delay and other categories for children, (d )

PAGE 77

77 the eligibility criteria that are applied across states, (e) the polices that support transition from Part C to Part B, and (f) the within state variation resulting from LEA discretion on the adoption of developmental delay Illustration of Concerns an d Implications for Research Because of the variation in state eligibility systems, states might use the same disability category to identify children with different disabilities and characteristics (Chambers Perez, et al., 2004) For example, in State A, the IDEA eligibility determination system identifies all young children ages 3 through 6 with developmental delay In this state, a 3 year old child with significant multiple impairments including intellectual disabilities and physical disabilities is id entified for services under the disability category of developmental delay In the same state, a 4 year old child with speech articulation difficulties also is identified for services under the disability category of developmental delay Despite the diff characteristics, both children would be identified with developmental delay. To illustrate further the variation related to specific disability categories such as developmental delay, descriptive information about chi are shown in Appendix A The IDEA related disability category for each of the four related to different functional skills are shown In this example, childr en resided in different states As each child category for each child was devel opmental delay. Moreover, children with the same disability and functional ability characteristics might be identified by different disability categories across states because of the

PAGE 78

78 variation in state eligibility systems ( Chambers, Perez, et al., 2004; Dunst, Trivette, Appl, & Bagnato, 2004) For example, the children in the above referenced example move to State B where the eligibility determination system permits use of most IDEA related categories for preschool children In this state, the 3 year ol d child with significant multiple impairments, including intellectual disabilities and physical disabilities, is identified for services under the disability category of multiple disabilities while the 4 year old child with speech articulation difficultie s is identified for services under the disability categor y of speech or language impairments In this example, each characteristics es because of the policies of the state in which they reside. Differences in the identification of children with disabilities that are a result of variability in eligibility systems across states is further illustrated by comparing state data on children r eceiving services Table 2 2 shows data for the number of children, 3 through 5 years of age, served under IDEA by disability category for three example states (Minnesota, Washington, Wisconsin) These states were identified for similarities in total number of preschool children served across all disability categories As shown in Table 2 2, the percentages of children identified by different disability categories are different across the three states For example, Minnesota and Washington identified approximately 50% to 57% of children with developmental delay, while Wisconsin only identified approximately 19% of children with developmental delay Wisconsin, however, identified 67% of children with speech or language impairments but Minnesota and W ashington identified 32% and 27% of children with this category,

PAGE 79

79 respectively Minnesota has a higher percentage of children with autism, but lower percentage of children with other health impairments (an IDEA disability category used in some states when a child does not have a medical diagnosis of autism) compared to Washington or Wisconsin These data might illustrate patterns of differential prevalence rates in the population of children with disabilities across states or these data might illustrate so me of the issues related to comparing state level data when disability categories result from different state eligibility systems. As shown in the previous examples and noted by Florian and colleagues (2006), there is as much within category variation as t here is a between category variation for the IDEA disability categories This variation potentially introduces several limitations when aggregating data for use in nationally representative data sets and using disability category as a predictor or explana tory variable of child outcomes (M a cMillian & Resch ly, 1998) Limitations related to the use of IDEA categorical descriptions of disability include (a) compar ing children by disability category from different states when disabilities categories, definitio ns, and criteria differ across states and (b) distinguish ing differences between children within a disability category (Florian et al., 2006; Simeonsson, Simeonsson, & Hollenweger, 2008) Moreover, disability category alone provides limited information a functioning and subsequent support needs and provides no information about secondary conditions or contextual factors that might facilitate or inhibit competence (Florian et al 2006 ; Forhan, 2009 ) Summary Related to Categorical Descriptions of Disability The v ariations in state eligibility determination systems have resulted in disability categories that are idiosyncratic to individual states (Danaher, 2007; Muller &

PAGE 80

80 Markowitz, 2004 ) Given the issues related to variations within and between disability categories, the use of IDEA disability category to characterize children as a homogenous group in research is not recommended ( Florian et al 2006 ; Lollar & Simeonsson, 2005 ) Alter natively, researchers have recommended using a functional Based on concerns identified in the present review and recommendations using for a functional approach, the present study exam ined the use of functional ability profile subgroup membership as an alternative to disability category as a correlate of child outcomes. The ICF CY Framework This section of literature review focuses on the International Classification of Functioning, Dis ability, and Health for Children and Youth (ICF CY) The ICF CY was the conceptual framework that guided the present study The ICF CY is part of the Family of International Classifications (WHO FIC) promulgated by the World Health Organization The cl assification system is designed to categoriz e relevant dimensions of health and well being for the purposes of systematic recording or analysis of data across health and related sectors both internationally and nationally with the aim of improving health, well being, and related services (Madden, Sykes, & Bedirhan 2007) Three classifications are part of the primary reference classification system: the International Classification of Disease, 10th edition (ICD 10); International Classifica tion of Functioning, Disability, and Health (ICF); and International Classification of Health Interventions (ICHI) The ICHI is currently under development The ICF CY is considered a derived classification system, based on the framework of the ICF, designed spe cifically for children birth through adolescence The ICF CY is the classification of interest in the present study.

PAGE 81

81 The ICF CY is not used to identify or classify the etiology of health condition, disability, disease, or disorder; this classification has been the domain of the ICD 10 (WHO, 1992) One potential use of the ICF CY is to be used in conjunction with the ICD 10 to understand functioning in relation to a disability/health condition (Peterson, 2005) For example, the ICD 10 has classification c odes can be used to denote if a child has conditions such as Down syndrome (code Q90.9), blindness (code H54.0), asthma (J45.9), or other diagnoses T he ICF CY can be used to document the level of functioning related to cognitive function (code b117.0), s ensory functions (code b156.0), language functions (code b167.0), motor activities such as walking or running (code b770.0) or other domains to enhance the information that is known about a child identified with a condition, such as Down syndrome through the ICD 10 When used in combination, the two systems (i.e., ICD 10 and ICF CY) are designed to provide a (WHO, 2010) History of the ICF CY The ICF framework, which i s the basis for the ICF CY, is a revision of the original I nternational Classification of Impairments Disability, and Handicaps ( ICIDH ) framework proposed by the World Health Organization in 1980 At the time it was first disseminated, the ICIDH presente d four novel approaches to common conceptualizations of disability (Simeonsson et al., 2003) First, it conceptualized disability as the consequence of an underlying health condition or disorder Second, it differentiated the consequences of disability a cross the body, the person, and society Third, it introduced a multi dimensional view of disability, in which functioning might be

PAGE 82

82 activities, and a handicap or disadvantage related to participation in society Fourth, it introduced a numeric coding system that could be used to document the impact of disability across the body, the person, and society Despite the significant advancements that the ICIDH present ed, it was released as an experimental document and it was not widely disseminated or adopted (Florian et al., 2006; Simeonsson et al., 2003; Simeonsson, Lollar, Hollowell, & Adams, 2000) Following its release, the ICIDH was criticized for not acknowledg ing contextual factors al., 2000) and for its linear approach from health condition to impairment to disability to handicap ( Figure 2 1; Florian et al., 2006; Simeon sson et al., 2000) These concerns, in addition to other concerns that were raised about the ICIDH, resulted in revisions to the framework in the 1990s During this time, two beta versions of the revised ICIDH were released for continued feedback and rev isions before the current ICF was released in 2001 (Simeonsson et al., 2000; Simeonsson et al., 2003) The ICF framework, illustrated in Figure 2 2, incorporates aspects of body functions and structures, activities, and participation to describe disabili ty and functioning, however, the framework also highlights the influence of health conditions and contextual factors (i.e., environmental and personal) on functioning The ICF framework focuses on components of health instead of disease, emphasizes the im portance of health and wellness, and emphasizes the universal nature of functioning and disability, such that health and wellness of all persons are influenced by the components of body functions and structures, activities, participation, and contextual

PAGE 83

83 fa ctors (Lollar & Simeonsson, 2005 ; Raghavendra, Bornman, Granlund, & Bjorck Akesson, 2007) The ICF framework offers descriptors that emphasize positive aspects (e.g., integrity, facilitators) or negative aspects (e.g., limitation, restriction, barriers) of health and functioning The framework aims to reorient users on the health condition and the consequences of health conditions on functioning instead of a focus on etiology or diagnosis (Carlson, Benson, & Oakland, 2010) A key aspect of the framework is the emphasis on the multi dimensional and interactive features of functioning and disability including contextual factors (Raghavendra et al., 2007) The ICF is designed to provide a broader conceptual framework related to disability as contrasted, for example, with IDEA disability categories The ICF is considered a classification system for functioning and disability; the ICF is not a specific measurement instrument The ICF classification system presents chapters and codes for related domains fo r the framework components As shown in Table 2 3, the ICF classification system is organized in chapters related to body functions, body structures, activities and participation, and environmental factors The ICF classification system does not provide domains related to personal factors Within each chapter, there are second level and third level codes, for some domains there are also fourth level codes Table 2 4 shows an example of second, third and fourth level codes for C hapter 1 of the body fun ctions components For each ICF code, a generic rating or qualifier can be applied to describe the level of functioning A rating of 0 is applied to indicate normal function related to the code Normal function is defined by what is typical of the avera ge, same aged peer (WHO, 2007) A rating of

PAGE 84

84 1 is applied to indicate mild functional impairment, limitation, or restriction related to the code A rating of 2, 3, or 4, is applied to indicate moderate, severe, or complete functional impairment, limitatio n, or restriction, respectively, related to the code This qualifier system is used to describe the body functions, body structures, and activities and participation codes For the environmental codes, a (+) is assigned if the code is considered a facili tator and a ( ) is assigned if the code is considered a barrier Since the release of the ICF, some have offered continued feedback and critique (Dahl, 2002; Imrie, 2003) Others have highlighted its strengths and focused on its contributions to the co nceptualization of disability and the potential for a common language that might be used across countries and cultural contexts (Florian et al., 2006; Peterson, 2005; Rosenbaum & Stewart, 2004; Simeonsson, 2003; Simeonsson et al., 2000; Simeonsson et al., 2003) Many authors, however, have noted the ICF was not developmentally appropriate for children and adolescence Based on concerns about using the ICF with children and youth, a derived classification system was released in 2006, the International Clas sification of Function for Children and Youth (ICF CY; WHO, 2007) The ICF CY provides a classification system with content appropriate for children and youth from birth through adolescence ( Bjorck Akesson et al., 2010) The ICF CY accounts for the expe cted variations and developmental changes during childhood Moreover, environmental factor domains were expanded to emphasize the role of family and school factors during these years of development Lollar and Simeonsson (2005) noted the ICF CY emphasize s the unique nature of child development and provides

PAGE 85

85 codes that can account for the pattern of change in the nature, intensity, and consequence of functional ability or disability during childhood. ICF CY Applications The ICF CY manual indicates the ICF C Y might be used to develop statistical tools, research tools, clinical tools, social policy tools, or educational tools (WHO, 2007; 2010) Simeonsson and colleagues (2006) identified seven specific applications of the ICF CY Four of these applications h ave particular relevance for the present study First, the ICF CY provides a framework for interdisciplinary practice Many authors allied health, mental health, nursing, ps ychology and psychiatry, education and special education, and social and family services Lollar and Simeonsson (2005) noted, 327) These auth ors suggested the ICF CY might provide a second language that could be used to facilitate communication and understanding across disciplines Second, the ICF CY can yield profiles of child functioning The focus on profiles of functioning switches emphas is from classifying the child using an IDEA related disability category, which might have stigmatizing effects, to describing the variation in child specific characteristics in the context of related factors that influence functioning A profile helps ide ntify needs and supports that are unique to the child Third, the ICF CY profiles can help clarify clinical diagnoses and co morbidit ies of traditional categorical systems For example, the diagnosis of autism might be related to impairment in social Fourth, the focus of the ICF

PAGE 86

86 CY cla ssification system is on functional characteristics related to difficulties and strengths in meeting the demands of daily life profile can have practical implications for individualizing interventions in educationa l and clinical treatment planning In the context of the present study, information about CY model, were used to empirically derive subgroups of children with similar functional ability profiles and to examine relationships between subgroup membership and social competence. ICF CY Usability The ICF CY is a classification system to document functioning and disability Evidence for ICF CY codes might be gathered from direct measur ements, observation, respondent interviews, or professional judgment (WHO, 2007) Because ICF CY is not a measurement instrument, many have focused their efforts on identifying existing measures or developing new measures that can be used to assess inform ation related to ICF CY codes For example, Morris, Kurinczak, and Fitzpatrick (2005) examined the extent to which seven different instruments identified for use with families to complete self assessment of child functioning provided items related to code s in the activities and participation component These authors noted that no existing instruments included items related to all domains and four instruments only provided items related three of the domains in the activities and participation component O gonowski, Kronk, Rice, and Feldman (2004) examined the extent to which two raters could apply ICF codes to scores from existing measures of child functioning These authors reported interrater reliability was higher (kappa >.80) for instrument items that corresponded directly with a single ICF code Kronk, Ogonowski, Rice, and Feldman (2005) further examined the

PAGE 87

87 extent to which two raters could assign ICF codes based on content from an open ended structured interview with families These authors reported two raters had high interrater reliability when assigning binary codes (kappa >.85) but lower interrater reliability when assigning a severity rating (Mean kappa .72; range .25 1.0) The ICF CY provides over 1,400 alphanumeric codes across the four list ed components ( Table 2 3) It is neither expected nor feasible to document every code listed on the ICF CY ( WHO, 2007) Instead, researchers have focused efforts on identifying measures and developing practical tools to document ICF CY codes for differen t purposes Researchers have suggested generic tools that provide a brief sample of ICF CY codes to gain an overall picture related a variety of codes, comprehensive checklists to assess the level two and three codes on specific domains, or core sets of c odes identified for specific purposes or specific health conditions (Simeonsson, 2009; Stucki, 2005; Stucki & Cieza, 200 4 ) ICF CY in Educational Contexts The ICF CY has been used in education and related fields such as special education, related service s, and early intervention Despite its strengths, researchers have not suggested the ICF CY replace disability categories for educational eligibility, clinical, or other administrative purposes however, many researchers have noted ICF CY based functional profiles provide information useful for practice, research, and policy (Simeonsson, 2009; Simeonsson & Lollar, 2005) For example, Florian and colleagues (2006) suggested ICF CY based functional profiles give educators relevant information for providing educational services including planning, individualizing interventions, assessment, and evaluations These authors proposed the information related to a

PAGE 88

88 disability c ategories functional abilities T he ABILITIES Index (Simeonsson & Bailey, 199 1 ) is a judgment based rating scale based on previous versions of the ICF It was designed to profile the functional abilities of children across nine developmental and performance domains : audition (hearing), behavior and social skills, Intellectual functioning, use of limbs, intentional communication, tonicity, integrity of health, eyes (vision), and structural status. Simeonsson, Bailey, Smith, and Buysse (1995) examine d the utility of the ABILITIES Index to create functional ability profiles. The authors used a sample of 91 children receiving early childhood intervention services to generate subgroups of children with similar functional ability profiles The sample of children were eligible for services based on the following disability categories: speech language impaired ( n = 53); mentally handicapped ( n = 15); behaviorally emot ionally handicapped ( n = 14); multiply handicapped ( n = 3); autism ( n = 2); orthopedically impaired ( n = 2); other health impaired ( n = 1); and learning disabled ( n = 1) Using a hierarchical cluster analysis, a six cluster model was identified (i.e., six subgroups with distinct and interpretable functional ability profiles were identified in the sample) Simeonsson et al described the differences between groups based on each ability profile For example, Cluster 1 which incl uded 42 children, was characterized by mean ratings reflecting normal or suspected disabilities across all nine domains Cluster 2 ( n = 18) was characterized by ratings that reflected

PAGE 89

89 substantial disability in three domains: social skills and behavior, in telligence, and intentional communication Cluster 3 ( n = 9) and Cluster 4 ( n = 14) were characterized by ratings that reflected substantial disability related to intentional communication or social skills and behavior, respectively Both Cluster 5 ( n = 5) and C luster 6 ( n = 3) were characterized by ratings that reflected substantial disability in use of limbs and tonicity, with substantial disability related to intelligence, health, and vision for cluster 5 The authors reported children identified by different disabilit y categories were distributed across the six subgroups Other examples of using the ICF in education contexts include the use of ICF codes in research conducted with large scale, national data sets Simeonsson, Scarborough, and Hebb e l er (2006) mapped ICF codes to eligibility descriptions provided by early intervention service providers in the National Early Intervention Longitudinal Study (NEILS) The authors used the ICF and ICD codes to describe variations in the population of child ren receiving services, beyond the three broad Part C eligibility categories (i.e., developmental delay, diagnosed medical condition, or at risk) The authors reported 71% of the sample had documented eligibility related to ICF body functions, 41% related to ICD health conditions, 10% related to ICF activities or participation, and 5% related to ICF environmental factors In addition, Chambers Perez, et al (2004) and Daley, Simeonsson, and Carlson (2009) have conducted research with nationally represent ative data sets in which an adapted version of the ABILITIES Index was used These studies are described in further detail under the S tudies that Used Functional Ability Composite Scores

PAGE 90

90 Summary Related to ICF CY The ICF CY provides a framework to describe and quantify abilities A primary strength of the ICF CY is the inclusion of contextual factors and their Proponents of the ICF CY have noted se veral possible applications The utility and feasibility of these applications continues to be tested and examined in research and in practice The ICF CY framework was used in the present study to frame the research questions and inform the selection o f variables from the PEELS data set This framework, therefore, guided the secondary analyses conducted Constructs identified in the ICF CY framework are consistent with constructs identified in the ext a nt literature on social competence, although diffe rent terms and labels are used For example, malleable and non malleable child characteristics described by the Institute of Education Sciences are consistent with aspects of body functions and activities, and personal factors, respectively, in the ICF CY framework R isk factors and promotive or protectiv e factors described in relation re consistent with the use of contextual hindrances or barriers and facilitators in the ICF CY framework PEELS Data Set In this section of the literature review, i nformation about the Pre Elementary Educational Longitudinal Study (PEELS) data set and findings from previous studies conducted with the PEELS data set are provided to set the context for the present study The PEELS study was a prospective, longitudinal study focused on preschool children with disabilities. The PEELS data set provides a nationally representative sample of 3,10 0 children with disabilities ages 3, 4, and 5 at the start of the study

PAGE 91

91 Children were followed for four consecutive years from 2003 04 through 2006 07, with a follow up year completed during 2009 10. Public Information from the PEELS Reports This section briefly describes public information about the PEELS sample Westat researchers who implemented PEELS restricted use data set (Carlson, Posner, & Lee, 2008); three overview reports from the first three waves of data collection (Markowitz et al., 2006; Carlson Daley, et al., 2008; Carlson et al ., 2009); one selected findings report (Carlson, Bitterman, & Daley, 2010); and a variety of two page briefs on PEELS findings Information about the PEELS data set and PEELS reports are available on the PEELS website ( www.peels.org ) Findings from the first wave of data collection focused on the characteristics of young children with disabilities and their families, the educational services they received, transitions from early intervention settings, and performance on measures of school readiness and social competence (i.e., social and behavioral skills ; Markowitz et al., 2006) The following paragraphs highlight findings from Markowitz et al (2006) wave 1 report In the PEELS sample, children identified with dis abilities were disproportionately male (70%) Children represented diverse families: White (67%), Hispanic (22%), and Black (11%) Due to small sample sizes, data for other races were not included in report ed findings related to race/ethnicity Twenty s even percent of all children with disabilities were from households with family incomes of $20,000 or less Nearly half the children were identified by the IDEA disability category speech or language impairments (47%), followed by developmental delay (27% ) All other disability categories combined made up the remaining 26% of the PEELS sample : autism (7%),

PAGE 92

92 mental retardation (4.5%), other health impairment (3%), learning disability (2.5%), orthopedic impairment (2%), emotional disturbance (1%), and a comb ined low incidence category (6%; multiple disabilities, deaf blindness, deafness, hearing impairment, visual impairment, and traumatic brain injury) Eighty six percent of parents indicated they thought their child spent the right amount of time with pee rs without disabilities The majority of children with disabilities had a teacher with a graduate degree (55%) or degree (38%) For children with an individualized family service plan before age 3, 31% of these children had a gap in services d uring the transition from Part C early intervention services to Part B preschool services Overall, children with disabilities were within one standard deviation of the population mean on ratings of social skills and problem behaviors as measured by the P KBS 2 (Merrell, 2002) Deviations were noted for specific disability categories; for example, mean scores on social skills for children with autism or mental retardation were significantly lower than other disability groups Overall, females had higher social skills ratings and fewer problem behaviors than boys Differences by race/ethnicity for social skills and differences by race/ethnicity and income for problem behaviors were also noted Findings from the second wave of PEELS data collection focus ed on changes in eligibility and classification status, changes in educational services, and growth in skills (Carlson Daley, et al., 2008) The following paragraphs highlight findings from the Carlson Daley, et al (2008) wave 2 report At the time o f recruitment, all children had an active IEP or IFSP, however, during the course of the study, some children no longer required special education services

PAGE 93

93 and were declassified Children who were declassified from special education services remained in t he PEELS sample Twenty one percent of children with disabilities were declassified at some point between recruitment and wave 2 data collection Of those, 2% were declassified before wave 1 data collection and then re classified by wave 2 data collectio n Declassification differed by primary disability category, for example, 37% of children identified with speech or language impairments were declassified, and 21% of children identified with emotional disturbance were declassified Children transitionin g between settings (22 % 24%) were more likely to be declassified than children who remained in the same setting (6%) between wave 1 and wave 2 Of children who continued to receive special education services, 23% were reclassified with a ne w primary dis ability category. Between wave 1 and wave 2, the number of special education services that children with disabilities received decreased, while the mean number of hours spent in general education settings with children without disabilities increased Chil performance on standardized assessments increased for letter word identification and applied problems, but remained stable for picture word vocabulary Ratings of problem behaviors ratings were not rep orted. experiences and social competence (Carlson et al., 2009) Between wave 2 and wave 3 data collection, 82% of children with disabilities made a transition between progra ms, schools, or grade Twenty one percent of children who transitioned were declassified compared to 9% of children who were declassified but did not undergo a transition

PAGE 94

94 her transition practices Teachers who indicated using more transition strategies were associated with improved transition outcomes Special educators were more likely to use more transition strategies than general educators The numbers of transition s trategies were also associated with district size and wealth (e.g., medium and small districts used more strategies than large or very large districts). low but positive ( r = .12 for males and r = .06 for females); however, correlations for problem behaviors were moderate and negative ( r = .39 for males and r = .52 for females) Different measures, however, were used to gather perspectives from parents and teachers Parent ratings were based on a scale created from 29 parent interview items (Carlson et al., 2009) and the teacher ratings were based on the PKBS 2 (Merrell, 2002) behaviors Ch ildren with more advanced social skills and fewer problem behaviors were more likely to be declassified. community and recreational activities and educational services for kinderga rten children focused on children age s 5 through 7 (i.e., wave 3 data) Based on parent report, 50% of children with disabilities participated in organized athletic activities, 20% in clubs or performances, and 10% in music lessons by disability category (e.g., 55% of children with speech or language impa irments

PAGE 95

95 participated in organized athletic activity compared to 28% with mental retardation) income, parent reports of neighborhood safety, and access to transport ation classroom was the primary education placement for 73% of kindergarteners receiving special education services Differences in general education or special education pla cements were associated with district size, metropolitan status, and district wealth For kindergarteners receiving special education services, teachers reported 44% accessed grade level materials without modifications, 29% required some modifications, 13 % required substantial modifications, and 14% used specialized curriculum or materials Teachers reported the amount of time spent in different types of activities for kindergarteners receiving special education services Instructional time was allocated in the following ways: 39% adult directed whole class activities, 23% adult directed small group activities, 16% adult directed individual activities, 13% child directed activities, and 9% instructional or therapy services Studies Conducted with the PE ELS Data Set S earch procedures described earlier in this chapter resulted in the identification of six published studies that were conducted using the PEELS data set These studies include at least one author from the Westat PEELS project team A brief s ynopsis of each study, followed by a detailed description of studies related to the present study, is provided. Li, Lee, Lo, and Norman (2008) conducted a study to investigate bias in teacher questionnaire data that were imputed using the Auto Impute softw are Bitterman, Daley, Misra, Carlson, and Markowitz (2008) conducted a study to examine parent

PAGE 96

96 satisfaction and educational services for children identified with autism (i.e., a subsample of 186 children from the larger PEELS sample) In 2009, Daley and Carlson reported information about the correlates of change in eligibility status (i.e., declassification) for preschool children in special education Daley, Simeonsson, and level of functioning using parent interview information In 2010, Carlson, Bitterman, and Jenkins St Clair, and the extent to which these factors were associated with educational performance Of the six studies listed above, Daley, Simeonsson, and Carlson (2009) is most closely related to the present study Daley and colleagues discussed the importance of the ICF CY for describing and quantifying The authors noted that the ABILITIES Index developed by Simeonsson and Bailey (1991) was a useful tool for describing a range of child abilities across developmental d omains Using the ABILITIES Index, parent s functioning (1) normal ability to (6) profound lack of ability on individual items (e.g., vision, hearing, communication) The authors used PEELS parent interview information from wave 1 data collection to create an index to characterize the nature and severity of Using 24 parent interview questions, the authors created 15 items that described children 4 point scale: ( 1) normal or typical functioning to (4) severe limitation in functioning Additional information about scale items is described in Chapter 3.

PAGE 97

97 The 15 items were mapped to relevant codes in the body functions and activities /participation sections of ICF CY To examine the utility of the items, the authors conducted multivariate regression analysis using the 15 variables (i.e., all items entered simultaneously) on eight outcome variables (i.e., picture vocabulary, letter wor d identification, applied problems, social skills, problem behaviors and three scales of adaptive functioning conceptual, practical, and social) Five items from the functional index, cognition, communicating with others, understanding, overall health, a nd regulation of activity, had statistically significant associations with at least four outcomes All other items were differentially related to varying outcomes Based on these analyses, t he authors created five possible disability severity indices co mposed of 15 items, 6 item s 7 item s 7 items, and 8 item s, respectively To examine each index, the authors calculated Pearson product moment correlations between a composite score from each index and the eight outcome variables The authors noted the composite score from the 15 item and 6 item indices had the highest correlations (range .2 2 through .53) with the outcome variables The authors compared the correlations between the indices and reported no statistically significant differences were ident ified Given the comparability of associations between outcomes using the composite score from 15 item and 6 item indices, the authors selected the 6 item index to create the PEELS Disability Severity Index This index had items related to: cognition, co mmunication, overall health, activity level, attention, and understanding. To examine the valid ity of the Disability Severity I ndex, the authors used correlations and t test to compare the 6 item index with additional data reported from parents and teacher s For example, index scores were significantly and positively

PAGE 98

98 related to the number of modifications children needed to access curricula and materials and the number of related services children received at school Moreover, statistically significant di fferences in mean index scores were identified for children who took the alternate assessment (i.e., higher mean index score) and those who received regular assessment ( i.e., lower mean index score ) and children who were declassified from special education services (i.e., lower mean index score) and those who remained in special education ( i.e., higher mean index score ) The authors also compared whether Disability Severity Index resulted in improved explanation of variance when compared to the use of disa bility category alone These an S tudies that used subheading Summary Related to Peels Data Set Findings from the PEELS reports and published studies using the PEELS data set highlight trends and patterns in educational services, transition experiences, and characteristics of young children with disabilities The findings from these reports and published studies might aid interpretation of analyses and findings for the present study F or example, 21% of all children were declassified between recruitment and wave 2 B y wave 1 data collection some children were no longer receiving special education services based on a primary disability classification and were not assigned a primary dis ability category Moreover, the published studies have offered informa tion to inform secondary analyse s conducted with the PEELS data set Given the recent availability of the PEELS data through restricted use licenses, those associated with the PEELS project have been the primary authors of studies published to date

PAGE 99

99 Empirical Studies with Direct Relevance for the Present Study In this section, empirical studies with direct relevance for the present study are described Studies reviewed in this sect ion were identified as part of the systematic search process previously described A study was included in this section of the review if the study was conducted with a large sample (i.e., more than 500 participants) that included young children ages 3 thr ough 5 and the study examined (a) the contribution of child functioning over disability categor y in relat ion to examining child outcomes, or (b) the use of person oriented analytic techniques to identify subgroups of children with similar profiles of abili ties to examine variations in child outcomes. A total of eight studies were selected for this review and are shown in Table 2 5 The table shows the extent to which each study included eight key features identified as important for addressing the researc h questions in the present study The eight key features were that the study (a) was conducted with a U.S based nationally representative sample, (b) was conducted with children with disabilities ages 3 through 5, (c) was based on the ICF CY framework, ( oriented analytic techniques to identify subgroups with similar profiles, (f) compared functional ability to disability category, (g) examined relat ionships between child characteristics and outcome variables, and (e) considered contextual factors as part of the analysis This type of analysis was conducted because no single study included all eight features that were directly relevant for the presen t study Nonetheless, these studies were illustrative of research on key topics related to different aspects of the present study. To aid in critical analysis of the research, studies were grouped based on similar characteristics Two studies examined th e use of child functional ab ility composite

PAGE 100

100 scores as a correlate to child outcomes and as an alternative to using disability categories (Chambers Perez, et al., 2004; Daley et al., 2009) Six studies used person oriented analytic techniques to iden tify subgroups of children with similar profiles of abilities and examined relationships between these profiles and child outcomes (Hair, Halle, Terry Humen, Lavelle, & Calkins, 2006; Haapasalo, Tremblay, Boul erice, & Vitaro, 2000; Janson & Mathiesen, 2008 ; Konold & Pianta, 2005; Sanson, Letcher, Smart, Prior, Toumbourou, & Ober k laid, 2009; Stephen s Petras, Fabian, & Walrath, 2009) Additional information about the type of variables and analyses used across the eight studies are shown in Table 2 6 Studies that Used Functional Ability Composite Scores Two studies were identified in this group Daley and colleagues (2009) develop ed a Disabilit y Severity Index using the PEELS data set As previously described, the authors examined the creation of fo ur possible disability indices (i.e., 15 item index, 6 item index, 7 item index, and 8 item index) The authors determined a 6 item composite index was sufficiently representative (see previous description of this Studies Conducted with the PEELS subheading) ability composite score and disability category to explain variations academic/cognitive, social, and adaptive functioning skills Stepwise regression models were used to compare the adjusted R squared values for each model and the change in R squared when (a) disability category was used alone, (b) functional ability composite score was used alone, and (c) disability category and functional abil ity composite score were used simultaneously

PAGE 101

101 For measures related to pre academic/cognitive and social skills, the authors composite score alone accounted for more va riance in scores than disability category alone (i.e., between 2 % to 6% more variance accounted for across measure s ) For measures related to adaptive functioning skills, va riance in scores than functional ability composite score alone (i.e., between 1 % to 3% more variance accounted for across subscales of an adaptive functioning measure) The authors noted, however, that adaptive functioning skills were assessed as part of the alternate assessment and the authors indicated the reduced sample size for this analysis might have impacted reported findings functional ability composite score were used in combination, the amount of variance accounted for in the outcomes measures was greater than when either indicator was used alone (i.e., between 4 % to 13% additional variance accounted for in each outcome measure) The authors noted the increase in variance explained when the two indicators are different constructs, with less overlap than might be predicted given traditional ideas (Daley et al., 2009, p. 548) The second study was conducted as part of the Special Education Expenditure Project (SEEP ; http://csef.air.org/ ) for the Center for Special Educa tion Finance (CSEF) The SEEP proje ct provides a nationally representative data set about general educ ation and special education expenditures for a sample of over 10,27 0 students with disabilities served under IDEA Part B Students were from a sample of 1,7 70 schools in 4 50 school districts or 30 affiliate d intermediate education units and 20 state run s pecial

PAGE 102

102 education schools Students with low incidence disabilities were oversampled to ensure adequate sample sizes for less common disability categories Students, schools, and districts completed a series of surveys to gather information about fiscal p olicy, programs, personnel and students For more information about the SEEP study see Chambers, Parrish, Shkolnik, Levine, and Makris (2003) SEEP reports have indicated that schools spend an average of $12,639 on each student with a disability compared to $6,556 for each student without a disability (Chambers, Parrish, & Harr is 2004) Moreover, schools spend different amounts of money to provide educational services for students from different disability categories (e.g., $10,058 for students with lea rning disabilities compared to $20,095 for students with multiple disabilities; Chambers, Shkolnik, & Harris 2003) These reports, however, have noted there has been a tremendous amount of variation in spending within disability categories. In a SEEP stu dy conducted by Chambers Perez, et al ( 2004 ), t he ABILITIES Index was used to examine the extent to which variation in a composite score of The researchers used a series of multivariate regression analyses to explore the relationships between ex information The authors reported adjusted R squared to indicate the amount of variance accounted for by each variable T he researchers reported 10 % of the variance in sp disability category alone The researchers then included in the model the number of

PAGE 103

103 age, gende r, and ethnicity (accounted for 2% of the variance), and information about the district size and district fiscal policies (accounted for 15% of the variance) Taken together, these variables and primary disability category accounted for 27% of the varianc e functional ability composite score in the previous regression model, 40% of the variance in special education expenditure was accounted for by the six variables Functional ability resulted in a 15 % increase in the R squared value The authors noted that when disability category and functional ability were included in the model, 42% of the variance was explained, and the combination of these seven variables resulted in the most variance explained related to expenditure on educational services for students with disabilities Person Oriented Analytic Approaches to Examine Child Outcomes Six studies were identified that used person oriented analyses to examine child outcomes None of these studies were exclusively conducted with children with disabilities; however, children with disabilities might have been represented in the study sample Hair, Halle, Te rry Humen, Lavelle, and Calkins (2006) identified subgroups of childr en based on profiles of school readiness in a sample of kindergarten children from the Early Childhood Longitudinal Study Kindergarten (ECLS K) They used profile membership to examine the extent to which membership differentially predicted academic and social outcomes in first grade The ECLS K data set is a nationally representative data set of 17,2 20 children who were enrolled in kindergarten and

PAGE 104

104 followed through the eighth grade The authors identified five dimensions of school readiness for young children: physical health, social/emotional development, approaches to learning, language, and cognitive The authors noted that previous researc h examined how different school readiness factor s or dimensions predicted later in life outcomes, but no previ ous research had examined their combined influence The authors used multiple indicators from the ECLS K data set to construct scales that represented each school readiness dimension The scales were dichotomized for each indicator dimension using liberal and conservative cut points The authors reported there was very little variance in scores related to one indicator dimension (i.e., approaches to learning) Using liberal cut s to learning; therefore, this indicator was not included in subsequent analysis In the first part of the study, the authors examined the creation of subgroups based on similar profiles of school readiness using K means cluster analysis (MacQueen, 1967) for both the liberal and conservative scales The authors reported that a four cluster group model had the best conceptual and statistical fit across both the liberal and conservative cut points Cluster 1 was described as comprehensive positive develop ment on the four dimensions: physical health, social/emotional development, language, and cognitive (30% of sample using liberal cut offs, 16% of the sample using conservative cut offs) Cluster 2 was described as social/emotional and health strengths (34 % of sample using liberal cut offs, 37% of the sample using conservative cut offs) Cluster 3 was described as below the mean on all four dimensions with the greatest risk in social/emotional development (13% of sample using liberal cut offs, 27%

PAGE 105

105 of the s ample using conservative cut offs) Cluster 4 was described as health risks, but was also below the mean on language and cognitive (23% of sample using liberal cut offs, 20% of the sample using conservative cut offs) The authors examined the distributio n of children across cut points and determined that the clusters generated from scales with liberal cut population. Using the subgroups based on liberal cut points, the authors examined how school readiness profile membership from kindergarten predicted academic ability and social adjustment outcomes in first grade while controlling for child, family, and classroom variables rst across all measures Children in the comprehensive positive profile performed significantly better than all children on academic measures, but no differences were found on social adjustment measures for these children and children with the social/emot ional and health strengths profile. As part of this study, the authors examined whether child and family variables differed across profiles and used logistic regression models to examine whether child and family variables were associated with profile membe rship The authors noted that children from ., lowest incomes, parents with the least education, single or teen mothers, low birth weight) were more likely to be identified in one of the two risk profiles Haapasalo, Trem placement in preadolescence in a sample of 1,034 boys from Canada The authors

PAGE 106

106 reported the sample was rel atively homogenous with majority of participants reported to be White, low SES, non immigrant, and French speaking families living in the same urban environment In this study, the authors compared two analytic techniques: a variable oriented approach and person oriented approach In the variable oriented approach, the authors used logistic regression models to examine the relationships pro social behavior to outcom e variables in preadolescence Outcome variables included self reported delinquency in preadolescence, peer rated social withdrawal, and placement in a specialized school setting Findings from this analysis showed that physical aggression, hyperactivity and low anxiety in kindergarten were the best predictors of delinquency in preadolescence, high anxiety and low pro social behavior in kindergarten were the best predictors of social withdrawal in preadolescence, and inattention and low pro social behavi or were best predictors of school placement in preadolescence. In the person aggression, hyperactivity, inattention, anxiety, and pro social behavior to create subgroups of children with similar profiles of behavioral dimensions The authors reported using the QUICK clustering techniques (i.e., K means clustering in SPSS), which resulted in eight clusters that were replicated in previous samples Cluster groups were described by the following c haracteristics Cluster 1 was identified as competent with low aggression, hyperactivity, inattention, and anxiety, and high pro social behavior ( n = 265) Cluster 2 was described as anxious ( n = 63; i.e., high anxiety and low pro social behavior), C lust er 3 was described as passive ( n = 120; i.e., high on

PAGE 107

107 inattention and low pro social behavior), C luster 4 was described as inattentive ( n = 145; i.e., high inattention and hyperactivity), and C luster 5 was described as nervous ( n = 157; i.e., high inattent ion and anxiety) Cluster 6, 7, and 8 were all associated with high aggression and high hyperactivity Cluster 6 was described as bullying and included low pro social skills ( n = 46), C luster 7 was described as undercontrolled and included high anxiety a nd inattention (n = 73), and C luster 8 was described as multi problem and included high anxiety, inattention and low pro social behaviors ( n = 74) Using the cluster sub groups, the authors used logistic regression models to examine how cluster membership predicted outcomes in adolescence Findings from this analysis showed that membership in the inattentive, bully, undercontrolled, or multi problem groups in kindergarten were the most noteworthy predictors of delinquency in preadolescence Membership in the multi problem group in kindergarten was the most noteworthy predictor of social withdrawal in preadolescence Membership in the passive, inattention, nervous, undercontrolled, or multi problem groups were the most noteworthy predictors of school plac ement in preadolescence The authors noted that results from the cluster analysis revealed that boys with multi problems were at the greater risk of negative outcomes. As part of this study, the authors examined the relative predictive accuracy of both va riable oriented and person oriented analytic approaches Using a ROC curve to examine the sensitivity and specificity of prediction, the authors reported that these approaches had near similar predictive accuracy (i.e., within 2% points for each outcome v ariable) The authors noted, however, a key distinction between the approaches was the ability to make conclusions related to prevention and intervention

PAGE 108

108 efforts For example, from a variable oriented approach, physical aggression and high anxiety were p redictors of delinquency From a person oriented approach, subgroups with high physical aggression were predictive of delinquency, but sub groups associated with inattention and hyperactivity were also predictive of delinquency even though these variables were not identified in the variable oriented approach Moreover, subgroups associated with high anxiety that were not also associated with physical aggression were not predictive of delinquency, even though this variable was predictive in the variable or iented approach predictive value of [variables] Tremblay, Boulerice, & Vitaro, 2000, p 163) In addition to comparisons between the two analytic appro aches, the authors noted the importance of examining the contribution of family adversity in kindergarten to the prediction of outcomes in adolescence The authors combined four variables on family characteristics to create an index of family adversity ( T able 2 6) In all models, family adversity in kindergarten was predictive of delinquency and school placement in adolescence, but not social withdrawal The authors also examined whether subgroups identified from the person oriented analysis differed by family variables Janson and Mathiesen (2008) created subgroups of children based on a profile of emotionality, and shyness They subsequently examined associations between subgroup membership The authors were also interested in the distributions of subgroup membership and individual stability of profile characteristics over time The study was conducted wit h 921 children

PAGE 109

109 and mothers over 9 years Data were collected on children at 18 months, 30 months, 4 to 5 years, and 8 to 9 years; samples at each time point were 921, 784, 737, and 512, respectively On each occasion, mothers completed a survey about the temperament related to emotionality, shyness, sociability, and activity, and a rating scale on problem behaviors. The authors completed an I States as Object Analysis (ISOA; Bergman, Magnusson, & El Khouri, 2003), which uses cluster analysis techn iques to generate subgroups based on shared profiles over each period of interest The authors used a two means cluster method The authors selected a five cluster model across the time points and subsequently assigned each participant to the nearest cluster for each time point (i.e., total 2,594 profiles across entire sample) Clusters 1, 2, and 3 were described as undercontrolled ( n = 345), confident ( n = 394), and unremarkab le ( n = 378), respectively Clusters 4 and 5 were described as inhibited ( n = 213) and uneasy ( n = 391), respectively The authors noted that the proportion of children assigned to different clusters changed over time For example, the confident profile was most common at 18 months (32% of sample), but the uneasy profile was most common when children were 4 to 5 years (32% of sample) The authors reported between 33% and 46% of children stayed in the same profile between adjacent time points. The author s reported that subgroup membership was consistently associated with For example, membership in the undercontrolled profile was associated with high externalizing

PAGE 110

110 behaviors at each t ime point, while membership in the inhibited and easy profiles were associated with high internalizing behaviors at each time point Konold and Pianta (2005) used a sample of 964 children assessed at 54 months to create subgroups of children based on sim ilar profiles of school readiness related to three measures of social functioning and three measures of cognitive functioning The sample was identified through a larger study (i.e., National Institute of Child Health and Human Development [NICHD] Early C hild Care Research Network [ECCRN], 2002), which recruited families from hospital visits shortly after the birth of a child at 10 locations in the United States The number of males and females was approximately equal in the sample The sample included m ore children classified as Anglo (83%) than any other race (African American 11%, Asian or Pacific Islander 1.5% or other 4.5%) The sample was predominately non Hispanic (96.5%) The authors reported 25% of the sample was below the poverty line based on 1.85 threshold for an income to need ratio Using a two step cluster analysis technique, the authors generated profiles of followed by K means cluster method The authors reported six subgroups for the final cluster model The groups were distinguished in the following ways: P rofile 1 children with attention problems (10% of sample), P rofile 2 children with low cognitive ability (7% of sample), P rofile 3 childr en with low/average social and cognitive skills (20% of sample), P rofile 4 children with social difficulties (17% of sample), P rofile 5 children with high social competence (24% of sample), and P rofile 6 children with high cognitive ability and mild extern alizing behaviors (22% of sample)

PAGE 111

111 subgroup membership with first grade academic achievement scores (i.e., picture vocabulary, letter word identification, and applied problems) In general, the high cognitive group outper formed other subgroup s on all measures and the low cognitive group had the lowest performance on all achievement measures The authors cautioned, however, that the subgroups based on standardized assessment scores at 54 months, accounted for less than 20 % of the variance in scores in first grade outcomes (R squared .08 for letter word identification to .18 for applied problems) The authors noted that the multi dimensional profiles provide oderate predictors of academic ability in first grade Sub groups were compared on child and family variables and the authors reported income to needs ratio For exampl e, on average, children with younger mothers, less parental education, and lower income to needs ratios were more likely to be associated with a risk or below average skills profile. Sanson, Letcher, Smart, Prior, Toumbourou, and Oberklaid (2009) created s ubgroups of children based on profiles of temperament during infancy and early childhood and examined the extent to which subgroup membership was associated with differences in problem behaviors, social skills, and academic competence in later childhood an d preadolescence The sample included 2,443 infants and families who were recruited when children were infants between the ages of 4 and 8 months from urban and rural areas in a large state in Australia Families were surveyed every 1 to 2 years until ch ildren were 11 to 12 years of age In later childhood and preadolescence,

PAGE 112

112 teachers of child participants were also surveyed At the final stage of data collection 70% of sample remained Surveys were designed to collect information about economic status between infancy and early economic status between later childhood (i.e., 7 to 8 years) and preadolescence Teachers were surv social skills, and academic competence Temperament measures included three dimensions: reactivity, inhibition, and self regulation To identify subgroups of children with similar temperamen t profiles, the authors used a two step cluster analysis using means cluster method The authors determined a four cluster model was the most appropriate representation of the data Cluster 1 was described as nonreact ive/outgoing (25% of sample), C luster 2 was described as high attention regulation (27% of sample), C luster 3 was described as poor attention regulation (28% of sample) and C luster 4 was described a reactive/inhibited (20% of sample) The authors examined differences between clusters by gender and SES SES was represented as a composite score that accounted for parents occupation and education level The authors reported that there were more males than females in C luster 3 (i.e., poor attention regulatio n) but all other clusters had similar gender proportion Related to SES, C luster 1 (i.e., nonreactive/ outgoing) contained more children from higher SES and C luster 4 (i.e., reactive/ inhibited) contained more children from lower SES The authors report ed that sub group membership in early childhood was differe ntially associated with outcome variables in later childhood The patterns of

PAGE 113

113 associations, however, were different across parent and teacher reported measures For example, high levels of physica l aggression were associated with C lusters 3 and 4 when reported by parents, but associated with C lusters 1 and 3 when reported by teachers SES was found to moderate physical aggression (i.e., low SES resulted in higher levels of physical aggression) whe n reported by parents, but not when reported by teachers The authors noted possible conclusions based on the differential patterns between teacher and parent report s of outcomes but also noted that more research was needed to explore this finding further Stephens, Petra, Fabian, and Walrath (2009) examined patterns of functional impairment based on subgroups of youth identified for community mental health services Children were identified for this study as part of the Comprehensive Community Health S ervices for Children and Their Families National Evaluation The sample included 9,461 children and youth between the ages of 5 and 18 years whom had complete data across relevant predictors used for this study The sample for this study included more ma les than females (i.e., 68% male), over half the sample was identified as White, and just under half the sample reported a family income less than $ 15,000 a year (i.e., 45% of sample) Trained raters, at intake of services, assessed functional impairment Raters evaluated child functioning across eight life domains: home role, school role, community role, behavior toward others, mood and emotions self harmful behavior, substance abuse, and thinking The authors also created child risk and family risk com posite s cores based on descriptions of children and families at intake (Table 2 6) Composite scores were created based on the total number of risk factors for each child and family variable The authors used latent class analysis to

PAGE 114

114 identify subgroups o f children based on similar profiles from the eight observed domain scores related to functioning Models were estimated separately for males and females. A 3 class model was identified for males and females based on statistical indices and theoretical ra tionales Latent class descriptions (i.e., subgroups or profile names) were similar across males and females Profile 1 was identified as high impairment externalizing behavior (17% of males and 19% of females) Profile 2 was identified as high impairme nt internalizing behavior (50% of males and 51% of females) Profile 3 was identified as low impairment (33% of males and 30% of females) Latent class regressions were used to examine whether child and family composites were related to profile membershi p Males with high risk composites for child and family factors were more likely to be associated with P rofile 1 or 2 over P rofile 3 Females with high risk composites for child factors were more likely to be associated with P rofile 1 or 2 over P rofile 3 ; however, statistically significant relationships for family factors were not identified In this study, the authors did not examine associations between profile membership and other outcome variables T hey did examine the stability of profile membership using latent transition analysis Based on domain scores obtained 6 months after intake, the authors reported that males maintained higher levels of stability across profiles over time (i.e., 74% and 80% remained in same profile for high impair ment profiles), while females displayed lower levels of stability (i.e., 65% and 71% remained in same profile for high impairment profiles) Summary of Empirical Studies Related to Present Study The eight studies described in this review provided empiric al research to guide the present study The ways in which these studies informed the present study is grouped

PAGE 115

115 functional ability compared to disability category, (c) use of person oriented approaches to identify subgroups based on shared profiles, and (d) examination of contextual factors Social competence Five of the eight studies examined the relationship between child characteristics and outcomes related to soc ial competence (i.e., social skills, problem behaviors, delinquency, social withdrawal, or social adjustment) Four of these studies were conducted with young children ages 3, 4, or 5 Taken together, the studies suggest that children with lower function al skills, profiles associated with lower school readiness Four of these studies also suggest that the unique make in differences rel ated to social skills and problem behaviors outcomes. Functional a bility c ompared to d isability c ategory Two studies were conducted with nationally representative data sets on children with disabilities receiving special education services under IDEA (Cham bers Perez, et al., 2004; Daley et al., 2009) In these studies, researchers examined the use of disabilities category and functional ability to examine outcomes In both studies, the functional ability was represented as a composite score guided by mea sures based on the ICF as a guiding framework Findings from Daley et al (2009) and Chamber Perez, et al (2004) demonstrated that a functional approach or a combination of functional ability and disability category might explain more variance in outcom es over traditional disabilit y categories alone The present study extended findings from these previous studies by using person oriented analytic techniques to identify subgroups of

PAGE 116

116 children based on shared profiles of functional ability and to examine t he extent to which subgroup membership explained variance in social competence outcomes over traditional disabilities categories alone Person o riented a pproaches to i dentify s ubgroups Six studies used person oriented analytic techniques to identify subgroups of children with more homogeneous set of characteristics (i.e., similar profiles) within a large heterogeneous population Two of these studies illustrated the emerging use of these methods in early childhood research in the United States (Hair et al., 2006; Konold & Pianta, 2005) In addition, the six studies provided information about the types of analytic methods and the decision processes that have been used to generate subgroups based on shared profiles, the numbers of skills/characteristic s (i.e., variables) that have been used to create profiles, and the number of sub groups or model s that have been identified in previous studies on related topics Although these studies were not conducted specifically with children with disabilities, the processes and findings were informative for the present study. Related to the types of methods used to create subgroups, five studies used a type of cluster analysis and only one study used latent class analysis (Table 2 5) Of the studies that used clust er analysis, two studies used K mean cluster analysis and three studies used a two step procedure, which combined a hierarchical cluster analysis method such as Ward s cluster analysis with K means cluster analysis The two step procedure has been favored in the literature to protect against the critiques of each approach used individually (i.e., hierarchical approach restricts movement once assigned to cluster and K means sensitive to the quality of starting values; Keller, Spieker, & Gilchrist, 2005) S everal authors, however, have noted the advantages of

PAGE 117

117 latent class analysis over traditional cluster analysis methods ( Magidson & Vermunt, 2006 ; Vermunt & Magidson 2002) Latent class analysis is a model based approach that assigns cases to classes based on estimated probabilities (McCutcheon, 1987 ) This differs from traditional cluster analysis, which relies on algorithms to identify maximum separation between groups and minimal difference within groups (Everitt, Landau, & Lesse, 2001) With the incre ase in computer processing abilities and software programs to conduct latent class analysis, the use of latent class analysis has been recommended over cluster analysis methods (Vermunt & Magidson 2002) Related to the numbers of variables used to gener ate profiles of child characteristics, researchers entered between three and eight variables into the analytic method to create profiles ( Table 2 6) In Konold and Pianta (2005) the authors used six measures across two domains of school readiness (i.e., s ocial and academic) Ha ir et al (2006) used measures from four domains related to school readiness (i.e., physical health, social/emotional development, language, and cognitive) and these authors also reported that one variable had no variance and was th erefore removed from analysis The present study used 1 5 variables to create profiles and examine relationships to social competence Related to the number of subgroups or model s identified, previous studies identified between three and eight distinct s ubgroup s (Table 2 5) As noted by many authors of the reviewed studies model s were identified based on statistical indices and theoretical decisions to select a model that was interpretable and defensible The number of variables used to create profiles did not appear to correspond to the number of subgroups identified Studies that examined profiles of similar characteristics ( e.g.,

PAGE 118

118 school readiness or temperament) did not identify the same number of subgr oups Although no studies examined person skills referred to as school readiness in previous studies are related to functional abilities, and these studies suggest ed between four and six subgroups migh t be identified in the present study Examination of c ontextual f actors As shown in Table 2 6, the extent to which studies examined contextual factors, the contextual variables used in each study, and the types of analyses conducted varied across the six studies that considered contextual factors Related to the types of contextual factors considered, five studies included child factors, five studies included family factors, one study included district factors, and one study included classroom factors Across these factors, researchers used a range of variables Three studies examined relationships based on individual variables only, while three studies examined relationships based on a composite of child or family factors How composite scores were cr eated was not explicitly described in two of these studies (Haapasalo et al., 2000; Sanson et al., 2009) Related to analys e s, two studies controlled contextual variables as part of the analysis, four studies examined relationships between subgroup member ship and contextual variables, and one study examined contextual variables as moderators. In summary, these reviewed studies provide an emerging evidence base to justify the research questions posed and analyses used in the present study They highlight further the need to examine relationships between functional ability profile subgroup membership and social competence for preschool children with disabilities, while

PAGE 119

119 examining contextual factors that might moderate sub group membership and their social competence Summary This review of the literature outlined key issues related to (a) social competence (b) concerns about using IDEA disability catego ry to characterize children (c) the ICF CY framework, (d) findings from PEELS studies, and (e) empirical research related to the present study The existing literature related to these topics suggest ed that social competence is an select child and contextual To examine further social competence of young children with disabilities, a profile of children functional abilities, related to the ICF CY framework, might provide a correlate that explain s more variance in social competence over the use of disability categor y To date, no studies have used a person oriented approach to identify subgrou ps of young children with disabilities that have similar functional ability profiles to examine associations with social competence The PEELS data set provides a unique opportunity to examine social competence for children with disabilities related to fu nctional ability profiles, disability categories, and contextual factors.

PAGE 120

120 Table 2 1 Evaluative j udgments about c s ocial c ompetence Skills u sed to a chieve s ocial g oals Skills n ot u sed or s kills i neffective for a chiev ing s ocial g oals Appropriate b ehaviors for s ocial c ontext Good s ocial s kills Few p roblem b ehaviors Poor s ocial s kills Few p roblem b ehaviors Inappropriate b ehaviors for s ocial c ontext Good s ocial s kills Many p roblem b ehaviors Poor s ocial s kills Many p roblem b ehaviors

PAGE 121

121 Table 2 2 Population e stimates a cross d isability c ategories by s tate: Minnesota, Washington, and Wisconsin Disability category Minnesota ( n = 14,361 ) Washington ( n = 14,006 ) Wisconsin ( n = 15,153 ) Developmental d elay 50 57 19 Speech or language impairments 32 27 67 Intellectual disability 1 >1 1 Autism 11 5 5 Multiple disabilities >1 >1 n/a Orthopedic impairment 1 >1 1 Other health impairments >1 3 4 Hearing impairments 2 >1 1 Visual impairments >1 >1 >1 Deaf blindness >1 n/a 0 Traumatic brain injury >1 >1 >1 Emotional disturbance 2 >1 1 Specific learning disability >1 n/a >1 Note Data retrieved from IDEA Data (2008) A vailable at https://www.ideadata.org n/a refers to category not used in state eligibility system for children ages 3 through 5.

PAGE 122

122 Table 2 3 International Classification of Functioning, Disability, and Health (ICF) c hapters Ch apter Body f unction Body s tructure Activity and p articipation Environmental f actors 1 Mental Functions Structures of the nervous system Learning and applying knowledge Products and technology 2 Sensory functions and pain Eye, ear and related features General tasks and demands Natural environment and human made changes to environment 3 Voice and speech functions Structures involved in voice and speech Communication Support and relationships 4 Functions of the cardiovascular, immunologic and respiratory systems Structures of the cardiovascular, immunologic and respiratory systems Mobility Attitudes 5 Functions of the digestive, metabolic and endocrine systems Structures related to the digestive, metabolic and endocrine systems Self care Services, systems and policies 6 Genitourinary and reproductive systems Structures related to the genitourinary and reproductive systems Domestic life 7 Neuromusculoskeletal and movement related functions Structures related to movement Interpersonal interactions and relationships 8 Functions of the skin and related structures Skin and related structures Major life areas 9 Community, social and civic life Adapted from the International C lassification of F unctioning, D isab ility and Health: Children and Youth V ersion (pp. 31 42), by the World Health Organization, 2007, Geneva, Switzerland: WHO Press.

PAGE 123

123 Table 2 4 Selected c odes of ICF c lassification s ystem: Body f unctions c omponent Chapter Level 1 c lassifica tion Level 2 c lassification Level 3 c lassification Level 4 c lassification Mental functions Global mental functions Consciousness functions State of consciousness Quality of consciousness Orientation functions Orientation to time Orientation to place Orientation to person Orientation to self Orientation to others Intellectual functions Dispositions and intra personal functions Adaptability Responsively Activity level Specific mental functions Attention functions Sustaining attention Shifting attention Memory functions Short term memory Long term memory Adapted from the International C lassification of F unctioning, D isab ility and Health: Children and Youth V ersion (pp. 46 48), by the World Health Organization, 2007, Geneva, Switzerland: WHO Press.

PAGE 124

124 Table 2 5 Studies r elated to a spects of p resent s tudy Citation Sample CWD Age at recruitment ICF Child c haracteristics Person o riented t echniques CDC Relationship to o utcomes Contextual f actors c onsidered Chambers Perez, et al., 2004 9,000 (SEEP) YES S chool age a YES Composite score of functional ability N/A YES Special education expenditure Child factors District factors Daley, Simeonsson, and Carlson, 2009 3,10 0 (PEELS) YES 3 4, 5 yrs YES Composite score of functional ability N/A YES Pre academic skills S ocial skills Adaptive skills NO Hair, Halle, Terry Humen, Lavelle, and Calkins, 2006 17,2 20 (ECLS K) NO 5 yrs NO Profiles of school readiness K means cluster analysis 4 cluster model N/A Academic ability S ocial adjustment Child factors Family factors Classroom factors Haapasalo, Tremblay, Boul erice, and Vitaro, 2000 1,034 ( Canada; boys only) NO 6 yrs a b NO Profiles of behavior dimensions QUICK cluster analysis c 8 cluster model N/A Delinquency Social withdrawal S chool placement Family factors Janson and Mathiesen, 2008 939 (Norway) NO 18 m o a b NO Profiles of child temperament Two step cluster analysis means) 5 cluster model N/A Externalizing and internalizing behaviors NO Konold and Pianta, 2005 964 (NICHD) NO 4.5 yrs NO Profiles of school readiness Two step cluster analysis means) 6 cluster model N/A A cademic ability Child factors Family factors

PAGE 125

125 Table 2 5 Continued Citation Sample CWD Age at recruitment ICF Child c haracteristics Person o riented t echniques CDC Relationship to o utcomes Contextual f actors c onsidered Sanson, Letcher, Smart, Prior, Toumbourou, and Oberklaid, 2009 2,443 (Australia) NO 4 m o a b NO Profiles of child temperament Two step cluster analysis means) 4 cluster model N/A Behavior problems Social skills Academic competence Child factors Family factors Stephen s Petras, Fabian, and Walrath, 2009 4,161 (CMHI) NO d 5 18 yrs NO Profile s of functional impairment in life domains e Latent class analysis 3 class model f N/A NO g Child factors Family factors Note CWD refers to sample of children with disabilities ICF refers to study based on International Classification of Functioning CDC refers to methods to compare descriptions of children to disability category PEELS = Pre Elementary Educ ation Longitudinal Study; SEEP = Special Education Expenditure Project; ECLS K = Early Childhood Longi tudinal Study Kindergarten; NICHD = National Institute of Health and Human Development Study of Early Child Care; a Children age s 3, 4, or 5 years included in larger sample of school age children b Longitudinal study, children recruited at early age and followed through preadolescence. c Refers to K means cluster analysis in SPSS d Children and youth receiving mental health services. e F unctioning across eight life domains: home role, school role, community role, behavior toward others, mood and emotions self harmful behavior, substance abuse, and thinking f Separate models conducted for males and females. g Latent class transition analysis used to examine stability of profile.

PAGE 126

126 Table 2 6 Predictor, c riterion, and c ontextual v ariables and r elated a naly sis i ncluded in s tudies Citation Predictor v ariables (PV) Criterion v ariables a (CV) Contextual f actors Contextual v ariables Variables or c omposite Analysis r elated to c ontextual v ariables Chambers Perez, et al., 2004 19 variables to create composite score of functional ability Measured at various ages Special education expenditure Measured concurrently Child District Child age Child gender Child ethnicity District size Cost of education index State indicators (fiscal) Variables Variables Control child and district variables in multivariate regression analysis of PV to CV Daley, Simeonsson, and Carlson, 2009 6 variables to create composite score of functional ability Measured in preschool or kindergarten Pre academic skills S ocial skills Adaptive skills Measured concurrently N/A N/A N/A N/A Hair, Halle, Terry Humen, Lavelle, and Calkins, 2006 4 variables to create p rofiles of school readiness Measured in kindergarten Academic ability S ocial adjustment Measured in first grade Child Family Classroom Child age Child race/ethnicity Child gender Child disability Child birth weight Family type Family race/ ethnicity Family education Family income Family language Class day (full/half) Class size Teacher experience Teacher training School type (public/private) Variables Variables Variables Control child, family, and classroom variables in OLS regression analysis of PV to CV Bivariate analysis to examine wheth er subgroups differed by child and family variables Logistic regression to examine whether child and family variables predicted subgroup membership

PAGE 127

127 Table 2 6 Continued. Citation Predictor v ariables (PV) Criterion v ariables a (CV) Contextual f actors Contextual v ariables Variables or c omposite Analysis r elated to c ontextual v ariables Haapasalo, Tremblay, Boul erice, and Vitaro, 2000 b 5 variables to create p rofiles of behavior dimensions Measured in kindergarten Delinquency Social withdrawal S chool placement Measured in preadolescence Family Parent occupation Parent education Parent age Family structure Composite index of family adversity Bivariate analysis to examine whether subgroups differed by family variables Logistic regression to examine whether family adversity predicted CV Logistic regression to examine whether PV and family adversity predicted CV Janson and Mathiesen, 2008 4 variables to create profiles of child temperament Measured in infancy through late childhood Externalizing and internalizing behaviors Measured concurrently N/A N/A N/A N/A Konold and Pianta, 2005 6 variables to create profiles of school readiness Measured in preschool Academic ability Measured in first grade Child Family Child gender Child race Family income to needs ratio Mother age Mother education Partner education Variables Variables Bivariate analysis to examine whether subgroups differed by child and family variables

PAGE 128

128 Table 2 6 Continued. Note a Criterion variables referred to as outcome variables in narrative. b Study conducted with boys only. c Dimensions measured by age appropriate variables at each measurement occasion. d Latent class transition analysis used to examine stability of profile over 6 month period. Citation Predictor v ariables (PV) Criterion v ariables a (CV) Contextual f actors Contextual v ariables Variables or c omposite Analysis r elated to c ontextual v ariables Sanson, Letcher, Smart, Prior, Toumbourou, and Oberklaid, 2009 3 dimensions c to create profiles of child temp erament Measured in infancy and early childhood Behavior problems Social skills Academic competence Measured late childhood and preadolescence Child Family Child gender Parent education Parent occupation Variable Composite of Socio economic status (SES) Bivariate analysis to examine whether subgroups differed by gender and SES Child gender and SES composite entered as moderators in MANOVA Stephen s Petras, Fabian, and Walrath, 2009 8 variables to create profiles of functional impairment in life domains N/A d Child Family Child Family Child gender Child age Child race/ethnicity Family income History of physical/ sexual abuse Substance abuse Run away history Suicide attempt Sexual assault on others Caregiver felony Caregiver substance abuse Caregiver psychiatric hospitalization Family violence Caregiver mental health Sibling in foster care Variables Variable Composite of child risk factors Composite of family risk factors Separate analysis conducted for boys and girls Latent class regression to examine whether child and family variables and composites predicted subgroup membership

PAGE 129

129 Figure 2 1 International Classifica tion of Impairments, Disability, and Handicaps (ICIDH) framework.

PAGE 130

130 Figure 2 2 International Classification of Functioning, Disability, and Health framework (WHO, 2007) Reprinted with permission from the International Classification of Functioning, Disability and Health: Children and Youth Version (pp. 17), by the World Health Organization, 2007, Geneva, Switzerland: WHO Press.

PAGE 131

131 CHAPTER 3 M ETHOD OLOGY In the present study, secondary analyses were conducted using the Pre Elementary Education Longitudin al Study (PEELS) data set, a large scale nationally representative data set involving young children with disabilities A correlational study design was used to explore and examine relationships among factors associated with ence Variables from the PEELS data set were used to identify subgroups of young children with disabilities with similar functional ability profiles and to examine relationships between subgroup membership and social competence Findings from the body of research reviewed in Chapter 2 suggest ed that for children without disabilities The present study focused on young children with disabilities and one purpose of th e study was to examine associations between profile subgroup membership as well as their disability category and their social competence lity profile subgroup member ship or disability category, findings from previous research indicated that additional non malleable child factors, and contextual factors are associated with In the present study, the moderating influences of non malleable c hild factors as well as contextual factors ability profile subgroup membership in relation to their social competence were examined This chapter presents the research questions, research design, and hypothesized relationships am ong variables that were examined in this study Information about the

PAGE 132

132 PEELS study and data set, the variables examined, the methodological procedures, and analyses employed are described Research Questions The following research questions guided the seco ndary analyses conducted in the present study: 1. What distinct and interpretable functional ability profile subgroups emerge when using person oriented analytic techniques to examine functional ability variables contained in the PEELS data set for young chil dren with disabilities? 2. What is the strength of the relationship between functional ability profile subgroup membership and social competence? 3. What are the individual and combined contributions of functional ability profile subgroup membership and disabili ty category membership to the explanation of social competence? 4. To what extent do non malleable child factors and contextual factors moderate the relationship between functional ability profile subgroup membership and social competence? Each research question was identified to contribute to the growing body of aspects of the ICF CY framework: child functioning, disability/health condition, and contextual factors Resea rch Design The present study is a correlational study The study can be classified as a cross sectional, non experimental research design with research questions focused on exploration and explanation of relationships between and among variables (Johnson, 2000) Correlational research has been identified a s an appropriate research method to explore associations between malleable child factors and educational outcomes and explore factors or conditions that mediate or moderate these relationships (IES, 2011 )

PAGE 133

133 Correlational research contributes to the evidence base in early education and early childhood special education by examining the magnitude of associations; generating hypotheses related to underlying processes; informing theories of change; and identi fying variables to inform the design of interventions and experimental research (IES, 2011; Snyder & Kaiser, 2008; Thompson, Diamond, McWilliam, Snyder, & Snyder, 2005) A c orrelational research design was appropriate to examine the guiding research quest ions posed In the present study, the PEELS data set was used to explore whether subgroups of children with similar functional ability profiles could be identified ( Research Question 1) In addition, the study was designed to explicate relationships betwe en young disability category membership, and contextual factors that have been associated with R esearch Q uestions 2 through 4) V ariables of interest selected from the PEELS data set were examined at one time point during preschool or kindergarten (i.e., wave 1 data for children 3, 4 or 5 years of age ), thus the study is considered cross sectional Associations between and among variables are based on aged peers included in the PEELS data set Hypothesized Relationships The ICF framework suggests complex, transactional relationships exist among child functioning, disability/health conditions, and context ual factors The figures below show the hypothesized relationships between or among key variables for each research question It was important to specify these relationships a priori It would not have been feasible to test all possible relationships no r would it have been scientifically scale data set looking for statistically significant or

PAGE 134

134 noteworthy relationships (Pedhauzer, 1982) The models, therefore, represent the hypothesized relationships that were empir ically examined in this study based on a review of the relevant extant literature In the figures, squares represent observed variables (i.e., manifest variables) in the data set, circles represent latent variables created from observed variables, and arr ows represent the direction of the relationships between variables. Figure 3 1 illustrates the model examined to investigate the first and second research question s The left side of the model displays the observed variables in the PEELS data set used to identify subgroups with similar functional ability profiles The first research question addressed whether subgroups with similar functional ability profiles (i.e., categorical latent classes) would emerge from the observed data To identify subgroups, e xploratory analyses were used to examine the patterns of similarities and differences among children on observed functional ability variables Using these types of analyses, all children were associated with a subgroup that represented a group of children who share d similar patterns of characteristics but had different patterns of characteristics from other subgroups Given the exploratory nature of these analyses, both statistical indices and theoretical rationales were used to determine the appropriate class model (i.e., determine the number of subgroups that were defined for this study) The right side of the model displays the hypothesized relationship between subgroup membership, once identified, and social competence ( Research Question 2) The foc us of this investigation was to examine the strength of the relationship between subgroups (based

PAGE 135

135 Figure 3 2 illustrates the model used to investigate the individual and combined co R esearch Q uestion 3 ) To investigate the individual contribution of disability cat egory, social competence status was examined The individual contribution of functional ability profile subgroup membership was examined as part of Research Question 2 To investigate the combined contribution of each variable, the strength of the relationships among group memberships (i.e., functional ability profile and disability category) and Functional ability su bgroup profile membership and disability category membership were hypothesized to have an association The purpose of these analyses competence status was gained by adding ch ability profile subgroup membership to disability category in the model. Figure 3 3 illustrates the model examined to investigate the fourth research question The hypothesized relationship shown in the figure focuses on the extent to membership and their social competence In the present study, moderators were select variables and included non malleable child factors and contextual factors (i.e., family fac tors and environmental factors ) The scope and sequence of the research questions presented in this study focused on systematically examining relationships between functional ability profile

PAGE 136

136 subgroup membership and social competence for children with disabilities while considering the influence of disability category and contextual factors that were hypothesized to be related to functional ability profiles and to social competence The research questions and variables selected from the PEELS data set to address the study research questions were identified based on key aspects of the ICF CY framework and the literature review PEELS Study and Data Set The PEELS study was a longitudinal, prospective investigation to examine characteristics of children receiving early childhood special education, the programs and services they receive, their transitions from preschool to school age settings, and how children with disabilities function and learn in preschool and school age settings The PEELS data set pr ovides a nationally representative sample of 3,10 0 young children with disabilities Children in the PEELS sample were 3, 4, or 5 years of age and had an active individualized education program (IEP) or individualized family service plan (IFSP) at the tim e they were recruited into the study The sample is disproportionately male (70%; Markowitz et al., 2006 ) Children included in the sample are from diverse racial/ethnic and socioeconomic backgrounds At the time of enrollment, children attended early c hildhood education programs or kindergarten and transitioned into elementary school during the course of the longitudinal study Information in the data characteristics, educational services, and academic achievement as measured by direct child assessments or gathered from teacher and administrator questionnaires family members, family context, and local community and spectives of their child and educational services were o btained from parent interview

PAGE 137

137 Data for the PEELS study were collected in four waves from the 2003 2004 through 2006 2007 school years, and follow up data were collected during the 2009 2010 school year (Markowitz et al., 2006) Data for the present study were from the first wave of data collection so analyses are cross sectional in nature The PEELS data set was selected for use in the present study because it offers descriptive information on the characteristics, functioning, and performance of y oung children with disabilities during early childhood Moreover, the data set includes information related to family, school, and community environments This data set is unique from other large d uring early childhood because only young children with disabilities were included in the study PEELS Sampling Strategy The PEELS data set is a stratified sample of young children with disabilities receiving early childhood special education services at the time of enrollment PEELS data can be weighted to represent national estimates that can be generalized to the entire U.S population of children with disabilities ages 3 through 5 years of age The present study employed weighting for each analysis Weight files were applied as appropriate based on the variables of interest in the analysis. The sample consists of three age cohorts: Cohorts A, B, and C Children in cohort A were 3 years old (date of birth 3/1/00 through 2/28/01); children in cohort B were 4 years old (date of birth 3/1/99 through 2/28/00); and children in cohort C were 5 years old (date of birth 3/1/98 through 2/28/99) when recruited into the PEELS study Table 3 1 shows

PAGE 138

138 The prese nt study used wave 1 data to investigate the research questions with a sample of young children with disabilities who were 3, 4, and 5 years of age During wave 1, all children were enrolled in early childhood, preschool, or kindergarten settings The P EELS study used a two stage sampling design In the first stage, a nationally representative sample of local education agencies (LEA) was selected The LEA sample was stratified by geographic region, preschool special education enrollment size, and distr ict poverty level A total of 22 0 local education agencies (LEA) participated in the study The participating LEAs were recruited through three processes: (a) the main sample recruited in 2001, (b) a non response sample recruited in 2003 from non respond ers in 2001, and (c) a supplemental sample recruited in 2004 to address the representation of a key state originally banned from participation by state regulations (Carlson et al., 2008) In the second stage, a sample of preschoolers with disabilities, a ges 3, 4, or 5 years of age were selected from the participating LEAs Children were stratified by age cohorts Children were selected by age; therefore, children may have participated in preschool programs, kindergarten programs, or other early childho od special education programs at the time of recruitment Children were eligible for recruitment if they were age eligible (i.e., 3, 4, or 5 years of age ) and had an IEP or IFSP with the district PEELS had three additional eligibility criteria: (a) an E nglish or Spanish speaking adult or an adult who used signed communication who could respond to the household interview, (b) the child was the first in the family to be sampled for PEELS, and (c) the enrolment in the PEELS study

PAGE 139

139 (Markowitz et al., 2006) Children were sampled from district lists of eligible children recruitment packet Families that agreed to partici pate in the study completed and returned signed consent and enrollment forms A total of 3,10 0 children participated in the study. PEELS Instrumentation Data for PEELS were collected by parent interview, teacher questionnaire, direct child assessments, administrator questionnaire, local education agencies ( LEA ) questionnaire, and state education agencies (SEA) questionnaire in wave 1 In subsequent waves of data collection data were collected by parent interview, teacher questionnaire, and direct child assessments In addition, the PEELS project staff developed a demographic data file that contains information about child participants primary disability category (Carl son & Lowe, 2009) The present study used data from the parent interview, the teacher questionnaire the administrator questionnaire and the demographic file Parent or guardian interviews were conducted by phone using computer assisted telephone interview (CATI) technology Interviews were approximately 1 hour in length ; interviews used a protocol with identified skip patterns to ensure parents were not asked to respond to questions that were not applicable (e.g ., parent indicates a child has typical vision and does not wear glasses, remaining questions about vision acuity and service history related to vision were skipped; Markowitz et al., 2006 ) Parents were asked questions about their ity, behavior, special education services related services, transitions between school settings, and out of school

PAGE 140

140 activities In addition, parents were asked about their family, and their local community environment s including fam ily and community resources and family background. Three versions of the teacher questionnaire were used in PEELS : the Early Childhood Teacher Questionnaire (used in wave s 1, 2, 3); the Kindergarten Teacher Questionnaire (used in wave s 1, 2, 3, 4); and the Elementary Teacher Questionnaire (used in wave s 2, 3, 4 ; Carlson & Lowe, 2009 ) All teacher questionnaires were mailed Teachers reported information about classroom environment room interaction with peers, their philosophies of early childhood education, and their transition practices for children entering or leaving their progr am The teacher questionnaires also included selected teacher rating scales related to gross motor skills adaptive behavior, social skills, and problem behaviors An programs and related services Either the classroom teacher or the spec ial education service provider c ompleted questionnaire items There were two versions of the administrator questionnaire : the Elementary School Principal Questionnaire or Early Childhood Program Director Questionnaire (Carlson & Lowe, 2009) The appropriate questionnaire was sent to p rincipals or Only one administrator questionnaire was sent to each school or program, regardless of the number of PEELS participating children s school or program including staff, programs, and resources, and community characteristics including parent involvement The administrator questionnaire was sent during the first

PAGE 141

14 1 wave of data collection; a new questionnaire was sent if a child moved to a new school or center without a previous administrator questionnaire completed. PEELS Response Rates and Imputation for Missing Data The PEELS data set was affected by the response rates of parents, teachers, and administrators Table 3 2 displays the res ponse rates for the parent interviews and teacher questionnaires across waves Parent interview response rates were at or above 93% for the first two waves of data collection; these rates declined to at or above 80% response rates for the following two wa ve of data collection Teacher response rates were generally comparable across the four waves of data collection between 79% and 84% The response rate for the administrator questionnaire was 72% During data preparation for the restricted use data set missing data were imputed for selected items on child assessments, administrator questionnaire teacher questionnaires, and parent interviews Different methods of imputation were used depending on the nature of the data; methods included hot deck imputat ion, regression, external data source, and deterministic or derivation method, based on the internal consistency principle of interrelated variables ( Markowitz et al., 2006 ) More information on the data imputation and data imputation methods is available in t Manual ( Carlson, Posner, & Lee, 2008) O n average less than 10 % of data were imputed (Carlson & L owe 2009) Imputed data were recorded in the data set with imputation flags PEELS Variables Selected for Analysis in the Present Study Multiple sources informed the selection of variables for this study First, the social competence definition described in Chapter 1 and Chapter 2 influenced the selection of the criterion variables to measure social competence Second, the Daley, S imeonsson,

PAGE 142

142 and Carlson (2009) study described in Chapter 2 was used to identify variables from the abilities; these variables were used to identify subgroups of childr en with similar functional ability profiles and functional ability profile membership was used as an explanatory variable to examine relationships with social competence Third, disability categories, used as explanatory variables, were constructed to be consistent with IDEA based disability categories Fourth, findings from the literature review described in Chapter 2 influenced the selection of non malleable child factors and contextual factors (described in the section entitled descriptive variables) which were used to examine relationships among these factors, functional ability profile membership, and social competence. Criterion Variables Social competence was measured by two domains of the Preschool and Kindergarten Behavior Scale Second Edition ( PKBS 2; Merrell, 2002 ) and these domains (social skills and problem behavior) were used as the criterion variable s in this study To assess c the PKBS 2 was collected by teacher report in wave 1 and wave 2 Only wave 1 data we re used in the present study The PKBS 2 is a judgment based summated rating scale designed to evaluate social skills and problem behaviors in preschool and kindergarten children, age 3 through 6 years Home based or school based raters can complete th e rating scale and separate normative information is provided for home and school raters PEELS used the school rater form of the measure The PKBS 2 includes 76 items across the social skills and problem behaviors domain scales Each domain scale was examined separately in this study The Social Skills scale includes 34 items ( v =34) and the

PAGE 143

143 Problem Behavior s scale includes 42 items ( v =42) The Social Skills scale consists of three subscales: Cooperation, Interaction, and Independence The Problem Be havior s scale consists of two subscales: Internalizing Behaviors and Externalizing Behaviors Subscales were derived empirically through exploratory factor analysis (EFA), and verified through confirmatory factor analysis (CFA) procedures (Merrell, 2002). Directions to the teachers on the PEELS questionnaire indicate d rating of items should be based on observations of the child during the past 3 months Scoring is based on a 0 3 scale: 0 = never 1 = rarely 2 = sometimes and 3 = often Subscale raw scores are converted to standards scores Standard scores are summed and converted to composite standard score or percentile rank for each domain scale One score conversion table is used for children ages 3 t hrough 6 years The test developer repor ted that score conversion tables for each age were not warranted due to the small difference in means for each age Standard scores are based on a normal distribution with a mean of 100 and a standard deviation of 15 Interp re tation of scores is differen t for each scale High scores on social skills scale are desirable and associated with desired behaviors Low scores on the problem behaviors scale are desirable and associated with the absence of behavior labeled as challenging or problematic The PKBS 2 was standardized based on a sample of 3,313 children ages 3, 4, 5, or 6 The sample was constructed to approximate the general U.S population based on 2000 C ensus data with respect to geographic region, gender, race/ethnicity, special education st atus, and socioeconomic status (Merrell, 2002) The test develo per reported Cronbach alpha internal consistency score reliability coefficients between .88 and .97 for all subscales, a .97 for problem behaviors total score, and a .96 for social

PAGE 144

144 skills to tal score for the school rater form Additionally, test retest score reliabilit y coefficients between .58 and .87 for all subscales on 3 week retest and between .69 and .78 for all subscales on 3 month retest were reported Factors loadings for items on each subscale identified through EFA procedures were between .45 and .82 for social skills subscales and between .49 and .80 for problem behavior subscales. PKBS 2 scores have moderate to strong correlations with other measures of young r. For example the PKBS 2 scores had a .76 and .83 correlation with score s on the Social Skills Rating Scale (Gresham & Elliot, 1990) for social skills and problem behaviors scales, respectively. In the PEELS data set, the domain scores (social skills and problem behaviors ) and sub domain scores (c ooperation, i nteraction, i ndependence i nternalizing b ehaviors and e xternalizing b ehaviors ) were available as standard scores Raw scores or i tem level data were not available for users with the restricted use l icense. Explanatory Variables ICF r elated f unctional p rofiles ICF related functional ability profiles represent ed a latent class categorical variable (i.e., subgroup membership) created from 15 variables in the PEELS data set that describe d The 15 variables used in the present study were derived from observed variables in the PEELS data set that describe d functional ability and were based on the 15 item PEELS Disability Severity Index described by Dal ey, Simeonsson, and Carlson (2009) The PEELS Disability Severity Index was adapted from the ABILITIES Index (Simeonsson & Bailey, 1991) Daley and level of functio ning or severity of disability across 15 domains related to the ABILITIES

PAGE 145

145 Index This section presents information about the original ABILITIES Index, the PEELS Disability Severity Index and the 15 variables used in the present study. The ABILITIES Index developed by Simeonsson and Bailey (1991), is used to assess a a udition (i.e., hearing left and right ear) b ehavior and social skills, i ntellectual function, l imbs (i.e., right and left hand, arm, and leg) i nten tional communication (i.e., understanding and communication with others) t onicity (i.e., tightness and looseness) i ntegrity of physical health, e yes (i.e., vision right and left eye) and s tructural status The ABILITIES Index is a judgment based rating system The rating scale consists of 19 items related to the nine domains Each item is scored on a 6 point scale ranging from (1) normal ability to (6) profound lack of ability Parents, caregivers, or practitioners who are familiar with the child can complete the ABILITIES Index Ratings should be based on (a) knowledge about the assessment findings, or (c) other available sources of information such as documentation or records The ABILITIES Index can be used to create a profile of child functioning across the nine domains or compute a composite score The composite Bailey, Simeonsson, Buysse, and Smith (1993) examined interrater agreement and test retest score reliability for the ABILITIES Index The authors compared ratings for 130 parent teacher pairs, 130 parent specialist pairs, and 130 teacher specialist pairs for a sample of youn g children with disabilities Exact agreement across the 19 items for parent teacher pairs was 68.5% (range = 39 85); for parent specialist pairs was 65.5% (range = 32 84); and for teacher specialist pairs was 67.8% (range = 42 91)

PAGE 146

146 Interrater exact agre ement was lowest for items related to social skills, behavior, and communication skills The authors reported interrater agreement within one point across the 19 items: for parent teacher pairs was 85.1% (range = 76 95); for parent specialist pairs was 84 .9% (range = 66 97); and for teacher specialist pairs was 88.5% (range = 78 98) To examine test retest score reliability, 40 teachers completed a second ABILITIES Index for young children with disabilities 34 days after the first rating Exact agreement between ratings was 67.8% while agreement within one point was 90.6% The mean intraclass correlation coefficient was .70, and mean weighted kappa was .77 As described in Chapter 2, Simeonsson, Bailey, Smith, and Buysse, (1995) conducted a hierarchical cluster analysis that yielded a 6 cluster model of subgroups sharing similar ability profiles Consumer validation of the ABILITIES Index conducted by Buysse, Smith, Bailey, and Simeonsson (1993) showed that a range of early childhood stakeholders (e.g., parents, other family members, early interventionists, preschool teachers, therapists) indicated the index w as a useful assessment tool that was feasible to use and the index produced socially valid scores functioning related to the nine dom ains. The PEELS Disability Severity Index described by Daley, Simeonsson, and Carlson (2009) is based on the original ABILITIES Index (Simeonsson & Bailey, 1991) For the severity index, however, the authors used items from the PEELS parent interview to c reate scores for variables related to seven of the original nine domains (i.e., tonicity and structural status were not included ) Daley and colleagues reported the ABILITIES Index has been expanded since its original publication with an additional area c a

PAGE 147

147 activity level, regulation of feeling/emotions, academic skills, motivation and impulse control From these expanded domains, four domains were selected to be included on the PEELS Disability Severity Index: regulation of attention, regulation of activity level, regulation of feeling/emotions, and motivation The PEELS Disability Severity Index has 15 items related to eleven domains selected from the original ABILITIES Index and expanded ABILITIES Index The 15 items are hearing v ision overall health, use of hands, use of arms, u se of legs cognition, communicating with others understanding regula tion of attention, r egul ation of feelings and emotions, regulation of activity level, motivation, social skills, and inappropriate or unusual behavior To generate items for these areas, PEELS researchers used 24 questions from the parent interviews In some cases, an item was represented by a single question, ( e.g item use of arms For other items multiple questions w ere asked For example, without glass es were used to create the item for vision (Daley et al., 2009 ) Daley and colleagues (2009) reported their procedural decisions to select or combin e parent interview variables to derive final items for PEELS Disability Severity Index was based on theoretical information about the skills and domains Questions from the parent interview and corresponding item coding used to derive final items for the PEELS Disability S everity Index are presented in Appendix B Overall, nine items were created from a single question; of these, seven required recoding for use as part of the Disability Severity Index In addition, six items were derived from multiple questions As shown

PAGE 148

148 in Appendix B, final items are based on a 4 point scale : (1) normal or typical functioning, (2) mild limitation in functioning, (3) moderate limitation in functioning, and ( 4 ) severe limitation in functioning. Daley and colleagues (2009) reported they analyzed the extent to which the PEELS Disability Severity Index items mapped to codes of the ICF CY Table 3 3 shows the ICF CY related codes identified by Daley and colleagues The authors reported there were correspondences between index items with se ven Body Function codes and six Activities/Participation codes; however, no clear ICF CY codes were identified for inappropriate or unusual behavior and overall health Daley and colleagues examined the relationship between the each of the five possible PEELS Disability Severity Index composite score s (described in Chapter 2) and other variables identified in the PEELS data set by examining P earson product assessment scores Fo r the 15 item index, t he authors reported Pearson product moment correlation coefficients on the following measures: Peabody Picture Vocabulary Test (PPVT; r = .32); Woodcock Johnson Letter Word Identification subtest ( r = .22); Woodcock Johnson Applied Problems subtest ( r = .40); conceptual domain of the ABAS ( r = .53); practical domain of the ABAS ( r = .53); social domain of the ABAS ( r = .40); PKBS 2 social skills composite ( r = .47); and PKBS 2 problem behaviors composite ( r = .35) For the 6 item index, which was the selected index in the Daley et al. study, the authors reported the following Pearson product moment correlation coefficients: Peabody Picture Vocabulary Test (PPVT; r = .3 6 ); Woodcock Johnson Letter Word

PAGE 149

149 Identification subtest ( r = .2 6 ); Woodcock Johnson Applied Problems subtest ( r = .4 5 ); conceptual domain of the ABAS ( r = 46 ); practical domain of the ABAS ( r = 43 ); social domain of the ABAS ( r = 35 ); PKBS 2 social skills composite ( r = 43 ); and PKBS 2 problem behaviors composite ( r = .35). In the present study, the 15 items from the PEELS Disability Severity Index created by Daley and colleagues were used to identify subgroups of children with similar functional ability profiles The items used to create th e profiles are aligned with select ICF As noted previously, the subgroups were created with exploratory analyses (i.e., latent class analysis) to examine the patterns of similarities and differences among chi ldren on these 15 variables Children identified with the same profile (i.e., belong to the same subgroup) share similar characteristics of strengths and needs across the 15 functional ability variables The number of subgroups that were defined for this study was based on both statistical indices and theoretical rationales considered during the exploratory analyses. Disability c ategory s primary disability category is available in the demographic file provided with the restricted use PEELS data set This variable was derived from multiple sources The PEELS team used the disa bility reported on the teacher and related service questionnaire parent interview, and enrollment forms submitted by district personnel to determine a primary disability category for each wave (Carlson & Lowe, 2009) The PEELS demographic dat a file includes 15 possible disabilit y categories These categories are consistent with IDEA disab ility categories (i.e., 14 categories); the additional category in the PEELS data set is due to the inclusion of mild mental retardation and moderate mental retardation (2 categories) to represent the IDEA

PAGE 150

150 category of mental retardation F or the present s tudy mild mental retardation and moderate mental retardation were grouped into one disability category (i.e., mental retardation) Sample size in some disability categories were too small for analysis, therefore, a low incidence disability category was c reated for the present study Low incidence include d hearing impairment, deaf/blind, deafness, multiple disabilities, orthopedic impairment, other health impairments, traumatic brain injury, and visual impairment Although sample size was small for emoti onal behavioral disturbance, this category was not included in the low incidence disability category, because of possible strong associations between As a result, seven disability categories were used in this study The categories in the present study were speech or language impairments developmental disability, autism, emotional behavioral disturbance, mental retardation, learning disability, and low incidence disability. Descriptive Variables N on malleable child and c ontextual (i.e., family and environmental) factors were used to explore relationships in two ways in the present study First, variables were used to describe children within and across functional ability profile subgroups Second competence w as examined with variables associated with non malleable child factors as well as with family factors and environmental factors i ncluded as moderators in the model Not all descriptive variables were included as moderators. Child, family, and environmental variables used in the present study are briefly described in this section D etailed information about PEELS questions used to derive these variables and how these variables were used in the present study are shown in Appendix D

PAGE 151

151 Child f actors Child gender was identified as male or female in the PEELS demographic file and was used in the present study Child a ge was identified as age in months at th e time of assessment in the PEELS demographic file and was used in the present study. Race/ e thnicity was derived from two questions on the parent interview The first question asked the respondent to indicate if the child was of Hispanic, Latino, or oth er Spanish speaking origin The second question asked the respondent to indicate the Five possible categories were listed : White, African American or Black, American Indian or Alaska Native, Asian, or Native Hawaiian or other Pacific Islander English as a second language was derived from one question in the parent interview. Respondents were asked to indicate if the child spoke any language other than English at home IFSP before age 3 was derived from o ne question in the parent interview. Respondents were asked to identify if the child had an IFSP before the age of 3 years old. Birth weight was derived from one question in the parent interview. Respondents in ounces Ounces were converted to grams for the present study. Weeks premature was derived from one question in the parent interview. Respondents were asked to indicate the number of weeks premature the child was born I f children were not born prema ture the inapplicable code was applied.

PAGE 152

152 Family f actors Family factors were identified in one of two ways: family characteristics or parent child interactions Family characteristic variables were home living environment, marital status, family income, and parent education Parent child in teraction variables were parent school activities, child activities, regular child activities, parent child activities child participation in activities regularly, family meals per week and the extent families read t o the child Respondent role was derived from two questions in the parent interview. First, respondents were asked to identify if they were biological, adoptive, step parents, foster parents or guardians. Second, respondent s were asked what type of relationship they had to the child (e.g., mother, father, grandmother, sibling, non relative). Home living e nvironment was derived from one question in the parent interview R espondents were asked to identify if the child lived at home or in another setting (e.g., hospital, care facility) Marital status was derived from one question in the parent interview Respondents were asked indicate if there were married, never married, widowed, separated, or divorced Family i ncome was derived from one question in the parent interview Respondents were asked to indicate the total income for all persons in the household. Parental e ducation was derived from one question in the parent interview Respondents were asked to indicate their le vel of education Parent school activities w ere derived from multiple questions in the parent interview Respondents were asked to indicate whether they had participated in

PAGE 153

153 Seven possible activities were described Respondents answered yes or no to each possible activity Child activities w ere derived from multiple questions in the parent interview. Respondents were asked to indicate whether their child had ever participated in after school or communit y activities. Seven possible activities were described. Respondents answered yes or no to each possible activity. Regular child activities w ere derived from multiple questions in the parent interview Respondents were asked to indicate whether their c hild participated in after school or community activities on a monthly basis Seven possible activities were described Respondents answered yes or no to each possible activity Parent child activiti es w ere derived from multiple questions in the parent interview Respondents were asked to indicate whether they participated in community based activities with their child on a monthly basis Seven possible activities were described Respondents answered yes or no to each possible activity C hild parti cipation in activities regularly was derived from one question in the parent interview. Respondents were asked to indicate whether their child participated in any group activities on a monthly basis Respondents answered yes or no to this question. Meals per week were derived from one question in the parent interview. Respondents were asked to indicate how many times a week the family had a meal together

PAGE 154

154 Extent famil y read s to the child was derived from one question in the parent interview. Respondents were asked to i ndicate how often someone in the family reads to the child. Environmental f actors Environmental factors included community and school related factors. N eighborhood safet y was derived from one question in the parent interview Respondents were asked to indicate the extent to which they felt their neighborhood was safe for children to play outside during the day. School/neighborhood income was derived from one question in the administrator questionnaire Respondents were asked to indicate what percentage of their students or children lived in low income households School/program quality was derived from two questions in the administrator questionnaire For school princ ipals, respondents were asked to indicate if the school was designated low performing or in need of improvement under the No Child Left Behind Act For early childhood programs, respondents were asked to indicate if the program was licensed or accredited Parent satisfaction with special education services was derived from one question in the parent questionnaire. Respondents were asked to rate their level of satisfaction Number of children with IEP s in cla ss was derived from one question in the teacher questionnaire. Respondents were asked to indicate the number of children with

PAGE 155

155 Number of children without IEP s in class was derived from one question in the teacher questionnaire. Respondents were asked to indicate the number of children Classroom intervention to support social interaction was derived from one question in the teacher questionnaire Respondents were asked to indicate whether their program facilitated interactions between children with disabilities and children without disabilities was derived from multiple questions in the teacher questionnaire. Respondents were asked to indicate the top prioriti goals Eight possible priority areas were described Respondents selected the top three priority areas. Procedures This section describes the methodological procedures used in the present study to conduct the secondary analyses of th e PEELS data set. Data File Preparation PEELS data files are available in cross sectional formats for each data source (i.e., child assessments, administrator questionnaire, parent interview). For teacher questionnaires, data from questionnaires answered by different teachers (e.g., early child hood or kindergarten teacher) are presented in separate files. Data files for the present study were prepared by selecting variables of interest from each individual ide ntification number in the study. The unique ID was used to merge reduced files from each data source.

PAGE 156

156 To derive variables for the present study that were created from multiple variables in the PEELS data files, variable re coding and variable merging sy ntax were used to transform or create final variables. A final data file containing all variables used in the present study was exported into a file. Final data files were used in statistical software programs using program specific syntax to conduc t analyses. All data files used in the present study were kept on a secure, password protected computer authorized for use under the restricted use licensing agreement. Variable coding and re coding syntax and analytic program syntax are presented in App endix D. Primary Sampling Unit, Stratification, and Sampling Weights The PEELS data set is based on complex sampling techniques. First, the sample was identified from a two stage sampling process. The primary sampling units were school districts, which w ere selected from lists of eligible districts. Following district selection, children were selected from lists of eligible children within districts. Throughout the recruitment process, historical (i.e., child eligible at the beginning of recruitment) an d on going (i.e., child became eligible during recruitment) lists of children were used. Second, the PEELS sample was stratified. Districts were stratified by geographic region, preschool special education enrollment size, and district poverty level. Ch ildren were stratified by age cohort. Third, the PEELS data set can be weighted to represent a nationally representative sample of children. Analyses reported in the present study were conducted by accounting for the stratification of the sample, cluste r sampling (i.e., sampling districts prior to sampling children), and the sampling weights. The Taylor Series linearization method accounts for the effect of cluster sampling and was used to calculate standard errors (Rust, 1985) This was achieved by us ing appropriate analytic procedures in 9.2 or

PAGE 157

157 Mplus 6.1 (e.g., SurveyReg). Taylor weight files in the PEELS data set include parent child weights and parent child teacher weights. The parent child weights are appropriate for analyses conducted with data from the parent interview and the child assessments (Carlson, Posner, & Lee, 2008) and include 2 870 available cases. The parent child teacher weights are appropriate for analyses conducted with data from the parent interview, the child assessments, and the teacher questionnaire (Carlson, Posner, & Lee, 2008) and include 2 180 available cases. Weight files for the present s tudy were selected to maximize the number of available cases for each analysis using appropriate weight files. Analyses for R esearch Q uestion 1 were conducted with the cross sectional wave 1 parent child weights Analyses for R esearch Q uestion 2 and 3 we re conducted with the cross sectional wave 1 parent child teacher weights Based on advice from PEELS staff the parent child teacher weights were also used for analyses conducted with wave 1 parent interview, wave 1 teacher questionnaire, and wave 1 admi nistrator questionnaire (i.e., R esearch Q uestion 4 ). Missing Data Large scale, longitudinal, prospective data sets have missing data. When data were missing, a missing data indicator was used in the data set so the analysis appropriately takes into accoun t observed and missing data. To address missing data for the analyses conducted as part of the present study, models were analyzed using 9.2 or Mplus 6.1 (Muthen & Muthen, 2007) with maximum likelihood estimation. and Mplus with maximum likelih ood estimation analyses that were conducted in th e present study use d all available data to generate estimates.

PAGE 158

158 Analys e s Research Q uestion 1 To examine whether distinct subgroups based on similar functional ability profiles emerged in the PEELS data set (R esearch Q uestion 1 ) latent class analysis (LCA) in Mplus 6.1 was used. LCA uses observed predictor variable scores (i.e., manifest variables) to create subgroups with similar patterns of scores. These subgroups are referred to as latent classes. In t functional abilities (i.e., observed variable scores) were used to create subgroups with similar functional ability profiles (i.e., the latent classes). Appendix D shows Mplus syntax. All models we re estimated with 500 random starts and the top 40 log likelihood values were examined to investigate whether local m axima were avoided in the estimation procedure. The following model fit indices were used to inform selection of the optimal latent class model : log likelihood (LL) value and r eplication of LL Bayesian information criterion (BIC), and e ntropy. Parameter estimates are found that maximize the LL. Demonstrating that the same LL is obtained for different starting values (replication of LL) pr ovides evidence that the estimates are for a global maximum BIC is a recommended measure of fit for latent class models and lower BIC numbers are desirable ( Nylund, Asparouhov, & Muthen, 2007 ). Entropy refers to the classification uncertainty (Vermunt, & Magidson, 2002) In Mplus, relative entropy is reported and values near one indicate a high certainty in classification To report information about the functional ability profile subgroups, basic descriptive analyses were conducted in 9.2 using m ost likely class membership.

PAGE 159

159 Research Q uestion 2 To examine the relationships between functional ability profile subgroup membership and social competence, two possible analytic methods were considered. The first method was to use the mixture modeling feature of Mplus 6.1 to model the relationships between latent class membership (i.e., subgroup with similar functional ability profile) and PKBS 2 standard scores. This method uses posterior probability based imputation in testing equality of mean PKBS 2 scores for the latent classes. The second method was to use Mplus 6.1 to export the assigned class membership for each case after running the latent class analysis. This method created a data file in which each child was assigned to a most likely class membership based on the posterior probabilities. The assigned latent class membership becomes a categorical variable represented by a number. I n the present study these categorical variables represent subgroups of children with similar functional abilit y profiles and were referred to by profile number (i.e., Profile 1, 2, 3, 4, or 5). The file was then merged with a data file that included the explanatory variables, descriptive variables, and the criterion variables to estimate a regression model to exa mine the relationship s between functional ability profile subgroup membership and and problem behaviors standard scores using ProcSurveyReg in 9.2. The concern with using the first method was related to the sampling weights. The recommended PEELS sampling weights for an analysis using functional ability items and PKBS 2 scores are different than the recommended PEELS sampling weights for an analysis using functional ability items alone. Therefore, to use posterior probabili ty based imputation, it would have been necessary to use different sampling weights in the analysis to investigate whether latent class membership was associated with PKBS 2

PAGE 160

160 scores than the weights used in the analysis to determine the latent classes. Thi s variation in weights could have resulted in different latent classes in the two analyses. To avoid this possibility, the second method was selected. This method avoided the problem with the sampling weights and was appropriate given that mean posterior probabilities for class membership were all above .86, with a range of .86 to .92 over the five latent classes. Regression analyses were completed using assigned functional ability profile subgroup membership (i.e., latent class) as a categorical variable in ProcSurveyReg in 9.2 and social skills and problem behavior s standard scores as criterion measures Analyses were conducted separately for social skills and problem behaviors standard scores Research Q uestion 3 To examine the relationships betwe en functional ability profile subgroup membership, disability category, and social competence, three regression models were estimated using ProcSurveyReg in 9.2. The first model included disability category as an explanatory variable, the second mode l included profile membership as an explanatory variable (same as R esearch Q uestion 2 ), and the third model included both disability category and profile membership as explanatory variables. Analyses were conducted separately using social skills and probl em behaviors standard scores as criterion measures To report information about the relationship between functional ability profile subgroup membership and disability category, basic descriptive analyses were conducted in 9.2.

PAGE 161

161 Research Q uestion 4 To examine the extent to which select contextual factors moderated the relationship between functional ability profile subgroup membership and social competence, ProcSurveyReg in 9.2 was used. For this research question, select non malleable child va riables as well as contextual variables were examined through a series of regression models E ach model included an interaction term as a test of moderation. Variables were grouped by child factors, family characteristic s parent child interactions, and environmental factors. Within each group of factors, the Bonferonni Holm criterion (Holm, 1979) was applied to account for the multiple inferential tests and control Type 1 error rate. When significant moderation was identified, follow up analyses includ ed comparisons of all possible profile combinations to identify how the differences in the profiles varied across scores on the moderator variables. Again, the Bonferonni Holm criterion was applied to control Type 1 error rate. Analyses were conducted se parately using social skills and problem behaviors standard scores as criterion measures Statistical models for each research question are presented in Appendix E. Statistical Software Data file preparation was conducted with SPSS Version 19.0 or Ve rsion 9.2. Analyses were conducted with Version 9.2 or Mplus Version 6.1. All procedures and analyses were conducted in accordance with the restricted use data license agreement. Summary S econdary analys e s of the Pre Elementary Education Longitud inal Study (PEELS) data set were conducted in the present study. Research questions and hypothesized

PAGE 162

162 models guided the analys e s Statistical techniques used included latent class analys i s, regression models, and regression models with interaction terms The primary focus of th e present study was to identify subgroups of young children with disabilities with similar profiles of functional ability and to examine the relationship between subgroup membership and s In addition, the subgroup membership and disability category membership to the explanation of children social skills and problem behavior s w ere examined. Finally, the exte nt to which non malleable child factors as well as contextual factors moderated relationship s functional ability profile subgroup membership and their social skills and problem behavior s was examined.

PAGE 163

163 Table 3 1 a ge by c ohort and w ave in the PEELS sample Cohort Wave 1 (2003 2004) Wave 2 (2004 2005) Wave 3 (2005 2006) Wave 4 (2006 2007) A 3 4 5 6 B 4 5 6 7 C 5 6 7 8

PAGE 164

164 Table 3 2 Response r ates in the PEELS d ata s et Wave 1 Wave 2 Wave 3 Wave 4 Freq. % Freq. % Freq. % Freq. % Parent i nterview 2,80 0 96 2,89 0 93 2,7 20 88 2,4 90 80 Teacher q uestionnaire 2,2 90 79 2,59 0 84 5,51 0 81 2,50 0 81 Early childhood teacher 2,0 20 79 1,320 86 3 50 82 n/a n/a Kindergarten teacher 2 70 73 9 60 79 99 0 81 4 20 79 Elementary teacher n/a n/a 3 10 86 1,1 80 81 2,08 0 81 Note Freq refers to the frequency of responses collected (%) refers to the percentage response rate from the total sample Access to educational and community activities for young Carlson, A Bitterman, and T Daley, 2010, W estat Report Available at www.peels.org.

PAGE 165

165 Table 3 3 PEELS Disability Severity Index and r elated ICF CY c odes Disabilities i ndex v ariable ICF CF c ode ICF CY c ode n ame Audition b230 Hearing function Vision b210 Seeing function Overall Health n/a n/a Use of Arms d445 Hand and arm use Use of Hands d440 Fine hand use Use of Legs d435 Moving lower extremities Cognition b117 Intellectual functions Communicate with Others d349 Communicating producing Understanding d329 Communicating receiving Regulation of Attention b140 Attention Regulation of Feeling and Emotions b1521 Responsivity Regulation of Activity Level b1252 Activity level Motivation b1301 Motivation Social Skills d710 Basic interpersonal interactions Inappropriate or Unusual Behavior n/a n/a Note Alpha numeric codes from ICF CY (WHO, 2007) B refers to body functions codes and D refers to Activity/Participation codes Adapted from Constructing and T esting a D isability I ndex in a US S ample of P reschoolers with D isabilities T Daley, R Simeonsson, & E Carlson, 2009, Disability and Rehabilitation, 31 p 543

PAGE 166

166 Figure 3 1 Hypothesized relationship between functional ability v ariables, functional

PAGE 167

167 Figure 3 2 Hypothesized relationships between functional profile membership, disability category membership, an d social competence

PAGE 168

168 Figure 3 3 Hypothesized moderat or relationships involving functional ability profile subgroup membership and competence

PAGE 169

169 CHAPTER 4 RESULTS The present study used a cross sectional correlational design to explore and examine competence through secondary analyses of the Pre Elementary Education Longitudinal Study (PEELS) data set. The purposes of the study were to (a) determine whether a set of functional abil ity variables included in the PEELS data set would be useful for empirically deriving distinct and interpretable latent classes that represent subgroups of children who share similar functional ability profiles, (b) explore the relationships between childr and problem behaviors, (c) examine the individual and combined contributions of disability category as correlates of their social skills and problem behaviors, and (d) explore whether select non malleable child variables and contextual variables moderate the relationships problem behaviors. These secondary analyses were exploratory and analytic decisions were made as part of the conduct of the study. In this chapter, analytic decisions made in order to conduct analyses for each research question and the findings associated with e ach of research question are described. Context for Reporting and Interpreting Findings The PEELS data were collected using a complex sampling design that used stratified cluster sampling and unequal probability of selection. Sampling weights were used to address the unequal probability of selection and Taylor Series linearization was used to calculate standard errors that take into account the cluster sampling (Rust,

PAGE 170

170 1985). As part of exploratory procedures, analyses also were conducted with the unweight ed sample and the sample weighted in relation to the primary sampling unit. Similar results were found across estimation methods. All findings reported in this chapter were estimated with the Taylor Series linearization method using either Mplus 6.1 (Mut hen & M uthen, 2007) or the Proc SurveyReg procedure in SAS 9.2. Findings related to each research question are reported below. Steps taken to prepare the data files, changes made to the structure of data variables, or additional analytic procedures compl eted to conduct or interpret the analyses to address each research question are described, when appropriate. All analyses were weighted and thus findings are nationally representative of children ages 3, 4, or 5 years in fall of 2003 who were receiving pr eschool or early elementary special education services under the Individuals with Disabilities Education Act. To meet IES reporting requirements for restricted use data sets, all sample estimates reported are rounded to the nearest tens and the (#) symbol is used in table cells to designate findings when less than 3% of the reference sample (e.g., proportion within a profile) is represented. Zero is reported as a finding when sample estimates have no cases represented, except when noted in a table. Resear ch Question 1 The purpose of the secondary analyses conducted to address this research question were to explore whether distinct and interpretable latent classes could be identified and to use both statistical indices and substantive interpretations to sel ect the optimal latent class model (i.e., distinct, interpretable, and defensible). In the present study, latent classes represent an unobserved categorical variable that define

PAGE 171

171 subgroups of children with similar functional ability profiles that were empi rically derived from the 15 observed functional ability variables included in the PEELS Disability Severity Index. Data Analyses to Conduct Latent Class Analyses Exploratory and descriptive analyses were conducted to evaluate the 15 functional ability var iables that would be used to conduct the latent class analyses. These analyses showed the sample of children was not evenly distributed across the 3 or 4 possible ordinal response categories for the 15 observed variables related to functional ability that are included in the PEELS Disability Severity Index (Daley, Simeonsson, & Carlson, 2009). Response categories were (1) normal or typical functioning, (2) mild limitation in functioning, (3) moderate limitation in functioning, and ( 4 ) severe limitation in functioning Table 4 1 shows the proportion of the weighted sample associated with each variable and response category in PEELS Disability Severity Index. Five variables (hearing, vision, use of arms, use of legs, and use of hands) were identified for which less than .5% of the sample was rated as having a severe limitation in function (fourth response category). During initial exploratory latent class analyses using Mplus, the models would not converge or error messages related to estimating model par ameters were identified due to the small sample size in these five cells. Review of the response category criteria for these five variables showed that the fourth response category was identified by descriptors related to no function (e.g., child has no u se at all of one or both arms for the use of arms variable) or severe limitations in functional ability even when adaptive equipment or devices were used (e.g., child h as a lot of trouble seeing or can no t see at all, even with glasses for the vision variab le). The third response

PAGE 172

172 categories for these five variables were associated with a lot of difficulty or difficulty even with the use adaptive equipment or devices. For these five variables, the descriptions for the third response category were more simil ar to descriptions for the fourth response category than for the other 10 variables. A decision was made to combine the fourth and third response categories for these five variables to represent a moderate/severe limitation in functioning related to the f unctional ability variable and the combined response category was coded as a 4 to be consistent with coding categories for the other 10 variables. Following the re coding of response options for these five functional ability variables, exploratory models for the latent class analyses were rerun. The revised models converged and error messages associated with parameter estimates for these five variables were not identified. Table 4 2 shows the proportion of the weighted sample for each variable and respo nse category after hearing, vision, use of arms, use of legs, and use of hands were recoded for the present study. Results for the latent class analyses conducted in Mplus were generated as probability scales representing the proportion of cases associated with each response category for each variable within a latent class. These proportions were subsequently transformed into model of the functional ability profiles associated with each latent c lass (Appendix D shows the implied means for each variable within each latent class were calculated by multiplying the proportion of each response category by the ordinal value of the response cate gory and summing these values. The model implied means for each functional ability variable within each latent

PAGE 173

173 class were inspected to describe the latent classes in terms of unique functional ability profiles. Some of the initial exploratory latent class analyses produced error messages associated with estimation of the threshold values for five variables (use of arms, use of hands, inappropriate or unusual behavior, understanding, and communication) in some of the latent classes. The default in Mplus is to set 15 as the lowest possible threshold and 15 as the highest possible threshold and to issue a warning when thresholds have been estimated to be equal to these default values. When these warning occurred, the defaults were modified to be 45 and 45 in order to determine if estimated values below 15 and above 15 would occur for any thresholds. In general, estimates below 15 and above 15 did not occur and thresholds that had been estimated at the 15 or 15 default values by Mplus were estimated at t he 45 and 45 default values by Mplus. Results are based on the model estimated by using the 45 and 45 default values. The choice between 15 and 15 default values and 45 and 45 default values has a trivial effect on the estimation of model implied mea ns and standard deviations because with either set of values when a threshold is set at the lower default (i.e., 15 or 45) the expected proportion of response in the lowest response category is substantially below .000001 and when a threshold is set at t he higher default (i.e., 15 or 45) the expected proportion of response in the highest response category is substantially below 000001 (Algina, personal communication, October 22, 2011). Generating and Evaluating Latent Class Models As part of exploratory analyses to identify the latent class model that would be interpreted and used to address other research questions in the present study, models

PAGE 174

174 with 2 to 7 latent classes were estimated. Models beyond a 7 class model were not explored because statistical indices of model fit and replication began to decline. Table 4 3 shows the fit measures for the models that were generated. The log likelihood (LL) value increased and the Bayesian information criterion (BIC) decreased with the addition of latent classe s; both of which are desirable. For the 7 class model, however, the BIC began to increase. Entropy stayed above the desirable level of .80 for the 2 through 6 class models. The numbers of replications that yielded the same LL was sufficiently large (37 40 replications for the 2 through 4 class models, 11 13 replications for the 5 and 6 class models, and 6 replications for the 7 class model) to indicate that a model at a local maximum was not likely to have occurred for any of the models. The 2 t hrough 6 class models were examined to evaluate them with respect to which, if any, were interpretable and logical from a substantive perspective. Based on the statistical indices, however, the 4 5 and 6 class models were considered the statistically defensible models for selection. Within each model, each latent class represents a subgroup of children with similar functional ability profiles. To aid interpretation, each subgroup (latent class) was labeled with a profile number and a descriptive inte rpretation of each profile was generated. For the remainder of this chapter, profile numbers refer to specific subgroups and the term profile is used when describing these unique subgroups. The phrases functional ability profile subgroup membership or subgroup membership are used when referring to the subgroups (latent classes) as a categorical variable. Substantive interpretations focused on examining shared features of functional ability variables within a profile and the distinguishing features of t hese variables across

PAGE 175

175 profiles. Functional ability profiles could be quantified related to the (a) level or severity of limitations on the functional ability variables (e.g., mild, moderate, or severe limitations), (b) number of functional ability variabl es with limitations (e.g., a few, many, all), and (c) nature or type of functional ability variables with limitations (e.g., limitations associated with a similar cluster of functional ability variables). Severity or level of limitations was examined by i nspecting the model implied mean score on each functional ability variable that were calculated from the proportion of cases associated with response categories for each variable. To describe the profiles and to interpret the severity of limitations acros s profiles, model implied means were grouped into ranges to account for the standard deviations. M oderate to severe limitations were associated with mean scores from 2.5 and above, mild to moderate limitations were mean scores from 1.5 to 2.49, and no to mild limitations were mean scores from 1.49 and below. As noted previously, to determine the optimal class model, model fit indices were combined with substantive interpretations of the profiles. The 5 class model was selected because (a) the 5 class m odel had improved model fit indices over the 4 class model; (b) the 5 class model provided a logical and interpretable model that offered distinctions among profiles; and (c) the 6 class model did not provide substantial fit improvement or notably differen t substantive information over the 5 class model. In the following section, information about the severity of limitations, number of limitations, and type of functional ability limitations identified in the five distinct profiles are de scribed. In additi on, information about the classification probabilities is provided. For reference, similar information about the non selected class models is provided in Appendix E.

PAGE 176

176 Interpreting the Selected Latent Class Model The selected 5 class model represented five subgroups of children with similar functional ability profiles on the 15 functional ability variables that were distinct from other subgroups. As noted earlier, each subgroup was labeled with a profile number Descriptions of each profile are based on th e (model implied) mean functional ability scores for the subgroup of children that the profile represents. Children assigned to a profile have, in general, similar functional abilities in terms of the severity of limitations, number of limitations, and ty pes of limitations across the 15 functional ability variables Severity and number of limitations on functional ability variables Figure 4 1 shows the means for each profile on the 15 functional ability variables. Table 4 4 shows the means and standard d eviations on the 15 functional ability variables for each profile. The means in Table 4 4 have been bolded or italicized to indicate the different severity or levels of limitations across the functional ability variables. Means show that Profile 1 and Pr ofile 2 had more moderate to severe limitations across functional ability variables Profile 1 (5% of the sample) was associated with limitations on 13 of 15 functional ability variables. This included moderate to s evere limitations on 11 functional ability variables: use of arms, use of hands, use of legs, communication, understanding, cognition, overall health, social skills, regulation of activity level, regulation of attention, and motivation ; and mild to moderate limitations on two fu nctional ability variables: vision and behavior. Profile 2 (15% of the sample) was associated with limitations on 11 of 15 functional ability variables. This included moderate to severe limitations across eight functional ability variables: communication cognition, use of hands, social skills, behavior, regulation of activity level, regulation of attention, and motivation ; and mild to moderate limitations

PAGE 177

177 associated with three functional ability variables: overall health use of hands, and regulation of emotions. Means show that Profile 3 and Profile 4 had more mild to moderate limitations across functional ability variables Profile 3 (7% of sample) was associated with limitations on 10 of 15 functional ability variables. This included moderate to s e vere limitations on two functional ability variables: cognition and use of hands; and mild to moderate limitations across eight functional ability variables: communication, overall health, use of arms, use of legs, social skills, regulation of activity lev el, regulation of attention, and motivation. Profile 4 (33% of sample) was associated with limitations on 9 of 15 functional ability variables This included moderate to s evere limitations on five functional ability variables: communication, cognition, regulation of activity level, regulation of attention, and motivation ; and mild to moderate limitations across four functional ability variables: understanding, overall health, social skills, and behavior. Profile 5 (40% of the sample) was associated with limitations on 5 of 15 functional ability variables This included mild limitations with communication cognition, regulation of activity level, regulation of attention, and motivation and no limitations on other functional abi lity variables. Types of limitations on functional ability variables A further distinction among the profiles was the types of limitations; specifically, there were clusters of variables that might affect similar aspects of functioning that were common within a profile. For example, there were five profiles with limitations related to regulatory functional ability variables but only three profiles with limitations related to physical functional ability variables Moreover, there were five profiles with limitations

PAGE 178

178 related to communication or cognition variables but only four profiles had limitations related to social competence variables (i.e., social skills and behavior). Regulatory variables were identified as regulation of activity, regulation of at tention, motivation, and regulation of emotions, although the emotions variable did not lead to distinctions between sub groups. Limitations on these variables might affect self regulation. Both Profile 1 and Profile 2 had moderate to severe limitations related to these variables. Model implied means were somewhat higher on these variables, however, for Profile 2 than the means for Profile 1. In addition, Profile 3 and Profile 4 had mild to moderate limitations related to these variables. Model implied means were somewhat higher on these variables for Profile 4 than the means for Profile 3. Profile 5 had mild limitations related to these variables. Physical indicators were identified as overall health, use of arms, use of hands, and use of legs. Limit ations on these variables might affect motor function. Profile 1 had moderate to severe limitations related to all of these variables and Profile 3 had mild to moderate limitations related to all these variables. Functional ability variables related to co mmunication, understanding, and cognition were similar to the differences between profiles associated with the severity of limitations. Model implied means indicated moderate to severe limitations for Profiles 1 and 2, mild to moderate limitations for Pro files 3 and 4, and mild or no limitations for Profile 5 on these variables. Profile 1, however, had slightly higher means (indicating somewhat greater limitations in function) on these variables than Profile 2, while Profile 4 had slightly higher means on these variables than Profile 3.

PAGE 179

179 Social skills and behavior variables that might be associated with social competence also distinguished differences between profiles associated with the severity of the limitations. Limitations on these parent rated func tional abilities associated with social competence might correspond with performance based teacher ratings of social skills and problem behaviors. Model implied means in Table 4 4 show the profiles with limitations related to physical variables (Profiles 1 and 3) had slightly lower means (indicating somewhat fewer limitations in function) than Profiles 2 and 4, respectively. Probability of being assigned to a profile As noted in Chapter 3, latent class analysis is a model based statistical procedure. As part of the analysis, the probability that each child would be assigned to each latent class (subgroup) is calculated and reported as posterior probabilities The mean posterior probability estimates for each of the five latent classes was greater than 85 %. Table 4 5 shows the mean profile (latent class) probabilities for most likely class membership. Children in the Profile 1 had a 91%, 6%, 3%, 0%, and 0% probability of being assigned to the first, second, third, fourth, and fifth profile, respectively Children in Profile 2 had a 2%, 89%, 1%, 8%, and 0% probability of being assigned to the first, second, third, fourth, and fifth profile, respectively, while children in Profile 3 had a 3%, less than 1%, 88%, 7%, and 2% probability of being assigned to the first, second, third, fourth, and fifth profile, respectively. For Profile 4, children had a 4%, 0%, 3%, 86%, and 7% probability of being assigned to the first, second, third, fourth, and fifth profile, respectively. Children in Profile 5 had a 0%, 0 %, less than 1%, 7%, and 92% probability of being assigned to the first, second, third, fourth, and fifth profile, respectively.

PAGE 180

180 Functional ability profile summary The 5 class model resulted in distinct subgroups with similar functional ability profiles that could be meaningfully interpreted with adequate fit indices and good replication. Severity, number, and types of limitations distinguished the five profiles across functional ability variables. The consistent pattern of distinction between children with moderate to severe limitations and children with mild to moderate limitations was substantively meaningful, as was the distinction between limitations related to different aspects of functioning. Specifically, the 5 class model permitted compar isons of functional ability limitations for subgroups of children with moderate to severe limitations (Profiles 1 and 2), for subgroups of children with mild to moderate limitations (Profiles 3 and 4), and for a subgroup of children associated with no to mild li mitations across the 15 functional ability variables (Profile 5). In addition, the 5 class model identified two subgroups of children whose limitations included physical functional abilit y limitations that might affect motor function (Profiles 1 and 3) Demographic and Descriptive Information for Members of Each Subgroup The sample distribution was not evenly allocated across the subgroups Profiles 1 and 3 were the smallest groups, 5% and 7% of the sample, respectively. Profile 5 was the largest grou p (40% of the sample). Profiles 2 and 4 were 15% and 33% of the sample, respectively. Table 4 6 shows demographic and descriptive information about the children in each profile. Children were assigned to a profile based on their most probable latent clas s membership. There were more boys than girls in each of the profiles, consistent with the gender dist ribution in the PEELS data set (i.e., 70% males and 30% females;

PAGE 181

181 Markowitz et al., 2006). Profiles 1, 2, and 4 have mid 70 to mid 20 percentage splits ( e.g., 74% boys and 26% girls for Profile 1). Profile 5, however, has 67% boys and 33% girls and Profile 3 has 58% boys and 42% girls. Mean age across profiles was between 54 and 56 months ( SD = 8 9 months). Within and across profiles, the distribution a cross 3, 4, and 5 years of age was generally evenly spread. Children from different racial backgrounds were generally evenly spread across profiles, with slightly less Caucasian children and more Hispanic children in Profiles 2 and 4. Profile 5 also had slightly more Caucasian children and fewer Hispanic children. Children, whose primary home language was not English, referred to as children with English as a second language (ESL), were generally equally distributed across profiles with a slightly smalle r proportion of children designated as ESL in Profile 5. Proportions of children who had an individualized family service plan (IFSP) before the age of 3 years were larger in profiles associated with limitations related to physical functional abilities, 82% of children and 72% of children within Profiles 1 and 3, respectively. In comparison, 50%, 36%, and 26% of children in Profiles 2, 4, and 5, respectively, had an IFSP before the age of 3 years. A similar pattern across profiles was identified related to the mean number of weeks premature or mean birth weight (i.e., greater proportions of children in Profiles 1 and 3 were born premature and had a lower mean birth weight). Table 4 7 shows demographic and descriptive information about the families of ch ildren in each profile. As noted earlier, profiles represent subgroups of children with similar functional abilities on the 15 functional ability variables and children were assigned to a subgroup based on their most probable latent class membership. As

PAGE 182

182 shown in Table 4 7, nearly 100% of the sample lived with the parent or guardian respondent who completed the PEELS data collection phone interview and provided satisfac the community in which they resided. Respondents were generally mothers, biological or adoptive, across the profiles (88% to 96%). Other respondent roles included biolo gical fathers (2% to 6%) or other respondents that might have been grandparents or foster parents (2% to 8%). Across profiles, children generally lived in two parent families, with 17%, 38%, 32%, 32% and 23% of children in Profiles 1, 2, 3, 4, and 5, resp ectively, living in single parent families. profiles with the exception of a larger proportion of parents with some college or a 2 year degree for children in Profile 1 and a sm aller proportion of parents with no high school diploma or equivalent for children in Profile 5. Distribution of reported family income also was similar across profiles with the exception of a larger proportion of families above $50,000 for children in Pr ofile 1 and a smaller proportion of families below $10,000 for children in Profile 5. Overall, the greatest proportion of respondents indicated the neighborhood in which they lived was safe or very safe for their child to play outside across all profiles. Related to parent satisfaction with special education services, parents generally indicated they were satisfied or very satisfied with special education services across all profiles. Table 4 8 shows descriptive information about child and family particip ation in a range of activities for each profile. Activity variables were generated from a series of

PAGE 183

183 questions asked as part of the parent interview. Child activities were defined as the number of extra curricular activities a child had ever participated in, the extent to which the child participated in any group activities on a monthly basis, and which group activities the child participated in on a monthly basis. Family school activities related to the number of different school activities that the pare Child family activities related to the number of community based activities that someone in the family engaged in with the child on a monthly basis, the number of meals the family ate together during the week, and the nu mber of times a family member read to the child during the week. Distributions of child, family school, and child family activities were similar across profiles, with one exception. For the extent to which a child participated in any group activities on a monthly basis, the proportion of children in Profile 1 (36%) was smaller than Profiles 2, 3, and 4 (45% to 47%) and the proportion of children in Profile 5 was the largest (55%). Table 4 programs or schools for each profile. As reported by school administrators, the extent to which children attended sch ools that served families from low income household s was generally equally distributed across profiles. Children attended preschool prog rams or elementary schools that had preschool or kindergarten classrooms. For children enrolled in a preschool program, the extent to which the preschool was accredited was similar across Profiles 1, 2, 4, and 5 (42% 50% accredited), while the proportio n of preschools that were accredited for Profile 3 was 62%. The extent to which schools that children attended met No Child Left Behind (NCLB) standards was generally similar

PAGE 184

184 across profiles (83% 93% met standards). Using information from the teacher i nterview, the extent to which teachers reported their program supported or facilitated interactions among children with and without disabilities was 69% for Profile 1, 54% for Profile 2, 64% for Profile 3, 60% for Profile 2, and 53% for Profile 5. The mea n number M = 6 8, SD = 3 classrooms, however, differed across profiles. Children in Profiles 1 and 2 had the fewest number of children without an IEP in their classroom ( M = 4, S D = 6), children in Profiles 3 and 4 had a slightly larger number of children without an IEP in their classroom ( M = 6, S D = 6 7), and children in Profile 5 had the larges t number of children without an IEP in their classroom ( M = 8, S D = 8). domains was different across profiles (Table 4 9). For Profile 1, 19% of children had a school read iness goal, 6% had a pre academic goal, 20% had a social goal, 23% had a behavior goal, 31% had an adaptive goal, 66% had a communication goal, 34% had a fine motor goal, and 38% had a gross motor goal. For Profile 2, 48% of children had a school readines s goal, 12% had a pre academic goal, 36% had a social goal, 32% had a behavior goal, 22% had an adaptive goal, 71% had a communication goal, 21% had a fine motor goal, and 6% had a gross motor goal. For Profile 3, 40% of children had a school readiness go al, 16% had a pre academic goal, 17% had a social goal, 8% had a behavior goal, 16% had an adaptive goal, 56% had a communication goal, 35% had a fine motor goal, and 38% had a gross motor goal. For Profile 4, 48% of children had a school readiness goal, 9% had a pre academic goal, 32% had a social goal, 22% had a

PAGE 185

185 behavior goal, 10% had an adaptive goal, 74% had a communication goal, 17% had a fine motor goal, and 4% had a gross motor goal. For Profile 5, 27% of children had a school readiness goal, 5% ha d a pre academic goal, 19% had a social goal, 6% had a behavior goal, 3% had an adaptive goal, 83% had a communication goal, 10% had a fine motor goal, and 4% had a gross motor goal. Distribution across curricular content domains within and across profile s corresponded with limitations on related functional ability variables associated with profile descriptions. For example, children in Profile 5 had the highest proportion of goals associated with communication, while children in Profiles 1 or 3 had highe r proportions of goals focused on fine motor or gross motor compared to other profiles. Overall, descriptive analyses of the 24 demographic variables related to children, families, their activities, and their school programs showed similar proportions dis tributed across the five latent class profiles for 17 of these variables. Noteworthy differences across profiles were found for four child characteristic variables (i.e., gender, IFSP before 3, weeks premature, birth weight), one contextual variable (part icipation in regular group activities), and two school related variables (i.e., the number of children are discussed further in Chapter 5. Research Question 2 The purpose of this research question was to examine relationship s between problem behaviors In addition, analyses were conducted to examine whether subgroups of children w ith different functional ability profiles differe d in their social skills and problem behaviors

PAGE 186

186 Using Most Probable Class Membership to Examine Social Competence The unique functional ability profile of each subgroup provided an opportunity to compare and contrast how different profiles of functional abilities might or might not pre viously, functional ability profile subgroup membership was estimated as a categorical variable (1, 2, 3, 4, 5) based on most probable latent class membership assigned as part of the latent class analysis. This approach addressed the issue with selecting the weight file for LCA with mixture modeling and was appropriate given that mean posterior probabilities for latent class membership were all above .86, with a range of .86 to .92 over the five latent classes. As noted earlier, the term profile refers to specific subgroups while the phrase subgroup membership refers to the categorical variable used in the analyses. To conduct regression analyses the parent child teacher weights were used. The application of parent child teacher weights to address the sec ond research question resulted in a reduced number of available cases to conduct analyses. The application of these weights ( N = 2,180), in combination with missing PKBS 2 scores, resulted in 2,090 cases available for these analyses. Descriptive analyses to examine and compare information about basic demographic characteristics of the children in the three different samples were conducted (i.e., full sample for LCA, reduced sample for parent child teacher weights, and reduced sample fo r PKBS 2). Table 4 10 shows the mean age, gender proportions, and race/ethnicity proportions for children in each profile and the total profile were equivalent acr oss the three samples. Two exceptions are shown in Table

PAGE 187

187 4 10. In the sample used to conduct analyses with the PKBS 2, children in Profile 1 included proportionally more females than males compared to the other two samples, and children in Profile 2 incl uded proportionally more males than females compared to the other two samples. Differences in gender were not observed when comparing the total sample across each of the three samples. Relationships Between Subgroup Membership and Social Skills and Proble m Behaviors membership and social skills and problem behaviors the R squared variance explained p accounted for 20% of the variance in PKBS 2 social skills standard scores and accounted for 11.5% of the variance in problem behaviors standard scores. In addition to the regression model, social skills and problem behaviors standard scores and differen ces between standard scores for each functional ability profile subgroup were examined. Table 4 1 1 shows the means, standard errors, and 95% confidence interval around the standard score mean estimate for social skills and problem behavior s for each functi onal ability profile subgroup Figure 4 2 shows the means for each subgroup. T he PKBS 2, which was used to measure social skills and problem behaviors, is standardized to a mean of 100 and standard deviation of 15 with higher scores on social skills asso ciated with more social skills (positive outcome) and higher scores on problem behavior s associated with more problem behaviors (negative outcome). As shown in the table, children in Profile 1 had the lowest mean social skills scores ( M = 69.99 SEM = 3.2 2 ) but not th e highest mean problem behaviors scores ( M

PAGE 188

188 = 100.6 1 SEM = 2. 1 2). Children in Profile 2 had the second lowest mean social skills scores ( M = 80.28 SEM = 1.95 ) and th e highest mean problem behaviors scores ( M = 106. 45 SEM = .96 ). Children in Profile 4 had lower mean social skills scores ( M = 91. 6 1, SEM = .95 ) a nd higher mean problem behaviors scores ( M = 100.81 SEM = .92 ) than children in Profile 3, ( M = 95.28 SEM = 1.94 for social skills scores and M = 95. 7 1, SEM = 1.31 for problem behaviors scores ). Children in Profile 5 had the highest mean social skills scores ( M = 102.08 SEM = 89 ) and lowest mean problem behavior s scores ( M = 92.4 3 SEM = .6 8 ). The relationships between functional ability profile subgroup membership a nd social skills and problem behaviors were examined further for PKBS 2 subscales related to social cooperation, social interaction, social independence, externalizing behaviors, and internalizing behaviors. Table 4 12 shows the means, standard errors, an d 95% confidence interval around the mean estimate for each subscale and profile. Figure 4 2 shows the means for each subscale and profile. Across all profiles, 78.15, 88.98, 91.12, 98.06 for Profiles 1, 2, 4, 3, and 5, respectively. For social cooperation and social independence, respectively, Profile 1 had the lowest mean scores (75.13 and 75.75), Profile 2 had the next lowest mean scores (87.66 and 81.96), followed by Profile 4 (95.77 and 93.07). Profile 3 had the second highest mean scores for social cooperation and social independence (100.43 and 96.09), and Profile 5 had the highest mean scores (105.58 and 101.53). For externalizing problem behaviors, children in Profile 2 followed by children in Profile 4 had the highest mean externalizing behaviors scores ( M = 104.01 and M =

PAGE 189

189 100.9 for Profiles 2 and 4, respectively). Children in Profiles 1, 3, and 5 had a mean score of 98.83, 93.58, and 91.94, respectively. For internalizing behaviors scores, children in Profile 2 followed by children in Profile 1 had the highest mean internalizing behaviors scores, ( M = 107.76 and M = 102.35 for Profiles 2 and 1, respectively). Children in Profiles 3, 4, and 5 had a mean sc ore of 98.84, 100.69, and 94.38, respectively. The magnitudes of differences between the mean standard scores for social skills and problem behaviors across the five subgroups are shown in Table 4 13. This table provides the standardized mean difference e ffect sizes for the scores on the PKBS 2 social skills composite and problem behaviors composite. Pooled residual variance was used to estimate the effect sizes for each composite score. Effect sizes show that mean social skills scores for children in Pr ofile 1 were at least one standard deviation below children in Profile 3, 4, and 5 (i.e., 1.35, 1.15, and 1.71, respectively) and .55 standard deviation below children in Profile 2. For Profile 2, effect sizes show children in Profile 2 were .80, .60, and 1.16 standard deviation below children in Profile 3, 4, and 5, respectively. Effect size for Profile 3 was .20 standard deviation above Profile 4 and was 1.16 standard deviation below Profile 5. Profile 4 was .56 standard deviation below Profile 5. The se standardized mean differences between profiles were found to be statistically significant at a p critical value of .01 or smaller except for the difference between mean social skills scores for children in Profile 3 and 4 (ES = .20), which was not stati stically significant. In general, the magnitudes of differences between mean problem behaviors scores across functional ability profile subgroups were not as large as those for social

PAGE 190

190 skills. For problem behaviors, higher scores are associated with more problem behaviors. Effect sizes show that mean problem behaviors scores in Profile 1 was .41 standard deviation below Profile 2 and was .01 standard deviation below Profile 4. Profile 1 was .35 and .58 standard deviation above Profile 3 and 5, respectiv ely. For Profile 2, standardized mean difference effect sizes show children in Profile 2 were .76, .40, and .99 standard deviation above children in Profile 3, 4, and 5, respectively. Effect sizes show that Profile 3 was .36 standard deviation below Prof ile 4 and was .23 standard deviation above Profile 5. Profile 4 was .59 standard deviation above children Profile 5. Standardized mean differences between profiles were found to be statistically significant at a p critical value of .05 or smaller except for the difference between mean problem behaviors scores for children in Profile 1 and Profile 3 ( ES = .35) or Profile 1 and 4 ( ES = .01), which were not statistically significant. Research Question 3 The purpose of this question was to examine whether functional ability profile subgroup membership accounted for more explained variance in social skills and problem behaviors scores than disability category. To explore these relationships, the individual and combined contributions of functional ability pr ofile subgroup membership disability category were examined by using these variables as models. The following explanatory variables were used: IDEA disability category alone, functional ability profile subgroup membership alone, and both variables entered simultaneously in the regression model. The following disability categories were used: autism, developmental delay, emotional behavioral disturbanc e, mental retardation, learning disabilities, low incidence, and speech or language impairments.

PAGE 191

191 Associations Between Explanatory Variables Associations between functional ability profile subgroup membership and disability category were examined before est imating regression models to examine variance accounted for when the variables were entered individually and together. Table 4 14 shows the percentage of the total sample cross classified by profile and disability category. Table 4 15 shows the distribut ion of disability categories across the profiles. Of note, 72% of the children in Profile 5 were identified for special education services with the speech or language impairments disability category. This cross classified group consisted of 29% of the to tal sample. PEELS staff has reported that a large portion of the sample included children with speech or language impairments who qualified for services due to speech (articulation) issues (Carlson & Lowe, 2009 ). As shown in Table 4 14, nearly half the sample was identified with speech or language impairments and one quarter was identified with developmental delay. Speech or language impairments made up 72% of the children in Profile 5, however, it was also represented across all other profiles: 43%, 15 %, 25%, and 5% of Profiles 4, 3, 2, and 1, respectively. The distribution of developmental delay (DD) was more similar across profiles; DD represented 26%, 38%, 37%, 32%, and 14% of Profiles 1, 2, 3, 4, and 5, respectively. The distribution of other disa bility categories varied across profiles. For example, 35% of children in Profile 1 were identified with a low incidence disability, 26% with mental retardation, and 8% with autism. For children in Profile 2, 23% were identified with autism, 6% with a lo w incidence disability, and 4% with mental retardation. Thirty seven percent of children in Profile 3 were identified with a low incidence disability and 3% with mental retardation. For children in Profile 4, 8% were identified with a low incidence disab ility, 6% with autism, 4% with a learning disability,

PAGE 192

192 and 3% with mental retardation. Less than 10% of children in Profile 5 were identified for services with disability categories other than speech or language impairments or developmental delay, nonethel ess, most or all disability categories were represented. As shown in Table 4 14, 9.2% of the total sample were children identified for special education services related to a low incidence disability. Table 4 16 shows the primary IDEA disability category by profile for children who represented the low incidence disability category used in the present study. As shown in this table, children with multiple disabilities, orthopedic impairments, or other health impairments were generally represented in all pr ofiles with more representation in Profile 1 or Profile 3 compared to other profiles. Estimating Individual and Combined Contributions of Explanatory Variables Distribution of disability categories across subgroup membership and within the total sample ind icates that there is some relationship or crossover between subgroup membership and disability category, particularly for speech or language impairments and Profile 5. For this reason, the individual and combined contributions of functional ability profil disability category as correlates of their social skills and problem behaviors were examined in two ways. First, the explanatory power of subgroup membership and disability category were examined by estimating thr ee regression models using all five subgroups and all seven disability categories. Model 1 included disability category as an explanatory variable. Model 2 included subgroup membership as an explanatory variable. Model 3 included subgroup membership and disability category entered together as explanatory variables. Second, a holdout analysis was conducted. Children with speech or

PAGE 193

193 language impairments were removed from the sample and the three regression models were reestimated. Individual and Combined Contributions of Explanatory Variables Table 4 17 shows the R squared variance explained value for social skills standard scores and problem behaviors standard scores for each explanatory variable and when entered together in the regression model In gene ral, more variance was explained for social skills scores than problem behaviors scores. In the first regression model, disability category accounted for 16.5% of the variance in social skills scores. In the second regression model (same as Research Ques tion 2), subgroup membership accounted for 20% of the variance in social skills scores. When subgroup membership and disability category were entered simultaneously in the third regression model, they accounted for 25.2% of the variance in social skills s cores. For PKBS 2 problem behaviors standard scores, the explanatory variables accounted for less variance than for social skills. In the first regression model, disability category accounted for 10.8% of the variance in problem behaviors scores. In th e second regression model (same as Research Question 2), subgroup membership accounted for 11.5% of the variance in problem behaviors scores. The most variance was accounted for when subgroup membership and disability category were entered simultaneously in the third regression model. This accounted for 15.9% of the variance in problem behavior scores. Again, these data indicate the explanatory variables were not perfectly uncorrelated, as was anticipated. Table 4 18 shows the R squared variance explaine d value for the holdout analyses in which children with speech or language impairments (SLI) were removed from the sample. More variance was explained for social skills scores than problem

PAGE 194

194 behaviors scores. In the first regression model, disability categ ory accounted for 11.4% of the variance in social skills scores. In the second regression model, subgroup membership, however, accounted for 20.1% of the variance in social skills scores. The most variance was accounted for when subgroup membership and d isability category were entered simultaneously in the third regression model. This accounted for 23.1% of the variance in social skills scores. Even without SLI in the sample, the explanatory variables were not perfectly uncorrelated. For problem behav iors, disability category accounted for 7.4% of the variance in problem behaviors standard scores in the first regression model. In the second regression model, subgroup membership accounted for 9% of the variance in problem behaviors scores. In the thir d regression model when subgroup membership and disability category were entered simultaneously, 10.4% of the variance was accounted for in problem behaviors scores. Research Question 4 The purpose of this question was to explore whether select non malleab le child factors and contextual (i.e., family and environmental) variables moderate relationships problem behaviors. This was examined by inspecting the extent to whic h non malleable child factors and contextual factors had an interactive influence on the relationships problem behaviors. Contextual factors were identified as family factors, which included family characteristics and parent child interaction factors, and other environmental factors, which included school and community variables.

PAGE 195

195 Examination of Non Malleable Child Factors and Contextual Factors Sixteen variables relate d to the four types of factors were selected from the PEELS data set and examined: child factors ( v = 3), family characteristics ( v = 3), parent child interaction factors ( v = 7), and other environmental factors ( v = 3). As shown in Appendix C, seven of t hese variables were continuous, four were dichotomous, and five were categorical. As part of exploratory work for the moderation analyses, frequencies of the categorical variables by subgroup membership were ity, parent education, number of times a week a family member reads to the child) were identified with no cases or a small proportion of the sample in at least one cross classified response category (i.e., profile by categorical response). To examine the moderating influence of these categorical variables on the relationships between functional ability profile subgroup membership and social skills and problem behaviors, categories for these variables were recoded to ensure each cross classified cell had ad equate sample sizes to examine moderation. Table 4 19 shows how the variables were recoded for moderation analyses. Child race ethnicity included six possible response categories: (1) Caucasian/ Non Hispanic; (2) Hispanic; (3) African American; (4) Ameri can Indian or Alaskan Nativ e; (5) Asi an, Native Hawaiian, or Pacific Islander ; and (6) Multi racial. Due to the small sample sizes in (1) American Indian or Alaskan Nativ e; (2) Asi an, Native Hawaiian, or Pacific Islander ; and (3) Multi racial, these were grouping is consistent with other PEELS studies and reports that have reported Markowitz et al., 2006). Parent education included seven possible response categories : ( 1 ) l ess than High School with no GED (2) h igh s chool diploma or GED ( 3 ) s ome college/post secondary vocational

PAGE 196

196 course ( 4 ) 2 or 3 year college degree or vocational school diploma, ( 5 ) 4 year college degree ( 6 ) s ome graduate work/no graduate degree and ( 7 ) g raduate degree Parent education was regrouped into six categories by combining (1) 4 year college degree (2) s ome graduate work/no graduate degree due to the sample size for some graduate work/no graduate degree response category. The number of times a family member reads to the child each week was originally coded as: ( 1 ) n ever ( 2 ) o nce or twice (3) 3 to 6 times, and ( 4 ) e very day Due to the small sample size for the response category never, this variable was re coded into 3 response cate gories: (1) 0 to 2 times, (2) 3 to 6 times, and (3) e very day Moderation was examined with 13 of the originally coded variables and the three recoded variables. If moderation was identified as statistically significant, follow up analyses included graphi ng the slopes by profile for continuous variables or tabling the social skills or problem behavior mean scores for each response category by profile for categorical variables. Moderating Influence of Contextual Factors on the Relationship Between Subgroup Membership and Social Skills Table 4 20 shows the extent to which variables associated with child factors, family characteristics, parent child interactions, and other environmental factors regressio n model and the extent to which these variables moderated the relationship between subgroup membership and social skills scores. As shown in the table, child factors of race/ethnicity [ F ( 12, 61 ) = 1.29, p = .24], age [ F ( 4, 61 ) = 1.71, p = .15], and gende r [ F ( 4, 61 ) = .15, p = .96] did not have a moderating influence on the relationship between subgroup membership and social skills scores nor

PAGE 197

197 did family factors of parent education [ F (20 61 ) = 1.58, p = .08], marital status [ F (4 61 ) = .75, p = .55], and f amily income [ F (4 61 ) = .74, p = .56]. For parent child interaction factors, the number of child activities a child had ever participated in, did have a statistically significant moderating influence on the relationship between subgroup membership and so cial skills scores [ F ( 4, 61 ) = 3.93, p = .0066], while other parent child interaction variables did not moderate this relationship. Environmental factors, school community income [ F ( 12, 61 ) = 1.50, p = .14], program support for social interaction [ F ( 4, 61 ) = 1.64, p = .17], and neighborhood safety [ F ( 8, 61 ) = 1.07, p = .39], also did not have a moderating influence on the relationship between subgroup membership and social skills. To examine further the moderating influence of the number of child activitie s that a child had ever participated in on the relationship between subgroup membership and social skills scores, the difference between slopes for each profile comparison (i.e., 10 comparisons) was examined. Follow up analyses revealed that there were st atistically significant differences in the slope between Profile 4 and Profile 5 [estimated slope difference = 2.91 t (61 ) = 2.91, p = .0051], however this difference in slope was not statistically significant at the Bonferonni Holm adjusted (Holm, 1979) p critical criterion of .005. Nonetheless, the nature of the interaction was of substantive interest and a graph of the interaction was examined and interpreted. Figure 4 3 shows the and 5 by the number of extra curricular child activities that a child has participated in. The figure shows that for children who have participated in fewer activities during their life (i.e., 0 1 activities), children in Profile 4 have lower social skill s scores than children in Profile 5.

PAGE 198

198 For children who have participated in more activities (i.e., 6 activities), children in Profile 4 have higher social skills scores than children in Profile 5. For children who have participated in 3 or 4 different act ivities, there is no difference in social skills scores between children in Profile 4 and Profile 5. Moderating Influence of Contextual Factors on the Relationship Between Subgroup Membership and Problem Behaviors Table 4 21 shows the extent to which vari ables associated with child factors, family characteristics, parent child interactions, and other environmental factors in the model and the extent to which these variables moderated the relationship between subgroup membership and problem behaviors scores. As shown in the table, the child factor of race/ethnicity [ F (12, 61) = 2.72, p = .005] was a statistically significant moderator on the relationship between subgroup memb ership and problem behaviors scores, while, age [ F ( 4, 61 ) = 1.63, p = .17], and gender [ F ( 4, 61 ) = .32, p = .86] did not have a moderating influence on this relationship. Family factors of parent education [ F (20 61 ) = 1.97, p = .02], family income [ F (4 61 ) = 2.28, p = .07], and marital status [ F (4 61 ) = 1.19, p = .32] did not have a moderating influence on the relationship between subgroup membership and problem behavior scores. For parent child interaction factors, the number of regular child activit ies [ F ( 4, 61 ) = 5.07, p = .0014] and the number of parent child activities [ F ( 4, 61 ) = 4.07, p = .0054], did have a statistically significant moderating influence on the relationship between subgroup membership and problem behaviors scores, while other par ent child interaction variables did not moderate this relationship. For other environmental factors, neighborhood safety [ F ( 8, 61 ) = 3.14, p = .0049], was a statistically significant

PAGE 199

199 moderator on the relationship between subgroup membership and problem be haviors scores, but school community income [ F ( 12, 61 ) = 1.83, p = .06] and program support for social interaction [ F ( 4, 61 ) = .42, p = .79] were not statistically significant moderators of the relationship between subgroup membership and problem behaviors scores. To examine further the moderating influence of the continuous variables, follow up analyses on the number of parent child ac tivities and number of regular child activities on the relationship between subgroup membership and problem behaviors scores were conducted for each variable. For the number of parent child activities, follow up analyses involving comparisons of the diffe rences between the slopes for 10 possible comparisons showed that there were statistically significant differences in the slope between Profile 1 and Profile 3 [estimated slope difference = 4.46 t (61 ) = 3.59, p = .0007] and slope between Profile 1 and Pro file 5 [estimated slope difference = 4.24, t (61) = 3.56, p = .0007]. Figure 4 4 shows problem behaviors scores between Profile 1 and Profile 3 and between Profile 1 and Profile 5 by the number of parent child ac tivities. For children who have participated in fewer parent child activities (i.e., 0 1 activities), the figure shows that the difference between problem behaviors scores for children in Profile 1 compared to children in Profile 3 or 5 is smaller than th e difference between problem behaviors scores for children who have participated in more parent child activities (i.e., 6 7 activities). Visual inspection of the graphed data revealed that outliers in Profile 1 influenced the slope of the line for Profile 1, and thus the moderation effect is likely due to these outliers. For the number of regular child activities, follow up analyses showed that there were statistically significant differences in the slope between Profile 1 and Profile 5

PAGE 200

200 [estimated slope d ifference = 6.83 t (61 ) = 3.09, p = 003]. Figure 4 5 shows the Profile 5 by number of regular child activities. The figure shows that the difference in problem behavi ors scores between children in Profile 1 and Profile 5 gets larger (higher ratings of problem behaviors for Profile 1) as the number of regular child activities increases. To examine further the moderating influence of the categorical variables (i.e., chi membership and problem behaviors scores, the problem behavior s standard score means for each profile for every response category of the categorical variable were calculated a nd tabled. Examination of statistically significant, mean difference comparisons for every possible comparison were not conducted due to the large number of comparisons for each variable (i.e., 30 comparisons for race/ethnicity and 15 comparisons for neig hborhood safety). Table 4 22 shows the mean problem behaviors scores for children in each profile for race/ethnicity. Consistent with means identified for each profile under Research Question 2, children in Profile 2 tend to have the highest mean proble m behaviors scores while children in Profile 5 tend to have the lowest mean problem behaviors scores. A few descriptive differences in mean scores are noted to highlight what the moderation is evaluating, but, as previously indicated, comparisons were not examined for statistical significance. For example, the difference between problem behaviors mean scores for Caucasian/ Non Hispanic children and Hispanic children was 9, 3, 1, 2, and 1 for Profile 1, 2, 3, 4, and 5, respectively. The 9 point difference between

PAGE 201

201 Caucasian/Non Hispanic children and Hispanic children for Profile 1 was not consistent with the pattern of the relationship across other profiles, suggesting some moderation between these response categories across subgroup profiles. Table 4 23 s hows the mean problem behaviors scores for children in each profile for the neighborhood safety response categories. Descriptively, there was a difference between problem behaviors scores for children who live in a very safe neighborhood compared to child ren who live in a not safe neighborhood among the profiles, suggesting some moderation between these response categories across profiles. Profiles 1, 3, and 5 had a large difference in scores between children who live in a very safe neighborhood compared a not safe neighborhood (5 to 6 point difference), while there was a small difference between these groups for Profiles 2 and 4 (1 to 2 point difference). In contrast, the differences between problem behaviors scores for children who live in a very safe n eighborhood were very similar to children who live in a safe neighborhood across profiles. There was a 3 or 4 point difference between children who live in a very safe neighborhood or a safe neighborhood across all profiles. In this set of mean differen ce comparisons (safe vs. very safe), the similar mean differences across profiles suggest no moderation between these response categories across profiles. Summary Results from the latent class analyses conducted in the p resent study showed that functional ability variables included in the PEELS data set were useful for empirically deriving distinct and interpretable latent classes. A 5 class model was identified based on statistical and substantive interpretations. The distinctions between profiles were

PAGE 202

202 across functional ability indicators. Regression analyses to ability profile subgroup membership and the ir social skills and problem behaviors showed that functional ability profile subgroup membership had a small to moderate social skills and problem behaviors. Statistically significant and noteworthy differences in social skills and problem behavior scores were identified between subgroups. E xamin ation of the individual and collective contributions of functional ability profile subgroup membership and children disability category as correlates of their social skills and problem behaviors showed that functional ability profile subgroup membership provided somewhat more explanatory power than the use of disability categories. When disability category and functional ability profile subgroup membership were included together in the regression models, explanatory power was larger than when either variable was used alone. Regression analyses that included an interaction term were used to explore whether select contextual variables moderated relationships between functional ability Moderation analyses revealed that only one variable moderated the relationship between functional abil ity profile subgroup membership and social skills. This variable was the number of child activities. Three variables moderated the relationship between functional ability profile subgroup membership and problem behaviors. These variables were race/ethni city, number of regular child activities, and neighborhood safety.

PAGE 203

203 The secondary analyses conducted in the present study were exploratory. Findings generally supported empirically hypothesized relationships shown in Figures 3 1 to 3 3. The present study offers preliminary evidence to suggest there are subgroups of young children with disabilities served under IDEA who share similar functional ability profiles and these profiles might be examined further in future studies. Chapter 5 includes additional di scussion of primary findings and implications from the present study.

PAGE 204

204 Table 4 1 Proportion of s ample for each r esponse c ategory for PEELS Disability Severity Index No limitations (1) Mild limitations (2) Moderate l imitations (3) Severe l imitations (4) Hearing 94.3 1.0 4.3 0.4 Vision 91.1 6.2 2.0 0.7 Overall Health 61.8 22.1 10.8 5.3 Use of Arms 84.1 12.2 2.9 0.8 Use of Hands 60.3 28.9 10.2 0.6 Use of Legs 83.1 12.6 3.7 0.7 Cognition 10.6 43.5 32.0 13.9 Communicate with O thers 29.3 9.2 44.8 16.7 Understanding 51.0 35.4 12.9 0.7 Regulation of Attention 29.2 42.4 n/a 28.4 Regulation of Feeling and Emotions 81.6 12.8 n/a 5.5 Regulation of Activity Level 31.6 33.9 n/a 34.6 Motivation 22.2 42.1 n/a 35.7 Social Skills 56.3 7.0 27.1 9.6 Inappropriate or U nusual B ehavior 60.6 26.0 10.5 3.0 Note N = 2 870 Estimates weighted using sampling weights n/a indicates that only 3 response categories (1, 2, and 4) were used for the functional ability item in the PEELS Disability Severity Index.

PAGE 205

205 Table 4 2 P roportion of s ample for each r esponse c ategory used in latent class analysis No limitations (1) Mild limitations (2) Moderate l imitations (3) Severe l imitations (4) Hearing 94.3 1.0 4.7 Vision 91.1 6.2 2.7 Overall Health 61.8 22.1 10.8 5.3 Use of Arms 84.1 12.2 3.7 Use of Hands 60.3 28.9 10.8 Use of Legs 83.1 12.6 4.4 Cognition 10.6 43.5 32.0 13.9 Communicate with O thers 29.3 9.2 44.8 16.7 Understanding 51.0 35.4 12.9 0.7 Regulation of Attention 29.2 42.4 n/a 28.4 Regulation of Feeling and Emotions 81.6 12.8 n/a 5.5 Regulation of Activity Level 31.6 33.9 n/a 34.6 Motivation 22.2 42.1 n/a 35.7 Social Skills 56.3 7.0 27.1 9.6 Inappropriate or Unusual B ehavior 60.6 26.0 10.5 3.0 Note N = 2870 Estimates weighted using Taylor method. n/a indicates that only 3 response categories (1, 2, and 4) were used for the functional ability item in the PEELS Disability Severity Index. indicates that cases in the this response category were re coded to conduct analyses (i.e., response category 3 was re coded to 4).

PAGE 206

206 Table 4 3 Model f it s tatistics for t wo through s even c lass model s # of classes LL # of p arameters BIC Entropy # of r eplications 2 33955.832 73 68492.920 0.843 40 3 33231.445 1 10 67338.755 0.805 40 4 32808.467 147 66787.410 0.842 3 7 5 32577.949 184 66620.983 0.835 13 a 6 32420.620 221 66600.935 0.828 11 7 32275.788 258 66605.880 0.784 6 Note LL= log likelihood BIC = Bayesian Information Criterion Replications based on sets of 500 start values with display set to show top 40 values. a 13 r e plications 32577.949 and 2 1 replications 32577.955 Estimates for the models with the two largest log likelihood were quite similar with differences in the third decimal place or smaller.

PAGE 207

207 Table 4 4 Model implied means (standard deviations) for 5 class model s Profile 1 ( n = 1 40 ) Profile 2 ( n = 4 2 0 ) Profile 3 ( n = 200 ) Profile 4 ( n = 950 ) Profile 5 ( n = 1150 ) Communication 3.53 (0.45) 3.36 (0.47) 2.19 (1.28) 2.51 (0.92) 2.08 (1.11) Understanding 2.64 (0.40) 2.56 (0.34) 1.44 (0.33) 1.74 (0.31) 1.11 (0.10) Cognition 3.82 (0.17) 3.36 (0.62) 2.50 (0.62) 2.61 (0.47) 1.90 (0.24) Overall Health 2.58 (1.12) 1.97 (1.00) 2.12 (1.12) 1.53 (0.62) 1.29 (0.39) Use of Arms 2.93 (0.99) 1.27 (0.25) 2.02 (1.01) 1.06 (0.06) 1.00 (0.00) Use of Hands 3.68 (0.53) 2.30 (1.10) 2.55 (1.02) 1.41 (0.35) 1.08 (0.08) Use of Legs 2.75 (1.03) 1.28 (0.24) 2.30 (1.33) 1.05 (0.05) 1.03 (0.06) Social Skills 3.26 (0.87) 3.03 (0.92) 1.76 (0.87) 1.99 (1.08) 1.27 (0.43) Behavior 2.29 (1.02) 2.53 (0.69) 1.27 (0.26) 1.64 (0.48) 1.09 (0.11) Reg of Activity Lev. 2.79 (1.49) 3.25 (1.27) 2.09 (1.43) 2.84 (1.45) 1.67 (0.78) Reg of Attention 3.12 (1.27) 3.18 (1.26) 2.17 (1.36) 2.57 (1.15) 1.60 (0.64) Motivation 3.27 (1.32) 3.21 (1.20) 2.38 (1.33) 2.67 (1.34) 2.00 (1.03) Reg of Emotions 1.37 (0.58) 1.82 (1.23) 1.39 (0.84) 1.29 (0.50) 1.07 (0.09) Hearing 1.26 (0.68) 1.10 (0.27) 1.18 (0.51) 1.18 (0.47) 1.13 (0.37) Vision 2.24 (1.66) 1.10 (0.17) 1.42 (0.69) 1.10 (0.17) 1.04 (0.08) Note N = 2870, however, subgroup numbers do not sum to total due to rounding error Estimates weighted using sampling weights Bolded items range from 2.5 and above to denote ratings associated with moderate to severe limitations Italicized items range from 1. 5 to 2.49 to de note ratings with mild to moderate limitations

PAGE 208

208 Table 4 5 Average latent class posterior probabilities for most likely class mem bership Probability of b eing a ssigned to p rofile Actual a ssigned p rofile Profile 1 Profile 2 Profile 3 Profile 4 Profile 5 Profile 1 .906 .057 .036 .000 .000 Profile 2 .021 .891 .012 .076 .000 Profile 3 .026 .009 .881 .066 .019 Profile 4 .044 .000 .029 .860 .066 Profile 5 .000 .000 .005 .071 .923 Note N = 2870 Estimates weighted using sampling weights

PAGE 209

209 Table 4 6 D emographic and d escriptive i nformation for c hildren within e ach p rofile Profile 1 ( 5% of sample ) Profile 2 ( 15% of sample ) Profile 3 ( 7% of sample ) Profile 4 ( 33% of sample ) Profile 5 ( 40% of sample ) Child gender (%) Male 74 78 58 74 67 Female 26 22 42 26 33 Age distributions (%) 3 years of age 19 23 20 25 18 4 years of age 38 43 41 33 40 5 years of age 43 34 39 42 41 Child race/ethnicity (%) Caucasian/ Non Hispanic 60 55 68 55 71 Hispanic 15 24 20 20 13 African American 14 11 8 13 8 American Indian or Alaskan Native 0 # # # # Asian, Native Hawaiian, or Pacific Islander 0 # # # # Multi racial 11 9 4 10 6 English as a second language (%) 24 25 21 23 16 IFSP before 3 (%) 82 50 72 36 26 Mean age in months (S D) 55 (9) 54 (8) 55 (9) 55 (9) 56 (8) Mean no. weeks premature (S D) 3.9 ( 5.6 ) 1.6 ( 3.1 ) 3.3 ( 4.7 ) 1.5 (3.3 ) .9 ( 2.5 ) Mean birth weight in grams (S D) 2665 (1134) 3147 (765) 2722 (1049) 3147 (822) 3289 (709) Note N = 2870 Estimates weighted using sampling weights Al l variables reported by parent or guardian IFSP = i ndividualized f amily s ervice p lan

PAGE 210

210 Table 4 7 D e mographic and d escriptive i nformation for f amilies within e ach p rofile Profile 1 ( 5% of sample ) Profile 2 ( 15% of sample ) Profile 3 ( 7% of sample ) Profile 4 ( 33% of sample ) Profile 5 ( 40% of sample ) Children live with respondent (%) 100 <100 100 <100 <100 Respondent Role (%) Biological m other 88 84 93 84 88 Adoptive mother # 5 3 4 3 Biological father 3 3 # 6 5 Other role (e.g., grandparent, foster) 7 8 # 6 4 Single parent family (%) 17 38 32 32 23 Respondent education level (%) Less than high school or GED 16 19 17 17 9 High School or GED 22 31 20 36 28 Some college or 2 year degree 45 29 33 26 33 Four year degree or higher 17 21 30 21 29 Family income bracket (%) Less than $10,000 19 18 12 11 8 $10,000 $20,000 14 17 20 16 11 $20,000 $30,000 6 11 13 21 17 $30,000 $40,000 8 15 9 13 12 $40,000 $50,000 11 12 7 8 13 $50,000 and above 42 27 39 30 39 Parent ratings Av g rating of neighborhood safety Safe Safe Safe Safe Safe Avg. rating of satisfaction with special education services Satisfied Satisfied Satisfied Satisfied Satisfied Note N = 2870 Estimates weighted using sampling weights All variables reported by parent or guardian.

PAGE 211

211 Table 4 8 D escriptive i nformation about c hild and f amily a ctivities within e ach p rofile Profile 1 ( 5% of sample ) Profile 2 ( 15% of sample ) Profile 3 ( 7% of sample ) Profile 4 ( 33% of sample ) Profile 5 ( 40% of sample ) Child Activities Children who regularly participate in group activities (%) a 36 47 46 45 55 Child Family Activities Family member reads to child (%) 0 2 times a week 26 28 19 25 16 3 6 times a week 26 29 34 35 41 Everyday 47 43 48 40 43 Child Activities Mean no. extra curricular activities children have ever participated in ( S D ) b <1 (1) <1 (1) 1 (1) 1 (1) 1 (1) Mean no. of regular group activities c 1.2 (.5 ) 1.3 ( .7 ) 1.6 ( .7 ) 1 .3 (.6 ) 1 .4 ( .7 ) Family School Activities school (S D) d 3 (2) 3 (2) 3 (2) 3 (2) 4 (2) Child Family Activities Mean no. of different family activities (S D) e 5 (1) 5 (1) 5 (1) 5 ( 2 ) 6 (1) Mean no. of family meals each week (S D) 5 ( 2 ) 5 ( 2 ) 5 ( 2 ) 5 ( 2 ) 5 ( 2 ) Note N = 2870 Estimates weighted using sampling weights Al l variables reported by parent or guardian a Represents the percentage of parents who indicated the child participate s in group activities on a monthly basis. b Represents the sum of responses to whether child has ever participat ed in 7 possible extra curricular activities dance lessons, athleti cs, clubs, music lessons, drama classes, art classes, or other performance activities. c Represents the sum of responses to whether child participat es monthl y in 6 possible group activities play group, story hour, religious group lessons, athletic team, or recreational clubs ; for those who indicated yes to any group activity d general school meetings, attend class events, volunteer in c lassroom, field trips, parent teacher conferences, policy council or similar, or fundraising e Represents the sum of responses to whether someone in the family and child go to 7 possible locations grocery store, shopping mall, restaurant, public park/pla yground, place of worship, library, or movie.

PAGE 212

212 Table 4 9 D escriptive i nformation about p rograms or s chools for c hildren and f amilies within e ach p rofile Profile 1 ( 5% of sample ) Profile 2 ( 15% of sample ) Profile 3 ( 7% of sample ) Profile 4 ( 33% of sample ) Profile 5 ( 40% of sample ) Percentage of school population from low income families (%) Less than 25% 37 32 32 28 40 25 50% 20 24 36 32 27 50 75% 29 13 8 13 12 More than 75% 14 31 24 27 21 Program quality (%) Preschool accredited ( n = 1 030 ) 42 47 62 48 50 School met NCLB standards ( n = 1 15 0) 89 92 83 91 93 Program supports social interactions among children with and without disabilities (%) 69 54 64 60 53 Mean no. children in class room ( STD ) With IEP 8 ( 4 ) 8 ( 4 ) 7 ( 3 ) 8 ( 4 ) 6 ( 4 ) Without IEP 4 ( 6 ) 4 (6) 6 ( 6 ) 6 (7) 8 ( 8 ) Ch ildren have IEP goal focused on (%) School readiness 19 48 40 48 27 Pre academic 6 12 16 9 5 Social 20 36 17 32 19 Behavior 23 32 8 22 6 Adaptive 31 22 16 10 3 Communication 66 71 56 74 83 Fine motor 34 21 35 17 10 Gross motor 38 6 38 4 4 Note N = 2870 for variables completed by parents or guardians N = 2180 for variables completed by teachers and program administrators or principals Estimates weighted using sampling weights NCLB = No Child Left Behind Act; IEP = Individualized Education Program

PAGE 213

213 Table 4 1 0 Comparative demographic information for samples used in analyses Profile 1 Profile 2 Profile 3 Profile 4 Profile 5 Total Sample Sample with p arent c hild w eights; (N = 2870) Percent of Sample (%) 5 15 7 33 40 100 Mean age in months 55 mo. 54 mo. 55 mo. 55 mo. 56 mo. 56 mo. Child gender (%) Male 74 78 58 74 67 71 Female 26 22 42 26 33 29 Child race/ethnicity (%) Caucasian/ Non Hispanic 60 55 68 55 71 62 Hispanic 15 24 20 20 13 18 African American 14 11 8 13 8 10 Other 11 10 4 12 8 10 Sample with p arent c hild t eacher w eights; (N = 2180) Percent of Sample (%) 6 15 7 32 40 100 Mean age in months 5 6 mo. 54 mo. 55 mo. 55 mo. 56 mo. 55 mo. Child gender (%) Male 74 78 56 74 66 70 Female 26 22 44 26 34 30 Child race/ethnicity (%) Caucasian/ Non Hispanic 58 58 65 56 72 63 Hispanic 15 21 21 19 12 17 African American 14 9 9 12 7 10 Other 12 11 4 13 9 10 Sample with PKBS 2 s cores; (N = 2090) Percent of Sample (%) 6 14 7 32 40 100 Mean age in months 5 6 mo. 54 mo. 55 mo. 55 mo. 56 mo. 56 mo. Child gender (%) Male 59 89 52 74 66 70 Female 41 11 48 26 34 30 Child race/ethnicity (%) C aucasian/ Non Hispanic 61 59 65 58 73 64 Hispanic 12 19 21 18 12 15 African American 15 9 9 12 6 10 Other 13 12 5 13 9 10 Note Estimates weighted using sampling weights

PAGE 214

214 Table 4 1 1 PKBS 2 c omposite s cores for e ach profile M St. Error CI Social skills composite score Profile 1 ( n = 130) 69.99 3.22 63.55, 76.43 Profile 2 (n = 29 0 ) 80.28 1.95 76.38, 84.18 Profile 3 ( n = 150) 95.28 1.94 91.40, 99.16 Profile 4 ( n = 670) 91.61 0.95 89.7 0 93.51 Profile 5 ( n = 8 4 0) 102.08 0.89 100.30, 103.86 Problem behavior s composite score Profile 1 ( n = 130) 100.61 2.12 96.37, 104.85 Profile 2 (n = 290) 106.45 0.96 104.53, 108.37 Profile 3 ( n = 150) 95.71 1.31 93.09, 98.33 Profile 4 ( n = 670) 100.81 0.92 98.97, 102.65 Profile 5 ( n = 830) 92.43 0.68 91.07, 93.79 Note N = 2090 however, profile numbers do not sum to total due to rounding error Estimates weighted using sampling weights

PAGE 215

215 Table 4 1 2 PKBS 2 s ub s cale s cores for e ach profile M S t. Error CI Social c ooperation Profile 1 75.13 3.9 0 67.33, 82.93 Profile 2 87.66 1.73 84.2 0 91.12 Profile 3 100.43 1.85 96.73, 104.13 Profile 4 95.77 0.93 93.91, 97.63 Profile 5 105.58 0.79 104.00, 107.16 Social i nteraction Profile 1 66.49 3.22 60.05, 72.93 Profile 2 78.15 1.88 74.39, 81.91 Profile 3 91.12 2.14 86.84, 95.40 Profile 4 88.98 0.99 87.00, 90.96 Profile 5 98.06 0.95 96.16, 99.96 Social i ndependence Profile 1 75.45 3.18 69.09, 81.81 Profile 2 81.96 1.85 78.26, 85.66 Profile 3 96.09 1.7 0 92.69, 99.49 Profile 4 93.07 0.92 91.23, 94.91 Profile 5 101.53 0.78 99.97, 103.09 Externalizing b ehaviors Profile 1 98.83 2.07 94.69, 102.97 Profile 2 104.01 1.06 101.89, 106.13 Profile 3 93.58 1.25 91.08, 96.08 Profile 4 100.9 0 0.99 98.92, 102.88 Profile 5 91.94 0.62 90.70, 93.18 Internalizing b ehaviors Profile 1 102.35 2.12 98.11, 106.59 Profile 2 107.76 1.07 105.62, 109.9 Profile 3 98.84 1.35 96.14, 101.54 Profile 4 100.69 0.8 0 99.09, 102.29 Profile 5 94.38 0.71 92.96, 95.80 Note N = 2090 Estimates weighted using sampling weights

PAGE 216

216 Table 4 1 3 PKBS 2 c omposite s cores s tandardized m ean d ifference e ffect s izes Profile 2 Profile 3 Profile 4 Profile 5 Social skills composite score Profile 1 0. 5 5 ** 1. 3 5 *** 1. 1 5 *** 1. 71*** Profile 2 -. 80 *** 60 *** 1. 1 6 *** Profile 3 --0. 20 0.3 6 ** Profile 4 ---0.5 6 *** Problem behavior s composite score Profile 1 0.4 1* 0.3 5 0.0 1 0.58 *** Profile 2 -0.7 6 *** 0. 40 *** 0.9 9 *** Profile 3 --0.36 *** 0.2 3* Profile 4 ---0. 59*** Note N = 2090 Estimates weighted using sampling weights Effect sizes estimated for row vs column For social skills, a positive effect size is associated with more social skills (positive outcome) and a negative effect is associated with less social skills For problem behavior s a positive effect size is associated with more problem behavior s ( negative outcome) and a negative effect is associated with less problem behavio rs Effect sizes were estimated by subtracting the means of column group from the row group and dividing the difference by the square root of the residual variance ***p < .001 p < .01 p < .05

PAGE 217

217 Table 4 1 4 P ercentage of total sample by profile a nd disability category SL I DD LI AUT MR LD EBD Missing Total Profile 1 # # # # # 0 # 0 5 Profile 2 3.7 5.5 # 3.4 # # # # 14.7 Profile 3 # 2.5 2.5 # # # # # 6.8 Profile 4 14.2 10.7 2.6 2.1 # # # # 33.3 Profile 5 29.1 5.6 # # # # # 2.4 40.2 Total 48.3 25.6 9.2 6.2 3.5 2.4 # 3.7 100 Note N = 2870 Estimates weighted using sampling weights SLI = speech or language impairments ; DD = developmental disability; LI= low incidence disability ; AUT = autism; MR = mental retardation ; LD = learning disability; EBD = emotional behavioral disturbance ; and Missing = disability category not identified.

PAGE 218

218 Table 4 1 5 P ercentage of profile s ample by d isability c ategory Profile 1 ( 5% of sample ) Profile 2 ( 15% of sample ) Profile 3 ( 7% of sample ) Profile 4 ( 33% of sample ) Profile 5 ( 40% of sample ) SL I 5 25 15 43 72 DD 26 38 37 32 14 LI 35 6 37 8 4 AUT 8 23 # 6 # MR 26 4 3 3 # LD # # # 4 # EBD # # # # # MISSING # # 4 3 6 Total 100 100 100 100 100 Note N = 2870 Estimates weighted using sampling weights. SLI = speech or language impairments ; DD = developmental disability; LI= low incidence disability; AUT = autism; MR = mental retardation; LD = learning disability; EBD = emotional behavioral disturbance ; and Missing = disability category not identified. In this table the # symbol is used to indicate cells with less than 3% of the profile sample, including cells with zero cases.

PAGE 219

219 Table 4 16 P ercentage of total s ample for disability categories with in the low incidence d isability c ategory by disability category and profile Profile 1 Profile 2 Profile 3 Profile 4 Profile 5 Total H earing impairment # # # 6 # 9 D eaf/blind # # # # # # D eafness # # # # # 4 M ultiple disabilities 8 4 6 # # 19 Orthopedic impairments 4 # 18 # 4 27 Other health impairments 7 5 11 6 4 32 T raumatic brain injury # # # # # 4 V isual impairment # # # # 3 4 Total 20 10 37 18 15 100 Note N = 160 Estimates weighted using sampling weights Low incidence makes up 9.2 % of the total sample In this table the # symbol is used to indicate cells with less than 3% of the profile sample, including cells with zero cases.

PAGE 220

220 Table 4 17 Variance a ccounted w ith the a ddition of e xplanatory v ariables: R 2 PKBS 2 c omposite score R 2 Model Explanatory v ariable Social s kills Problem b ehavior s 1 Disability category .165 .108 2 Subgroup membership .200 .115 3 Subgroup membership and disability category .252 .159 Note N = 2090 Estimates weighted using sampling weights

PAGE 221

221 Table 4 18 Hold out a nalyses for R 2 PKBS 2 c omposite score R 2 Model Explanatory v ariable Social s kills Problem b ehavior s 1 Disability category .1 1 4 0 74 2 Subgroup membership 201 0 9 0 3 Subgroup membership and disability category 2 31 1 04 Note N = 1090 Estimates weighted using sampling weights. Th ese analyses used a subset of the sample to examine the variance explained after excluding cases with speech or language impairments indicated as the disability category

PAGE 222

222 Table 4 19 Modified categorical coding for moderation analyses Variable Original c oding Coding for m oderation Child race ethnicity 1. Caucasian / Non Hispanic 2. Hispanic 3. African American 4. American Indian or Alaskan Nativ e 5. Asi an, Native Hawaiian, or Pacific Islander 6. Multi racial 1. Caucasian / Non Hispanic 2. Hispanic 3. African American 4. Other Parent education 1. L ess than High School with no GED 2. High School diploma or GED 3. S ome college/ post secondary vocational course 4. 2 or 3 year college degree or vocational school diploma 5. 4 year college degree 6. S ome graduate work/no graduate degree 7. G raduate degree 1. L ess than High School with no GED 2. High School diploma or GED 3. S ome college/ post secondary vocational course 4. 2 or 3 year college degree or vocational school diploma 5. 4 year college degree with or without some graduate work (no graduate degree) 6. G raduate degree Read to child each week 1. N ever 2. O nce or twice 3. 3 to 6 times 4. E very day 1. 0 to 2 times 2. 3 to 6 times 3. E very day

PAGE 223

223 Table 4 2 0 Moderation of difference s among profiles on social skills by non malleable child factors and contextual factors Prediction Moderation Predicts on its own (p value) Predicts with profile (p value) df F value p value p value criterion Moderation Non Malleable Child Factors Race/ethnicity Yes (.012) No (.103) 12,61 1.29 .2454 a .017 No Age Yes (<.001) Yes (<.001) 4,61 1.72 .1577 .025 No Gender Yes (.021) No (.096) 4,61 .15 .9637 .05 No Contextual Factors Family Characteristic Parent e ducation No (. 472 ) No (.2 17 ) 2 0 ,61 1. 58 0894 .017 No Marital s tatus No (.198) No (.395) 4,61 .75 .5597 .025 No Family i ncome No (.367) No (.432) 4,61 .74 .5665 .05 No Parent Child Interaction Child a ctivities (#) Yes (.004) Yes (.006) 4,61 3.93 .0066 .007 Yes Parent child activities (#) Yes (.031) Yes (.008) 4,61 1.76 .1490 .008 No Regular c hild a ctivities (#) Yes (.019) Yes (.011) 4,61 1.66 .1715 .01 No Child p articipates in a ctivities r egularly (yes/no) Yes (.017) Yes (.036) 4,61 .89 .4737 .0125 No Parent school activities (#) No (.137) No (.128) 4,61 1.02 .4048 .017 No Meals per week No (.231) No (.535) 4,61 1.00 .4156 .025 No Read to child No (. 467 ) No (. 928 ) 8 ,61 39 924 0 .05 No Environmental School community income No (.339) No (.292) 12,61 1.50 .1494 .017 No Program supports social interaction No (.767) No (.398) 4,61 1.64 .1761 .025 No Neigh borhood s afety Yes (.011) No (.095) 8,61 1.07 .3947 .05 No Note N = 2090 Estimates weighted using sampling weights The extent to which each variable was associated with social skills, referred to as prediction, was determined based on an alpha .05 For moderation analyses of interest related to child, family, and contextual factors, the p value criterion for determining moderation was based o n Bonferonni Holm Criterion within each group of variables. a p calculated value reflect s moderation with four categories of race/ethnicity M oderation was also examined for 6 categories of race/ethnicity Moderation for race/ethnicity for 6 categories resulted in a p calculated value of 0655

PAGE 224

224 Table 4 2 1 Moderation of difference s among profiles on problem behavior s by non malleable child factors and contextual factors Prediction Moderation Predicts on its own (p value) Predicts with profile (p value) df F value p value p value criterion Moderation Non Malleable Child Factors Race/ethnicity Yes (<.001) No (.122) 12,61 2.72 .0053 a .017 Yes a Age No (.361) No (.561) 4,61 1.63 .1784 .025 No Gender Yes (.012) Yes (.008) 4,61 .32 .8611 .05 No Contextual Factors Family Characteristics Parent e ducation No (. 112 ) Yes ( .006) 2 0 ,61 1.97 .0 227 .017 No Family i ncome Yes (.002) N o (.100) 4,61 2.28 .0706 .025 No Marital s tatus Yes (<.001) Yes (.054) 4,61 1.19 .3228 .05 No Parent Child Interaction Regular c hild a ctivities (#) No (.259) No (.164) 4,61 5.07 .0014 .007 Yes Parent child activities (#) No (.355) No (.784) 4,61 4.07 .0054 .008 Yes Child p articipates in a ctivities r egularly (yes/no) No (.481) No (.148) 4,61 2.85 .0313 .01 No Child a ctivities (#) No (.061) No (.341) 4,61 2.37 .0622 .0125 No Read to child No (. 409 ) No (.470) 8 ,61 1. 70 1177 .017 No Meals per week No (.372) No (.516) 4,61 1.27 .2910 .025 No Parent school activities (#) No (.193) No (.107) 4,61 .77 .5492 .05 No Environmental Factors Neigh borhood s afety No (.139) No (.062) 8,61 3.14 .0049 .017 Yes School community income No (.097) Yes (.040) 12,61 1.83 .0636 .025 No Program supports social interaction No (.162) No (.289) 4,61 .42 .7954 .05 No Note N = 2090 Estimates weighted using sampling weights The extent to which each variable was associated with social skills, referred to as prediction, was determined based on an alpha .05. For moderation analyses of interest related to child, family, and contextual factors, the p value criterion for determining moderation was based on Bonferonni Holm Criterion within each group of variables. a p calculated value reflect s moderation with 4 categories of race/ethnicity M oderation was also examined for 6 categories of race/ethnicity Moderation for race/ethnicity for 6 categories resulted in a p calculated value of .00 02

PAGE 225

225 Table 4 2 2 Mean problem behavior s standard score s by race/ethnicity and profile Profile 1 Profile 2 Profile 3 Profile 4 Profile 5 Caucasian/N on Hispanic 97 107 96 101 92 Hispanic 106 104 95 99 91 African American 107 106 96 107 96 Other a 102 106 99 97 93 Note N = 2090 Estimates weighted using sampling weights Significant moderation interaction between subgroup membership and problem behaviors by race/ethnicit y identified Differences between individual race/ethnicity categories across profiles not examined for significance due to the large number of comparisons. a Other = A merican Indian or Alaskan Native Asian, Native Hawaiian, or Pacific Islander and Multi racial.

PAGE 226

226 Table 4 2 3 Mean problem behavior s standard score s by parent rating of neighborhood safety and profile Profile 1 Profile 2 Profile 3 Profile 4 Profile 5 Not s afe 107 107 99 99 97 Safe 97 108 98 103 91 Very safe 101 105 94 100 92 Note N = 2090 Estimates weighted using sampling weights Significant moderation interaction between subgroup membership and problem behaviors by neighborhood safety identified Differences between individual neighborhood safety categories across profiles not examined for significance due to the large number of comparisons.

PAGE 227

227 Figure 4 1 Profile means across functional ability variables Functional ability score : (1) no limitation, (2) mild limitation, (3) moderate limitation, and (4) severe limitation.

PAGE 228

228 Social Skills Composite Score and Subscales Problem Behaviors Composite Score and Subscales Figure 4 2 Mean PKBS 2 score for each profile. PKBS 2 standardized to M = 100, SD = 15

PAGE 229

229 Figure 4 3 Moderation of differences on social skills standard scores between profiles 4 and 5 by child activities

PAGE 230

230 Figure 4 4 Moderation of differences on problem behaviors standard scores between profiles 1 and 3 and between profiles 1 and 5 by parent child activities. Visual inspection of the graphed data revealed that outliers in profile 1 influenced the slope of the line for profile 1, and thus the moderation effect is likely due to these outliers Outliers were a small cluster of cases in Profile 1 with 0 parent child activities and problem behavior scores at or near 80.

PAGE 231

231 Figure 4 5 Moderation of difference s on problem behaviors standard scores between profiles 1 and 5 by regular child activities.

PAGE 232

232 CHAPTER 5 DISCUSSION The purpose of the present study was to explore relationships between empirically derived subgroups of children with similar functional a bility pr ofiles and their social competence outcomes Study research questions were addressed by conducting secondary analys e s using cross sectional data from the Pre Elementary Education Longitudinal Study (PEELS) data set. The PEELS data set includes i nformation about young children with disabilities who were receiving early childhood special education services and supports under the Individuals with Disabilities Education Improvement Act (IDEA) All secondary a nalyses were conducted using appropriate weights so reported findings are nationally representative of young children with disabilities receiving early childhood special education services and supports in the fall of 2003 L atent class analyses were conducted to deter mine whether a set of 15 functional ability variables included in the PEELS data set would be useful for empirically deriving distinct and interpretable subgroups of children who share similar functional ability profiles. Following the selection of the 5 class model, regression analyses were used to explore relationship s and their social skills and problem behaviors Additional regression analyses were conducted to examine the explanatory p ower of functional ability profile subgroup membership and IDEA disability category with the variables used individually and together, as correlates of social skills and problem behaviors. R egression analyses that included an interaction term were used to examine whether select non malleable child factors and contextual factors moderate d relationships between

PAGE 233

233 behaviors. The purpose of this chapter is to int erpret findings, discuss implications of the findings, and provide recommendations for future research. Findings associated with each study research question are integrated within each section. Interpretation of Study Findings An underlying assumption of the present study was that 15 functional ability variables included in the PEELS data set could be used to create distinct and interpretable latent classes that represent subgroups of children with similar functional ability pro files. This assumption was supported by theory and previous research on the functional abilities of young children (e.g., Simeonsson, 2009; Simeonsson, Bailey, Smith, & Buysse, 1995; WHO, 2007) but had not been examined using a large scale data set f ocuse d on young children with disabilities. Overall, the hypotheses and relationships modeled as part of each research question were empirically supported including the identification of distinct and interpretable latent classes The sequence of research que stions help ed unpack the relationships amon g child functioning, disability categor y non malleable child factors, and contextual factors the International Classification of Functioning, Disabil ity, and Health for Children and Youth ( ICF CY; WHO, 2007) as a guiding framework. Characteristics of Latent Class Subgroups Based on Functional Ability Profiles Given the exploratory na ture of the latent class analyse s it was not known a priori which dimensions of function would lead to distinctions between subgroup s (latent classe s) The 5 class model was selected for interpretation because it was supported

PAGE 234

234 by model fit indices and it provided logical and interpretable distinctions between subgroups that highlighted salient patterns of functioning within and across subgroup s Each subgroup represent s a group of children with a similar functional ability prof ile and the functional ability profile for e ach subgroup is different than the profiles for other subgroups Across the 15 functional ability variables, t he dimensions of function that were identified in the profiles related to the (a) severity of the limitations (e.g., mild, moderate, or severe li mitations), (b) number of limitations, and (c) type of limitations (e.g., limitations associated with common clusters of variable s). S ubstantive interpretations of the profiles focused on examining patterns of functioning, including shared features of func tional ability variables within a subgroup and the distinguishing features of these variables across subgroups. To aid interpretation, each subgroup was labeled with a profile number and limi t ations across functional ability variables w ere examined P rof ile numbers refer to specific subgroups and the term profile is used when describing the functional abilities for these subgroups. The terms functional ability profile subgroup membership or subgroup membership are used when referring to the subgroups (latent classes) as a categorical variable. S everity of limitations for each subgroup was examined by inspecting the model implied mean score for each func tional ability variable. Model implied means were calcula ted from the proportion of cases associated with the ordinal response categories for each functional ability variable T he PEELS Disability Severity Index used a 4 point scale and r esponse categories were (1) normal or typical functioning, (2) mild limita tion in functioning, (3) moderate limitation in functioning, and ( 4 ) severe limitation in functioning To interpret the severity of limitations model implied means were grouped

PAGE 235

235 into ranges to account for variation among children in a subgroup when describing the profiles In the present study, m oderate to severe limitations were associated with mean scores from 2.5 and above, mild to moderate limit ations were mean scores from 1. 5 to 2.49, and no to mild limitations were mean scores from 1. 49 and be low. Findings from the latent class analyses suggest that five distinct and interpretable subgroups with similar functional ability profiles can be used to characterize the nationally representative sample of young children with disabilities included in th e PEELS data set Profiles were interpreted in terms of patterns of functional ability limitations across the 15 functional ability variables. The patterns of functioning and distinctions between subgroups w ere logical in relation to what is known about variations in the functional abilities of y oung children with disabilities ( Simeonsson, 2009; Simeonsson, Bailey, Smith, & Buysse, 1995 ). The profiles provide information about how various aspects of function co occur in children with disabilities. Profiles were examined individually and relative to other profiles. Profile 1 was comprised of 5% of children in the PEELS data set and the fu nctional ability profile indicated that these children had limitations on 13 of 15 functional ability variables including moderate to severe limitations related to communication/cognition, motor function social competence, self regulation and vision Pro file 2 was comprised of 15% of children in the PEELS data set and the functional ability profile indicated that these children had limitations on 1 1 of 15 functional ability variables including moderate to severe limitations related to communication/cognit ion, social competence, and self regulation Profile 3 was comprised of 7% of children in the PEELS data set and the functional ability profile indicated that these children had limitations on 10 of 15

PAGE 236

236 functional ability variables with mild to moderate li mitations that related to communication/cognition, motor function, social competence, and self regulation. Profile 4 was comprised of 33% of children in the PEELS data set and the functional ability profile indicated that these children had limitations on 9 of 15 functional ability variables including mild to moderate limitations that related to communication/cognition, social competence, and self regulation. Profile 5 was comprised of 40% of children in the PEELS data set and the functional ability profi le indicated that these children had limitations on 5 of 15 functional ability variables including mild limitations that related to communication/cognition and self regulation. P rofile patterns in relation to functional ability variables In this section, patterns in the functional ability p rofiles are d iscussed in relation to severity of limitations a s well as individual and clusters of functional ability variables associated with each profile As shown in Table 4 4, t he severity of limitations was related to the mean ratings for each functional ability variable. Several variables had the same or similar severity of limitation ratings across profiles. For example, the s e verity of limitations for communication and cogn ition were consistent with the overall severity of limitations identified for each profile such that t hese ratings were not notably different between Profile 1 and Profile 2 (children with moderate to severe li mitations) or between Profile 3 and Profile 4 (children with mild to moderate limitations). however, was consistently 1 mean rating point lower (indicating fewer limitations or more ability) than the communication and cognition variables ac ross profiles This finding was unexpected, given that communication and understanding or cognition and understanding are typically considered related skills. One explanation for this finding is

PAGE 237

237 that key words used to define these variables might encompa ss multiple functional skills or define functional skills in ways that alter how parents rated these variables. For example, the variable for communication was defined by how the child communicates (i.e., expresses their needs) and the extent to which oth message This definition emphasizes expressive communication and clarity of speech or expression T he variable for cognition was defined by how the child learns, thinks, and solves problems. T he variable for understanding was defined by how well a child understands verbal messages from others which emphasizes receptive communication The consistently higher ratings (i.e., more limitations) for cognition and communication, when compared to the ratings for understanding, might be explained by the former two variables including several functional skills within one functional ability item. Alternatively, lower ratings (i.e., more ability) for understanding, in comparison to ratings for variables such as cognition and communication, might relate to parents functional abilities in communication and cognition For example, a parent might say, can not I dentifying distinct patterns across functional ability profiles or identif ying how functional ability variables cluster together within a profile was part of the rational e for using person oriented techniques instead of variable oriented techniques. T he complex to understanding relationships between than using individual functional ability va riables in a he terogeneous group of children (Haapasal o, Tremblay, Boulerice, & Vitaro 2000).

PAGE 238

238 S ome of the distinct patterns between profiles were based on clusters of variables associated with similar functional ability limitations, but manifested at di fferent levels of severity. One cluster of variables was limitations related to overall health, use of arms, use of hands, and use of legs. Profiles 1 and 3 were associated with limi t ations on all of these variables. Th is cluster of variables was viewed as functional abilities related to physical aspects of the body that might affect motor function in the context of activities and social participation. Another cluster of variables was limitations related to regulation of attention, regulation of a ctivity, and motivation Profiles 1, 2, 3, 4 and 5 were associated with limitations on these variables however, Profile s 2 and 4 had higher m ean ratings (i.e., more limita tions) on these variables compared to Profiles 1 and 3, respectively This cluster of variables was viewed as functional abilities t hat might affect self regulation in the context of activities and social participation By focusing on clusters of variables with limitations it was easier to interpret the ways in which these variables mi participation in ongoing activities (e.g., motor function, self regulation) compared to a focus on individual variables. Despite the clustering of some variables that lead to interpretable and logical distinctions between groups, othe r variables did not contribute to the distinctions between profiles. For example, the variable for regulation of emotions did not generally distinguish differences between profiles nor was it related to other variables associated with regulatory processe s. One possible explanation for this finding is that the regulation of emotions variable was defined by the extent to which children were frequently anxious or depressed. It is possible that limitation with this aspect of functioning is less prevalent am ong young children with disabilities (i.e., 82% were

PAGE 239

239 identified with no limitation) This functional variable might provide more distinction between subgroups of children who have known mental health concerns or challenging behaviors. Alternatively, it i s possible that concerns related to mental health of young children and young children with disabilities are underreported by primary caregivers (Knitzer, 2002; Woods, Smith, & Cooper, 2010). H earing and vision variables generally did not distinguish diffe rences between profiles (with the exception of Profile 1 and the vision variable) The hearing and vision variables were defined by children who have difficulty hearing or seeing, even with the assistance of adaptive equipment (e.g., hearing aids or glass es). The inclusion of adaptive equipment as part of these variables (e.g., child can see with glasses or child can hear with a hearing aid) might be related to why ratings were consistently lower (indicating more ability and fewer limitations) across prof iles. W ith the use of adaptive equipment fewer children experience severe limitations and children with complete vision or hearing loss (i.e., blindness or deafness) might represent a small number of children with disabilities. Although these variables did not lead to distinctions between subgroups, the ICF overall functioning and the se variables should be interpreted in the context of each profile, and in the context of personal and en vironmental factors (Lollar & Simeonsson, 2005; Simeonsson, 2003) In this section salient patterns of functioning for the latent class functional ability profile subgroups in relation to the functional ability variables in the PEELS data set were summarized The functional ability profile patterns provide information about the functional abilities within a profile and the differences in functional ability across profiles.

PAGE 240

240 A few of the functional ability variables did not lead to distinction s among profiles and p lausible explanations were offered to describe why these functional ability variables might not have contributed to distinctions among profiles. Nonetheless, o ne strength of person oriented profile approach es is the information gaine d about the multi dimensional and interactive features of functioning reflected in each profile (Raghavendra et al., 2007). Based on statistical fit indices and substantive interpretations of the profile patterns the five l atent class subgroups identifi ed in th e present study w ere defensible and interpretable T he profile patterns were examined with respect to how the functional ability variables were expressed in each profile, how the functional ability variables relate to each other within and across profiles and how operational definitions associated with each of the functional ability variables might have affected obtained profile patterns Profile patterns in relation to child, family, and school variables As noted in Chapter 4, descriptive analyses showed child factors, family factors, and school factors differed across profiles in ways that were logical and interpretable given the patterns of functioning represented in each profile For child factors, proportions of children w ho had an IFSP before the age of 3 were higher, mean number of weeks premature was larger, and mean birth weight was smaller for children whose functional ability profile included limitations related to all the physical functional abilities. The subgroup associated with moderate to severe limitations (Profile 1) had only slightly larger scores on these variables compared to the subgroup associated with mild to moderate limitations (Profile 3). This finding indicates the profiles distinguished children wit h limitations related to physical function who are often identified for services earlier

PAGE 241

241 and are likely born premature or with low birth weight (Saigal, Stoskopf, Streiner, & Burrows, 2001; Vohr et al., 2000). Child gender was represented in different pr oportions across profiles. In Profiles 1, 2, 4, and 5, 68% to 78% of the children were boys and 22% to 33% were girls. These within profile percentages generally approximate d the percentage of boys and girls in the PEELS data set (i.e., 70% male and 3 0% female). In Profile 3, however, 58% of the children who shared a similar functional ability profile ( characterized as mild to moderate limitations in most areas of function including physical limitations ) were boys and 42% of these children were girls. W hy a smaller percentage of boys and a larger percentage of girls were associated with this profile when compared to gender percentages in other profiles and in the entire PEELS data set is unclear, but might warrant additional investigation in future studies. Related to family factors, larger proportions of children participated in regular group activities in profiles that had fewer limitations associated with the functional ability variables F or example, 55% of children in Profile 5 participated in regular group activities compared to 36% of children in Profile 1. Between 45% and 47% of children in Profile 2, 3, 4 participated in regular group activities. T hese finding s are consi stent with expectations that children with less severe functional limitations have more access to and participation in family and community based activities (Brown & Gordon, 1987; King et al., 2003; Spiker, Boyce, & Boyce, 2002). Patterns of functional a bility w ere also examined in relation to school factors. children associated with profiles with fewer and milder functional limitations (Profiles 3,

PAGE 242

242 4, and 5) sugges ting that children with more moderate to severe functional limitations (Profiles 1 and 2) attended programs that have fewer peers without disabilities (i.e., more restricted or specialized settings). This finding is consistent with previous research on pl acement for children with disabilities (e.g., Buysse, Bailey, Smith, & Simeonsson, 1994; Etscheidt, 2006 ). Second, the type of IEP goals that teachers reported for children within a profile corresponded with specific limitations on functional ability vari ables for that profile For example, children in Profiles 1 and 3 which included children with more physical limitations had a larg er percentage of motor goals compared to other profiles and other curricular domains This section summarized how profil es differed across child factors, family factors, and school factors in ways that were not unexpected given the patterns of functioning represented in each profile and previous research related to these factors D ifferences across profiles with respect to child, family, and school factors highlight functional abilities might relate to their access to and participation in family and community based activities, as well as classroom settings with same aged peers without disabilities. Addition al variables available in the PEELS data set might be used to examine further these differences among profiles (e.g., teacher reported activities, classroom placement and amount of time spent with peers at school) Comparison of the selected latent class model with existing studies Another way to interpret findings from the present study is to examine the number and type of latent classes (subgroups) as well as functional ability profile patterns within and across each subgroup in relation to findings from other studies that used similar person oriented analytic technique s to derive subgroups with shared characteristics As

PAGE 243

243 noted in Chapter 2, no other studies involving a nationally representative sample of young children with disabilities have used this technique. Nonetheless, eight studies reviewed in Chapter 2 were used to inform analyses conducted in the present study. Specifically, six studies shown in Table 2 5 used a person oriented approach and shared similar features with the present study ( Haapa salo et al., 2000; Hair, Halle, Te rry Humen, Lavelle, & Calkins, 2006; Janson & Mathiesen, 2008; Konold & Pianta, 2006; Sanson et al., 2009; Stephens, Petra, Fabian, & Walrath, 2009) These studies are compared to findings from the present study related t o the number and type of subgroups obtained. T he number of subgroups identified in the present study was similar to the number of s ubgroups identified in the six studies with different samples of children (i.e., 4 t o 8 subgroups in previous studies). Several of these studies also identified severity of limitations or level of impairment as a distinguishing feature among subgroups. For example, Stephens, Petra, Fabian, and Walrath (2009) examined patterns of functional impairment based on subgroups of youth identified for community mental health services and identified two high impairment groups and one low impairment group. Several of the studies shown in Table 2 5 also distinguished subgroups based on clusters of similar skills. Konold and Pianta ( 2006) examined school readiness in relation to three measures of social function and three measures of cognitive function. Subgroups were associated with different patterns of abilities representing relative strength and weakness on social and cognitive v ariables. Hair, Halle, Te rry Humen, Lavelle, and Calkins (2006) include d physical health as a variable to examine school readiness in young children, along with social/emotional development, approaches to

PAGE 244

244 learning, language, and cognition. This study dis tinguished subgroups based on differences in physical health, similar to the present study in which a cluster of physical variables, including overall health distinguished several subgroups The se comparisons illustrate that the number of subgroups identified in the present study is generally consistent with those identified in the six studies In addition, profile patterns that distinguished subgroups in these studies were somewhat similar to profile patterns that disti nguished subgroups in the present study with respect to severity of limitations, to clusters of variables associated with profile patterns, and by the use of physical health variables. Although comparisons across large scale studies focused on children w ith disabilities were not possible, Simeonsson, Bailey, Smith, and Buysse (1995) conducted a small scale comparative study which is described in detail in Chapter 2 Although t his study was not one of the eight large scale studies reviewed as part of Tab le 2 5 it is particularly relevant to the present study Simeonsson et al. used functional ability variables from the ABILITIES Index (Simeonsson & Bailey, 1991) and conducted a hierarchical cluster analysis to identify subgroups of children with similar functional ability profiles. T he study sample was small (i.e., 91 children receiving early childhood intervention services) but this study permits comparison of the number and type of functional ability profiles obtained in the present study with those o btained by Simeonsson et al. using similar functional ability variables. A few caveats are noted before comparing the profiles identified in the present study with those from the Simeonsson et al. study. First, the Simeonsson et al. study used the original nine variables from the ABILITIES Index; in the present study 15

PAGE 245

245 variables derived from parent interview data in the PEELS data set were used. Many of these variables were similar to those from the ABILITIES Index. Second, the present study included the additional ABILITIES Index variables reported by Daley, Simeonsson, and Carlson (2009; i.e., regulation variables and motivation); these variables were not used in the Sime onsson et al. study. Third, respondents in the present study rated functional ability using a 4 point scale, while respondents in the Simeonsson et al. study used a 6 point scale for the functional ability variables so s everity of limitation is quantified differently across the studies and might not be directly comparable. Simeonsson et al. identified six subgroups with distinct and interpretable functional ability profiles The authors noted distinct ions among subgroups based on the severity of limitat ions and the types of limitations The largest group (46%) in the Si m eonsson et al. study was associated with mild limitations similar to Profile 5 in the present study. Two subgroups had functional ability profiles that showed limitations related to var iables associated with interaction skills (e.g., communication, social skills) and two subgroups had profiles that showed limitations related to variables associated physical skills (e.g., use of limbs) similar to Profiles 2 and 4 and Profiles 1 and 3, res pectively, in the present study Subgroups in the Simeons son et al. study whose functional ability profiles showed limitations in physical skills were the smallest proportion of the sample, which was consistent with findings in the present study related to the proportion of the PEELS sample in Profiles 1 and 3. A sixth subgroup was identified in the Simeonsson et al. study that was not identified in the present study. The authors described this subgroup

PAGE 246

246 as having mild limitations across most functional abilities with moderate limitations related to communication; this group wa s most similar to Profile 5 in the present study. The comparisons between these two studies show that the distinction s between characteristics of children with disabilities identified in the present study are similar to those identified in the Simeonsson et al. study For example, the severity and types of limitations resulted in somewhat similar profile patterns across these studies In addition, the prevalence of different profile patterns w as similar across these studies C hildren with fewer limitations were the largest proportion of children and children with limitations related to physical/health aspects of functi oning were the smallest proportion of children. A unique feature of the present study was the use of 15 functional ability variables. Previous large scale studies shown in Table 2 5, used between three to eight variables to create profiles, while the Si meonsson et al study used nine variables to create profiles. The selection of variables for person oriented approaches will depend on the purpose of the examination (e.g., different skills related to one domain of functioning or range of skills represent ing different domains of functioning). The present study and the Simeonsson et al. study both focused on a range of skills representing different domains of functioning. Additional regulation variables (i.e., regulation of activity level, regulation of a ttention, regulation of emotion, and motivation) that were included in the present study were not included in the Simeonsson et al study. It is possible that these variables helped to provide a more nuanced analysis of profile patterns in the present study. For example, if the regulatory variables had not been included patterns of

PAGE 247

247 functioning for Profiles 2 and 4 might have been in terpreted differently (e.g., focus on cognition and communication). The present study demonstrate d that the 15 functional abilities variables in the PEELS data set could be used to create distinct and interpretable subgroups of children with similar func tional ability profiles in a large scale data set focused on young children with disabilities. The 15 variables represent a range of functional abilities that might be important for identifying patterns of functioning Findings from the present study sug gest variables that include information about and variables related to range of functional skills including physical/health aspects of functioning and self regulation might be important for identifying distinct su bgroups and interpretable profile patterns in other samples of nationally representative children with and without disabilities. Association Betw een Functional Ability Profile Subgroup Skills and Problem Behaviors S ocial competence of young children with disabilities receiving early intervention and early childhood special education services and supports under the Individuals with Disabilities Education Improvement Act (IDEA) has been identified as a desired outcome Children with social competence have the necessary skills to achieve social goals know when to use appropriate behaviors for a given social context (i.e., social skills) and refrain from inappropriate behavior (i.e., problem behavior) in a social context (Odom, McConn social skills and problem behaviors in overall development and school success has been emphasized in the literature ( Shonkoff & Phillips, 2000; National C ouncil on the Developing Child, 2004a ) and in polic y (Early Childhood Outcome Center, 2009). Given

PAGE 248

248 social skills and problem behaviors it was hypothesized that functional ability profile subgroup membership would be associated with social skills and problem beha viors Findings from the present study showed that functional ability profile subgroup membership was somewhat social skills and problem behaviors ( R squared = .20 for social skills and .115 for problem behaviors ) Th e variance accounted for in social skills and problem behaviors by subgroup membership was similar to finding s reported in previous research that had used functional ability or composite scores or school readiness profiles to examine relationships with pre academic and social competence outcomes ( cf. Daley et al 2009; Konold & Pianta, 2006). For example, Daley et al., (2009) explained between 7% and 20% of the variance in pre academic/cognitive outcomes and between 13% and 19% of the variance in adaptive functioning outcomes when they examined relationships with a functional ability composite score Konold and Pianta (2006) explained between 12% and 19% of the variance in pre academic/cognitive outcomes when they examined relati onships with school readin ess profiles Konold and Pianta noted that even though their school readiness profiles of young children were only moderate predictors of educationally relevant outcomes, the profiles provide d useful description s characteristics for other pu rposes such as understanding the multi dimensional features and making decisions a bout instructional interventions Because functional ability profile subgroup membership in the present study was estimated as a categorical variable (1, 2, 3, 4, 5) based on most probable latent class membership t he restricted range of the categorical variable might have limit ed the

PAGE 249

249 explanatory power of this variable with respect to social competence outcomes. In future studies, examination of associations between functional ability profiles and desired outcomes might be conducted using the posterior probability scores rather than a categorical subgroup membership variable Differences Bet ween Functional Ability Profile Subgroup Social Skills and Problem Behaviors Although the number of subgroups and functional ability profile patterns across subgroups was not known a priori, examining differences in social skills and problem behaviors across the subgroups of children who shared similar functional ability profiles was also of interest in the present study N otable differences were found across subgroups related to their social skills and problem behavior s as measured by standard scores on the PKBS 2 (PKBS 2 normative sample M = 1 00, SD = 15) Children associated with Profile 1 had social skills that were more than two standard deviations below the normative mean ( M = 70) although they evidenced problem behavior scores very near the normative mean ( M = 101 ) C hildren in Profile 2 had the highest ratings on problem behavior s ( M = 106) and their social skills were more than one standard deviation below the normative mean ( M = 80 ) Children in Profiles 3 and 4 had mean social skills and problem behavior scores that were at or withi n half of a standard deviation unit of the normative mean. In contrast, children in Profile 5 had the highest social skills ratings ( M = 102 ; near the normative mean ) and the lowest ratings on problem behavior s ( M = 92 ; below the normative mean ) Differences between subgroups in relation to social skills and problem behavior standard scores were generally statistically significant and effect sizes (i.e., standardized difference effect sizes) were typically .40 or greater. Children in

PAGE 250

250 subgroups ass ociated with more moderate to severe limitations (Profile 1 and 2) had lower ratings of social skills particularly social interactions in comparison to children in subgroups associated with less severe limitations (Profiles 3, 4, and 5). S ocial skills s cores for children in Profile 1 were 1.35, 1.15, and 1.71 standard deviation units below children in Profiles 3, 4, and 5 respectively. For children in Profile 2, social skills scores w ere .80, .60, and 1.16 standard deviation unit s below social skills s cores for children in Profile s 3, 4, and 5, respectively. Children in subgroups associated with more limitations re lated to self regulation ( Profiles 2 and 4) had higher ratings of problem behaviors, particularly externalizing behaviors, in comparison to children in profiles associated with physical limitations (Profiles 1 and 3) or children with the fewest limitations (Profile 5) P roblem behaviors scores for children in Profile 2 were .41, .76, and .99 standard deviations units above children in Profiles 1, 3, and 5 respectively. Problem behaviors scores for c hildren in Profile 4 were 36 and 59 standard deviation units above problem behaviors scores for children in Profile 3, and 5, respectively. Although differences in social skil ls and problem behaviors among subgroups were identified, standard score means for Profiles 3, 4, and 5 were generally at or within half a standard deviat ion from the PKBS 2 normative sample This finding suggests that social skills and problem behaviors for children with disabilities in these subgroups are generally within an expected range of skills and behaviors for their peer group (Merrell, 2002) Given this information, ongoing monitoring of social competence, using prevention practices, and impleme nting targeted interventions when needed, are likely

PAGE 251

251 appropriate for children with these functional ability profiles (Fox, Dunlap, Hemmeter, Joseph, & Strain, 2003 ; Merrell, 2002 ). Standard score means for social skills and problem behaviors for Profil es 1 and 2, however, fall outside the normative range of social skills. The mean social skills score for Profile 2 was 1.33 standard deviation unit s from the PKBS 2 normative sample mean and the mean social skills score for Profile 1 was 2 standard deviat ion unit s from the PKBS 2 normative sample mean, suggesting the need for regular assessment and intensive and individualized interventions to promote the development of social skills for children with these functional ability profiles (Fox et al., 2003 ; Merrell, 2002 ) Literature that describes the characteristics of children with significant disabilities might be helpful for explaining the notably lower social skills for children in Profiles 1 and 2. For example, Profile 2 represents children with mod erate to severe limitations including limitations with communication/cognition, social competence, and self regulation. This profile likely includes children who have been clinically diagnosed with autism or other intellectual disabilities as described by Diagnostic and Statistical Manual of Mental Disorders, f ourth edition (DSM 4 ) or the American Association o n Intellectual and Developmental Disabilities (AAIDD 2010 ) Disabilities identified in this way might not correspond with IDEA disability classifi cations. Nonetheless it is well documented in the research literature that children who have been clinically diagnosed with autism or intellectual disability have difficulty with a range of social skills, including social initiation and reciprocity, spon taneous social exchanges interpreting social cues, sustaining proximity to peers, and participating in novel play sequences (Dawson et al., 2004; Matson & Shoemaker, 2009; McConnell, 2002; Wetherby & Prizant, 2000). For this

PAGE 252

252 subgroup, concerns with socia l skills likely relate to the severity of limitations and the types of limitations that might affect social communication and self regulation identified in their profile. In contrast, Profile 1 represents children with moderate to severe limitations includ ing limitations with communication/cognition, motor function, social competence, self regulation, and vision This profile likely includes children who are considered to have the most significant disabilities, multiple disabilities, or the most extensive support needs (Snell & Brown 2006; Westling & Fox, 2004 ) Research supports that children with significant and multiple disabilities who require extensive supports often (a) have less access to peer groups (King et al., 1997; King et al., 2009), (b) lack mobility to access peers without assistance ( Harper & McCluskey 2002 ), (c) lack communication skills to initiate and respond to social interactions ( Reichle, 1997 ; Snell, Chen, & Hoover, 2006), (d) lack motors skills to participate in play activities (Kin g et al., 2009), and (e) are more likely to be excluded by peers without disabilities (Buysse, Goldman, & Skinner, 2002; Diamond, Hong, & Tu, 2008). For this subgroup, concerns with social skills likely relate to the overall severity of limitations and th e multiplicity of limitations that might affect social communication and motor function identified in their profile. characteristics and social skills, it was not surprising that chi ldren in P rofiles 1 and 2 had significantly and notably fewer social skills than children in other profiles T he size of the gap might be somewhat unexpected, however, especially for children in Profile 1. Although research that describes the characteristics of children with significant disabilities suggests that children in Profiles 1 and 2 were likely to have limitations

PAGE 253

253 related to their social skills, an unexpected finding from the present study was that fewer chi ldren in these profiles had IEP goals targeting social skills compared to other curricular domains. For example, 20% of the children in Profile 1 had a social skills goal compared to 60% of children who had a communication goal or 34% or 38% who had a fin e or gross motor goal, respectively. For Profile 2, 36% of children had a social skills goal compared to 71% of children who had a communication goal or 48% who had a school readiness goal. Profiles 2 and 4 had the highest percentage of social skills goa ls on their IEPS compared to other profiles (36% and 32%, respectively), although these percentages are smaller than the percentage of school readiness or communication goals for these children. The present study suggests a need to consider social skills goals and related instructional interventions for children with characteristics similar to children associated with Profiles 1 and 2. Disability Category Compared to Subgroup Membership Concerns with the use of IDEA disability categories to describe and classify children have been discussed in the special education literature since the passage of IDEA (cf. Hobbs, 1975). To address concerns identified with IDEA disability categories, a functional approach has been proposed to complement information that c an be obtained from disability category ( Florian & McLaughlin, 2008 ; Simeonsson et al., 2008) Previous studies have used national ly representative data sets to examine the use of disability category in the prediction or explanation of educational service s and skills ( e.g., Chambers Perez, et al., 2004; Daley, Simeonsson, & Carlson, 2009) Both of these studies used functional ability variables from the ABILITIES Index to create a compos ite score that represented overall severity of limitations in functioning. Findings from t hese

PAGE 254

254 studies showed that the functional ability composite score s accounted for more variance in the criterion variable than disability category alone These studies also showed that when the functional ability composite and disability category were used together the most variance was accounted for in the criterion variable The present study extends these findings by using a categorical variable that represents subgroups of children who have similar functional ability profiles instead of an overall composite score of functional ability to examine similar associations An important aspect of using a f unctional approach, whether focused on subgroups that have similar profiles or a composite score of functional ability is the extent to which distinctions among the severity of limitations can be identified. As noted by Daley and colleagues, distinctions among the severity of limitations are something that might not be well specified within disability categories. For example, low incidence Descriptive information from th e present study however, shows that only 30% of children identified by disability categories typically referred to as low incidence were also assigned to a profile associated with moderate to severe limitations in functioning (Profiles 1 and 2) while 55% of these children were assigned to a profile associated with mild to moderate limitations in functioning (Profiles 3 and 4) In relation to the prediction or explanation of child outcomes, the present study profi le subgroup membership ( R 2 = .20 for social skills; R 2 = .115 for problem behaviors ) accounted for more variance in the criterion variable than disability category ( R 2 = .165 for social skills; R 2 = .108 for problem behaviors ). Th is resulted in a 3.5% difference for social skills and .7%

PAGE 255

255 difference for problem behavior s in variance explained by the two categorical variables. Similar to the previous studies, the present study also showed that when these variables were used together they accounted for the most variance in the criterion variables and accounted for more variance in social skills than in problem behavior s Direct comparisons to the Daley et al. study, which was conducted with the PEELS data set using similar variables of functional ability to create a 6 item disability severity composite score, show the functional ability profile subgroup membership variable in the present study accounted for slightly more variance in the criterion variables (e.g., 19% of variance in so cial skills explained by the disability severity composite, 20% of variance in social skills explained by the functional ability profile) The increase in variance explained between the two studies was small (1% difference). T he decision to use a functi onal ability composite score (similar to the Daly et al. study) or to use a subgroup membership variable based on similar functional ability profiles (similar to the present study) as a pr edictor or explanatory variable in relation to child outcomes, such as social skills and problem behaviors, might be based on the substantive questions of interest. For example, for questions related to the cumulative influence of functional limitations a composite score might be appropriate. For questions related to the patterns of f unctioning across ability areas a profile approach might be appropriate. Daley and colleagues noted that they compar ed the explanatory power of IDEA disability categor y and a composite score of functional ability limitations to illustrate that the se variables reflect distinct constructs with less overlap than might be predicted given traditional ideas about

PAGE 256

256 R esults from the present study showed that there was some overlap between profiles and disability categories. This overlap, however, is not likely sufficient to predict a functional ability profile subgroup known or vice versa, with the possible exception of speech or language impairments and Profile 5. Findings showed there were a large proportion of children identified with speech or language impairments (SLI) who were assigned to the profile with the fewest limitations (i.e., Profile 5) this made up 29 % of the total PEELS sample and 72% of Profile 5. All other overlap between disability category and profile subgroup membership were less than 14% of the total PEELS sample To examine further the effect of this overlap on the explanation of variance i n social skills and problem behavior s holdout analyses were conducted with children with SLI excluded from the analyses. Findings from these analyses suggest that when children with SLI were not included in the model, disability category had less explanatory power for social skills ( R 2 = .114) and problem behavior ( R 2 = .074) Functional ability profiles, however, maintain ed explanatory power for social skills ( R 2 = .201) and had less explanatory power for problem behavior ( R 2 = .090) Overall, these and previous findings suggest that to characterize childhood disability, functional ability profiles or functional ability composite scores should be used in place of or together with IDEA disability category when examining relationships to important outcome variables for young children with disabilities. Variables related to d relatively

PAGE 257

257 homoge nous groups of children compared to IDEA disability categor ies that result in heterogeneous groups of children Non Malleable Child Factors and Contextual Factors factors and a primary purpose of the study was to examine the extent to which subgroups, based on similar profiles of malleable functional ability variables were s Given research on factors associated with and hypothesized relationships in ICF CY framework the extent to which select non malleable child factors and contextual factors moderated relationships between functional ability profile subgroup membership and social skills and problem behavior s was examined in the present study. T he interaction between functional ability profile subgroup membership and non malleable child factors and contextual factors were examined through a series of regression based moderat ion analyses. Contextual factors included family factors and environmental factors. Family factors consisted of family characteristics and parent child interactions; environmental factors included both school and community variables. The inclusion of pa rent child interactions and school variables was an important part of the present study because these variables have not been routinely examined in other large scale studies and the ICF CY specifically emphasizes the role of family and school factors durin g childhood ( Bjorck Akesson et al., 2010). Moderation analyses were used to examine whether there were differences i n problem behaviors among profiles as a function of moderat or variables. An unexpected finding from the prese nt study was the extent to which non

PAGE 258

258 malleable child factors and contextual factors generally did not moderate relationships between functional ability profile subgroup membership and social skills and problem behavior s Of the 16 select modera tor variables, only one variable (child activities) moderated the relationship between subgroup membership and social skills, and only three variables (regular child activities, race/ethnicity, safety of neighborhood) moderated the relationship between sub group membership and problem behavior s C ompar isons between findings from the moderator analyses in the present study with previous research is somewhat limited, as only one large scale study has used a person oriented profile approach and included non mal leable child factors a s well as contextual factors as moderators. Sanson et al. (2009) used child gender and a composite score of socio economic status (SES) to examine moderation of the relationship between temperament profiles and problem behavior s social skills, and school adjustment. The authors found SES was a significant moderator of the relationship between temperament profiles and problem behavior s but other interaction effects were not identified. The authors also noted that identifying r easons why SES was a moderator of temperament profiles in relation to problem behavior s were difficult to pinpoint. In the present study, the number of child activities was a significant moderator of the relationship between subgroup membership and childr up analyses suggested that the number of activities a child had ever participated in was more related to chi for children in Profile 4 (children with mild to moderate limitations) than for children in Profil e 5 (children with fewer limitations) who had participated in the same number of activities This finding suggest s that

PAGE 259

259 participation in a range of extracurricular activities might support social skills development for children with functional ability lim itations similar to Profile 4 compared to children with functional ability limitations similar to Profile 5 Alternatively, this finding might suggest that children who have functional ability limitations similar to Profile 4 and who have better social sk ills have participated in more activities. Both explanations are plausible based on the correlational analyses conducted in the present study. The r egular child activities variable was a significant moderator of the relationship between subgroup membershi p and problem behavior s Follow up analyses showed that the more activities that children regularly participated in, problem behaviors scores increased for children in Profile 1 (children with the most functional ability limitations) c ompared to children in Profile 5 (children with the fewest limitations) who participated in the same number of activities. This finding suggests that regular participation in extracurricular activities might not prevent problem behavior for children with the most functional ability limitations (Profile 1) compared to children with the fewest functional ability limitations (Profile 5 ) Alternatively, this finding might suggest that children who have the most functional ability limitations and who have more problem behaviors are provided with more regular extracurricular activities. B oth explanations are plausible based on the correlational analyses conducted in the present study. Two categorical variables, race/ethnicity and neighborhood safety, were signi ficant moderators of the relationship between subgroup membership and problem behavior s Follow up analyses for these categorical variables were examined by comparing the mean problem behavior s scores for each profile for every response categ ory of the

PAGE 260

260 ca tegorical variable. Due to the large number of comparisons, follow up comparisons were not examined for statistical significance. For race/ethnicity, comparisons showed that mean differences in problem behavior s scores for children in some race/ethnicity categories were not consistent when comparing profiles. For example, in Profile 1, Caucasian /N on Hispanic children had lower mean problem behavior s scores than children in other race/ethnicity categories compared to differences for children in Profile 2 who were Caucasian/Non Hispanic For Profile 4, African American children had higher mean problem behavior s scores than children in other race/ethnicity categories when compared to differences for children in Profile 3 who were African American. For neig hborhood safety, children in Profile 1 who were from a not safe neighborhood had higher mean problem behavior s scores than children in other neighborhood safety categories when compared to children in Profile 2 who were from a not safe neighborhood These moderation findings suggest that there are some interaction s among behaviors, and some interaction s functional ability profil e subgroup membership, and problem behaviors It is possible that these differences are related to actual levels of problem behaviors reported by teachers or it is about problem behavio rs in relation to these variables ( Konold, & Pianta 2007 ; Reid et al., 1998) Given the limited research that has examined non malleable child factors and contextual factors as moderators o f the relationships between functional ability profile subgroup m embership and social skills and problem behavior s the explanations for w hy

PAGE 261

261 these variables might operate differently on social skills or problem behavior s across the functional ability profile subgroups are preliminary. In addition, the patterns and cons istency of associations between non malleable and contextual factors and analytic methods used to examine these relationships (Krishnakumar & Black, 2002; Raver et al., 2007; S ameroff & Seifer, 1983) therefore findings related to moderator analyses might not replicate in future studies A nalyses conducted in the present study did not account for potential correlations between or among moderator variables. Nonetheless, the present study provides preliminary evidence about the potential role of select non malleable child factors and contextual factors in relation to functional ability profiles and their social skills and problem behaviors Implica tions of the Present Study social skills and problem behaviors In addition, examining prof iles of functional abilities or limitations within and across person oriented latent class subgroups provides important information about functional characteristics shared by young children with disabilities Profile s provide information about multi dimen sional and interactive that might be informative for research and practice P rofiles and profile patterns that might be used for educational planning, informing decisions about services and supports, or designing specific interventions. Findings from the present study support existing research and theory that suggest their social skills and

PAGE 262

262 problem behaviors might have different manifestations or be enacted in different ways in the context of a profile cf. Hair et al., 2006; Haapasalo et al., 2000; Janson & Mathie sen, 2008; Sanson et al., 2009) The present study extends previous research to young children with disabilities through the conduct of secondary analyses using a large scale nationally representative data set F ive sub group s of children with distinct and interpretable functional abi lity profiles were derived empirically The study offers information about the patterns the five different functional ability profiles for young children ages 3 through 5 years who were receiving specia l education services in the fall of 2003. Conceptual Implications of the Findings The present study used the International Classification of Functioning, Disability, and Health for Children and Youth (ICF CY) framework described by the World Health Organi zation (WHO, 2007) to guide the conceptualization of the study research questions, the creation of the functional ability profiles, and the identification of contextual factors developin g characteristics within his or her surrounding environment, while noting the on the (Snyder, 2006). The ICF CY highlights the unique nature of child development and suggests that patte limitations will change in nature, intensity, and consequence over time (Lollar & Simeonsson, 2005) This study supports the use of the ICF CY framework in future research with children with disabilities. For exampl e, the complex relationships that were identified within and betw een functional ability profile s provide the opportunity to examine the co occurrence of functional limitations or salient patterns of

PAGE 263

263 functioning and development. Specifically, the profiles provided information related to the (a) level or severity of limitations (b) number of limitations and (c) nature or type of limitations across the 15 functional ability variables Of particular im portance for research, constructs identified in the ICF CY framework are consistent with constructs identified in the extant literature, particularly related to key priorities in educational research. For example, malleable and non malleable characteristi cs described by the Institute of Education Sciences are consistent with aspects of body functions and activities, and personal factors, respectively, in the ICF CY framework The ICF CY framework provides a common language for researchers across disciplin es to begin to identify, define, and examine these variables in future studies ( Bjorck Akesson et al., 2010; Simeonsson, 2009). Using the ICF CY as a framework researchers might create measures that reflect key functional ability variables ( cf. Simeonsso n & Bailey, 1991) or map existing measures to functional ability variables (cf. Morris, Kurinczuk & Fitzpatrick, 2005; Simeonsson, Scarborough, & Hebb e ler, 2006) that allows for comparison of findings across studies and measures. Methodological Implicatio ns of the Findings Researchers have begun using advanced analytic methods to create profiles of or skills and to group children with similar or shared characteristics into descriptive profiles Th ese analytic methods are characterized as person oriented techniques (Campbell, Shaw, & Gilliom, 2000; Konold & Pianta, 2005). A n important aspect of the present study was the use of latent class analysis (Muthen & Muthen, 2007) to conduct person oriented analyses and to identify plausible la tent classes based on the functional abilit y profiles of young children with disabilities

PAGE 264

264 P revious investigations have predominately used cluster analysis to identify subgroups of children with shared profiles La tent class models have gained popularity over other methods because they use model based approaches to estimate membership probabilities in order to classify cases into the appropriate subgroup ( Magidson & Vermunt, 2006) The present study, however, was not able to use all the features of the model most probable latent class membership and examined as a categorical variable (i.e., Profile 1, 2, 3, 4 or 5). T h e decision to use most probable latent class membership wa s appropriate given that mean posterior probabilities for latent class membership were all above .86 and analyses were designed to use PEELS weight files that used all available cases to examine each research question. T he study did not analyze relationsh ips between functional ability and s using latent class analysis with mixture modeling. This approach would have used posterior probability scores rather than most probable latent class membership categories A potential methodological contribution of the present study however, is the ability to generate the posterior probabilities (cf. Collins & Lanza, 2010) for children not included in the PEELS sample Using posterior probabilities other children with dis abilities could be assign ed to their most probable functional ability profile subgroup based on the five profiles identified in the present study Collins and Lanza (2010) introduced an equation that, conceptually can be generated from the latent class analysis to calculate posterior probabilities. This equation requires information about the item response probabilities and the prevalence of the five latent classes generated in the present study along with scores for the 15 functional ability variables used in the

PAGE 265

265 present study for a new sample of children with disabilities. This potential contribution is conceptual because additional exploration of Mplus is needed to generate the required information. Nonethele ss, p osterior probabilities might be used in future research or in practic e based applications to classify children into the appropriate functional ability profile subgroup and examine further the relationships between their functional ability profile subg roup membership and educationally relevant outcomes. Practical Implications of the Findings The present study identified five functional ability profiles for a nationally representative sample of young children with disabilities who were receiving early childhood special education services in 2003 Profiles provided information about the d to social skills and problem behavior s and differences in social skills and problem behaviors were identified between profiles. In general, mean scores on social skills and problem behaviors standard scores on the PKBS 2 reflected a typical range of exp ected skills and behaviors for young children associated with Profiles 3, 4, and 5. From a practical standpoint, this finding suggests that children whose functional skills are similar to these profiles likely require general curricular interventions that focus on promotion of social skills and prevention of problem behaviors Often associated with the primary and secondary level of tiered frameworks (cf. Simeonsson 1991; Snyder, McLaughlin, & Denney, 2011), these interventions might include monitoring o f social skills and problem behaviors, universal promotion and prevention practices provided to all children, and targeted interventions and prevention practices when needed (e.g., Brown, Odom, &

PAGE 266

266 Conroy, 2001; Fox, Dunlap, Hemmeter, Joseph, & Strain, 2003; Hemmeter, Fox, & Snyder, 2008, 2010). Findings also showed that two profiles (Profiles 1 and 2) were associated with extremely low or low social skills and that these children were not likely to have an instructional goal to improve social skills compar ed to other curricular areas F or children whose functional abilities are similar to these two profiles, teachers, parents, and other care providers might consider regular assessment of social skills and individualized interventions to promote the develop ment of social skills as needed. These types of interventions often are associated with the tertiary or individualized level of tiered curricular frameworks ( Simeonsson 1991 ; Snyder et al., 2011 ). For children who are identified to have a need for targeted or individualized intervention, the functional ability profile offers characteristics that might provide practical information to inform the design and delivery of intervention s (Konold & Pianta, 200 5 ) As noted by Simeonsson and colleagues (2006) t ability profile can guide practitioners in individualizing interventions in educational and clinical treatment planning For example, a child in Profile 1 might experience diffi culties with social skills because the child has limitations in mobility that restrict the child from running, playing, and moving exhibit social skills. In contrast, a child in Profile 2 might experience difficulties with social skills because the child does not sit still or remain engaged with a play activity or peer for sufficient periods of time to develop or exhibit social skills. In this example, both children h ave difficulties with gaining or maintaining access to peers to develop or

PAGE 267

267 exhibit social skills. The distinction in what leads to this difficulty has important implications for the design and delivery of prevention supports such as environmental arrangem ents or targeted or individualized interventions such as peer buddies ( English, Goldstein, Shafer, & Kaczmarek, 1997) or social skills training (Brown & Conroy, 2001). In this example, each child will need supports or instruction to help gain or maintai n access to peers, but the nature of the supports or instruction will likely be very different based on the knowledge of the child gained through the description of the functional ability profile. Policy and Research Implications of the Findings F indings from the present study might help inform policy recommendations related to characterizing disability and function for young children with disabilities when examining correlates of socially valid outcomes such as social competence The increase in variance explained in social skills and problem behavior s measures when functional ability profile subgroup membership was used compared to when disability category was used support s the use of functional ability variables or functional variables in combination wi th disability categor y to examine variations in outcomes. National efforts to collect and report information about educational services and outcomes for young children with disabilities (i.e., IES national studies, accountability requirements under IDEA Part B S ection 618) might include abilities as part of required data collection Using this information, outcome and performance data might be reported in relation to functional abili ty profile subgroup. The use of functional ability profiles might also have important implications for intervention research that helps inform policy and service provision National efforts to

PAGE 268

268 identify empirical evidence for the use of different educational interventions often focus on Clearinghouse). As noted by King et al. (1997), policy makers, researchers, and practitioners cannot assume that interventi ons that work for children with one set of abilities will work for other children. Given the range of functional characteristics across young children with disabilities, the use of a profile to describe the functional abilities of children for whom the in tervention was effective might help advance the development of interventions funding allocations for supports and services and related policies. Moreover, p a way to organize a range of malleable explana tory or predictor variable s in way s that account for the complex and transactional relationships among these variables. Characterizing children in relation to malleable factors to examine relationships that are associated with, moderate, or mediate childr educational research (IES, 2011). The functional ability profile approach might be helpful to identify children who are not achieving desired outcomes and to use information from their pr ofile to inform the types of supports these children might need. By using a profile of malleable child factors, interventions can be designed to target areas of identified need and to interpret intervention outcomes in relation to functional profile characteristics. Recommendations for Future Research The present study examin ed relationship s among child functioning, IDEA disability categor y non malleable child and contextual factors, and young competence. Findings from the present study suggest that additional examination of

PAGE 269

269 these relationships is warranted. This section highlights a few of the potential next steps for future research. P erson oriented analytic techniques including latent class analysis will likely be used incr easingly in future research. As noted earlier, based on most likely latent class membership instead of the posterior probability scores. Researchers might compare and contrast findings that use posterior probabilit y scor es to findings that use most likely latent class membership R esearchers might examine the extent to which the profiles obtained in the present study are replicated using (a) other samples of young children with and without disabilities, (b) other populati ons of individuals with disabilities (e.g., school age, adults), ( c ) different functional abili ty variables or ( d ) other domains of child functioning indicated on the ICF CY These types of comparative analyses might further enhance understanding s of rel ationships between functional abilities and desired early learning or educational outcomes. As noted by Lollar and Simeonsson ( 2005 ), or disability will change in nature, intens ity, and consequence over time. The present study used cross sectional data gathered at one wave (i.e., wave 1) from the PEELS data set to examine relationships between their social skills and problem behavior s Latent transition analysis (cf. Collins & Lanza, 2011; Stephen s Petras, Fabian, & Walrath, 2009) or a series of latent class analyses might be used to examine how the nature and prevalence of functional ability profiles change over time and how these cha nges might affect relationships with outcome variables In addition, latent transition analysis might be used to examine the stability of

PAGE 270

270 functional ability profile subgroup membership over time or how profiles change in relation to implemented intervention s The present study examine d whether non malleable child factors and contextual factors moderated relationship s subgroup membership and social competence outcomes. Although a few noteworthy mod erat or relationships were identified, interpretation s related to these relationships w ere preliminary given the limited research that has examined these variables as moderators particularly as moderators o f relationships between functional ability profile subgroup membership and social skills and problem behaviors (cf. Sanson et al. 2009 ). Additional research is need ed to better understand these relationships to explore whether these relationships replicate in future studies, and to examine relationships among moderator variables (e.g., create latent variables that represent clusters of moderators rather than using one variable at a time to examine moderation) Summary development and school success, it is critical to unpack how malleable child factors such as functional abilities, Information gained from a person oriented functional approach can inform the design and delivery of interventions and policies to provide efficient and effective supports and services for children and their families. The present study used a cross sectional correlational design to explore and examine relationships among young chi skills and problem behaviors and t heir functional abilities, IDEA disability category, and non malleable child and contextual factors through secondary analyses of the Pre Elementary Education Longitudinal Study (PEELS) data set. Fifteen fu nctional ability

PAGE 271

271 variables based on the ICF CY framework were used in latent class analys e s and these analyses identified five subgroups of children with distinct and interpretable functional ability profiles. Functional ability profiles were described in relation to the severity, number, and type of functional ability limitations. Profiles 1 and 2 were associated with moderate to severe limitations, Profiles 3 and 4 were associated with mild to moderate limitations, and Profile 5 was associated with n o to mild limitations across the 15 functional ability variables. Profiles 1 and 3 were different from other profiles because they included limitations related to all the physical/health variables. P rofile patterns identified in the present study were similar to profile patterns identified in existing research that has used person oriented methods to create subgroups of children with similar profiles of abilities Descriptive differences for select c hild, family, and school factors a cross the profiles were noted. These differences were logical in relation to the patterns of functioning within and across the profiles. Findings from the person oriented functional approach provide important information about the characteristics of young children with disabilities in relation to their functional ability profile patterns. Functional ability profile subgroup membership was moderately related to the relationship to problem beh aviors was small Although differences between subgroups were identified on social skills and problem behaviors, mean standard scores indicated that children associated with profiles with no to mild limitations or mild to moderate limitations were general ly within the normative range with respect to expected social skills and problem behaviors for their peer group Children with moderate to sever e limitations, however, were identified to have social skills

PAGE 272

272 notably lower than same aged peers suggesting th e need for targeted or individualized intervention supports. T he explanatory power of functional ability profile subgroup membership was greater than the explanatory power disability category as a correlate of their social skills and problem behaviors When the two variables were used together, functional ability profile subgroup membership and IDEA disability category accounted for the most variance in social skills and problem behaviors standard scores on the PKBS 2 These findings important when examining correlates of child outcomes in future research One of the 16 non malleable child or contextual variables moderated the relationship between functional ability profi le subgroup membership and their social skills (i.e., number of child activities). Three of the se 16 variables moderated the relationship between functional ability profile subgroup membership and their problem behaviors (i.e., race/ethnicity, number of regular child activities, and neighborhood safety). Findings related to moderation identified in the present study are preliminary but suggest relevant directions for further research. Th e present study highlight s the importance of considering t he diversity of young young children who receive special education and related services under the Individuals with Disabilities Education Improvement Act. Findings su pport that the ICF CY fr amework is useful to guide how s are defined and measured in educational research. The present study demonstrates how a functional approach can be combined with person oriented analytic methods to examine

PAGE 273

273 associations between subgroups of children with similar functional ability profiles and desired outcomes, including social competence. Using functional ability profiles to describe and quantify ch category, that can be used to d esign interventions and supports so children develop the skills they need to be social ly competent. More than 35 years ago, Nicholas Hobbs (1975) authored The Futu re of Children: Categories, Labels, and Their Consequences In this seminal report Hobbs described an alternative to traditional disability classification systems used in education H e to construct a profile of assets and liabilities of the child The profile should be the basis for specification of [educational planning] (p. 25). At the time, Hobbs was heartened that C omputer technology provides the means of organizing perfected system, data from all states could be aggregated to provide the federal government with information to plan legislation, not in terms of gross categories of exceptionality but in terms of specif ic requirements for services Suc h a system should be developed. (p. 25) Thir ty five years lat e r, Hobbs recommendations are still relevant today The present study highlights how frameworks such as the ICF CY can be used to move closer to the type of system Hobbs envisioned to help guide educational planning and service delivery for young children with disabilities

PAGE 274

274 APPENDIX A PROFILES OF CHILDREN Y

PAGE 275

275 Table A 1 Functional p rofile: Child A Child A Primary Disability: Developmental Delay Hearing Visions Use of Arm Use of hands Use of Legs Overall Health Cognition Com with Others Understand Others Social Skills Behavior Motivation Activity Level Attention Regulate Emotion 1 Typical Function X X X X X X X X X X 2 Mild I mpairment X X X 3 Moderate Impairment X X 4 Severe Impairment Table A 2 Functional p rofile: Child B Child B Primary Disability: Developmental Delay Hearing Visions Use of Arm Use of hands Use of Legs Overall Health Cognition Com with Others Understand Others Social Skills Behavior Motivation Activity Level Attention Regulate Emotion 1 Typical Function X 2 Mild I mpairment X X 3 Moderate Impairment X X X X X X X X 4 Severe Impairment X X X X

PAGE 276

276 Table A 3 Functional p rofile: Child C Child C Primary Disability: Developmental Delay Hearing Visions Use of Arm Use of hands Use of Legs Overall Health Cognition Com with Others Understand Others Social Skills Behavior Motivation Activity Level Attention Regulate Emotion 1 Typical Function X X X X X X X X X X X 2 Mild I mpairment X X 3 Moderate Impairment X X 4 Severe Impairment Table A 4 Functional p rofile: Child D Child D Primary Disability: Developmental Delay Hearing Visions Use of Arm Use of hands Use of Legs Overall Health Cognition Com with Others Understand Others Social Skills Behavior Motivation Activity Level Attention Regulate Emotion 1 Typical Function X X X X 2 Mild I mpairment X X X X X X X X X 3 Moderate Impairment X X 4 Severe Impairment

PAGE 277

277 APPENDIX B PEELS DISABILITY SEVERITY INDEX VARIABLES Table B 1 PEELS Disability Severity Index v ariables Item PEELS Parent Interview Questions Derived Items Hearing HEARCMP: Compared with children about same the same age, would you say {CHILD} 1: Hears normally 2: Might have a hearing problem 3: Does have a hearing problem WELHRDV: How well does {CHILD} seem to hear with the currently used hearing device(s) 1: Hears normally 2: Has a little trouble hearing 3: Has a lot of trouble hearing 4: Does not hear at all HRNGLSS: Is {CHILD}'s unaided hearing loss 1: Mild, ( 90 Dba hearing level) DP1PROBHEAR 1: No hearing loss and hears normally 2: Hears normally or has only a little trouble hearing regardless of level of hearing loss 3: Has a little or lot of trouble hearing with a range of hearing loss 4: Does not hear at all or severe/profound loss and has a lot of trouble hearing Vision Would you say {he/she} 1: Sees normally without glasses 2: Might have a vision problem 3: Does have a vision problem VSWTHGLS: How well can {CHILD} see with glasses? Would you say {he/she} 1: Sees normally 2: Has a little trouble seeing 3: Has a lot of trouble seeing 4: Does not see at all VSWOGLS: How well can {CHILD} see without glasses? Would you say {he/she} 1: Sees normally 2: Has a little trouble seeing 3: Has a lot of trouble seeing 4: Does not see at all DP 1PROBVISION 1: Sees normally without glasses 2: Sees normally with glasses 3: Has a little trouble seeing, even with glasses see at all, even with glasses

PAGE 278

278 Table B 1 Continued. Item PEELS Parent Interview Questions Derived Items Overall Health BHLTHCMP: Compared with other children general health is: 1: Excellent 2: Very good 3: Good 4: Fair 5: Poor any way because of a health problem? 1: Yes 2: No DP1PROBHEALTH 1: Excellent or very good health with no limitation in activities 2: Good health with no limitation in activities or excellent/very good health with limitation in activities 3: Good health with limitation in activities or poor health with no limitation in activities 4: Fair or poor health with limitations in activities or poor health with no limitations in activities Use of Arms BARMSGMS: How well does {CHILD} use {his/her} arms for things like throwing, lifting, or carrying? 1: Uses {his/her} hands and fingers normally 2: Has a little trouble using them 3: Has a lot of trouble using them 4: Has no use at all of {his/her} hands and fingers 5: missing one or both hands DP1BARMSGMS 1: Uses {his/her} hands and fingers normally 2: Has a little trouble using them 3: Has a lot of trouble using them 4: Has no use at all of {his/her} hands and fingers Use of Hands BARMSFMS: How well does {CHILD} use {his/her} hands and fingers for things like buttoning a shirt or using a spoon, pencil or scissors? 1: Uses {his/her} arms and hands normally 2: Has a little trouble using one or both 3: Has a lot of trouble using one or both 4: Has no use at all of one or both arms or hands 5: Missing one or both arms DP1 BARMSFMS 1: Uses {his/her} arms and hands normally 2: Has a little trouble using one or both 3: Has a lot of trouble using one or both 4: Has no use at all of one or both arms or hands Use of Legs BLEGSWEL: How well does {CHILD} use {his/her} feet? 1: Uses both legs and feet normally 2: Has a little trouble using one or both 3: Has a lot of trouble using one or both 4: Has no use at all of one or both legs or feet 5: Missing one or both legs DP1BLEGSWEL 1: Uses both legs and feet normally 2: Has a little trouble using one or both 3: Has a lot of trouble using one or both 4: Has no use at all of one or both legs or feet Cognition CBLEARN: Compared with other children about the same age, does {CHILD}, learn, think and solve problems 1: Be tter than other children {his/her} age 2: As well as other children 3: Slightly less well than other children 4: Much less well than other children DP1CBLEARN 1: Better than other children {his/her} age 2: As well as other children 3: Slightly less well than other children 4: Much less well than other children

PAGE 279

279 Table B 1 Continued. Item PEELS Parent Interview Questions Derived Items Communicate with others NDSKNWN: Compared with other children about the same age, how well does {CHILD} make {his/her} needs known to you and others? Communication can be any form, for example crying, pointing, or talking Would you say {he/she} 1: Communicates just as well as other children 2: Has a little trouble communicating 3: Has a lot of trouble commun icating 4: Does not communicate at all? EASYUNDR: When {CHILD} talks to people {he/she} doesn't know well, is {he/she} 1: Very easy to understand 2: Fairly easy to understand 3: Somewhat hard to understand 4: Very hard to understand 5: Does not or will not talk at all DP1PROBCOMC 1: Communicates just as well as other children and very easy to understand 2: Some difficulties communicating or being understood 3: Moderate difficulties communicating or being understood 4: Does not communicate at all or very Understanding VERBCOMM: Compared with other children about the same age, how would you describe nonverbal communication (signs, gestures, symbol systems)? Would you say {he/she} 1: Understands just as well as other children 2: Has a little trouble understanding 3: Has a lot of trouble understanding 4: Does not understand at all DP1VERBCOMM 1: Understands just as well as other children 2: Has a little trouble understanding 3: Has a lot of trouble understanding 4: Does not understand at all Regulation of Attention CBPYATTN: Some children are good at paying attention to things and staying focused on what they are doing Does this sound 1: Very much like {CHILD} 2: A little like {him/her} 3: Not like {him/her} DP1CBPYATTNR4 1: Very much like {CHILD} 2: A little like {him/her} 4: Not like {him/her} Regulation of Feeling and Emotions CBDEPRSD: Some children are frequently anxious or depressed Does this sound 1: Very much like {CHILD} 2: A little like {him/her} 3: Not like {him/her} DP1CBDEPRSDR4 1: Not like {him/her} 2: A little like {him/her} 4: Very much like {CHILD} Regulation of Activity Level CBRSTLSS: Some children are restless, fidget a lot, and have trouble sitting still. Does this sound 1: Very much like {CHILD} 2: A little like {him/her} 3: Not like {him/her} DP1CBRSTLSSR4 1: Not like {him/her} 2: A little like {him/her} 4: Very much like {CHILD}

PAGE 280

280 Table B 1 Continued. Item PEELS Parent Interview Questions Derived Items Motivation CBFINISH: Some children try to finish things, even if it takes a long time Does this sound 1: Very much like {CHILD} 2: A little like {him/her} 3: Not like {him/her} DP1CBFINISHR4 1: Very much like {CHILD}, 2: A little like {him/her} 4: Not like {him/her} Social Skills CBPLAYNG: Would you say that {CHILD} 1: Has no trouble playing with 2: Has some trouble playing 3: Has a lot of trouble playing 4: not around other children CBFRIEND: Some children have a lot of trouble making or keeping friends Does this sound 1: Not like {CHILD} 2: A little like {him/her} 3: Very much like {him/her} CBTKTURN: When some children are with other children their same age, they take turns and cooperate Does this sound 1: Very much like {CHILD} 2: A little like {him/her} 3: Not like {him/her} 4: child never interacts with peers DP1PROBSOC 1: No trouble playing with children, making friends, taking turns with other children 2: A little trouble playing with children, making friends, taking turns with other children 3: Moderate trouble playing with children, making friends, taking turns with other children 4: A lot of trouble playing with children, making friends, taking turns with other children Inappropriate or unusual behavior CBMANAGE: Would you say {CHILD} is 1: Easy to manage 2: Sometimes hard to manage 3: Often hard to manage behavior is 1: Is typical and appropriate for {his/her} age 2: Is mildly inappropriate 3: Is moderately inappropriate DP1PROBBEH 1: Child is easy to manage and behavior is appropriate for age 2: Some difficulty with managing behavior or inappropriate behavior for age 3: Moderate difficulty with managing behavior or inappropriate behavior for age 4: Child is difficult to manage and behavior i s moderately or severely inappropriate for age Note Alpha numeric codes represent variable ID in the PEELS data set All items from wave 1 parent interview file Constructing and testing a disability index in a US sample of preschoolers with disabilities Daley, R Sime o nsson, and E Carlson 2009, Disability and Rehabilitation, 31 pp 550 552

PAGE 281

281 APPENDIX C CHILD, FAMILY, AND ENVIRONMENTAL VARIABLES Table C 1 Child f actor v ariables from PEELS d ata s et Variable Related PEELS Question(s) Coding Child Factors Child sex a moderator variable CHDSEX 1 : Male 2 : Female Child age a *moderator variable ASSESSAGMW1 (Age at Assessment for Wave 1) Valid Responses: 37 76 (Months) Race/ ethnicity b *moderator variable a list of categories Please choose race 1: White (CHRACEWH) 2: African American or Black (CHRACEBL) 3: American Indian or Alaska Native (CHRACEAI) 4: Asian (CHRACEAS) 5: Native Hawa iian or other Pacific Islander (CHRACEPI) CHDETHN: Is {CHILD} of Hispanic, Latino, or other Spanish origin? 1 YES 2 NO 1 : Caucasian /N on Hispanic 2 : Hispanic 3 : African American or Black 4 : American Indian or Alaska Native 5 : Asian, or Native Hawaiian or other Pacific Islander 6 : Multi racial/ethnicity English as a second language b CHDLANG Any language other than English regularly spoken in child's home 1 : YES 2 : NO 1: Yes 2: No Child had IFSP b IFSPLAN Child had an IFSP before age 3 1 : YES 2 : NO 1 : Yes 2: No Birth weight b PBRTHOZ Birth weight in ounces Valid Responses: 16 229 Transformed to grams Weeks premature b ERLYNUM Number of weeks early child was born Valid Responses: 3 20 1 inapplicable Valid Responses: 0 20 (weeks) Notes Alpha numeric codes represent variable ID in the PEELS data set Coding associated with inapplicable; these responses will be coded as missing data or no based on response options for variable a refers to variables identified in demographic file. b refers to variables identified in parent interview file.

PAGE 282

282 Table C 2 Family c haracteristics v ariables from PEELS d ata s et Variable Related PEELS Question(s) Coding Family Factors: Family Characteristics Respondent role b RESPRNT Respondent type of parent/guardian 1: Biological 2: Adoptive 3: Step 4: Foster RESTYPE Respondent's relationship to child 1: Mother 2: Father 3: Brother 4: Sister 5: Grandmother 6: Grandfather 7: Aunt 8: Uncle 9: Cousin 11: Other relative 12: Non relative 1: Biological mother 2: Adoptive mother 3: Biological father 4: Other role Home living environment b CHDLVNOW : Child currently lives with respondent ? 1: Yes 2: No Same as indicated Marital status b *moderator variable MARSTATS : Respondent's legal marital status : 1: Never married 2: Married 3: Separated 4: Divorced 5: Widowed 1: Married 2: Not married Family income b *moderator variable INCME: Please tell me which group best describes the total income of all persons in your household over the past year, including salaries or other earnings, money from public assistance, child support, retirement, and so on, for all household members Was your household 1: $25,000 or less, or 2: More than $25,000?

PAGE 283

283 Table C 2 Continued. Variable Related PEELS Question(s) Coding Family Factors: Family Characteristics HOWMCH 1: $5,000 or less, or 2: $5,001 to $10,000, 3: $10,001 to $15,000, 4: $15,001 to $20,000, or 5: $20,001 to $25,000? INC25_50 1: $25,001 to $30,000, 2: $30,001 to $35,000, 3: $35,001 to $40,000, 4: $40,001 to $45,000, 5: $45,001 to $50,000, or 6: More than $50,000? 1: less than 5,000 2: 5,001 10,000 3: 10,001 15,000 4: 15,001 20,000 5: 20,001 25,000 6: 25,001 30,000 7: 30,001 35,000 8: 35,001 40,000 9: 40,001 45,000 10: 45,001 50,000 11: more than 50,000 (Variable used as a continuous variable for moderation analyses) Parent education b *moderator variable GRADE: What is the highest year or grade you finished in school? 1: Less than High School with no GED 2: High School diploma or GED 3: Some college/post secondary vocational course 4: 2 or 3 yearcollege degree (AA degree) or vocational school diploma 5: 4 year college degree (BA, BS degree) 6: Some graduate work/no graduate degree 7: Graduate degree (MA, MBA, Ph.D., JD, MD) Same as indicated Notes Alpha numeric codes represent variable ID in the PEELS data set Coding associated with inapplicable; these responses will be coded as missing data or no based on re sponse options for variable a refers to variables identified in demographic file. b refers to variables identified in parent interview file. c refers to variables identified in administrator questionnaire file d refers to variables identified in teacher questionnaire file.

PAGE 284

284 Table C 3 Parent child i n teraction v ariables from PEELS d ata s et Variable Related PEELS Question(s) Coding Family Factors: Parent Child In teraction Parent school activities b *moderator variable Preschool Questions Since the beginning of the school year, have you or another adult in the household done the following at [1=YES, 2= NO] PATNDMT: a Attended a general school or program meeting, for example, back to school night, or a meeting of a parent teacher organization? PATNDSE: b Attended a school or class event, such as a play, sports event, or science fair? PATNDVL: c minutes? PATNDTRP: d Helped with field trips or other special events? PATNPTC: e Attended parent teacher conferences? PATNPOL: f Participated in Policy Council, monitoring related activities, or other school or program planning groups? PATNFND: g Participated in f undraising activities? Kindergarten Questions Since the beginning of the school year, have you or another adult in the household done any of the [1=YES, 2= NO] KATNDMT: a Attended a general school meeting, for exam ple, back to school night, or a meeting of a parent teacher organization? KATNDSE: b Attended a school or class event, such as a play, sports event, or science fair? KATNDVL: c minutes? Sum of yes responses for each activity type; scored 0 7

PAGE 285

285 T able C 3 Continued. Variable Related PEELS Question(s) Coding Family Factors: Parent Child In teraction KATNDTRP: d Helped with field trips or other special events? KATNPTC: e Attended parent teacher conferences? KATNPOL: f Participated in Policy Council, monitoring related activities, or other school planning groups? KATNFND: g Participated in fundraising activities? Child extra curricular a ctivities b *moderator variable GDANCE: {Outside of school hours,} has {CHILD} ever participated in dance lessons? 1: YES 2: NO GATHLETE: {Outside of school hours,} has {he/she} ever participated in organized athletic activities, like gymnastics, soccer, baseball, or basketball? 1: YES 2: NO GCLUBS: {Outside of school hours,} has {CHILD} ever participated in organized clubs or recreational programs, like scouts? 1: YES 2: NO GMUSIC: {Outside of school hours,} has {he/she} ever participated in music lessons, such as piano, instrumental music, or singing lessons? 1: YES 2: NO GDRAMA: {Outside of school hours,} has {CHILD} ever participated in drama classes? 1: YES 2: NO GARTCLSS: {Outside of school hours,} has {he/she} ever participated in art or crafts classes or le ssons, such as painting, drawing, or sculpturing? 1: YES 2: NO Sum of yes responses for each activity type; scored 0 7

PAGE 286

286 Table C 3 Continued. Variable Related PEELS Question(s) Coding Family Factors: Parent Child In teraction GPERFORM: {Outside of school hours,} has {CHILD} ever participated in organized performing arts programs, performances? 1: YES 2: NO Child participation in group activities b *moderator variable GGRPACTV: Are there any hours, play groups, lessons, Sunday schools, gym programs, or other programs that {CHILD} goes to at least once a month? 1: YES 2: NO Same as indicated Regular child group activities b *moderator variable What gr oup activities does {he/she} go to at least monthly? GPLAYGRP: a babysitting with other children) GSTORYHR: b Story hour (at library) GSUNSCHL: c Sunday School/Church childcare GLESSONS: d Lessons (swimming, art, Gymboree) GTEAMS: e Athletic teams (soccer t ball) GSCOUTS: f GACTOTHR: g Other Sum of yes responses for each activity type; scored 0 7 Parent child activities b *moderator variable In the past month, has anyone in your family done the following things with {CHILD}? [1= YES, 2 = NO, 9 = DOES NOT APPLY] GPSTGRCY: a Gone to a grocery store? Sum of yes responses for each activity type; scored 0 7

PAGE 287

287 Table C 3 Continued. Variable Related PEELS Question(s) Coding Family Factors: Parent Child In teraction GPSTMALL: b Gone to a shopping mall, department store, or discount store? GPSTREST: c Gone to a restaurant or fast food place? GPSTPARK: d Gone to a public park or playground? GPSTCHRH: e Gone to a church, synagogue, or place of worship? GPSTLBRY: f Gone to the library? GPSTMVIE: g Gone to a movie? Family meals b *moderator variable GEATMEAL: How many days out of a typical week does your family eat the evening meal together? Number indicated 0 7 Read to b *moderator variable GREADTO: How many times have you or someone in your family 1: Never 2: Once or twice 3: 3 to 6 times 4: Every day Same as indicated Notes Alpha numeric codes represent variable ID in the PEELS data set Coding associated with inapplicable; these responses will be coded as missing data or no based on response options for variable a refers to variables identified in demographic file. b refers to variables identified in parent interview file. c refers to variables identified in administrator questionnaire file d refers t o variables identified in teacher questionnaire file.

PAGE 288

288 Table C 4 Environmental f actor v ariables from the PEELS d ata s et Variable Related PEELS Question(s) Coding Environmental Factors (Community and School) Neighborhood Safety b *moderator variable GPLYSAFE : How safe is it for children to play outside during the 1: Not at all safe 2: Somewhat safe 3: Very safe Same as indicated School/ Neighborhood Income c *moderator variable Kindergarten Elementary School Questions S1LB3M1: from low income families (e.g., have a child in the free or reduced price lunch program)? 1: Less than 25% 2: 25% 50% 3: 51 75% 4: More than 75% Preschool Questions S1EB3M1: What percentage of the children ages 3 through 5 whom you serve live In low income households (e.g., receive income assistance or food stamps)? 1: Less than 25% 2: 25% 50% 3: 51 75% 4: More than 75% Same as indicated School/ program quality c Kindergarten Elementary School Questions Has your school been designated as a school in need of improvement or a low performing school under the No Child Left Behind Act? Is this a.. S1LA7_1: School in need of improvement? 1:YES 2: NO S1LA7_2: Low p erforming school? 1:YES 2: NO Preschool Questions S1EA8AM1: Is your program licensed or accredited? 1:YES 2: NO For Kindergarten Elementary 1: Meet NCLB standard 2: Low performing or in need of improvement For Preschool Same as indicated

PAGE 289

289 Table C 4 Continued. Variable Related PEELS Question(s) Coding Environmental Factors (Community and School) Parent satisfaction with services b ESATISFD Overall respondent satisfaction with special education services 1: Very satisfied 2: Satisfied 3: Dissatisfied 4: Very dissatisfied Same as indicated Classroom intervention to support social competence d *moderator variable T1 KA47 or T1EA29 Does your program support social interaction between this child and children without disabilities? 1: YES 2: N/A children without disabilities not enrolled in class 3: N/A child does not have contact with children without disabilities 4: N/A no support needed 5: NO 1: Yes 2: No Number of child with IEP in classroom d T1EA2_1 or T1KA7A Number of children with IEP in class Valid Responses: 0 26 Same as indicated Number of child without IEP in classroom d T1EA2_2 or T1KA7B Number of children without IEP in class Valid Responses: 0 33 Same as indicated Focus of IEP goals d For this school year, what are the three most important IEP goals for this child? Please check up to three. T1ED5A or T1KB3B a: Improve overall school readiness T1ED5B or T1KB3C b: Improve pre academic performance in a specific area T1ED5C or T1KB3D c: Improve social skills T1ED5D or T1KB3E D: Improve appropriateness of general behavior T1ED5E or T1KB3F e: Improve adaptive behavior or self help skills For each goal type 1: Yes 0: No

PAGE 290

290 Table C 4. Continued. Variable Related PEELS Question(s) Coding Environmental Factors (Community and School) T1ED5F or T1KB3G f: Improve speech/communication skills T1ED5G or T1KB3H g: Improve fine motor skills T1ED5H or T1KB3I h: Improve gross motor skills T1ED5I or T1KB3J i: Other Notes Alpha numeric codes represent variable ID in the PEELS data set Coding associated with inapplicable; these responses will be coded as missing data or no based on response options for variable a refers to variables identified in demographic file. b refers to variables identified in parent interview file. c refers to variables identified in administrator questionnaire file d refers to variables identified in te acher questionnaire file.

PAGE 291

291 APPENDIX D VARIABLE CODING AND ANALYTIC SYNTAX Variable Coding Syntax C oding for Disability Index (Provided by Daley, Carlson, & Simeonsson, 2009) LIBNAME DATA 'C: \ DATA'; DATA DATA.W1_PARENT; SET WORK.W1parent; RUN; DATA data.W1_dis_indx_15VAR; set data.W1_PARENT; DP1PROBCOMC=0; DP1PROBHEALTH=0; DP1CBRSTLSSR4= 0; DP1CBPYATTNR4=0; DP1PROBHEAR=0; DP1PROBSOC=0; DP1PROBBEH=0; DP1BARMSGMS=0; DP1BARMSFMS=0; DP1BLEGSWEL=0; DP1PROBVISION=0; DP1CBDEPRSDR4=0; DP1CBFINISHR4=0; D P1SEVERINDX15VAR=0; /*Create DP1PROBCOMC by combining P1NDSKNWN and P1EASYUNDR*/ If P1NDSKNWN=1 and P1EASYUNDR=1 then DP1PROBCOMC=1; Else If P1NDSKNWN=1 and P1EASYUNDR=2 then DP1PROBCOMC=1; Else If P1NDSKNWN=2 and P1EASYUNDR=1 then DP1PROBCOMC=2; Else I f P1NDSKNWN=2 and P1EASYUNDR=2 then DP1PROBCOMC=2; Else If P1NDSKNWN=3 and P1EASYUNDR=1 then DP1PROBCOMC=2; Else If P1NDSKNWN=1 and P1EASYUNDR=3 then DP1PROBCOMC=3; Else If P1NDSKNWN=1 and P1EASYUNDR=4 then DP1PROBCOMC=3; Else If P1NDSKNWN=3 and P1EASYUNDR=2 then DP3PROBCOMC=3; Else If P1NDSKNWN=2 and P1EASYUNDR=3 then DP1PROBCOMC=3; Else If P1NDSKNWN=3 and P1EASYUNDR=3 then DP1PROBCOMC=3; Else If P1NDSKNWN=4 and P1EASYUNDR=1 then DP1PROBCOMC=4; Else If P1NDSKNWN=4 and P1EASYUNDR=2 then DP1PROBCOM C=4; Else If P1NDSKNWN=4 and P1EASYUNDR=3 then DP1PROBCOMC=4; Else If P1NDSKNWN=4 and P1EASYUNDR=4 then DP1PROBCOMC=4; Else If P1NDSKNWN=1 and P1EASYUNDR=5 then DP1PROBCOMC=4; Else If P1NDSKNWN=2 and P1EASYUNDR=5 then DP1PROBCOMC=4; Else If P1NDSKNWN=3 and P1EASYUNDR=5 then DP1PROBCOMC=4;

PAGE 292

292 Else If P1NDSKNWN=4 and P1EASYUNDR=5 then DP1PROBCOMC=4; Else If P1NDSKNWN=3 and P1EASYUNDR=4 then DP1PROBCOMC=4; Else If P1NDSKNWN=2 and P1EASYUNDR=4 then DP1PROBCOMC=4; label DP1PROBCOMC = INDEX child communication with others Wave 1; /*Create D1PROBHEALTH by combining P1BHLTHCMP and P1ACTLMTD*/ If P1BHLTHCMP=1 and P1ACTLMTD=2 then D1PROBHEALTH=1; Else If P1BHLTHCMP=2 and P1ACTLMTD=2 then D1PROBHEALTH=1; Else If P1BHLTHCMP=3 and P1ACTLMTD=2 then D1PROBHEALTH=2; Else If P 1BHLTHCMP=1 and P1ACTLMTD=1 then D1PROBHEALTH=2; Else If P1BHLTHCMP=2 and P1ACTLMTD=1 then D1PROBHEALTH=2; Else If P1BHLTHCMP=3 and P1ACTLMTD=1 then D1PROBHEALTH=3; Else If P1BHLTHCMP=4 and P1ACTLMTD=2 then D1PROBHEALTH=3; Else If P1BHLTHCMP=4 and P1ACTLMT D=1 then D1PROBHEALTH=4; Else If P1BHLTHCMP=5 and P1ACTLMTD=1 then D1PROBHEALTH=4; Else If P1BHLTHCMP=5 and P1ACTLMTD=2 then D1PROBHEALTH=4; label D1PROBHEALTH = INDEX Child overall health Wave 1; /*create new variable DP1CBRSTLSSR4 reverse coding*/ If P1CBRSTLSS=1 then DP1CBRSTLSSR4 = 4; Else if P1CBRSTLSS=2 then DP1CBRSTLSSR4 = 2; Else if P1CBRSTLSS=3 then DP1CBRSTLSSR4 = 1; label DP1CBRSTLSSR4 = INDEX Child regulation of activity level Wave 1; /*create new variable DP1CBPYATTNR4 shift 3 value to 4*/ If P1CBPYATTN = 1 then DP1CBPYATTNR4 = 1; Else if P1CBPYATTN = 2 then DP1CBPYATTNR4 = 2; Else if P1CBPYATTN = 3 then DP1CBPYATTNR4 = 4; label DP1CBPYATTNR4 = INDEX Child regulation of attention Wave 1; /*Rename cognition*/ DP1CBLEARN=P1CBLEARN; LABEL DP1C BLEARN = INDEX Child cognition Wave 1; /*Rename understanding*/ DP1VERBCOMM=P1VERBCOMM; label DP1VERBCOMM = INDEX Child understanding communication Wave 1; /*Create DP1PROBHEAR by com bining P1HEARCMP, P1WELHRDV and P1HRNGLSS*/ If P1HEARCMP=1 Then DP1PROB HEAR=1; Else If P1HEARCMP=2 and P1HEARTSTD=2 Then DP1PROBHEAR=2; Else If P1HEARCMP=2 and P1HEARTSTD=3 Then DP1PROBHEAR=2; Else If P1HEARCMP=2 and P1HEARTSTD= 8 Then DP1PROBHEAR=2; Else If P1HEARCMP=2 and P1HEARTSTD= 9 Then DP1PROBHEAR=2; Else If P1HEARCMP= 2 and P1DIAGPROF=1 Then DP1PROBHEAR=3;

PAGE 293

293 Else If P1HRNGLSS=1 and P1WELHRDV=1 Then DP1PROBHEAR=3; Else If P1HRNGLSS=1 and P1WELHRDV=2 Then DP1PROBHEAR=3; Else If P1HRNGLSS=2 and P1WELHRDV=1 Then DP1PROBHEAR=3; Else If P1HRNGLSS=2 and P1WELHRDV=2 Then DP1PROBHEAR=3; Else If P1HRNGLSS=3 and P1WELHRDV=1 Then DP1PROBHEAR=4; Else If P1HRNGLSS=4 and P1WELHRDV=1 Then DP1PROBHEAR=4; Else If P1HRNGLSS=1 and P1WELHRDV=3 Then DP1PROBHEAR=4; Else If P1HRNGLSS=2 and P1WELHRDV=3 Then DP1PROBHEAR=4; Else If P1HRNGLSS =3 and P1WELHRDV=2 Then DP1PROBHEAR=4; Else If P1HRNGLSS=4 and P1WELHRDV=2 Then DP1PROBHEAR=4; Else If P1HRNGLSS=1 and P1WELHRDV=4 Then DP1PROBHEAR=4; Else If P1HRNGLSS=2 and P1WELHRDV=4 Then DP1 PROBHEAR=4; Else If P1HRNGLSS=3 and P1WELHRDV=4 Then DP1PROBH EAR=4; Else If P1HRNGLSS=4 and P1WELHRDV=4 Then DP1PROBHEAR=4; Else If P1HRNGLSS=4 and P1WELHRDV=3 Then DP1PROBHEAR=4; Else If P1HRNGLSS=3 and P1WELHRDV=3 Then DP1PROBHEAR=4; label DP1PROBHEAR = INDEX Child hearing Wave 1; /* Problems with social skills: Create DP1PROBSOC */ If P1CBPLAYNG = 1 then P1CBPLAYNG = .; If P1CBPLAYNG = 7 then P1CBPLAYNG = .; If P1CBPLAYNG = 8 then P1CBPLAYNG = .; If P1CBPLAYNG = 9 then P1CBPLAYNG = .; If P1CBPLAYNG = 4 then P1CBPLAYNG = .; If P1CBPLAYNG = 2 then P1CBPL AYNG = .; If P1CBTKTURN = 1 then P1CBTKTURN = .; If P1CBTKTURN = 7 then P1CBTKTURN = .; If P1CBTKTURN = 8 then P1CBTKTURN = .; If P1CBTKTURN = 9 then P1CBTKTURN = .; If P1CBTKTURN = 4 then P1CBTKTURN = .; If P1CBTKTURN = 2 then P1CBTKTURN = .; If P1CBFRIEND=1 then DP1CBFRIENDR=3; If P1CBFRIEND=2 then DP1CBFRIENDR=2; If P1CBFRIEND=3 then DP1CBFRIENDR=2; If P1CBPLAYNG=1 and DP1CBFRIENDR=1 and P1CBTKTURN=1 Then DP1PROBSOC=1; Else If P1CBPLAYNG=1 and DP1CBFRIENDR=2 and P1CBTKTURN=1 Then DP1PROBSOC=1 ; Else If P1CBPLAYNG=1 and DP1CBFRIENDR=1 and P1CBTKTURN=2 Then DP1PROBSOC=1; Else If P1CBPLAYNG=1 and DP1CBFRIENDR=1 and P1CBTKTURN=3 Then DP1PROBSOC=1; Else If P1CBPLAYNG=1 and DP1CBFRIENDR=2 and P1CBTKTURN=2 Then DP1PROBSOC=1;

PAGE 294

294 Else If P1CBPLAYNG=1 and DP1CBFRIENDR=2 and P1CBTKTURN=3 Then DP1PROBSOC=1; Else If P1CBPLAYNG=1 and DP1CBFRIENDR=3 and P1CBTKTURN=2 Then DP1PROBSOC=1; Else If P1CBPLAYNG=1 and DP1CBFRIENDR=3 and P1CBTKTURN=3 Then DP1PROBSOC=1; Else If P1CBPLAYNG=1 and DP1CBFRIENDR=3 and P1CBTKTUR N=1 Then DP1PROBSOC=1; Else If P1CBPLAYNG=2 and DP1CBFRIENDR=1 and P1CBTKTURN=1 Then DP1PROBSOC=2; Else If P1CBPLAYNG=2 and DP1CBFRIENDR=1 and P1CBTKTURN=2 Then DP1PROBSOC=2; Else If P1CBPLAYNG=2 and DP1CBFRIENDR=2 and P1CBTKTURN=1 Then DP1PROBSOC=2; Else If P1CBPLAYNG=2 and DP1CBFRIENDR=1 and P1CBTKTURN=3 Then DP1PROBSOC=2; Else If P1CBPLAYNG=2 and DP1CBFRIENDR=3 and P1CBTKTURN=1 Then DP1PROBSOC=2; Else If P1CBPLAYNG=2 and DP1CBFRIENDR=2 and P1CBTKTURN=2 Then DP1PROBSOC=3; Else If P1CBPLAYNG=2 and DP1CBFRI ENDR=2 and P1CBTKTURN=3 Then DP1PROBSOC=3; Else If P1CBPLAYNG=2 and DP1CBFRIENDR=3 and P1CBTKTURN=2 Then DP1PROBSOC=3; Else If P1CBPLAYNG=2 and DP1CBFRIENDR=3 and P1CBTKTURN=3 Then DP1PROBSOC=3; Else If P1CBPLAYNG=3 and DP1CBFRIENDR=1 and P1CBTKTURN=1 Then DP1PROBSOC=4; Else If P1CBPLAYNG=3 and DP1CBFRIENDR=1 and P1CBTKTURN=2 Then DP1PROBSOC=4; Else If P1CBPLAYNG=3 and DP1CBFRIENDR=1 and P1CBTKTURN=3 Then DP1PROBSOC=4; Else If P1CBPLAYNG=3 and DP1CBFRIENDR=2 and P1CBTKTURN=1 Then DP1PROBSOC=4; Else If P1CBPLAYNG=3 and DP1CBFRIENDR=2 and P1CBTKTURN=2 Then DP1PROBSOC=4; Else If P1CBPLAYNG=3 and DP1CBFRIENDR=2 and P1CBTKTURN=3 Then DP1PROBSOC=4; Else If P1CBPLAYNG=3 and DP1CBFRIENDR=3 and P1CBTKTURN=2 Then DP1PROBSOC=4; Else If P1CBPLAYNG=3 and DP1CBFRIEND R=3 and P1CBTKTURN=3 Then DP1PROBSOC=4; Else If P1CBPLAYNG=3 and DP1CBFRIENDR=3 and P1CBTKTURN=1 Then DP1PROBSOC=4; label DP1PROBSOC = INDEX Child Social Skills Wave 1;

PAGE 295

295 /*Create DP1PROBBEH by combining P1CBMANAGE and P1CBBEHAVR*/ If P1CBMANAGE=1 a nd P1CBBEHAVR=1 Then DP1PROBBEH=1; Else If P1CBMANAGE=1 and P1CBBEHAVR=2 Then DP1PROBBEH=1; Else If P1CBMANAGE=2 and P1CBBEHAVR=1 Then DP1PROBBEH=1; Else If P1CBMANAGE=2 and P1CBBEHAVR=2 Then DP1PROBBEH=2; Else If P1CBMANAGE=2 and P1CBBEHAVR=3 Then DP1PROB BEH=2; Else If P1CBMANAGE=3 and P1CBBEHAVR=2 Then DP1PROBBEH=2; Else If P1CBMANAGE=1 and P1CBBEHAVR=3 Then DP1PROBBEH=3; Else If P1CBMANAGE=3 and P1CBBEHAVR=1 Then DP1PROBBEH=3; Else If P1CBMANAGE=1 and P1CBBEHAVR=4 Then DP1PROBBEH=3; Else If P1CBMANAGE=2 and P1CBBEHAVR=4 Then DP1PROBBEH=3; Else If P1CBMANAGE=3 and P1CBBEHAVR=3 Then DP1PROBBEH=3; Else If P1CBMANAGE=3 and P1CBBEHAVR=4 Then DP1PROBBEH=4; label DP1PROBBEH = INDEX Child Inappropriate behavior Wave 2; /* Create new variable: DP1BARMSGMS shift 5 value to 4*/ If P1BARMSGMS = 1 then DP1BARMSGMS = 1; Else if P1BARMSGMS = 2 then DP1BARMSGMS = 2; Else if P1BARMSGMS = 3 then DP1BARMSGMS = 3; Else if P1BARMSGMS = 4 then DP1BARMSGMS = 4; Else if P1BARMSGMS = 5 then DP1BARMSGMS = 4; label DP1BARMSGMS = INDEX Child use of arms Wave 1; /* Create new variable: DP2BARMSFMSL shift 5 value to 4*/ If P1BARMSFMS = 1 then DP1BARMSFMS = 1; Else if P1BARMSFMS = 2 then DP1BARMSFMS = 2; Else if P1BARMSFMS = 3 then DP1BARMSFMS = 3; Else if P1BARMSFMS = 4 then DP1BAR MSFMS = 4; Else if P1BARMSFMS = 5 then DP1BARMSFMS = 4; label DP1BARMSFMS = INDEX Child use of hands Wave 1; /* Create new variable: DP1BLEGSWEL shift 5 value to 4*/ If P1BLEGSWEL = 1 then DP1BLEGSWEL = 1; Else if P1BLEGSWEL = 2 then DP1BLEGSWEL = 2; Els e if P1BLEGSWEL = 3 then DP1BLEGSWEL = 3; Else if P1BLEGSWEL = 4 then DP1BLEGSWEL = 4; Else if P1BLEGSWEL= 5 then DP1BLEGSWEL = 4; label DP1BLEGSWEL= INDEX Child use of legs Wave 1; /* Create new variable: DP1PROBVISION from P1CHDEYEST, P1VSWOGLS, and P1V SWGLS*/ If P1CHDEYEST=1 Then DP1PROBVISION=1; Else If P1VSWOGLS=1 Then DP1PROBVISION =1; Else If P1VSWTHGLS=1 Then DP1PROBVISION=2; Else If P1VSWTHGLS=2 Then DP1PROBVISION=3; Else If P1VSWTHGLS=3 Then DP1PROBVISION=4;

PAGE 296

296 Else If P1VSWTHGLS=4 Then DP1PROBVISIO N=4; label DP1PROBVISION = INDEX Child vision Wave 1; /*create new variable DP1CBDEPRSDR reverse coding*/ If P1CBDEPRSD = 1 then DP1CBDEPRSDR = 3; Else if P1CBDEPRSD = 2 then DP1CBDEPRSDR = 2; Else if P1CBDEPRSD = 3 then DP1CBDEPRSDR = 1; If DP1CBDEPRSDR = 3 then DP1CBDEPRSDR4=4; Else DP1CBDEPRSDR4 = DP1CBDEPRSDR; label DP1CBDEPRSDR4 = INDEX Child regulation of feelings emotions Wave 1; /*create new variable DP1CBFINISHR4 replace 3 with 4*/ If P1CBFINISH = 1 then DP1CBFINISHR4 = 1; Else if P1CBFINISH = 2 then DP1CBFINISHR4 = 2; Else if P1CBFINISH = 3 then DP1CBFINISHR4 = 4; label DP1CBFINISHR4 = INDEX Child motivation Wave 1; /*alternate code for 3 to 4 shifts; If P1CBFINISH = 3 then DP1CBFINISH4=4; Else DP1CBFINISH4 = P1CBFINISH;*/ TMP1= DP1CBLEARN+DP1VERBCOMM+DP1PROBCOMC+D1PROBHEALTH+DP1C BRSTLSSR4+DP1CBPYATTNR4; TMP2 = DP1PROBHEAR+DP1PROBSOC+DP1PROBBEH+DP1BARMSGMS+DP1BAR MSFMS+DP1BLEGSWEL; TMP3 =DP1PROBVISION+DP1CBDEPRSDR4+DP1CBFINISHR4; DP1SEVERINDX15VAR = TMP1+TMP2+TMP3; label DP1SEVERIND X15VAR = Var 15 Severity of disability index temp(unweighted)W1; keep DP1PROBHEAR DP1PROBSOC DP1PROBBEH DP1BARMSGMS DP1BARMSFMS DP1BLEGSWEL DP1PROBVISION DP1CBDEPRSDR4 DP1CBFINISHR4; keep DP1CBLEARN DP1VERBCOMM DP1PROBCOMC D1PROBHEALTH DP1CBRSTLSSR4 DP1CB PYATTNR4; keep DP1SEVERINDX15VAR CHILDID P1CURMMS P1CHCURGRD; run; DATA data.index_mplus; set data.W1_dis_indx_15VAR; comunicate=DP1PROBCOMC; health=D1PROBHEALTH; activty=DP1CBRSTLSSR4; attention=DP1CBPYATTNR4; hear=DP1PROBHEAR;

PAGE 297

297 social=DP1PROBSOC; behavior=DP1PROBBEH; arms=DP1BARMSGMS; hands=DP1BARMSFMS; legs=DP1BLEGSWEL; vision=DP1PROBVISION; emotion=DP1CBDEPRSDR4; motivate=DP1CBFINISHR4; cognition=DP1CBLEARN; understand=DP1VERBCOMM; age=P1CURMMS; KEEP CHILDID age comunicate health activty attentio n hear social behavior arms hands legs vision emotion motivate cognition understand; Run; Coding for Non Malleable Child Factors and Contextual Factors Variables DATA work.parent_var_all; SET tmp2.w1parent; ID = CHILDID + 0; keep id p1marstats P1INCME P1HOWMCH P1INC25_50 P1GRADE P1CHDLANG p1resprnt p1restype DP1PBRTHOZ; keep p1chdlvnow p1ifsplan p1erlynum; keep p1chracewh p1chracebl p1chraceai p1chraceas p1chracepi p1chdethn; /*RACE ETHNICTIY*/ K EEP P1ESATISFD; /*SATISFACTIO WITH SERV ICE*/ KEEP P1PATNDMT P1PATNDSE P1PATNDVL P1PATNDTRP P1PATNPTC P1PATNPOL P1PATNFND; /*PARENT PARTICPATION PRESCHOOL*/ KEEP P1KATNDMT P1KATNDSE P1KATNDVL P1KATNDTRP P1KATNPTC P1KATNPOL P1KATNFND; /*PARENT PARTICPATION KINDERGARDEN*/ KEEP P1GDANCE P1GATHLETE P1GCLUBS P1GMUSIC P1GDRAMA P1GARTCLSS P1GPERFORM ; /*CHILD PARTICPATION COMMUNITY ACTIVITIES*/ keep P1GGRPACTV P1GPLAYGRP P1GSTORYHR P1GSUNSCHL P1GLESSONS P1GTEAMS P1GSCOUTS P1GACTOTHR;/*CHILD PARTICPATION OTHER GR OUP ACTIVITIES*/ KEEP P1GPSTGRCY P1GPSTM ALL P1GPSTREST P1GPSTPARK P1GPSTCHRH P1GPSTLBRY P1GPSTMVIE; /*ACTIVTIES FAMILIES DO TOGETHER*/ keep P1GEATMEAL P1GREADTO; /*# O F TIMES FAMILY MEAL AND READ TO CHILD IN WEEK*/ KEEP P1GPLYSAFE; /* NEIGHBORHOOD SAFE*/ RUN; DATA work.principal_var;

PAGE 298

298 SET tmp2.w 1principal; ID = CHILDID + 0; KEEP ID S1LA7_1 S1LA7_2 S1LB3M1; RUN; DATA work.progdir_var; SET tmp2.w1progdir; ID = CHILDID + 0; KEEP ID S1EA8AM1 S1EB3M1; RUN; DATA work.ECTEACHER_var; SET tmp2.w1ecteacher; ID = CHILDID + 0; KEEP ID t1ea29 t1ea2_1 t1ea2_2; Keep t1ed5a t1ed5b t1ed5c t1ed5d t1ed5e t1ed5f t1ed5g t1ed5h t1ed5i; RUN; DATA work.kteacher_var; SET tmp2.w1kteacher; ID = CHILDID + 0; KEEP ID t1ka47 t1ka7A t1ka7B; keep t1kb3b t1kb3c t1kb3d t1kb3e t1kb3f t1kb3g t1kb3h t1kb3i t1kb3j; RUN; DAT A WORK.MOD_VAR_ALL; MERGE work.kteacher_var work.ECTEACHER_var work.progdir_var work.principal_var work.parent_var_all; BY ID; RUN; DATA WORK.MOD_VAR_ALL_REVISE; SET WORK.MOD_VAR_ALL; birth_wt = DP1PBRTHOZ; esl = P1CHDLANG; live_w_res = p1chdlvnow; sat_ses =P1ESATISFD; wk_prem = p1erlynum; if p1erlynum = 1 then wk_prem = 0; ifsp_pre3 = p1ifsplan; if p1ifsplan = 1 then ifsp_pre3 =2; /*IEP goals*/ g_sch_red = .; if t1ed5a = 0 then g_sch_red = 0; if t1ed5a = 1 then g_sch_red = 1; if t1kb3b = 0 t hen g_sch_red = 0;

PAGE 299

299 if t1kb3b = 1 then g_sch_red = 1; g_pre_ac = .; if t1ed5b = 0 then g_pre_ac = 0; if t1ed5b = 1 then g_pre_ac = 1; if t1kb3c = 0 then g_pre_ac = 0; if t1kb3c = 1 then g_pre_ac = 1; g_soc_sk_1 = .; if t1ed5c = 0 then g_soc_sk = 0; if t1ed5c = 1 then g_soc_sk = 1; if t1kb3d = 0 then g_soc_sk = 0; if t1kb3d = 1 then g_soc_sk = 1; g_beh_1 = .; if t1ed5d = 0 then g_beh = 0; if t1ed5d = 1 then g_beh = 1; if t1kb3e = 0 then g_beh = 0; if t1kb3e = 1 then g_beh = 1; g_adapt_1 = .; if t1ed5e = 0 then g_adapt = 0; if t1ed5e = 1 then g_adapt = 1; if t1kb3f = 0 then g_adapt = 0; if t1kb3f = 1 then g_adapt = 1; g_com_1 = .; if t1ed5f = 0 then g_com = 0; if t1ed5f = 1 then g_com = 1; if t1kb3g = 0 then g_com = 0; if t1kb3g = 1 then g_com = 1; g_Fmotor_1 = .; if t1ed5g = 0 then g_Fmotor = 0; if t1ed5g = 1 then g_Fmotor = 1; if t1kb3h = 0 then g_Fmotor = 0; if t1kb3h = 1 then g_Fmotor = 1; g_Gmotor_1 = .; if t1ed5h = 0 then g_Gmotor = 0; if t1ed5h = 1 then g_Gmotor = 1 ; if t1kb3i = 0 then g_Gmotor = 0; if t1kb3i = 1 then g_Gmotor = 1; g_other_1 = .; if t1ed5i = 0 then g_other = 0; if t1ed5i = 1 then g_other = 1; if t1kb3j = 0 then g_other = 0; if t1kb3j = 1 then g_other = 1; Res_role =.; if p1resprnt = 1 and p1 restype= 1 then res_role =1; /* biological mother*/ if p1resprnt = 1 and p1restype= 2 then res_role =2; /* biological father*/ if p1resprnt = 2 and p1restype= 1 then res_role =3; /*adoptive mother*/ if p1resprnt = 2 and p1restype= 2 then res_role =4; /*adoptive father*/

PAGE 300

300 if p1resprnt = 1 and p1restype= 5 then res_role =5; /* biological grandmother*/ if p1resprnt = 3 then res_role =6; /*other respondent*/ if p1resprnt = 4 then res_role =6; /*other respondent*/ if p1resprnt = 1 then res_role =6; /*oth er respondent*/ MAR_STATS =.; IF p1marstats = 1 THEN MAR_STATS = 2; /*NOT MARRIED*/ IF p1marstats = 2 THEN MAR_STATS = 1; /*MARRIED*/ IF p1marstats = 3 THEN MAR_STATS = 2; /*NOT MARRIED*/ IF p1marstats = 4 THEN MAR_STATS = 2; /*NOT MARRIED*/ IF p1marstats = 5 THEN MAR_STATS = 2; /*NOT MARRIED*/ MRACE = 6; if p1chracewh =1 and p1chdethn =2 AND p1chracebl =2 AND p1chraceai=2 AND p1chraceas =2 AND p1chracepi =2 then MRACE = 1; /*WHITE NON HISPANIC*/ if p1chdethn =1 and p1chracewh =1 AND p1chracebl =2 AND p1c hraceai=2 AND p1chraceas =2 AND p1chracepi =2 then MRACE = 2; /*HISPANIC*/ if p1chracebl = 1 and p1chracewh =2 AND p1chdethn =2 AND p1chraceai=2 AND p1chraceas =2 AND p1chracepi =2 then MRACE = 3; /*BLACK*/ if p1chraceai = 1 and p1chracewh =2 AND p1chrace bl =2 AND p1chdethn =2 AND p1chraceas =2 AND p1chracepi =2 then MRACE = 4; /* ALASKAN OR AMERICAN INDIAN*/ if p1chraceas = 1 and p1chracewh =2 AND p1chracebl =2 AND p1chraceai=2 AND p1chdethn =2 AND p1chracepi =2 then MRACE = 5; /*ASIAN */ if p1chracepi = 1 and p1chracewh =2 AND p1chracebl =2 AND p1chraceai=2 AND p1chraceas =2 AND p1chdethn =2 then MRACE = 5; /*PACFICI ISLAND*/ IF p1chracepi = THEN MRACE=.; Mpar_edu = .; /*parent education for moderation*/ if P1GRADE = 1 then Mpar_edu = 1; /*no HS*/ if P1 GRADE = 2 then Mpar_edu = 2; /* HS*/ if P1GRADE = 3 then Mpar_edu = 3; /*SOME COLLEGE*/ if P1GRADE = 4 then Mpar_edu = 4; /*2 OR 3 YR DEGREE*/ if P1GRADE = 5 then Mpar_edu = 5; /*4 YEAR DEGREE */ if P1GRADE = 6 then Mpar_edu = 6; /*SOME GRADUATE */ if P1GRADE = 7 then Mpar_edu = 7; /* GRADUATE DEGREE */ if P1GRADE = 1 then Mpar_edu = .; if P1GRADE = 9 then Mpar_edu =.; Dpar_edu = .; /*parent education for descriptive*/ if P1GRADE = 1 then Dpar_edu = 1; /*no HS*/ if P1GRADE = 2 then Dpar_edu = 2; /* HS*/ if P1GRADE = 3 then Dpar_edu = 3; /*SOME COLLEGE OR 2 OR 3 YR DEGREE*/ if P1GRADE = 4 then Dpar_edu = 3; /* SOME COLLEGE OR 2 OR 3 YR DEGREE*/ if P1GRADE = 5 then Dpar_edu = 4; /*4 YEAR DEGREE OR MORE */ if P1GRADE = 6 then Dpar_edu = 4; / *4 YEAR DEGREE OR MORE */ if P1GRADE = 7 then Dpar_edu = 4; /*4 YEAR DEGREE OR MORE */ if P1GRADE = 1 then Dpar_edu = .; if P1GRADE = 9 then Dpar_edu =.; Mincome = .; /*income for moderation*/

PAGE 301

301 if P1HOWMCH = 1 then Mincome = 1; /*LESS THEN 5,000*/ if P1HOWMCH = 2 then Mincome = 2; /* 5,000 10,000*/ if P1HOWMCH = 3 then Mincome = 3; /* 10,000 15,000*/ if P1HOWMCH = 4 then Mincome = 4; /* 15,000 20,000*/ if P1HOWMCH = 5 then Mincome = 5; /* 20,000 25,000*/ if P1INC25_50 = 1 then Mincome = 6; /* 25,000 30,000*/ if P1INC25_50 = 2 then Mincome = 7; /* 30,000 35,000*/ if P1INC25_50 = 3 then Mincome = 8; /* 35,000 40,000*/ if P1INC25_50 = 4 then Mincome = 9; /*40,000 45,000*/ if P1INC25_50 = 5 then Mincome = 10; /* 45,000 50,000*/ if P1INC2 5_50 = 6 then Mincome = 11; /* 50,000 AND ABOVE*/ dincome = .; /*income for descriptive*/ if P1HOWMCH = 1 then dincome = 1; /*LESS THEN 10,000*/ if P1HOWMCH = 2 then dincome = 1; /*LESS THEN 10,000*/ if P1HOWMCH = 3 then dincome = 2; /* 10,000 20,000*/ if P1HOWMCH = 4 then dincome = 2; /* 10,000 20,000*/ if P1HOWMCH = 5 then dincome = 3; /* 20,000 30,000*/ if P1INC25_50 = 1 then dincome = 3; /* 20,000 30,000*/ if P1INC25_50 = 2 then dincome = 4; /* 30,000 40,000*/ if P1INC25_50 = 3 then dincome = 4; /* 30,000 40,000*/ if P1INC25_50 = 4 then dincome = 5; /*40,000 50,000*/ if P1INC25_50 = 5 then dincome = 5; /* 40,000 50,000*/ if P1INC25_50 = 6 then dincome = 6; /* 50,000 AND ABOVE*/ MSAFE_NEIG = .; iF P1GPLYSAFE = 1 THEN MSAFE_NEIG = 1; /* NOT SAFE*/ iF P1GPLYSAFE = 2 THEN MSAFE_NEIG = 2; /* SOMEWHAT SAFE*/ iF P1GPLYSAFE = 3 THEN MSAFE_NEIG = 3; /* VERY SAFE*/ /*set up for summing activities*/ PPT1 =0; PPT2 =0; PPT3 =0; PPT4 =0; PPT5 =0; PPT6 =0; PPT7 =0; PPT8 =0; PPT9 =0; PPT10 =0; PPT11 =0 ; PPT12 =0; PPT13 =0; PPT14 =0; CA1 = 0; CA2 = 0; CA3 = 0; CA4 = 0; CA5 = 0; CA6 = 0; CA7 = 0; CA8 = 0; OT1 = 0; OT2 = 0; OT3 = 0; OT4 = 0; OT5 = 0; OT6 = 0; OT7 = 0; FA1 = 0; FA2 = 0; FA3 = 0; FA4 = 0; FA5 = 0; FA6 = 0; FA7 = 0; IF P1PATNDMT = 1 THEN PPT1=1; IF P1PATNDSE = 1 THEN PPT2=1; IF P1PATNDVL = 1 THEN PPT3=1; IF P1PATNDTRP = 1 THEN PPT4=1; IF P1PATNPTC = 1 THEN PPT5=1; IF P1PATNPOL = 1 THEN PPT6=1; IF P1PATNFND = 1 THEN PPT7=1; IF P1KATNDMT = 1 THEN PPT8=1; IF P1KATNDSE = 1 THEN PPT9=1; IF P1KATNDVL = 1 THEN PPT10=1; IF P1KATNDTRP = 1 THEN PPT11=1; IF P1KATNPTC = 1 THEN PPT12=1; IF P1KATNPOL = 1 THEN PPT13=1;

PAGE 302

302 IF P1KATNFND = 1 THEN PPT14=1; P_PART_SCH = PPT1 + PPT2 + PPT3 + PPT4 + PPT5 + PPT6 + PPT7 + PPT8 + PPT9 + PPT10 + PPT11+ PPT12 + PPT1 3 + PPT14; /*parent participation in school*/ IF P1GDANCE = 1 THEN CA1=1; If P1GATHLETE = 1 THEN CA3=1; IF P1GCLUBS = 1 THEN CA4=1; IF P1GMUSIC = 1 THEN CA5=1; IF P1GDRAMA= 1 THEN CA6=1; IF P1GARTCLSS = 1 THEN CA7=1; IF P1GPERFORM = 1 THEN CA8=1; C_PART_A CT = CA1 + CA3 + CA4 + CA5 + CA6 + CA7 + CA8; /*CHILD ACTIVITIES EVER*/ IF P1GGRPACTV= 1 THEN GEN_ACT_PART=1; /*other activities yes/no*/ if P1GPLAYGRP = 1 then OT1 =1; IF P1GSTORYHR = 1 then OT2 =1; IF P1GSUNSCHL = 1 then OT3 =1; IF P1GLESSONS = 1 the n OT4 =1; if P1GTEAMS = 1 then OT5 =1; if P1GSCOUTS = 1 then OT6 =1; if P1GACTOTHR = 1 then OT7 =1; C_OTH_ACT = OT1 + OT2 + OT3 + OT4 + OT 5 + OT6 + OT7; /*R EGULAR CHILD ACTIVITIES */ IF P1GPSTGRCY = 1 THEN FA1=1; IF P1GPSTMALL = 1 THEN FA2=1; IF P1GPST REST = 1 THEN FA3=1; IF P1GPSTPARK = 1 THEN FA4=1; IF P1GPSTCHRH = 1 THEN FA5=1; IF P1GPSTLBRY= 1 THEN FA6=1; IF P1GPSTMVIE= 1 THEN FA7=1; FAM_ACT = FA1 + FA2 + FA3 + FA4 + FA5 + FA6 + FA7; /*parent child activities*/ read_to = .; if P1GREADTO = 1 then read_to = 1; /*never*/ if P1GREADTO = 2 then read_to = 2; /*1 2 times a week*/ if P1GREADTO = 3 then read_to = 3; /*3 6 times a week*/ if P1GREADTO = 4 then read_to = 4; /*everyday*/ if P1GREADTO = 8 then read_to = .; if P1GREADTO = 9 then read_to = .; eat_meal = P1GEATMEAL; if P1GEATMEAL = 8 then eat_meal = .; if P1GEATMEAL = 9 then eat_meal = .; NEIG_INC =.; IF s1lb3m1 = 1 then NEIG_INC =1; /*less than 25% free reduced lunch*/ IF s1lb3m1 = 2 then NEIG_INC =2; /* 25% 50% free reduced lunch*/ IF s1lb 3m1 = 3 then NEIG_INC =3; /* 50 75% free reduced lunch*/ IF s1lb3m1 = 4 then NEIG_INC =4; /*more than 75% free reduced lunch*/ IF s1eb3m1 = 1 then NEIG_INC =1; /*less than 25% free reduced lunch*/

PAGE 303

303 IF s1eb3m1 = 2 then NEIG_INC =2; /* 25% 50% free reduced l unch*/ IF s1eb3m1 = 3 then NEIG_INC =3; /* 50 75% free reduced lunch*/ IF s1eb3m1 = 4 then NEIG_INC =4; /*more than 75% free reduced lunch*/ NCLB = 3; IF S1LA7_1 = 1 THEN NCLB =2; /* IN NEED OF IMPROVEMENT*/ IF S1LA7_2 = 1 THEN NCLB =1; /* LOW PERFORMING* / IF S1LA7_1 = THEN NCLB =.; IF S1LA7_2 = THEN NCLB =.; PRE_ACRD = S1EA8AM1; if S1EA8AM1 = 3 then PRE_ACRD = 2; SSUP_SS = .; IF t1ea29 = 1 then SSUP_SS = 1; IF t1ea29 = 2 then SSUP_SS = 2; IF t1ea29 = 3 then SSUP_SS = 2; IF t1ea29 = 4 then SSUP_SS = 2; IF t1ea29 = 5 then SSUP_SS = 2; IF t1ka47 = 1 then SSUP_SS = 1; IF t1ka47 = 2 then SSUP_SS = 2; IF t1ka47 = 3 then SSUP_SS = 2; IF t1ka47 = 4 then SSUP_SS = 2; IF t1ka47 = 5 then SSUP_SS = 2; DROP PPT1 PPT2 PPT3 PPT4 PPT5 PPT6 PPT7 PPT8 PPT9 PPT10 PPT1 1 PPT12 PPT13 PPT14; drop CA1 CA2 CA3 CA4 CA5 CA6 CA7 CA8; Drop OT1 OT2 OT3 OT4 OT5 OT6 OT7; Drop FA1 FA2 FA3 FA4 FA5 FA6 FA7; RUN; /* test all revised variable frequencies*/ proc freq data =work.MOD_VAR_ALL_REVISE; table birth_wt ; table esl ; table liv e_w_res ; table sat_ses ; table wk_prem ; table ifsp_pre3 PRE_ACRD ; run; proc freq data =work.MOD_VAR_ALL_REVISE; table g_sch_red g_pre_ac g_soc_sk g_beh; table g_adapt g_com g_Fmotor g_Gmotor g_other; table Res_role; table mar_stats; table mrace; table mpar_edu ; run; proc freq data =work.MOD_VAR_ALL_REVISE;

PAGE 304

304 table P_PART_SCH fam_act ; table C_PART_ACT ; table GEN_ACT_PART C_OTH_ACT; table read_to eat_meal; table NEIG_INC nclb ; table SSUP_SS PRE_ACRD; run; /*get number of children wit h and without IEP previous CM file*/ DATA TMP1.temp_iep; SET tmp1.Cm_descriptive_combine; n_ch_IEP = n_iep_class ; IF n_iep_class = 1 THEN n_ch_IEP = .; IF n_iep_class = 9 THEN n_ch_IEP = .; n_ch_nIEP = n_nep_class ; IF n_nep_class = 1 THEN n_ch_nIEP = .; IF n_nep_class = 9 THEN n_ch_nIEP = .; KEEP ID n_ch_IEP n_ch_nIEP; RUN; LIBNAME DATA 'C: \ DATA'; proc sort data=tmp1.Model_cm_sc_remove; by ID; RUN; proc sort data=work.MOD_VAR_ALL_REVISE; by ID; RUN; proc sort data=TMP1.temp_iep; by ID; RUN; Data data.MODERATOR_FILE; merge work.MOD_VAR_ALL_REVISE data.Model_cm_sc_remove TMP1.temp_iep; by id; run; LIBNAME DATA 'C: \ DATA'; data DATA.moderator_file ; set DATA.Moderator_file; drop g_soc_sk_1 g_beh_1 g_adapt_1 g_com_1 g_Fmotor_1 g_Gmotor_1 g_OTHER _1; run;

PAGE 305

305 Analytic Coding Syntax Mplus Latent Class Analyse s 2 through 7 Class Model s Research Question 1 Title: LCA SYNTAX 2 CLASS DATA: FILE IS "C:/data/Wt_use_lca.txt" ; VARIABLE: NAMES ARE hear id com heal act att soc beh arm leg hand vis emo mot cog und SOC_COOP SOC_INT SOC_IND BEH_EXT BEH_I NT PROB_BEH autism dd ebd ld mr sl li age CHDSEX SOC_SKL Varstrat Varunit PCT1CW0 PC1CW0 ; USEVARIABLES ARE hear com heal act att soc beh arm leg hand vis emo mot cog und ; CLASSES = c (2) ; CA TEGORICAL = hear com heal act att soc beh arm leg hand vis emo mot cog und ; missing are all (0); STRATIFICATION IS Varstrat; CLUSTER IS Varunit; WEIGHT IS PC1CW0; define: if arm==3 then arm=4; if hand==3 then hand=4; if leg==3 then leg=4; if vis== 3 then vis=4; if hear==3 then hear=4; ANALYSIS: Type = MIXTURE COMPLEX; STARTS = 500 40 ; loghigh = 45; loglow = 45; OUTPUT: TECH1 ; !sampstat; Title: LCA SYNTAX 3 CLASS DATA: FILE IS "C:/data/Wt_use_lca.txt" ; VARIABLE: NAMES ARE hear id com heal act att soc beh arm leg hand vis emo mot cog und SOC_COOP SOC_INT SOC_IND BEH_EXT BEH_I NT PROB_BEH autism dd ebd ld mr sl li age CHDSEX SOC_SKL Varstrat Varunit PCT1CW0 PC1CW0 ;

PAGE 306

306 USEVARIABLES ARE hear com heal act att soc beh arm leg hand vis emo mot cog und ; CLASSES = c (3) ; CATEGORICAL = hear com heal act att soc beh arm leg hand vis emo mot cog und ; missing are all (0); STRATIFICATION IS Varstrat; CLUSTER IS Varunit; WEIGHT IS PC1CW0; define: if arm==3 then arm=4; if hand==3 then hand=4; if leg==3 then leg=4; if vis==3 then vis=4; if hear==3 then hear=4; ANALYSIS: Type = MIXTURE COMPLEX; STARTS = 500 40 ; loghigh = 45; loglow = 45; OUTPUT: TECH1 ; !sampstat; Title: LCA SYNTAX 4 CLASS D ATA: FILE IS "C:/data/Wt_use_lca.txt" ; VARIABLE: NAMES ARE hear id com heal act att soc beh arm leg hand vis emo mot cog und SOC_COOP SOC_INT SOC_IND BEH_EXT BEH_INT PROB_BEH autism dd ebd ld mr sl li age CHDSEX SOC_SKL Varstrat Varunit PCT1CW0 PC1CW 0 ; USEVARIABLES ARE hear com heal act att soc beh arm leg hand vis emo mot cog und ; CLASSES = c (4) ; CATEGORICAL = hear com heal act att soc beh arm leg hand vis emo mot cog und ; missing are all (0); STRATIFICATION IS Varstrat; CLUSTER IS Varunit; WEIGHT IS PC1CW0; define: if arm==3 then arm=4; if hand==3 then hand=4; if leg==3 then leg=4;

PAGE 307

307 if vis==3 then vis=4; if hear==3 then hear=4; ANALYSIS: Type = MIXTURE COMPLEX; STARTS = 500 40 ; loghigh = 45; loglow = 45; OUTPUT: TECH1 ; !sampstat; Title: LCA SYNTAX 5 CLASS DATA: FILE IS "C:/data/Wt_use_lca.txt" ; VARIABLE: NAMES ARE hear id com heal act att soc beh arm leg hand vis emo mot cog und SOC_COOP SOC_INT SOC_IND BEH_EXT BEH_I NT PROB_BEH autism dd ebd ld mr sl li age CHDSEX SOC_SKL Varstrat Varunit PCT1CW0 PC1CW0 ; USEVARIABLES ARE hear com heal act att soc beh arm leg hand vis emo mot cog und ; CLASSES = c (5) ; CATEGORICAL = hear com heal act att soc beh arm leg hand vis emo mot cog und ; missing are all (0); STRATIFICATION IS Varstrat; CLUSTER IS Varunit; WEIGHT IS PC1CW0; define: if arm==3 then arm=4; if hand==3 then hand=4; if leg==3 then leg=4; if vis==3 then vis=4; if hear==3 then hear=4; ANALYSIS: Type = MIXTURE COMPLEX; STARTS = 500 40 ; loghigh = 45; loglow = 4 5; OUTPUT: TECH1 ; SAVEDATA: FILE = fiveclassout.dat ; SAVE = CPROB ;

PAGE 308

308 Title: LCA SYNTAX 6 CLASS DATA: FILE IS "C:/data/Wt_use_lca.txt" ; VARIABLE: NAMES ARE hear id com heal act att soc beh arm leg hand vis emo mot cog und SOC_COOP SOC_INT SOC_IND BEH_EXT BEH_I NT PROB_BEH autism dd ebd ld mr sl li age CHDSEX SOC_SKL Varstrat Varunit PCT1CW0 PC1CW0 ; USEVARIABLES ARE hear com heal act att soc beh arm leg hand vis emo mot cog und ; CLASSES = c (6) ; CA TEGORICAL = hear com heal act att soc beh arm leg hand vis emo mot cog und ; missing are all (0); STRATIFICATION IS Varstrat; CLUSTER IS Varunit; WEIGHT IS PC1CW0; define: if arm==3 then arm=4; if hand==3 then hand=4; if leg==3 then leg=4; if vis==3 then vis=4; if hear==3 then hear=4; ANALYSIS: Type = MIXTURE COMPLEX; STARTS = 500 40 ; loghigh = 45; loglow = 45; OUTPUT: TECH1 ; !sampstat; Title: LCA SYNTAX 7 CLASS DATA: FILE IS "C:/data/Wt_use_lca.txt" ; VARIABLE: NAMES ARE hear id com heal act att soc beh arm leg hand vis emo mot cog und SOC_COOP SOC_INT SOC_IND BEH_EXT BEH_I NT PROB_BEH autism dd ebd ld mr sl li age CHDSEX SOC_SKL Varstrat Varunit PCT1CW0 PC1CW0 ; USEVARIABLES ARE hear com heal act att soc beh arm leg hand vis em o mot cog und ; CLASSES = c (7) ; CATEGORICAL = hear com heal act att soc beh arm leg hand

PAGE 309

309 vis emo mot cog und ; missing are all (0); STRATIFICATION IS Varstrat; CLUSTER IS Varunit; WEIGHT IS PC1CW0; define: if arm==3 then arm=4; if hand==3 then hand=4; if leg==3 then leg=4; if vis==3 then vis=4; if hear==3 then hear=4; ANALYSIS: Type = MIXTURE COMPLEX; STARTS = 500 40 ; loghigh = 45; loglow = 45; OUTPUT: TECH1 ; !sampstat; to Calculate Model Implied Means 5 clas s model Research Question 1 ( Calculations p rovided by Algina ) DATA ; INPUT VAR $CLASS SOLUT PROP1 PROP2 PROP4 ; MEAN=1*PROP1+2*PROP2+4*PROP4; VARIANCE=PROP1*(1 MEAN)**2+PROP2*(2 MEAN)**2+PROP4*(4 MEAN)**2; DATALINES; HEAR 1 5 0.955 0.016 0.029 ARM 1 5 0.743 0.249 0.008 HAND 1 5 0.195 0.557 0.248 LEG 1 5 0.739 0.254 0.007 VIS 1 5 0.924 0.064 0.013 HEAR 2 5 0.909 0.010 0.082 ARM 2 5 0.000 0.537 0.463 HAND 2 5 0.000 0.159 0.841 LEG 2 5 0.032 0.581 0.388 VIS 2 5 0.412 0.263 0.325 HEAR 3 5 0.940 0.000 0.060 ARM 3 5 0.327 0.501 0.172 HAND 3 5 0.072 0.613 0.314 LEG 3 5 0.271 0.442 0.287

PAGE 310

310 VIS 3 5 0.731 0.194 0.075 HEAR 4 5 0.929 0.017 0.054 ARM 4 5 0.941 0.059 0.000 HAND 4 5 0.624 0.358 0.018 LEG 4 5 0.946 0.054 0.000 VIS 4 5 0.927 0.059 0.014 H EAR 5 5 0.954 0.004 0.042 ARM 5 5 0.997 0.002 0.000 HAND 5 5 0.918 0.081 0.001 LEG 5 5 0.976 0.019 0.005 VIS 5 5 0.971 0.022 0.007 PROC PRINT; RUN; DATA ; INPUT VAR $CLASS SOLUT PROP1 PROP2 PROP4 ; MEAN=1*PROP1+2*PROP2+4*PROP4; VARIANCE=PROP1*(1 MEAN)**2+PROP2*(2 MEAN)**2+PROP4*(4 MEAN)**2; DATALINES; ACT 1 5 0.112 0.208 0.680 ATT 1 5 0.098 0.262 0.640 EMO 1 5 0.540 0.280 0.180 MOT 1 5 0.080 0.274 0.646 ACT 2 5 0.178 0.338 0.484 ATT 2 5 0.096 0.295 0.609 EMO 2 5 0.744 0.198 0 .058 MOT 2 5 0.132 0.168 0.700 ACT 3 5 0.419 0.326 0.255 ATT 3 5 0.348 0.392 0.260 EMO 3 5 0.811 0.087 0.101 MOT 3 5 0.236 0.455 0.309 ACT 4 5 0.159 0.343 0.498 ATT 4 5 0.111 0.549 0.340 EMO 4 5 0.809 0.141 0.049 MOT 4 5 0.149 0.441 0.409 ACT 5 5 0.520 0 .387 0.093 ATT 5 5 0.530 0.404 0.066 EMO 5 5 0.936 0.059 0.004 MOT 5 5 0.345 0.483 0.172 PROC PRINT; RUN; DATA ; INPUT VAR $CLASS SOLUT PROP1 PROP2 PROP3 PROP4; MEAN=1*PROP1+2*PROP2+3*PROP3+4*PROP4;

PAGE 311

311 VARIANCE=PROP1*(1 MEAN)**2+PROP2*(2 MEAN)**2+PROP3*(3 MEAN)**2+PROP4*(4 MEAN)**2; DATALINES; COM 1 5 0.014 0.077 0.441 0.467 HEAL 1 5 0.414 0.313 0.170 0.104 SOC 1 5 0.141 0.023 0.497 0.339 BEH 1 5 0.082 0.449 0.329 0.140 COG 1 5 0.052 0.038 0.405 0.505 UND 1 5 0.024 0.416 0.535 0.025 COM 2 5 0.000 0.099 0.276 0.625 HEAL 2 5 0.184 0.304 0.259 0.253 SOC 2 5 0.101 0.036 0.367 0.496 BEH 2 5 0.272 0.295 0.301 0.132 COG 2 5 0.005 0.000 0.159 0.835 UND 2 5 0.018 0.392 0.521 0.069 COM 3 5 0.434 0.06 9 0.368 0.129 HEAL 3 5 0.356 0.317 0.175 0.152 SOC 3 5 0.563 0.144 0.270 0.024 BEH 3 5 0.759 0.215 0.026 0.001 COG 3 5 0.109 0.358 0.457 0.077 UND 3 5 0.606 0.351 0.043 0.000 COM 4 5 0.230 0.133 0.530 0.107 HEAL 4 5 0.633 0.227 0.118 0.022 SOC 4 5 0.492 0.073 0.381 0.053 BEH 4 5 0.479 0.411 0.103 0.007 COG 4 5 0.064 0.313 0.573 0.050 UND 4 5 0.321 0.622 0.058 0.000 COM 5 5 0.455 0.066 0.417 0.061 HEAL 5 5 0.788 0.153 0.045 0.015 SOC 5 5 0.832 0.075 0.084 0.010 BEH 5 5 0.921 0.066 0.013 0.000 CO G 5 5 0.173 0.756 0.069 0.002 UND 5 5 0.895 0.102 0.003 0.000 PROC PRINT; RUN; LIBNAME DATA 'C: \ DATA'; DATA DATA.class5solfinal; SET WORK.data20 work.data21 work.data22 ; RUN; Quit;

PAGE 312

3 12 Descriptive Analyses Research Question 1 /* OPEN LIBRARY TO ACCESS FILE*/ LIBNAME DATA 'C: \ DATA'; DATA DATA.Moderator_file; SET tmp1.Moderator_file; RUN; /*CM =latent class membership Results were interpreted using the following key CM 1 = Profile 2 CM 2 = Profile 1 CM 3 = Profile 3 CM 4 = Profile 4 CM 5 = Profile 5 */ /*GENERAL DESCRIPTIVES*/ Proc sort data=DATA.Moderator_file; by cm; run; PROC SURVEYMEANS data=DATA.Moderator_file VARMETHOD=Taylor MEAN VAR clm; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PC1CW0; VAR AGE ; BY CM; RUN; /* CODE E XAMPL E FOR DESCRIPTIVES use to get S D for weighted analyses vardef=wdf*/ proc MEANS vardef=wdf data=DATA.Moderator_file ; WEIGHT T_PC1CW0; VAR age; BY CM; RUN; /* CROSS CLASSIFICATION FOR CM AND DIS -PARENT CHILD WEIGHTS*/ proc SURVEYfreq data=DATA.Moderator_file VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PC1CW0; table cm*AUTISM; table cm*DD;

PAGE 313

313 table cm*LI; table cm*SL; table cm*EBD; table cm*MR; table cm*LD; run; DATA WORK.Moderator_file_ TEMP; SET DATA.Moderator_file; IF DIS = THEN DIS =8; RUN; /* CROSS CLASSIFICATION FOR CM AND DIS -PARENT CHILD WEIGHTS*/ proc SURVEYfreq data=WORK.Moderator_file_temp VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PC1CW0; table cm*DIS; /*1=aut, 2=dd, 3= li, 4= ebd, 5=mr, 6=ld, 7=sl, 8=missing*/ TABLE DIS*CM; run; /* CROSS CLASSIFICATION FOR CM AND Gender -PARENT CHILD WEIGHTS*/ proc SURVEYfreq data=DATA.Moderator_file VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; C LUSTER Taylor_VARUNIT; WEIGHT T_PC1CW0; table cm*chdsex; run; /* CROSS CLASSIFICATION FOR CM AND AGE COHORT -PARENT CHILD WEIGHTS*/ proc SURVEYfreq data=DATA.Moderator_file VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T _PC1CW0; table cm*COHORT; run; /* CROSS CLASSIFICATION FOR CM AND race ethnicity -PARENT CHILD WEIGHTS*/ proc SURVEYfreq data=DATA.Moderator_file VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PC1CW0; table cm*MRACE; run;

PAGE 314

314 /* CROSS CLASSIFICATION FOR CM AND ESL -PARENT CHILD WEIGHTS*/ proc SURVEYfreq data=DATA.Moderator_file VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PC1CW0; table cm*esl; run; /* CROSS CLASSIFICATION FOR CM AND ifsp pre 3 -PARENT CHILD WEIGHTS*/ proc SURVEYfreq data=DATA.Moderator_file VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PC1CW0; table cm*ifsp_pre3; run; /* CM AND weeks premature, birthweight -PARENT CHILD WEIGHTS*/ PROC SURVE YMEANS data=DATA.Moderator_file VARMETHOD=Taylor MEAN VAR clm; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PC1CW0; VAR wk_prem birth_wt; BY CM; RUN; /* use to get STD for weighted analyses vardef=wdf*/ proc MEANS vardef=wdf data=DATA.Mo derator_file ; WEIGHT T_PC1CW0; VAR wk_prem birth_wt; BY CM; RUN; /* CROSS CLASSIFICATION FOR CM AND home live environment -PARENT CHILD WEIGHTS*/ proc SURVEYfreq data=DATA.Moderator_file VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PC1CW0; table cm*live_w_res; run; /* average child activities -PARENT CHILD WEIGHTS*/ PROC SURVEYMEANS data=DATA.Moderator _file VARMETHOD=Taylor MEAN VAR clm;

PAGE 315

315 STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PC1CW0; VAR C_PART_ACT ; BY CM; RUN; /* use to get S D for weighted analyses vardef=wdf*/ proc MEANS vardef=wdf data=DATA.Moderator_file ; WEIGHT T_PC1CW0; VAR C_PART_ACT ; BY CM; RUN; /* CROSS CLASSIFICATION FOR CM AND activity information PARENT CHILD WEIGHTS*/ proc SURVEYfreq data=DATA.Moderator_file VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PC1CW0; table cm*GEN_ACT_PART; table cm*read_to; run; data work.moderator_activty_temp ; set DATA.Moderator_file; if GEN_AC T_PART = then C_OTH_ACT = .; run; /* average child activities for children who particpate -PARENT CHILD WEIGHTS*/ PROC SURVEYMEANS data=work.moderator_activty_temp VARMETHOD=Taylor MEAN VAR clm; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PC1CW0; VAR C_OTH_ACT ; BY CM; RUN; /* use to get S D for weighted analyses vardef=wdf*/ proc MEANS vardef=wdf data=work.moderator_activty_temp ; WEIGHT T_PC1CW0; VAR C_OTH_ACT; BY CM; RUN;

PAGE 316

316 /* average family activities PARENT CHILD WEIGHTS*/ PROC SURVEYMEANS data=work.moderator_activty_temp VARMETHOD=Taylor MEAN VAR clm; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PC1CW0; VAR fam_act eat_meal ; BY CM; RUN; /* use to get S D for weighted analyses vardef=wdf*/ proc MEANS varde f=wdf data=work.moderator_activty_temp ; WEIGHT T_PC1CW0; VAR fam_act eat_meal ; BY CM; RUN; /* CROSS CLASSIFICATION FOR CM AND martial status PARENT CHILD WEIGHTS*/ proc SURVEYfreq data=DATA.Moderator_file VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; C LUSTER Taylor_VARUNIT; WEIGHT T_PC1CW0; table cm*mar_stats; run; /* CROSS CLASSIFICATION FOR CM A ND respondent role PARENT CHILD WEIGHTS*/ proc SURVEYfreq data=DATA.Moderator_file VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PC1CW0; table cm*res_role; run; /* CROSS CLASSIFICATION FOR CM AND dincome PARENT CHILD WEIGHTS*/ proc SURVEYfreq data=DATA.Moderator_file VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PC1CW0; table cm*dincome; run; /* average parent satisfication PARENT CHILD WEIGHTS*/ PROC SURVEYMEANS data=work.moderator_FILE VARMETHOD=Taylor MEAN VAR clm;

PAGE 317

317 STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PC1CW0; VAR sat_ses ; BY CM; RUN; /* use to get SD for w eighted analyses vardef=wdf*/ proc MEANS vardef=wdf data=work.moderator_FILE ; WEIGHT T_PC1CW0; VAR sat_ses ; BY CM; RUN; /* CROSS CLASSIFICATION FOR CM AND school income area PARENT CHILD Teacher WEIGHTS*/ proc SURVEYfreq data=DATA.Moderator_file VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; table cm*neig_inc; run; /* CROSS CLASSIFICATION FOR CM AND school quality PARENT CHILD Teacher WEIGHTS*/ proc SURVEYfreq data =DATA.Moderator_file VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; table cm*PRE_ACRD; table cm*nclb; run; data work.moderator_GOAL_temp ; set DATA.Moderator_file; if g_sch_red = then g_sch_red = 0; if g_pre_ac = then g_pre_ac = 0; if g_soc_sk = then g_soc_sk = 0; if g_beh = then g_beh = 0; if g_adapt = then g_adapt = 0; if g_com = then g_com = 0; if g_Fmotor = then g_Fmotor = 0; if g_Gmotor = then g_Gmotor = 0; if g_other = then g_other = 0; IF SSUP_SS = THEN SSUP_SS = 2; run;

PAGE 318

318 /* CROSS CLASSIFICATION FOR CM AND goal type PARENT CHILD Teacher WEIGHTS*/ proc SURVEYfreq data=WORK.Moderator_GOAL_TEMP VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; table cm* g_sch_red; table cm* g_pre_ac; table cm* g_soc_sk; table cm* g_beh; table cm* g_adapt; table cm* g_com; table cm* g_Fmotor; table cm* g_Gmotor; table cm* g_other; run; /* CROSS CLASSIFICATION FOR CM AND support participation PARENT CHILD Teacher WEIGHTS* / proc SURVEYfreq data=work.moderator_GOAL_temp VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; table cm*SSUP_SS; run; /* CROSS CLASSIFICATION FOR CM AND support participation PARENT CHILD Teacher WEIGHTS*/ PROC SURVEYMEANS data=work.moderator_activty_temp VARMETHOD=Taylor MEAN VAR clm; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; VAR n_ch_IEP n_ch_nIEP ; BY CM; RUN; /* use to get S D for weighted analyses vardef=wdf*/ proc MEANS va rdef=wdf data=work.moderator_activty_temp ; WEIGHT T_PCT1CW0; VAR n_ch_IEP n_ch_nIEP ; BY CM; RUN;

PAGE 319

319 Regression Models Research Question 2 (Example for Social Skills Analyses ) LIBNAME DATA 'C: \ DATA'; DATA DATA.Model_cm_sc_remove; SET WORK.Model_cm_sc_remove ; RUN; /*MODEL TO REPORT ESTIMATES AND COMAPRSIONS;CM CENTERED ON AGE; NEED TO CALCULATE ES and CI*/ proc SURVEYREG data=data.Model_cm_sc_remove VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW 0; CLASS CM; MODEL SOC_SKL = CM /ANOVA SOLUTION; ESTIMATE 'CL ASS 1' INTERCEPT 1 CM 1 0 0 0 0 ; ESTIMATE 'CLASS 2' INTERCEPT 1 CM 0 1 0 0 0; ESTIMATE 'CLASS 3' INTERCEPT 1 CM 0 0 1 0 0; ESTIMATE 'CLASS 4' INTERCEPT 1 CM 0 0 0 1 0; ESTIMATE 'CLASS 5' INTERCEP T 1 CM 0 0 0 0 1; ESTIMATE '1 VS 2' CM 1 1 0 0 0 ; ESTIMATE '1 VS 3' CM 1 0 1 0 0 ; ESTIMATE '1 VS 4' CM 1 0 0 1 0 ; ESTIMATE '1 VS 5' CM 1 0 0 0 1 ; ESTIMATE '2 VS 3' CM 0 1 1 0 0 ; ESTIMATE '2 VS 4' CM 0 1 0 1 0 ; ESTIMATE '2 VS 5' CM 0 1 0 0 1 ; ESTIMATE '3 VS 4' CM 0 0 1 1 0 ; ESTIMATE '3 VS 5' CM 0 0 1 0 1 ; ESTIMATE '4 VS 5' CM 0 0 0 1 1 ; RUN; Regression Models Research Question 3 (Example for Social Skills Analyses) /* MODEL DIS ONLY FOR R2 */ proc SURVEYREG data=data.Model_cm_sc_remove VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS DIS; MODEL SOC_SKL = DIS /ANOVA SOLUTION; RUN;

PAGE 320

320 /* MODEL CM ONLY FOR R2*/ proc SURVEYREG data=data.Model_cm_sc_remo ve VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM; MODEL SOC_SKL = CM /ANOVA SOLUTION; RUN; /* MODEL CM AND DIS FOR R2 */ proc SURVEYREG data=data.Model_cm_sc_remove VARMETHOD=Taylor; STRATA Taylor_VARS TRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM DIS; MODEL SOC_SKL = CM DIS /ANOVA SOLUTION; RUN; /*HOLD OUT ANALYSES FOR r2*/ DATA WORK.Model_cm_sc_remove_HOLD_OUT; SET DATA.Model_cm_sc_remove ; if dis = 7 then delete; RUN; /* MODEL DIS ONLY FOR R2 */ proc SURVEYREG data=WORK.Model_cm_sc_remove_HOLD_OUT VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS DIS; MODEL SOC_SKL = DIS /ANOVA SOLUTION; RUN; /* MODEL CM ONLY FOR R2*/ proc SURVEYREG da ta=WORK.Model_cm_sc_remove_HOLD_OUT VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM; MODEL SOC_SKL = CM /ANOVA SOLUTION; RUN; /* MODEL CM AND DIS FOR R2 */

PAGE 321

321 proc SURVEYREG data=W ORK.Model_cm_sc_remove_HOLD_OUT VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM DIS; MODEL SOC_SKL = CM DIS /ANOVA SOLUTION; RUN; /* CROSS CLASSIFICATION FOR CM AND DIS -PARENT CHILD TEACHER WEIGHTS*/ DATA WORK.Moderator_file_TEMP; SET DATA.Moderator_file; IF DIS = THEN DIS =8; RUN; proc SURVEYfreq data=WORK.Moderator_file_temp VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; table cm*DIS; /*1=aut, 2=dd, 3= li, 4= ebd, 5=mr, 6=ld, 7=sl, 8=missing*/ run; Moderation Analyses Research Question 4 (Example for Social Skills Analyses) /*get mean to center continuous variables */ PROC MEANS vardef=wdf DATA= data.Moderator_file; WEIGHT T_PCT1CW0; VAR age MINCOME EAT_MEAL P_PART_SCH C_PART_ACT C_OTH_ACT FAM_ACT ; /*ALL CONTINOUS SHOULD BE CONETERED*/ RUN; /*create new centered continuous variables */ data data.Moderator_file_CENTERED; SET data.Moderator_file; AGE_CEN=AGE 55.77;*55.77 CALCULATED AS A WEIG HTED MEAN; Mincome_CEN=MINCOME 7.3144104; 7.3144104 CALCULATED AS A WEIGHTED MEAN; EAT_MEAL_CEN = EAT_MEAL 5.4065115; P_PART_SCH_CEN = P_PART_SCH 3.3415803;

PAGE 322

322 C_PART_ACT_CEN = C_PART_ACT 0.6884862; C_OTH_ACT_CEN = C_OTH_ACT 0.6741080; FAM_ACT_CE N = FAM_ACT 5.1572612; run; /* MODERATION CODE CHILD F ACTORS */ proc SURVEYREG data=data.Moderator_file_CENTERED VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PC1CW0; CLASS CM; MODEL SOC_SKL = CM AGE_CEN /ANOVA SOLUTION; RUN; proc SURVEYREG data=data.Moderator_file_CENTERED VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PC1CW0; CLASS CM; MODEL SOC_SKL = CM AGE_CEN CM*AGE_CEN/ANOVA SOLUTION; RUN; proc SURVEYREG data=data.Moderato r_file VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM CHDSEX; MODEL SOC_SKL = CM CHDSEX /ANOVA SOLUTION; RUN; proc SURVEYREG data=data.Moderator_file VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM CHDSEX; MODEL SOC_SKL = CM CHDSEX CM*CHDSEX/ANOVA SOLUTION; RUN; /* moderation with 4 groups for race/ethnicity*/ data work.moderator_file_race_temp; Set data.Moderator_file; mrace_new = .; if mrace = 1 then mrace_new = 1 ; if mrace = 2 then mrace_new = 2 ; if mrace = 3 then mrace_new = 3 ;

PAGE 323

323 if mrace = 4 then mrace_new = 4 ; if mrace = 5 then mrace_new = 4 ; if mrace = 6 then mrace_new = 4 ; run; proc SURVEYREG data=work.Moderator_file_race_temp VARMETHOD=Tay lor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM MRACE_new; MODEL SOC_SKL = CM MRACE_new /ANOVA SOLUTION; RUN; proc SURVEYREG data=work.Moderator_file_race_temp VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylo r_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM MRACE_new; MODEL SOC_SKL = CM MRACE_new CM*MRACE_new/ANOVA SOLUTION; RUN; /* MODERATION CODE FAMILY fACTORS */ /* moderation with 6 groups for PARENT EDUCATION*/ data work.moderator_file_PARENT_temp; Set data.Moderator_file; MPAR_EDU_new = .; if MPAR_EDU = 1 then MPAR_EDU_new = 1 ; if MPAR_EDU = 2 then MPAR_EDU_new = 2 ; if MPAR_EDU = 3 then MPAR_EDU_new = 3 ; if MPAR_EDU = 4 then MPAR_EDU_new = 4 ; if MPAR_EDU = 5 then MPAR_EDU_new = 5 ; if MPAR_EDU = 6 t hen MPAR_EDU_new = 5 ; if MPAR_EDU = 7 then MPAR_EDU_new = 6 ; run; proc SURVEYREG data=WORK.Moderator_file_PARENT_temp VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM MPAR_EDU_new; MODEL soc_skl = CM MPA R_EDU_new /ANOVA SOLUTION; RUN; proc SURVEYREG data=WORK.Moderator_file_PARENT_temp VARMETHOD=Taylor; STRATA Taylor_VARSTRAT;

PAGE 324

324 CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM MPAR_EDU_new; MODEL soc_skl = CM MPAR_EDU_new CM*MPAR_EDU_new/ANOVA SOLUTION; RUN; proc SURVEYREG data=data.Moderator_file VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM mar_stats; MODEL SOC_SKL = CM mar_stats /ANOVA SOLUTION; RUN; proc SURVEYREG data=data.Moderator_file VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM mar_stats; MODEL SOC_SKL = CM mar_stats CM*mar_stats/ANOVA SOLUTION; RUN; proc SURVEYREG data=data.Moderator_file_CENTERED VARMETHOD=Taylor; STRATA Taylor _VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM ; MODEL SOC_SKL = CM Mincome_Cen /ANOVA SOLUTION; RUN; proc SURVEYREG data=data.Moderator_file_CENTERED VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM ; MODEL SOC_SKL = CM Mincome_Cen CM*Mincome_Cen /ANOVA SOLUTION; RUN; /* MODERATION CODE FAMILY INTERACTION FACTORS */ proc SURVEYREG data=data.Moderator_file_CENTERED VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_V ARUNIT; WEIGHT T_PCT1CW0; CLASS CM ; /*PARENT PARTICIPATION IN SCHOOL */ MODEL SOC_SKL = CM P_PART_SCH_CEN /ANOVA SOLUTION; RUN;

PAGE 325

325 proc SURVEYREG data=data.Moderator_file_CENTERED VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGH T T_PCT1CW0; CLASS CM ; /*PARENT PARTICIPATION IN SCHOOL */ MODEL SOC_SKL = CM P_PART_ SCH_CEN CM*P_PART_SCH_CEN/ANOVA SOLUTION; RUN; proc SURVEYREG data=data.Moderator_file_CENTERED VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM ; /*child PARTICIPATION IN community activities EVER */ MODEL SOC_SKL = CM C_PART_ACT_cen /ANOVA SOLUTION; RUN; proc SURVEYREG data=data.Moderator_file_CENTERED VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUN IT; WEIGHT T_PCT1CW0; CLASS CM ; /*child PARTICIPATION IN community activities EVER */ MODEL SOC_SKL = CM C_PART_ACT_cen CM*C_PART_ACT_cen/ANOVA SOLUTION; RUN; proc SURVEYREG data=data.Moderator_file_CENTERED VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM ; /*child PARTICIPATION IN other activities REGULARLY */ MODEL SOC_SKL = CM C_OTH_ACT_cen /ANOVA SOLUTION; RUN; proc SURVEYREG data=data.Moderator_file_CENTERED VARMETHOD=Taylor; STRATA Taylor_VARSTR AT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM ; /*child PARTICIPATION IN other activities REGULARLY */ MODEL SOC_SKL = CM C_OTH_ACT_cen CM*C_OTH_ACT_cen/ANOVA SOLUTION; RUN; data work.moderator_file_activity_temp; Set data.Moderator_file;

PAGE 326

326 if GEN_ACT_PART = then GEN_ACT_PART = 0; run; proc SURVEYREG data=work.moderator_file_a ctivity_temp VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM GEN_ACT_PART; /*child PARTICIPATION IN other activities R EGULARLY YES/NO */ MODEL SOC_SKL = CM GEN_ACT_PART /ANOVA SOLUTION; RUN; proc SURVEYREG data=work.moderator_file_activity_temp VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM GEN_ACT_PART; /*child PARTICIP ATION IN other activities REGULARLY YES/NO */ MODEL SOC_SKL = CM GEN_ACT_PART CM*GEN_ACT_PART/ANOVA SOLUTION; RUN; proc SURVEYREG data=data.Moderator_file_CENTERED VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM ; /*family and child activities */ MODEL SOC_SKL = CM FAM_ACT_cen /ANOVA SOLUTION; RUN; proc SURVEYREG data=data.Moderator_file_CENTERED VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM ; /*family and child activities */ MODEL SOC_SKL = CM FAM_ACT_cen CM*FAM_ACT_cen/ANOVA SOLUTION; RUN; proc SURVEYREG data=data.Moderator_file_CENTERED VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM ; /*meals per wee k */ MODEL SOC_SKL = CM EAT_MEAL_cen /ANOVA SOLUTION; RUN; proc SURVEYREG data=data.Moderator_file_CENTERED VARMETHOD=Taylor;

PAGE 327

327 STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM ; /*meals per week */ MODEL SOC_SKL = CM EAT_MEAL _cen CM*EAT_MEAL_cen/ANOVA SOLUTION; RUN; proc SURVEYREG data=data.Moderator_file VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM READ_TO; /*READ TO CHILD */ MODEL SOC_SKL = CM READ_TO /ANOVA SOLUTION; RUN; proc SURVEYREG data=data.Moderator_file VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM READ_TO; /*READ TO CHILD */ MODEL SOC_SKL = CM READ_TO CM*READ_TO/ANOVA SOLUTION; RUN; data work.moderator_file_read _temp; Set data.Moderator_file; READ_TO_new = .; if READ_TO = 1 then READ_TO_new = 1 ; if READ_TO = 2 then READ_TO_new = 1 ; if READ_TO = 3then READ_TO_new = 2 ; if READ_TO = 4 then READ_TO_new = 3 ; run; proc SURVEYREG data=work.Moderator_file_read_temp VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM READ_TO_new; /*READ TO CHILD */ MODEL SOC_SKL = CM READ_TO_new /ANOVA SOLUTION; RUN; proc SURVEYREG data=work.Moderator_file_read_temp VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM READ_TO_new; /*READ TO CHILD */ MODEL SOC_SKL = CM READ_TO_new CM*READ_TO_new/ANOVA SOLUTION;

PAGE 328

328 RUN; /* MODERATION CODE ENVIRNMENTAL FACTORS */ proc SURVEYREG data=data.Moderat or_file VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM MSAFE_NEIG; /*NEIGBORHOOD SAFE */ MODEL SOC_SKL = CM MSAFE_NEIG /ANOVA SOLUTION; RUN; proc SURVEYREG data=data.Moderator_file VARMETHOD=Taylor; STRAT A Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM MSAFE_NEIG; /*NEIGBORHOOD SAFE */ MODEL SOC_SKL = CM MSAFE_NEIG CM*MSAFE_NEIG/ANOVA SOLUTION; RUN; proc SURVEYREG data=data.Moderator_file VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM SSUP_SS; /*SUPPORT SOCIAL INTERACTION */ MODEL SOC_SKL = CM SSUP_SS /ANOVA SOLUTION; RUN; proc SURVEYREG data=data.Moderator_file VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; W EIGHT T_PCT1CW0; CLASS CM SSUP_SS; /*SUPPORT SOCIAL INTERACTION */ MODEL SOC_SKL = CM SSUP_SS CM*SSUP_SS/ANOVA SOLUTION; RUN; proc SURVEYREG data=data.Moderator_file VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0 ; CLASS CM NEIG_INC; /*NEIGBORHOD INCOME */ MODEL SOC_SKL = CM NEIG_INC /ANOVA SOLUTION; RUN; proc SURVEYREG data=data.Moderator_file VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT;

PAGE 329

329 WEIGHT T_PCT1CW0; CLASS CM NEIG_INC; /*NEIGBORHOD INCOME */ MODEL SOC_SKL = CM NEIG_INC CM*NEIG_INC/ANOVA SOLUTION; RUN; /*explore significant moderator*/ proc SURVEYREG data=data.Moderator_file_CENTERED VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM ; / *child PARTICIPATION IN community activities EVER */ MODEL SOC_SKL = CM C_PART_ACT_cen CM*C_PART_ACT_cen/ANOVA SOLUTION; RUN; / INTERCEPTS AND SLOPES MODEL. USED IF A QUANTITATIVE INDEPENDENT VARIABLE HAS A SIGNI FICANT INTERACTION WITH PROFILE*/ proc S URVEYREG data=data.Moderator_file_CENTERED VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS CM ; MODEL SOC_SKL = CM C_PART_ACT_cen(cm)/ANOVA SOLUTION NOINT; ESTIMATE 'SLOPES FOR P1 VS P2' C_PART_ACT_cen(cm) 1 1 0 0 0; ESTIMATE 'SLOPES FOR P1 VS P3' C_PART_ACT_cen(cm) 1 0 1 0 0; ESTIMATE 'SLOPES FOR P1 VS P4' C_PART_ACT_cen(cm) 1 0 0 1 0; ESTIMATE 'SLOPES FOR P1 VS P5' C_PART_ACT_cen(cm) 1 0 0 0 1; ESTIMATE 'SLOPES FOR P2 VS P3' C_PART_ACT_cen(cm) 0 1 1 0 0 ; ESTIMATE 'SLOPES FOR P2 VS P4' C_PART_ACT_cen(cm) 0 1 0 1 0 ; ESTIMATE 'SLOPES FOR P2 VS P5' C_PART_ACT_cen(cm) 0 1 0 0 1 ; ESTIMATE 'SLOPES FOR P3 VS P4' C_PART_ACT_cen(cm) 0 0 1 1 0 ; ESTIMATE 'SLOPES FOR P3 VS P5' C_PART_ACT_cen(cm) 0 0 1 0 1 ; E STIMATE 'SLOPES FOR P4 VS P5' C_PART_ACT_cen(cm) 0 0 0 1 1 ; run; /*plot moderation and examine for outliers*/ /*plot moderation*/ data; do cm = 4 to 5 by 1; do c_part_act_cen=0 to 7; if cm = 4 then Yhat=91.846744+ 3.231931* c_part_act_cen; if cm = 5 then Yhat= 102.011277+ ( 0.313673)* c_part_act_cen; OUTPUT; END;

PAGE 330

330 PROC PRINT; RUN; PROC PLOT; PLOT YHAT*c_part_act_cen; RUN; /*outliers*/ DATA NEW; SET data.Moderator_file_CENTERED; IF CM=5; PROC PLOT; PLOT soc_skl*c_part_act; RUN; DATA NEW; SET data.Moderator_file_CENTERED; IF CM=4; PROC PLOT; PLOT soc_skl*c_part_act; RUN; proc SURVEYFREQ data=data.Moderator_file_CENTERED; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; TABLE CM*c_part_act ; RUN; ( Moderation Example for Problem Behavior When Categorical Variable Significant) /* create new variable to run all possible comparisons for race -4 categories White, Hispan, Black, Other*/ DATA work.Race_PROFILE; SET data.Moderator_file ; temp_race = .; if mrace = 1 then temp_race = 1; if mrace = 2 then temp_race = 2; if mrace = 3 then temp_race = 3; if mrace = 4 then temp_race = 4; if mrace = 5 then temp_race = 4; if mrace = 6 then temp_race = 4; NEWCLASS=(10*temp_race)+ CM; run; /*FOLLOWING TO ESTIMATE CELL MEANS WHEN M ODERATION IS SIGNIFICANT AND TO COMPARE CELL

PAGE 331

331 MEANS for categorical;*/ proc SURVEYREG data=race_PROFILE VARMETHOD=Taylor; STRATA Taylor_VARSTRAT; CLUSTER Taylor_VARUNIT; WEIGHT T_PCT1CW0; CLASS NEWCLASS; MODEL PROB_BEH = NEWCLASS/SOLUTION NOINT; /* 2 0 new class variables */ ESTIMATE 'P1white VS P1hisp' NEWCLASS 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0; ESTIMATE 'P1white VS P1black' NEWCLASS 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0; ESTIMATE 'P1white VS P1other' NEWCLASS 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 ; ESTIMATE 'P1hisp VS P1black' NEWCLASS 0 0 0 0 0 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0; ESTIMATE 'P1hisp VS P1other' NEWCLASS 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0; ESTIMATE 'P1black VS P1other' NEWCLASS 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 1 0 0 0 0; ESTIMATE 'P2white VS P2hisp' NEWCLASS 0 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 ; ESTIMATE 'P2white VS P2black' NEWCLASS 0 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 ; ESTIMATE 'P2white VS P2other' NEWCLASS 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 ; ESTIMATE 'P2hisp VS P2black' NEWCLASS 0 0 0 0 0 0 1 0 0 0 0 1 0 0 0 0 0 0 0 0 ; ESTIMATE 'P2hisp VS P2other' NEWCLASS 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 1 0 0 0 ; ESTIMATE 'P2black VS P2other' NEWCLASS 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 1 0 0 0 ; ESTIMA TE 'P3white VS P3hisp' NEWCLASS 0 0 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 ; ESTIMATE 'P3white VS P3black' NEWCLASS 0 0 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 ; ESTIMATE 'P3white VS P3other' NEWCLASS 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 ; ESTIMATE P3hisp VS P3black' NEWCLASS 0 0 0 0 0 0 0 1 0 0 0 0 1 0 0 0 0 0 0 0 ; ESTIMATE 'P3hisp VS P3other' NEWCLASS 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 1 0 0 ; ESTIMATE 'P3black VS P3other' NEWCLASS 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 1 0 0 ; ESTIMATE 'P4w hite VS P4hisp' NEWCLASS 0 0 0 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 ; ESTIMATE 'P4white VS P4black' NEWCLASS 0 0 0 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 ; ESTIMATE 'P4white VS P4other' NEWCLASS 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 ; ESTIMATE 'P4hisp VS P4black' NEWCLASS 0 0 0 0 0 0 0 0 1 0 0 0 0 1 0 0 0 0 0 0 ; ESTIMATE 'P4hisp VS P4other' NEWCLASS 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 1 0 ; ESTIMATE 'P4black VS P4other' NEWCLASS 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 1 0 ; ESTIMATE 'P5white VS P5hi sp' NEWCLASS 0 0 0 0 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 ; ESTIMATE 'P5white VS P5black' NEWCLASS 0 0 0 0 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 ; ESTIMATE 'P5white VS P5other' NEWCLASS 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 ; ESTIMATE 'P5hisp VS P5black' NEWC LASS 0 0 0 0 0 0 0 0 0 1 0 0 0 0 1 0 0 0 0 0 ; ESTIMATE 'P5hisp VS P5other' NEWCLASS 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 1 ; ESTIMATE 'P5black VS P5other' NEWCLASS 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 1 ; RUN;

PAGE 332

332 APPENDIX E STATISTICAL MODELS Question 1: What distinct and interpretable functional ability profile subgroups emerge when using person oriented analytic techniques to examine functional ability variables contained in the PEELS data set for young children with disabilities? Note s: denotes P observed indicators of K latent classes. denotes the probability of belonging to latent class denotes the distribution of the indicators conditional on the parameters of the distributions. denotes the distribution of the indicators conditional on the parameters of the distributions in the k th latent class. Question 2: What is the strength of the relationship between functional ability profile subgroup membership and social competence? Notes: denotes the PKBS standard score for the problem behavior or social skills scale denotes a dummy variable with the K th latent class used as the reference group. Subgroup membership will be determined by using the most likely class membership calculated based on the model in Question 1 The actual statistical method was to test equality of means across latent classes using ProcSurveyReg in which functional ability profile subgroup membership was indicated as a class variable

PAGE 333

333 Question 3: What are the individual and combined contributions of functional ability profile subgroup membership and disability category membership to the explanation of social competence? Examine individual contribution of disability category Notes: denotes the PKBS standard score for the problem behavior or social skills scale. denotes a dummy variable with the Speech or language impairments category used as the reference group and the j th disability category coded 1. Examine combined contribution of dis ability category and functional ability profile subgroup membership Notes: denotes the PKBS standard score for the problem behavior or social skills scale. denotes a dummy variable with the Speech or language impairments category used as the reference group and the j th disability category coded 1. denotes a dummy variable with the K th latent class used as the reference group Subgroup membership will be determined by using the most likely class membership calculated based on the model in Question 1 The actual statistical method wa s to test equality of means across latent classes or disability category using ProcSurveyReg in which functional ability profile subgroup membership and disability category were indicated as class variable s

PAGE 334

334 Question 4: To what extent do non malleable child factors and contextual factors moderate the relationship between functional ability profile subgroup membership and social competence? Examine moderation Notes: de notes the PKBS standard score for the problem behavior or social skills scale. denotes a dummy variable with the K th latent class used as the reference group Subgroup membership will be determined by using the most likely class membership calculated b ased on the model in Question 1 The actual statistical method wa s to test the interaction term using ProcSurveyReg in which functional ability profile subgroup membership and any categorical contextual variables were indicated as class variable s C denotes the variable of interest for non malleable child factors or contextual factors Non malleable child factors: gender age, race/ethnicity Contextual variables include family income parent education single parent household child activities, parent child activities, regular child activities, child participation in activities regularly, family meals, extent child is read to, SE S neighborhood neighborhood safety, and program support for social interaction.

PAGE 335

335 APPENDIX F LATENT CLASS MODELS NOT S ELECTED Interpreting Non selected Latent Class Model s The following models were not selected for the present study because they were not supported by model fit statistics or were not as optimal as the selected 5 class model. I nformation about the non sele cted class model s including, substantive interpretations and classification probabilities are provided for comparative purpose s As noted in Chapter 4, substantive interpretations of models focused on examining shared features of functional ability variables within a profile and the distinguishing features of these variables across profiles. Functional ability profiles could be quantified related to the (a) severity or level of the limitations across functional ability variables (e.g., mild, moderat e, or severe limitations), (b) number of functional ability variables with limitations (e.g., a few, many, all), and (c) natu re or type of functional ability variables with limitations (e.g., limitations associated with a similar cluster of functional abil ity variables). Severity or level of limitations was examined by inspecting the mean score on each functional ability variable (4 point scale). To describe the profiles, the ratings for severity of limitations were grouped to account for the standard dev iations. M oderate to severe limitations were associated with mean scores from 2.5 and above, mild to moderate limitations were mean scores from 1. 5 to 2.49, and no to mild limitations were mean scores from 1. 49 and below. Two Class Model Table F 1 shows t he model implied means and standard deviations for each variable in each profile (i.e., subgroup) in the model. Profile 1 consisted of 40% of the sample and Profile 2 consisted of 60% o f the sample. Profile 1 was associated with

PAGE 336

336 m oderate to severe limita tions on six functional ability variables: communication, cognition, social skills, regulation of activity level, regulati on of attention, and motivation. Profile 1 also had mild to moderate limitations on seven functional ability variables: understanding overall health, use of arms, use of hands, use of legs, behavior, and regulation of emotion. Profile 2 was associated with mild to moderate limitations on five functional ability variables that included communication, cognition, regulation of activity l evel, regulation of attention, and motivation, and no limitations on other functional ability variables Overall, the 2 class model seemed to distinguish between a subgroup of children whose functional ability profile was associated with more moderate to severe lim itations on functional ability variables (Profile 1) and a subgroup of children whose functional ability profile was associated with mild limitations on functional ability variables (Profile 2). Profile probability estimates were high for the 2 class model Children in Profile 1 had a 9 4 % probability of being assigned to the first profile. Children in Profile 2 had a 9 6 % probability of being assigned to the second profile. Based on the substantive and statistical examination of the 2 class mo del it appeared that this model provided distinct classes, that could be meaningfully interpreted, however, other class model s were identified that provided a better fit with data and more substantive aspects to examine associations with social competence. For these reason s the 2 class model was not selected. Three Class Model The 3 class model improved in model fit statistics over the 2 class model (i.e., increased log likelihood and decreased BIC). Table F 2 shows the model implied

PAGE 337

337 means and standard deviations for the 15 functional ability variables for each profile. Profile 1 consisted of 19% of the sample, Profile 2 consisted of 37% of the sample, and Profile 3 consisted of 44% of the sample. Profile 1 was a subgroup of children with mod erate to severe limitations on eight functional ability variables. These included communication, understanding, cognition, use of hands, social skills, regulation of activity level, regulati on of attention, and motivation. This profile also was associated with mild to moderate limitations on f ive fun ctional ability variables: overall health, use of arms, use of legs, behavior, and regulation of emotions Profile 2 was associated with moderate to severe limitations on five functional ability variables: com munication, cognition, regulation of activity level, regulation of attention, and motivation. This profile was also associated with mild to moderate limitations on five variables: understanding, overall health, use of hands, social skills, and behavior. Profile 3 was associated with mild to moderate limitations on five functional ability variables : communication, cognition, regulation of activity level, regulation of attention, and motivation. Profile probability estimates also were high for the 3 clas s model Children in Profile 1 had a 93% probability of being assigned to the first profile, 6% probability of being assigned to the second profile, and less than 1% probability of being assigned to the third profile. Children in Profile 2 had a 88% prob ability of being assigned to the second profile, 5% probability of being assigned to the first profile, and 7% probability of being assigned to the third profile. Children in Profile 3 had less than 1%, 7%, and 93% probability of being assigned to the fir st, second, and third profile, respectively. The 3 class model provided distinct classes that could be meaningfully interpreted, but again,

PAGE 338

338 other class model s were identified that provided a better fit with the data and more substantive aspects to examine in relation to social competence. Four Class Model The 4 class model had a larger log likelihood and a smaller BIC than the 3 class model Table F 3 shows the model implied means and standard deviations for the 15 variables for each profile in the model. Profile 1 (14% of the sample) was associated with moderate to severe limitations on nine functional ability variables including communication, understanding, cognition, use of hands, social skills, behavior, regulation of activity level, regulation of attention, and motivation, and with mild to moderate limitations on four functional ability variables: overall health use of arms, use of legs, and regulation of emotions Profile 2 (8% of sample) was associat ed with moderate to severe limitations on five functional ability variables: communication, cognition, use of hands, use of legs, and motivation. This profile also was associated with mild to moderate limitations on seven functional ability variables: unde rstanding overall health, use of arms, social skills, regulation of activity level, regulation of attention and vision Profile 3 (35% of sample) was associated with moderate to severe disabilities on five functional ability variables : communication, cog nition, regulation of activity level, regulation of attention, and motivation. In addit i on, this profile was associated with mild to moderate disabilities on four functional ability variables understanding, overall health, social skills, and behavior. Pr ofile 4 (43% of the sample) was as sociated with mild to moderate limitations on five functional ability variables : communication cognition, regulation of activity level, regulation of attention, and motivation.

PAGE 339

339 The 4 class also had good profile probabili ty estimates. Children in the Profile 1 had a 92%, 3%, 5%, and 0% probability of being assigned to the first, second, third and fourth profile, respectively. Children in Profile 2 had a 4%, 90%, 5.0% and 1% probability of being assigned to the first, se cond, third and fourth profile, respectively, while children in Profile 3 had a 2%, 1%, 89% and 7% probability of being assigned to the first, second, third and fourth profile, respectively. Children in Profile 4 had a 0% probability of being assigned t o Profile 1, less than 1% probability of being assigned to Profile 2, 7% probability of being assigned to Profile 3, and a 94% probability of being assigned to Profile 4. The 4 class model provided distinct classes that could be meaningfully interpreted w ith adequate fit indices, but the continued decrease in the BIC for the 5 and 6 class model s warranted the examination of theses model s therefore, the 4 class model was not selected Six Class Model The 6 class model had good fit statistics (increased LL and reduced BIC over previous models) with 11 replications. The reduction in BIC, however, was not significant compared to previous reductions between models (e.g., BIC dropped 500 points between 3 and 4 class models BIC dropped 170 points between 4 and 5 class models, but only dropped 5 points between the 5 and 6 class models). Table F 4 shows the model implied means and standard deviations for the 15 variable for each profile in the model. The 6 class model wa s similar to the 5 class model Profile 1 ( 4 % of the sample) was associated with the same severity and type of limitations in functional abilities as Profile 1 in the 5 class model Profile 2 ( 10 % of the sample) was associated with the

PAGE 340

340 same severity and type of limitations in functional abilities as Profile 2 in the 5 class model Profile 3 ( 7 % of sample) was associated with the same severity and type of limitations in functional abilities as Profile 3 in the 5 class model Profile 6 ( 38 % of the sample) associated with the same severity and type of limitations in functional abilities as Profile 5 in the 5 class model The distinction in the 6 class model was the splitting of the subgroup identified in the 5 class model that was identified as Profile 4 The addition of a sixth profile appeared to break this subgroup into two groups (Profile 4 and Profile 5 in the 6 class model) Both profiles were associated with limitations in communication, understanding, overall health, social skills, behaviors, regu lation of activity level, regulation of attention, and motivation. As shown in Table F 4 Profile 4 ( 29 % of the sample) was associated with lower means on these variable s except for regulation of activity level which had a slightly higher mean score than Profile 5 Profile 5 ( 12 % of the sample) was associated with higher means on these variables compared to Profile 4, with the highest mean scores on cognition and motivation variable s The 6 class model probability estimates dropped below 85% for Profiles 4 and 5. Children in Profile 1 had a 95%, 3%, 1%, 0 %, 1 %, and 0% probability of being assigned to first, second, third fourth fifth and sixth profile, respectively. Children in Profile 2 had a 2%, 90%, less than 1%, 2 %, 5 % and 0% probability of being assigned to the first, second, third fourth, fifth and sixth profile, respectively. For children in Profile 3, there was a less than 1%, less than 1%, 88%, 5 %, 3 %, and 3% probability of being assigned to the first, second, third fourth, fifth and sixth profile, respectively. For Profile 4, children had a 0 %, less than 1% 1 %, 8 4 %, 6 % and 8 % probability of being assigned to

PAGE 341

341 the first second, third fourth, fifth and sixth profile, respectively. C hildren in Profile 5 had a 1 %, 5 %, 2 %, 12 %, 79 % and less than 1% probability of being assigned to the first, second, third fourth, fifth and sixth profile, respectively. Children in class 6 had a 0%, 0%, less than 1%, 7 %, 0 % and 92 % probability of being assigned to the first, second, third fourth, fifth and sixth profile, respectively. The 6 class model presented an interesting distinction between subgroups of children with and without mode rate to sever e limitations in cognition among children with mild to moderate limitations in functional ability. Although cognition is likely associated with teacher ratings of social competence, the use of this single variable to distinguish between sub gr oups was not justified, particularly given the BIC was not notably smaller and than the 5 class model and the decrease in the probability of membership (i.e., dropped below 80%) for assignment between Profile s 4 and 5; therefore the 6 class model was not selected.

PAGE 342

342 Table F 1 Model implied means (standard deviations) for 2 class model s Profile 1 ( n = 11 50 ) Profile 2 ( n = 17 20 ) Communication 2.98 (0.90) 2.17 (1.09) Understanding 2.20 (0.47) 1.25 (0.21) Cognition 3.16 (0.58) 2.05 (0.35) Overall Health 1.92 (1.02) 1.37 (0.48) Use of Arms 1.50 (0.74) 1.06 (0.09) Use of Hands 2.25 (1.22) 1.19 (0.21) Use of Legs 1.52 (0.76) 1.09 (0.18) Social Skills 2.63 (1.21) 1.42 (0.61) Behavior 2.10 (0.77) 1.20 (0.22) Reg. of Activity Lev. 3.05 (1.41) 1.93 (1.15) Reg. of Attention 2.96 (1.31) 1.81 (0.85) Motivation 3.01 (1.34) 2.15 (1.16) Reg. of Emotions 1.55 (0.92) 1.12 (0.21) Hearing 1.19 (0.50) 1.13 (0.36) Vision 1.27 (0.53) 1.07 (0.12)

PAGE 343

343 Table F 2 Model implied means (standard deviations) for 3 class model s Profile 1 ( n = 5 60 ) Profile 2 ( n = 11 70 ) Profile 3 ( n = 12 50 ) Communication 3.32 (0.70) 2.56 (0.92) 2.07 (1.11) Understanding 2.51 (0.46) 1.77 (0.32) 1.12 (0.11) Cognition 3.46 (0.56) 2.66 (0.47) 1.90 (0.25) Overall Health 2.22 (1.14) 1.57 (0.68) 1.33 (0.44) Use of Arms 1.85 (1.09) 1.11 (0.11) 1.06 (0.11) Use of Hands 2.83 (1.31) 1.50 (0.40) 1.15 (0.21) Use of Legs 1.82 (1.04) 1.15 (0.22) 1.09 (0.21) Social Skills 3.00 (1.04) 2.05 (1.09) 1.28 (0.43) Behavior 2.35 (0.90) 1.67 (0.50) 1.10 (0.11) Reg. of Activity Lev. 3.01 (1.48) 2.84 (1.49) 1.69 (0.81) Reg. of Attention 3.14 (1.28) 2.57 (1.19) 1.62 (0.673 Motivation 3.20 (1.28) 2.67 (1.33) 2.02 (1.06) Reg. of Emotions 1.68 (1.10) 1.33 (0.58) 1.09 (0.14) Hearing 1.17 (0.45) 1.18 (0.47) 1.13 (0.35) Vision 1.42 (0.83) 1.12 (0.20) 1.06 (0.12)

PAGE 344

344 Table F 3 Model implied means (standard deviations) for 4 class model s Profile 1 ( n = 390 ) Profile 2 ( n = 240 ) Profile 3 ( n = 1010 ) Profile 4 ( n = 1220 ) Communication 3.50 (0.39) 2.54 (1.28) 2.58 (0.91) 2.09 (1.11) Understanding 2.71 (0.29) 1.75 (0.53) 1.81 (0.34) 1.12 (0.11) Cognition 3.52 (0.57) 2.91 (0.80) 2.69 (0.49) 1.92 (0.25) Overall Health 2.18 (1.11) 2.21 (1.15) 1.56 (0.65) 1.31 (0.42) Use of Arms 1.63 (0.80) 2.32 (1.22) 1.08 (0.07) 1.02 (0.03) Use of Hands 2.68 (1.31) 2.94 (1.08) 1.48 (0.40) 1.11 (0.13) Use of Legs 1.55 (0.64) 2.62 (1.26) 1.07 (0.06) 1.05 (0.09) Social Skills 3.30 (0.67) 2.08 (1.15) 2.07 (1.12) 1.29 (0.44) Behavior 2.72 (0.64) 1.37 (0.35) 1.71 (0.52) 1.10 (0.12) Reg. of Activity Lev. 3.31 (1.19) 2.12 (1.43) 2.90 (1.44) 1.69 (0.81) Reg. of Attention 3.32 (1.11) 2.38 (1.45) 2.62 (1.19) 1.63 (0.67) Motivation 3.34 (1.07) 2.57 (1.47) 2.71 (1.35) 2.02 (1.06) Reg. of Emotions 1.84 (1.20) 1.28 (0.60) 1.33 (0.58) 1.09 (0.13) Hearing 1.16 (0.44) 1.17 (0.48) 1.17 (0.45) 1.13 (0.37) Vision 1.25 (0.50) 1.72 (1.17) 1.10 (0.18) 1.04 (0.08)

PAGE 345

345 Table F 4 Model implied means (standard deviations) for 6 class model s Profile 1 ( n = 1 00 ) Profile 2 ( n = 290 ) Profile 3 ( n = 200 ) Profile 4 ( n = 830 ) Profile 5 ( n = 350 ) Profile 6 ( n = 1 090 ) Communication 3.49 (0.54) 3.51 (0.35) 1.98 (1.20) 2.49 (0.98) 2.94 (0.65) 2.08 (1.11) Understanding 2.53 (0.47) 2.77 (0.26) 1.26 (0.23) 1.68 (0.34) 2.15 (0.23) 1.11 (0.10) Cognition 3.81 (0.19) 3.44 (0.70) 2.28 (0.54) 2.49 (0.51) 3.26 (0.22) 1.91 (0.24) Overall Health 2.47 (0.97) 2.16 (1.13) 2.16 (1.13) 1.53 (0.61) 1.67 (0.81) 1.28 (0.37) Use of Arms 3.18 (1.00) 1.42 (0.39) 2.01 (1.09) 1.01 (0.01) 1.29 (0.21) 1.01 (0.00) Use of Hands 3.74 (0.45) 2.48 (1.30) 2.43 (1.01) 1.28 (0.26) 2.23 (0.73) 1.09 (0.09) Use of Legs 3.05 (1.05) 1.35 (0.32) 2.30 (1.41) 1.02 (0.02) 1.31 (0.23) 1.03 (0.06) Social Skills 3.13 (1.04) 3.40 (0.58) 1.78 (0.90) 2.03 (1.11) 2.15 (1.07) 1.24 (0.38) Behavior 1.93 (0.67) 2.90 (0.62) 1.31 (0.33) 1.67 (0.52) 1.73 (0.45) 1.07 (0.08) Reg. of Activity Lev. 2.44 (1.43) 3.48 (1.01) 2.15 (1.54) 2.87 (1.48) 2.71 (1.40) 1.63 (0.71) Reg. of Attention 2.98 (1.30) 3.36 (1.07) 2.13 (1.35) 2.55 (1.17) 2.73 (1.32) 1.59 (0.61) Motivation 3.15 (1.51) 3.27 (1.10) 2.30 (1.24) 2.58 (1.34) 3.09 (1.29) 2.00 (1.03) Reg. of Emotions 1.21 (0.30) 2.00 (1.29) 1.50 (1.05) 1.32 (0.55) 1.31 (0.58) 1.06 (0.06) Hearing 1.19 (0.51) 1.17 (0.43) 1.21 (0.59) 1.20 (0.53) 1.05 (0.15) 1.13 (0.36) Vision 2.50 (1.68) 1.16 (0.32) 1.44 (0.73) 1.10 (0.18) 1.08 (0.11) 1.04 (0.08)

PAGE 346

346 R EFERENCES American Association on Intellectual and Developmental Disabilities (2010) Intellectual disability: Definition, classification, and systems of supports Washington, DC: AAIDD. American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders (4th e d., text rev.). Washington, DC: Author. American Psychological Association (2010). Publication manual of the American Psychological Association (6th ed.) Washington, DC: American Psychological Association Bailey, D. B., Simeonsson, R. J. Buysse, V., & Smith, T. M. (1993). Reliability of an index of child characteristics. Developmental Medicine & Child Neurology, 35 806 815. Bailey, D. B., & Wolery M (199 2 ). Teaching infants and preschoolers with disabilities (2nd ed.) Upper Saddle River, NJ: Prenti c e Hall. Baker, B L., McIntyre, L L., Blacher, J., Crnic, K., Edelbrock, C., & Low, C (2003) Pre school children with and without developmental delay: B ehavior problems and parenting stress overtime Journal of Intellectual Disability Research, 47 217 230. Bergman, L., & Magnusson, D. (1997). A person oriented approach in research on developmental psychology. Developmental and Psychopathology, 9 291 319. Bergman, L., Magnusson, D., & El Khouri, B. (2003). Studying individual development in an interindividual context: A person oriented approach (Vol. 4). In D. Magnusson (Series Ed.), Paths through life Mahwah, NJ: Erlbaum. Bitterman, A., Daley, T., Misra, S., Carlson, E., & Markowitz, J (2008) A national sample of preschoolers with autism spectrum disorders: special education services and parent satisfaction Journal of Autism and Developmental Disorders, 38, 1509 1517. Bjorck Akesson, E., Wilder, J., Granlund, M., Pless, M., Simeons son, R J., Adolfsson, Lillvist, A (2010) The international classification of functioning, disability and health and the version for children and youth as a tool in child habilitation/early childhood intervention: Feasibility and usefulness as a common language and frame of reference for practice Disability and Rehabilitation, 32, 125 138. Blackorby, J., & Cameto, R. (2004). Changes in the school engagement and academic performance of students with disabilities Menlo Park, CA: SRI. Retrieved from SEEL S website: http://www.seels.net/infoproduct.htm

PAGE 347

347 Blackorby, J., Wagner, M., Cadwallader, T., Cameto, R., Leveine, P., & Marder, C. (2002). Behind the label: The functional implications of disability Menlo Park, CA: SRI. Retrieved from SEELS website: http:/ /www.seels.net/infoproduct.htm Blandon, A Y., Calkins, S D., & Keane, S P (2010) Predicting emotional and social competence during early childhood from toddler risk and maternal behavior Development and Psychopathology, 22, 119 132. Brown, M., & Gordon, W. A. (1987) Impact of impairment on activity patterns of children. Archives of Physical Medicine and Rehabilitation, 68, 828 832. Brown, W. & Conroy, M. A. (2001). Promoting peer related social communicative competence in preschool children with developmental delays. In H. Goldstein, L., Kaczmarek, & K. English (Eds.), Promoting social communication in children and youth with developmental disabilities (pp. 173 210). Baltimore MD : Brookes. Brown, W ., Odom, S., & Conroy, M. A. (2001). An interven tion hierarchy for promoting Topics in Early Childhood Special Education 21 162 175. Brown, W., Odom, S., & Holcombe, A (1996) Observational assessment of young children social behavior with p eers Early Childhood Research Quarterly, 11, 19 40. Brown, W., Odom, S., & McConnell, S (2008) Social competence of young children: Risk disability and intervention Baltimore, MD: Brookes. Brown, W., Odom, S., McConnell, S ., & Rathel, J. (2008) Peer i nteraction interventions for preschool children with developmental difficulties. In W. Brown, S. Odom, & S. McConnell (Eds.), Social competence of young children: Risk disability and intervention (pp 141 164) Baltimore, MD: Brookes. Burchinal, M., Vandergrift, N., Pianta, R., & Mashburn, A. (2010). Threshold analysis of associations between child care quality and child outcomes for low income children in pre kindergarten programs. Early Childhood Research Quarterly, 25, 166 176. Burke P J., & Ruedel, K (2008) Disability classificati on, categorization in education : A US perspective In L Florian & M McLaughlin (Eds.), Disability classification in education: Issues and perspectives (pp 68 76) Thousand Oaks, CA: Corwin Press Buy sse, V., Bailey, D. B., Smith, T. M., & Simeonsson, R. J. (1994). The relationship between child characteristics and placement in specialized versus inclusive early childhood programs. Topics in Early Childhood Special Education, 14, 419 435.

PAGE 348

348 Buysse, V., G oldman, B., & Skinner, M. (2002). Setting effects on friendship formation among young children with and without disabilities. Exceptional Children, 68, 503 517. Buysse, V., Smith, T. M., Bailey, D. B., & Simeonsson, R. J. (1993). Consumer validation of an index characterizing the functional abilities of young children with disabilities. Journal of Early Intervention, 17 224 238. Campbell, S B. (19 94 ) Hard to manage preschool boys: Externalizing behavior, social competence, and family context at two year follow up. Journal of Abnormal Child Psychology, 22 147 166 Campbell, S B., Breaux, A M., Ewing, L J., & Szumoski, E K (1986) Correlates and predictors of hyperactivity and aggression: A longitudinal study of parent referred problem preschoolers Journal of Abnormal Child Psychology, 14, 217 234. Campbell, S B., March, C., Pierce, E., Ewing, L J., & Szumoski, E K (19 91 ) Hard to manage preschool boys: Family context and the stability of externalizing behavior. Journal of Abnormal Child Psycholo gy, 19 301 318 Campbell, S B., Shaw, D S., & Gilliom, M (2000) Early externalizing behavior problems: Toddler and preschooler at risk for later maladjustment Developmental and Psychopathology, 12, 467 488. Campbell, S B., Spieker, S., Buchinal, M., Poe, M D., & the NICHD Early Child Care Research Network (2006) Trajectories of aggression from toddlerhood to age 9 predict academic and social functioning through age 12 Journal of Child Psychology and Psychiatry, 47, 791 800. Carlson, E., Bitterman, A., & Daley, T (2010) Access to educational and community activities for young children with disabilities Rockville, MD: Westat Retrieved from PEELS website : www.peels.org. Carlson, E., Bitterman, A., & Jenkins, F. (2010). Home literacy environments a nd its role in the achievement of preschoolers with disabilities. The Journal of Special Education, 43, 1 11. doi: 10.1177/0022466910371229 Carlson, E., Daley, T., Bitterman, A., Heinzen, H., Keller, B., Markowitz, J., & Riley, J (2009) Early school tran sitions and the social behavior of children with disabilities: Selected finding from the pre school elementary education longitudinal study wave 3 report Retrieved from PEELS website : https://www.peels.org /reports.asp Carlson, E., Daley, T., Bitterman, A., Riley, J., Keller, B., & Markowitz, J (2008) Changes in the characteristics, services, and performance of preschoolers with disabilities from 2003 04 to 2004 05: Wave 2 overview report from the pre school elementary education longitudinal study Retrieved from PEELS website : https://www.peels.org/reports.asp

PAGE 349

349 Carlson, E., & Lowe, A (2009, January) PEELS training institute Presentation at the National Center for Special Education Research PEELS Training Institute, Washington, DC. Carlson, E., Posner, D., & Lee, H (2008) Pre elementary education longitudinal study restricted Washington DC: U.S Department of Education, National Center for Spe cial Education Research. Carlson, J F., Benson, N., & Oakland, T (2010) Implications of the international classification of functioning, disability and health (ICF) for test development and use School Psychology International, 31, 353 371 doi:10.1177/0143034310377149 Center on the Developing Child at Harvard University. (2011). traffic control system: How early experiences shape the development of executive function : Working p aper no. 11 Retrieved from http://www.developingchild.harvard.edu Chambers, J., Parrish, T., & Harris, J. (200 4 ) What are we spending on special education in the United States, 1999 2000?. Palo Alto, CA: American Institutes for Research Report submitted to the Office of Special Education Programs Retrieved from http://csef.air.org/pub_seep_national.php Chambers, J., Parrish, T., Shkolnik, J., Levine, R., & Makris, F (2003) The p urpose and design of the special education expenditure project Palo Alto, CA: American Institutes for Research Retrieved from http://csef.air.org/pub_seep_national.php Chambers, J., Perez, M., Socias, M., Shkolnik, J., Esra, P, & Campbell Brown, S (2004 ) Educating students with disabilities: Comparing methods for explaining expenditure variation Palo Alto, CA: American Institutes for Research Retrieved from http://csef.air.org/pub_seep_national.php Chambers, J., Shkolnik, J., & Harris, J. (2003) Tota l expenditure for students with disabilities, 1999 2000: Spending variation by disability. Palo Alto, CA: American Institutes for Research Retrieved from http://csef.air.org/pub_seep_national.php Clements, M., Reynolds, A., & Hickey, E. (2004). Site school and social competence in the Chicago child parent centers. Early Childhood Research Quarterly, 19, 273 296. Collins, L., & Lanza, S. (2010). Latent class and latent transition analysis : With ap plications in the social, behavioral, and health sciences Hoboken, NJ: Wiley & Sons Inc. Conroy, M. A., Brown, W., & Olive, M. L. (2008 ). Social competence interventions for young children with challenging behavior. In W. Brown, S. Odom, & S. McConnell (E ds.), Social competence of young children: Risk disability and intervention (pp 205 232 ) Baltimore, MD: Brookes.

PAGE 350

350 Dahl T H (2002) International classification of functioning, disability and health: An introduction and discussion of its potential impac t on rehabilitation services and research Journal of Rehabilitation Medicine, 34 201 204. Daley, T. C. & Carlson, E (2009) Predictors of change in eligibility status among preschoolers in special education Exceptional Children, 74, 412 426. Daley, T C., Simeonsson, R J., & Carlson, E (2009) Constructing and testing a disability index in a US sample of preschoolers with disabilities Disability and Rehabilitation, 31 538 552. Danaher, J (2007) Eligibility policies and practices for young children under part B of IDEA (NECTAC Notes No 24) Chapel Hill: The University of North Carolina, FPG Child Development Institute, National Early Childhood Technical Assistance Center. Dawson, G., Toth, K., Ab bott, R., Osterling, J., Munson, J., Estes, A., & Liaw, J. (2004). Early social attention impairments in autism: Social orienting, joint attention, and attention to distress. Developmental Psychology, 40, 271 283. De Kleijn De Vrank r ijker, M W (2003) T he long way from the international classification of impairments, disabilities, and handicaps (ICIDH) to the international classification of functioning, disability and health (ICF) Disability and Rehabilitation, 25 561 564. Diamond, K., Hong, S. Y., & B competence in early childhood programs. In W. Brown, S. Odom, & S. McConnell (Eds.), Social competence of young children: Risk disability and intervention (pp 205 232 ) Baltimore, MD: Brookes. Diamond, K., Hong, S. Y., & Tu, H. (2008). Context influences preschool children's decisions to include a peer with a physical disability in play. Exceptionality, 16, 141 155. Domitrovich, C., Cortes, R., & curriculum. The Journal of Primary Prevention, 28 67 91. Drasgow, E., Lowery, A., Turan, Y., Halle, J., & Meadan, H. (2008 ). Soc ial competence interventions for young children with serve disabilities. In W. Brown, S. Odom, & S. McConnell (Eds.), Social competence of young children: Risk disability and intervention (pp 273 300 ) Baltimore, MD: Brookes. Dunlap, G., Strain, P S., Fo x, L., Carta, J J., Conroy, M., Smith, B J., (2006) summary of current knowledge Behavioral Disorders, 32, 29 45.

PAGE 351

351 Dunst, C J., & Trivette, C M (2004) Toward a ca tegorization scheme of child find, referral, early identification and eligibility determination practices Tracelines, 1 (2). Retrieved from http://www.tracecenter.info/tracelines.php Dunst, C J., Trivette, C M., Appl, D J., & Bagnato, S J (2004) Framework for investigating child find, referral, early identification and eligibility determination practices Tracelines, 1 (1). Retrieved from http://www.tracecenter.info/tracelines.php Early Childhood Outcome s Center. ( 20 09, November). The child outcomes ECO Resources. Retrieved from http://www.fpg.unc.edu/~eco/pages/outcomes.cfm Elias, M., Gara, M., Schuyler, T., Brandon Muller, L., & Sayet te, M. (1991). The promotion of social competence: Longitudinal study of preventive school based program. American Journal of Orthopsychiatry, 61, 409 417. English, K., Goldstein, H., Shafer, K., & Kaczmarek, L A (1997). Promoting interactions among pres choolers with and without disabilities: E ffects of a buddy skills training program. Exceptional Children, 63 229 243 Etscheidt, S. (2006). Least restrictive and natural environments for young children with disabilities: A legal analysis of issues. Topics in Early Childhood Special Education, 26, 167 178. Everitt, B S., Landau, S., & Leese, M (2001) Cluster analysis (4th ed.) London, UK: Wiley & Sons Ltd. Florian, L., Hollenweger, J., Simeonsson, R J., Wedell, K., Riddell, S., Terzi, L., & Holland, A (2006) Cross cultural perspectives on the classification of children with disabilities The Journal of Special Education, 40 36 45. Florian, L., & McLaughlin, M J (2008) Disability classification in education In L Florian & M McLaughlin (Eds.), D isability classification in education: Issues and perspectives (pp 3 10) Thousand Oaks, CA: Corwin Press Forhan, M (2009) An analysis of disability models and the application of the ICF to obesity Disability and Rehabilitation, 31, 1382 1388. Fox, L Dunlap, G., Hemmeter, M. L., Joseph, G., & Strain, P. (2003). The Teaching Pyramid: A model for supporting social competence and preventing challenging behavior in young children. Young Children 58 48 53. Fox, L., Dunlap, G., & Powell, D (2002) Young children and challenging behavior: Issues and considerations for behavior support Journal of Positive Behavior Interventions 4, 208 217.

PAGE 352

352 Gifford Smith, M., & Brownell, C. (2003). Childhood peer relationships: Social acceptance, friendships, and peer net works. Journal of School Psychology, 41 235 284. Gilliam, W S (2005) Prekindergarteners left behind: Expulsion rates in state prekindergarten systems Retrieved from http://www.preschool.org/documents/pk expulsion.pdf Granlund, M., Eriksson, L., & Ylven, R (2004) Utility of international classification of to items from extant rating instruments Journal of Rehabilitation Medicine, 36, 130 137. Greshman, F. M., & Eliot, S. N. (1990). Social skills rating system manual. Minneapolis, MN: Pearson. Guralnick, M J (1999) Family and child influences on the peer related social competence of young children with developmental delays Mental Retardation and Developmental Disabilities Research Reviews, 5, 21 29. Guralnick, M J (2006) Peer relationships and the mental health of young children with intellectual delays Journal of Policy and P ractice in Intellectual Disabilities 3, 49 56. Guralnick, M J., Hammond, M A., Connor, R T., & Neville, B (2006) Stability, change, and correlates of the peer relationships of young children with mild developmental delays Child Development, 77 312 324. Gutman, L. M. Sameroff, A. J. & Cole, R (2003) Academic growth curve trajectories from 1 st grade to 12 th grade: Effects of multiple social risk factors and preschool child factors Developmental Psychology, 39 777 790. Haapasalo, J., Temblay, R., Boue l rice, B., & Vitaro, F. (2000). Relative advantages of person and variable based approaches for predicting problem behaviors from kindergarten assessments. Journal of Quantitative Criminology, 16 145 168. Hair, E., Halle, T., Terry Humen, E., Lavell e, B., & Calkins, J (2006) readiness in the ECLS K: Predictions to academic, health, and social outcomes in first grade Early Childhood Research Quarterly, 21, 431 454. Haring, K., Farron Davis, F., Goetz, L., Karasoff, P., Sailor, W., & Zeph, L (1992) LRE and th e placement of students with sever e disabilities Journal of the Association for Persons with Severe Handicaps 17 145 153. Haring, K., Lovett, D., Haney, K., Algozzine, B., Smith, D., & Clarke, J. (1992). Labeling preschoole r as learning disabled: A cautionary tale. Topics in Early Childhood Special Education, 12, 151 173.

PAGE 353

353 Harper, L., & McCluskey, K. (2002). Caregiver and peer responses to children with language and motor disabilities in inclusive preschool programs. Early Ch ildhood Research Quarterly 17, 148 166. Hebbeler, K ., & Kahn, L (2008) Setting targets for child outcomes Early Childhood Outcomes Center Report Retrieved from http://www.fpg.unc.edu/~eco/pages/papers.cfm#settingtargets Hebert Myers, H., Guttentag, C L., Swank, P R., Smith, K E., & Landry, S H (2006) The importance of language, social, and behavioral skills across early and later childhood as predictors of social competence with peers Applied Developmental Science, 10, 174 187. Hemmeter, M. L., Fox, L., & Snyder, P. (2008). The Teaching Pyramid Observation Tool research edition Unpublished instrument. Vanderbilt University, TN. Hemmeter, M.L., Fox, L., & Snyder, P. (2010). Evaluating the potential efficacy of a classroom wide model for p romoting social emotional development and addressing challenging behavior in preschool [Data file and codebook]. Vanderbilt University, TN. Hobbs, N. (1975). The future of children: Categories, labels, and their consequences Retrieved from http://www.eric.ed.gov/ERICWebPortal/search/detailmini.jsp?_ nfpb =true&_&ERICExtSearch_SearchValue_0=ED115069&ERICExtSearch_ Sear chType_0=no&accno=ED115069 Holm, S. (1979). A simple sequentially rejective multiple test procedure. Scandinavian Journal of Statistics 6 65 70. Huffman, L. C., Mehlinger, S L., & Kerivan, A S (2001) Off to a good start: Research on the risk factors for early school problems : Paper n o 1 Retrieved from http://www.eric.ed.gov/ERICWebPortal/custom/portlets/recordDetails/deta ilmini.jsp ?_nfpb=true&_&ERICExtSearch_SearchValue_0=ED476378&ERICExtSearch_Se archType_0=no&accno=ED476378 IDEA Data (2008) Child count data tables Data Accountability Center. Retrieved from https://www.ideadata.org Imrie, R (2003) Demystifying disability: A review of the i nternational classification of functioning, disability, and health Sociology of Health and Illness, 26, 1 19. Individuals with Disabilities Education Act of 1991, Pub L No 102 119, (1991). Individuals with Disabilities Education Act of 1997, Pub L No 105 17, (1997). Individuals with Disabilities Education Improvement Act of 2004, Pub L No 108 446, (2004).

PAGE 354

354 Institute of Education Sciences. (2011). Request for applications: Special education research grants CFDA n umber: 84.324A Retrieved from http://ies.ed.gov/funding/ncser_rfas/ncser_earlyintervention.asp Janson, H., & Mathiesen, K. (2008). Temperament profiles from infancy to middle childhood: Development an d associations with behavior problems. Developmental Psychology, 44 1314 1328. Johnson, B. (2000, April). comparative and correlational research in educational research methods textbooks Paper presented at the Annual Conference of American Educational Research Association. New Orleans, LA. Keller, T., Spieker, S., & Gilchrist, L. (2005). Patterns of risk and trajectories of preschool problem behaviors: A person oriented analysis of attachment in context. Develop ment and Psychopathology, 17 349 384. King G Law M King S Rosenbaum P Kertoy M & Young N. L. (2003) A conceptual model of the factors affecting the recreation and leisure participation of children with disabilities. Physical & Occupationa l Therapy in Pediatrics, 23, 63 90. King, G., McDougall, J., De Witt, D., Hong, S., Miller, L., Offord, D., La Porta, J (2005) Roles of physical health status, environmental, family, a nd child factors International Journal of Disability, Development, and Education, 52 313 344. King, G., McDo u gall, J., DeWitt, D., Petrenchik, T., Hurley, P., & Law, M (2009). Predictors of change over time in the activity participation of children and youth with physical disabilities. Child Health Care, 38, 321 351. King, G., Specht, J., Schultz, I., Warr Leeper, G., Redekop, W., & Risebrough, N. (1997). Social skills training for withdrawn unpopular children with physical disabilities: A preliminary ev aluation. Rehabilitation Psychology, 42 47 60. Knitzer, J. (2002). Building services and systems to support the healthy emotional development of young children: An action guide for policymakers (Report) Retrieved from the National Center for Children in Poverty website : http://www.nccp.org/publications/mentalhealth_pubs.html Konold T R., & Pianta, R C (2005) Empirically driven, person orientated patterns of school readiness in typically developing children: Descriptions and prediction to first grade Applied Developmental Science, 9, 174 187. behavioral functioning: A latent variable approach. Journal of Psychoeducational Assessment, 25, 222 236.

PAGE 355

355 Kri s hnakumar, A. & Black, M M (2002) Longitudinal predictors of competence among African American children: The role of distal and proximal risk factors Applied Developmental Psychology, 23, 237 266. Kronk, R., Ogonowski, J., Rice, C., & Feldman, H. (2005). Reliabili ty in assigning ICF codes to children with special health care needs using a developmentally structured interview. Disability and Rehabilitation, 27, 977 983. Ladd G W., & Price, J M (1987) following t he transition from preschool to kindergarten Child Development, 58, 1168 1189. Lecavalier, L., Leone, S., & Witt, J (2006) The impact of behavior problems on caregiver stress in young people with autism spectrum disorders Journal of Intellectual Resear ch, 50, 172 183. Li, L., Lee, H., Lo, A., & Norman, G (2008) Imputation of missing data for the pre elementary education longitudinal study Proceedings for the American Statistical Association for the Joint Statistical Meetings in Denver Colorado Retri eved from www.amstat.org/sections/srms/proceedings/y2008/Files/301122.pdf Linehan, P (2001) Developmental delay: Review of research and future directions Proceedings from Project Forum at the National Association of State Directors of Special Education Retrieved from www.projectforum.org/docs/development_delay.pdf Loeber, R., & Hay, D (1997) Key issues in the development of aggression and violence from childhood to early adulthood Annual Review Psychology, 48 371 410. Lollar, D J., & Simeonsson, R J (2005) Diagnosis to function: Classification for children and youths Developmental and Behavioral Pediatrics, 26, 323 330. MacMillan, D L., & Reschly, D J (1998) Overrepresentation of minority students: The case for greater specificity or recons ideration of the variables examined The Journal of Special Education, 32, 15 24. MacQueen, J B (1967) Some m ethods for classification and a nalysis of m ultivariate observations. Proceedings of 5th Berkeley Symposium on Mathematical Statistics and Probab ility (pp. 281 297 ). Retrieved from http://projecteuclid.org/DPubS?service=UI&version=1.0&verb=Display&handle=eu clid.bsmsp/1200512992 Madden, R., Sykes, C., & Bedirhan U. T. (2007) World health organizations family of international classifications: Defin ition, scope and purpose WHO Report Retrieved from www.who.int/classifications/en/FamilyDocument2007.pdf Magidson J., & Vermunt, J.K (2006), Latent class models Document retrieved from http://www.statisticalinnovations.com/articles/

PAGE 356

356 Marder, C. (2009). Menlo Park, CA: SRI. Retrieved from SEELS website: http://www.seels.net/infoproduct.htm Markowitz, J., Carlson, E., Frey, W., Riley, J., Shimshak, A., Heinzen, H., Lee, H (2006) Preschoolers with disabilities, characteristics, services, and results: Wave 1 overview report from the pre school elementary education longitudinal study Retrieved from PEELS website : https://www.peels.org/reports.asp Matson, J., & Shoemaker, M. (2009). Intellectual disab ility and its relationship to autism spectrum disorders. Research in Developmental Disabilities, 30, 1107 1114. McCollum, J A., & Ostrosky, M (2008) related social competence In W Brown, S Odom, & S McC onnell (Eds.), Social competence of young children : Risk disability and intervention (pp 31 60 ) Baltimore, MD: Brookes. McConnell, S. (2002). Interventions to facilitate social interaction for young children with Autism: Review of available research and recommendations for educational intervention and future research. Journal of Autism and Developmental Disorders, 32, 351 372. McCrae, J., & Barth, R. (2008). Using cumulative risk to screen for mental health problems in child welfare. Research on Social Work Practice, 18 144 159. McCutcheon, A. (1987). Latent class analysis Newbury Park, CA: Sage Publications. McIntyre, L L., Blacher, J., & Baker, B L (2006) The transition to school: adaptation in young children with and without intellectual disability Journal of Intellectual Disability Research, 50 349 361. McLean, M., Smith, B J., Mc Cormick, K., Schakel, J., & McE voy, M (1991) Developmental delay: Establishing parameters for a preschool category of exceptionality (White Paper) Missoula MT: Division for Early Childhood. Mendez, J L., Fantuzzo, J., & Cicchetti, D (2002) Profiles of social competence among low income African American preschool children Child Development, 73, 1085 1100. Mendez, J L., McDermott, P., & Fantuzzo, J (200 2) Identifying and promoting social competence with African American preschool children: Developmental and contextual consideration Psychology in the Schools, 39 111 123. Merrell, K W (2002) al second edition Austin, TX: Pro E d.

PAGE 357

357 Missall, K., & Hojnoski, R. (2008 related social competence fro transition to school. In W. Brown, S. Odom, & S. McConnell (Eds.), Social competence of young chi ldren: Risk disability and intervention (pp 117 140 ) Baltimore, MD: Brookes. Morris, C., Kurinczuk, J., & Fitzpatrick, R. (2005). Child or family assessed measures of activity performance and participation for children with cerebral palsy: A structured i nterview. Child Care, Health, and Development, 31 397 407. Muller, E., & Markowitz, J (2004) Disability categories: State terminology, definitions, & eligibility criteria (Project FORUM Report) Retrieved from the National Association of State Directors of Special Education website: www.projectforum.org/docs/disability_categories.pdf Muller, E., Markowtiz, J., & Srivastava S (2005) Disability categories: Relation of state terms and eligibility criteria to the proportion of children receiving special education services (Project FORUM Report) Retrieved from the National Association of State Directors of Special Education website: www.projectforum.org/docs/DisabilityCategories_two.pdf Muthen, L K., & Muthen B O (2007) (5th ed.). Los Angeles, CA: Muthen & Muthen. National C ouncil on the Developing Child (2004a) Young children develop in an environment of relationship : Working paper no. 1 Retrieved from http://www.developingchild.net/pubs/wp.html National Counci l on the Developing Child (2004b) built into the architecture of their brain : Working paper no. 2 Retrieved from http://www.developingchild.net/pubs/wp.html National Institute of Child Health and Human Development Early Child Care Research American Educational Research Journal, 39, 133 164. National Institute of Child Health and Human Development Early Child Care Research Network (2003) Social functioning in first grade: Associati ons with earlier home and child care predictors and with current classroom experiences Child Development, 74, 1639 1662. NICHCY the National Dissem ination Center for Children with Disabilities (2009) Categories of disability under IDEA Retrieved from: http://www.nichcy.org/disabilities/categories/pages/Default.aspx Ny lund, K. L., Asparouhov, T., & Muthn, B. (2007). Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Structural Equation Modeling: A Multidisciplinary Journal, 14, 535 569.

PAGE 358

358 Odom, S., McCo nnell, S., McEvoy, M. Favazza, P. C. (1999). Relative effects of interventions supporting the social competence of young children with disabilities. Topics in Early Childhood Special Education, 19 75 91. Odom, S., & McConnell, S (1985) A performance based conceptualization of social competence of handicapped preschool children: Implications for assessment Topics in Early Childhood Special Education, 4, 1 19. Odom, S., McConnell, S., & Brown, W. (2008 ). Social com petence of young children: Conceptualization, assessment, and influences. In W. Brown, S. Odom, & S. McConnell (Eds.), Social competence of young children: Risk disability and intervention (pp 117 140 ) Baltimore, MD: Brookes. Odom, S., McConnell, S., & McEvoy, M (1992) Social competence of young children with disabilities: Issues and strategies for intervention Baltimore, MD: Brookes. Office of Special Education Program. (2006, April) 26 th Annual (2004) report to congress on the Implementation of the Individuals with Disabilities Education Act vol. 1. Washington, D.C. Ogonowski, J., Kronk, R., Rice, C., & Feldman, H. (2004). Inter rater reliability in assigning ICF codes to children with disabilities. Disability and Rehabilitation, 2 6 353 361. Olson S., & Hoza, B (1993) Preschool developmental antecedents of conduct problems in children beginning school Journal of Clinical Psychology, 22 60 67. Olson, S., & Lunkenheimer, E. (2009) Expanding concepts of self regulation to social relationships: Transactional processes in the development of early behavioral adjustment In A Sameroff (Ed.), The transactional model of development: How children and contexts shape each other (pp 55 76) Washington, DC: American Psychological Association. Oravecz, L M., Koblinsky, S A., & Randolph, S M (2008) Community violence, interpartner conflict, parenting, and social supports as predictors of the social competence of African American preschool children Journal of Black Psychology, 34, 192 216 doi: 10.1177 /0095798408314142 Pedha uze r, E. J. (1982). Multiple regression in behavioral research: Explanation and prediction New York: NY: CBS College Publishing. Perry, D F., Dunne, C M., McFadden, L., & Campbell, D (2008) Reducing the risk for preschool expul sion: Mental health consultation for young children with challenging behaviors Journal of Child and Family Studies, 17, 44 54 doi: 10.1007/s10826 007 9140 7

PAGE 359

359 Peterson, D B (2005) International classification of functioning, disability and health: An introduction for rehabilitation psychologists Rehabilitation Psychology, 50, 105 112. Qi, C., & Kaiser, A P (2003) Behavior problems of preschool children from low income families: Review of the literature Topics in Early Childhood Special Education, 23, 188 216. Raghavendra, P., Borman, J., Granlund, M & Bjorck Akesson, E. (2007) The world health organization international classification of functioning, disability and health: Implications for clinical and research practice in the field of augmentat ive and alternative communication Augmentative and Alternative Communication, 23, 349 361. Raver, C G., Gershoff, E T., & Aber, J L (2007) Testing equivalence of mediating models of income, parenting, and school readiness for white, black, and Hispan ic children in a national sample Child Development, 78 96 115. Raver, C G ., & Zigler, E F (1997) Social competence: An untapped dimension in Early Childhood Research Quarterly, 12, 363 385. Reichle, J. (1997) Communi cation intervention with persons who have severe disabilities. The Journal of Special Education, 31, 110 134. Reid, R., DuPaul, G. J., Power, T. J., Anastopoulos, A. D., Rogers Atkinson, D., Noll, M. B., & Riccio, C. (1998). Assessing culturally different students for AD/HD using behavior rating scales. Journal of Abnormal Child Psychology, 26, 187 198. Reschly, D. (1996). Identification and assessment of students with disabilities. The Future of Children: Special Education for Students with Disabilities, 6, 40 53. Romano, E., Kohen, D., & Findlay, L C (2010) Associations among child care, family, and behavior outcomes in a nation wide sample of preschool aged children International Journal of Behavioral Development, 34, 427 440 doi: 10.1177/0165025409351657 Rosenbaum, P., & Stewart, D (2004) The world health organization international classification of functioning, disability and health: A model to guide clinical thinking, practice, and research in th e field of cerebral palsy Seminars in Pediatric Neurology, 11, 5 10. Rouse, H., & Fantuzzo, J. (2009). Multiple risks and educational well being: A population based investigation of threats to early school success. Early Childhood Research Quarterly, 24, 1 14. Rust, K. (1985). Variance estimation for complex estimation in sample surveys. Journal of Official Statistics, 1, 381 397.

PAGE 360

360 Rutter, M (1979) Protective factors in children's responses to stress and disadvantage In M W Kent & J E Rolf (Eds.), Pr imary prevention of psychopathology Vol 3 Social competence in children (pp 49 74) Hanover, NH: University Press of New England. Saigal, S., Stoskopf, B. L., Streiner, D. L., & Burrows, E. (2001). Physical growth and current health status of infants who were of extremely low birth weight and controls at ad olescence. Pediatrics, 108, 407 415 Sameroff, A. (2009) The transactional model In A Sameroff (Ed.), The transactional model of development: How children and contexts shape each other (pp 3 22) Wa shington, DC: American Psychological Association. Sameroff, A ., & Seifer, R (1983) Familial risk and child competence Child Development, 54 1354 1268. Sameroff, A., Seifer, R., Baldwin, A., & Baldwin, C (1993) Stability of intelligence from preschool to adolescence: The influence of social and family risk factors Child Development, 64, 80 97. Sameroff, A., Seifer, R., & Zax, M (1982) Early development of children at risk for emotional disorders Monographs for the Society for Research in Child Deve lopment, 47 1 82. Sanford, C., Levine, P., & Blackorby, J. (2008). A national profile of students with autism. Menlo Park, CA: SRI. Retrieved from SEELS website: http://www.seels.net/infoproduct.htm Sanson, A., Letcher, P., Smart, D., Prior, M., Toumbourou, J., & Oberklaid, F. (2009). Associations between early childhood temperament clusters and later psychological adjustment. Merrill Palmer Quarterly, 55, 26 54. Version 9.2 [Computer Software] Cary, NC: SAS Institute. Schmidt, M E., Demulder, E K., & Denham, S A (2002) Kindergarten social emotional competence: Developmental predictors and psychosocial implication Early Child Development and Care, 172 451 462. Schneider, N., & Goldstein, H. ( 2008 ). Social competence interventions for young children with communication and language disorders. In W. Brown, S. Odom, & S. McConnell (Eds.), Social competence of young children: Risk disability and intervention (pp 205 232 ) Baltimore, MD: Brooke s. Shonkoff, J., & Phillips, D (2000) From neurons to neighborhoods Washington, DC : National Academy Press.

PAGE 361

361 Sigman, M., & Ruskin, E. (1999) Continuity and change in the social competence of children with autism, down syndrome, and developmental delay Monographs for the Society for Research in Child Development Malden, MA: Blackwell Publishers. Simeonsson, R J (1991) Primary, secondary, and tertiary prevention in early intervention Journal of Early Intervention, 15 124 134. Simeonsson, R J (2003) Classification of communication disabilities in children: Contribution of the international classification of functioning, disability, and health International Journal of Audiology, 42 2 8. Simeonsson, R J (2009) ICF CY: A universal tool for d ocumentation of disability Journal of Policy and Practice in Intellectual Disability, 6, 70 72. Simeonsson, R J., & Bailey, D B (1991) The ABILITIES index Retrieved from http://www.fpg.unc.edu/~publicationsoffice/fpgpdfs/AbilitiesIndex.pdf Simeonsson, R J., Bailey, D B., Smith, T. M. & Buysse, V (1995) Young children with disabilities: Functional assessment by teachers Journal of Development and Physical Disabilities, 7, 267 284. Simeonsson, R J., Leonardi, M., Bjorck Akesson, E., Hollenweger, J., Lollar, D., Martinuzzi, A & TenNapel, H (2006) ICF CY: A universal tool for practice, policy, and research Paper presented at Meeting of WHO Collaborating C enters for the Family of International Classifications, Tunis, Tunisia Retrieved from http://apps.who.int/classifications/apps/icd/meetings/2006meeting/documentlist.ht ml Simeonsson, R J., Leonardi, M., Lollar, D., Bjorck Akesson, E., Hollenweger, J., & M artinuzzi, A (2003) Applying the international classification of functioning, disability, and health (ICF) to measure childhood disability Disability and Rehabilitation, 25, 602 610. Simeonsson, R J., Lollar, D., Hollowell, J., & Adams, M (2000) Revi sion of the international classification of impairments, disabilities, and handicaps: Developmental issues Journal of Clinical Epidemiology, 53, 113 124. Simeonsson, R J., & Scarborough, A (2001) Issues in clinical assessment In R Simeonsson, & S Rosenthal (Eds.), Psychological and developmental assessment: Children with disabilities and chronic conditions (pp 17 31) New York, NY: Guilford Press. Simeonsson, R J., Scarborough, A A., & Hebbler, K M (2006) ICF and ICF codes provide a standard language of disability in young children Journal of Clinical Epidemiology, 59, 365 373.

PAGE 362

362 Simeonsson, R J., Simeonsson, N E., & Hollenweger, J (2008) International classification of functioning, disability, and health for children and youth: A common la nguage for special education In L Florian & M McL aughlin (Eds.), Disability classification in education: Issues and perspectives (pp 207 226) Thousand Oaks, CA: Corwin Press Smith, B J., & Schakel, J A (1986) Noncategorical identification of pre school handicapped children: Policy issues and options Journal of the Division for Early Childhood, 11, 78 86. Snell, M. E., & Brown, F. (Eds.). (2006). Instruction of students with severe disabilities. Upper Saddle River, New Jersey: Pearson. Snell, M. E Chen, L., & Hoover, K. (2006). Teaching augmentative and alternative communication to students with severe disabilities: A review of intervention research 1997 2003. Research & Practice for Persons with Severe Disabilities, 31, 203 214. Snyder, P (2006 July) A focus on function Presentation at the Center for Child Development Vanderbilt University, Nashville, Tennessee Snyder, P., Bailey, D B., & Auer, C (1994) Preschool eligibility determination for children with known or suspected learning disabilities under IDEA Journal of Early Intervention, 18 380 390. Snyder, P., & Kaiser, A. (2008, August). The role of correlational research evidence in an evidence based framework Paper presented to the Annual Conference of American Psychological Ass ociation Boston, MA. Snyder, P., McLaughlin, T., & Denney, M. (2011). Frameworks for guiding program focus and practices in early intervention. In J.M. Kauffman & D.P. Hallahan (Series Eds.) & M. Conroy (Section Ed.), Handbook of special education: Sectio n XII Early identification and intervention in exceptionality (pp.716 730) New York, NY: Routledge Spiker, D., Boyce, G., & Boyce, L. (2002). Parent child interactions when young children have disabilities. International Review of Research in Mental Reta rdation, 25, 35 70. St Clair, D., Heinzen, H., Jenkins, F., & Carlson, E. (2010). Defining risk for preschoolers with disabilities and predicting educational performance. Journal on Developmental Disabilities 15 10 24. Strain, P., Schwartz, I., & Bovey E. ( 2008 ). Social competence interventions for young children with autism. In W. Brown, S. Odom, & S. McConnell (Eds.), Social competence of young children: Risk disability and intervention (pp 253 272 ) Baltimore, MD: Brookes.

PAGE 363

363 Stephens, R., Petras, H., Fabian, A., & Walrath, C (2009) Patterns of functional impairment and their change among youth served in systems of care: An application of latent transition analysis Journal of Behavioral Health Services and Research, 37, 491 507. Stucki, G. (2005). I nternational classification of functioning, disability, and health: A promising framework and classification for rehabilitation medicine American Journal of Physical Medicine and Rehabilitation, 84, 733 740. Stucki, G., & Cieza, A. (2004). International c lassification of functioning, disability, and health core sets for rheumatoid arthritis: A way to specify functioning American Rheumatoid Disorders, 63, 40 45. Thompson, B., Diamond, K., McWilliam, R., Snyder, P., & Snyder, S. (2005). Evaluating the quali ty of evidence from correlational research for evidence based practice. Exceptional Children, 71 184 194. Vermu nt, J. K., & Magidson J, (2002). Latent class cluster analysis In J. Hagenaars & A. McCutcheon (Eds.), Applied latent class analysis (pp.89 106 ). Cambridge, England: Cambridge University Press. Vermu nt, J. K., & Magidson J, (2006). Latent class models Document retrieved from http://www.statisticalinnovations.com/articles/ Vohr, B., Wright, L., Dusick, A., Mele, L., Verter, J., Steichen, J., Sim M. (2000). Neurodevelopmental and functional outcomes of extremely low birth weight infants in the National Institute of Child Health and Human Development Neonatal Research Network, 1993 1994. Pediatrics, 105, 1216 1226. Wagner, M., Frie nd, M., Bursuck, W., Kutuash, K., Duchnowski, A., Sumi, C., & Epstein, M. (2006). Educating students with emotional disturbances: A national perspective on school programs and services. Journal of Emotional and Behavioral Disorders 14, 12 30. Ward, J. H. (1963). Hierarchical grouping to optimize an objective function. Journal of American Statistical Association, 58, 236 244. Webster Stratton, C., & Lindsay, D. W. (1999). Social competence and early onset conduct problems: Issues in assessment. Journal of Child Clinical Psychology, 28, 25 93. Wei, X., & Marder, C. (2011). Self concept development of students with disabilities: Disability category, gender, and racial difference from early elementary to high school Remedial and Special Education, 32, 1 11. Werner, E E., & Smith, R S (2001) Journeys from childhood to midlife: Risk, resilience, and recovery Ithaca, NY: Cornell University Press

PAGE 364

364 Westling, D. L., & Fox, L. (2004). Teaching students with severe disabilities. Columbus OH: Pearson Merrill Prentice Hall. Wetherby, A., & Prizant, B. (2000). Autism spectrum disorders: A transactional developmental perspective Baltimore, MD: Brookes. Woods, T. A., Smith, S., & Cooper, J. L. (2010). Promoting the social emotional wellbeing of infants and toddl ers in early intervention programs (Report) Retrieved from the National Center for Children in Poverty website : http://www.nccp.org/publications/mentalhealth_pubs.html World Health Organization ( 1992 ) International classification disease (10th ed.). Gen eva, Switzerland: WHO Press. World Health Organization (2007) International classification of functioning, disability and health: Children and Youth Version Geneva, Switzerland: WHO Press. World Health Organization (2010) International classification of functioning, disability and health (ICF) Retrieved from the United Nations World Health Organization Website: http://www.who.int/classifications/icf/en/ Wu, X., Hart, C., Draper, T., & Olsen, J. (2001). Peer and teacher sociomentrics for preschool children: Cross informant concordance, temporal stability, and reliability. Merril l Palmer Quarterly, 47 416 443.

PAGE 365

365 BIOGRAPHICAL SKETCH Tara McLaughlin earned her doctorate at the University o f Florida During her doctoral program she received a grant from the American Educational Research Association funded by the National Science Foundation to conduct secondary analyses al abilities, disability classification, and contextual factors social competence other research interests include (a) instructional and behavioral supports for childre n with disabilities in inclusive settings, (b) professional deve lopment for teachers supporting children with disabilities in inclusive settings, and (c) cross sector research and policy focused on supports and services for children with disabilities and their families In 2010, Tara received t he J David Sexton Docto ral Student Award from the Division for Early Childhood (DEC) The award is given to a DEC member and doctoral level student who has made contributions to young children with special needs and their families through their efforts in research, higher educa tion, publications, and policy. Prior to enrolling in the doctoral program in 2006, Tara was an early primary special education teacher (new entrant through year 3) in New Zealand She completed her m degree in s pecial e ducation at the University of Florida in 2003 and her b degree of a rts in p sychology and e ducation at Hobart and William Smith Colleges in 2000. Following the completion of her degree, Tara has worked as a researcher for the Center for Excellence in Early Childho od Studies at the University of Florida The Center is a campus wide interdisciplinary center focused on the science of early learning and development The Center includes two model training, demonstration, and research sites located at Baby Gator Child Development and Research Center.