Citation
Comparison of descriptive functional assessment instruments to experimental functional analyses for children with autism

Material Information

Title:
Comparison of descriptive functional assessment instruments to experimental functional analyses for children with autism
Creator:
Chait, Andrea
Publication Date:
Language:
English
Physical Description:
xi, 159 leaves : ill. ; 29 cm.

Subjects

Subjects / Keywords:
Autistic disorder ( jstor )
Classrooms ( jstor )
Disabilities ( jstor )
Educational environment ( jstor )
Functional analysis ( jstor )
Furniture tables ( jstor )
Schools ( jstor )
Special education ( jstor )
Syntactical antecedents ( jstor )
Tax noncompliance ( jstor )
Dissertations, Academic -- Educational Psychology -- UF ( lcsh )
Educational Psychology thesis, Ph. D ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 2002.
Bibliography:
Includes bibliographical references (leaves 152-158).
General Note:
Printout.
General Note:
Vita.
Statement of Responsibility:
by Andrea Chait.

Record Information

Source Institution:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
029186950 ( ALEPH )
50861098 ( OCLC )

Downloads

This item has the following downloads:


Full Text









COMPARISON OF DESCRIPTIVE FUNCTIONAL ASSESSMENT INSTRUMENTS TO EXPERIMENTAL FUNCTIONAL ANALYSES FOR CHILDREN WITH AUTISM















By

ANDREA CHAIT


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


2002













ACKNOWLEDGMENTS

First and foremost, I would like to thank my dissertation chair, Dr. Jennifer

Asmus, for her endless patience and unflagging support. I would also like to thank Drs. Maureen Conroy, Tina Smith-Bonahue, and Greg Valcante for their expertise and assistance on this research project. I thank the students and their families who donated their time to participate in this study. A special thank you goes to all of the members of the Autism Inclusion Project at the University of Florida who assisted in the data collection process. Also, I would like to thank my friends for believing in me and providing me with continuous encouragement throughout this process. Finally, I would like to thank my parents, Marvin and Rochelle Chait, for their love and support. This dissertation is dedicated to them.













TABLE OF CONTENTS

page

ACKNOWLEDGMENTS ..... ii LIST OF TABLES V LIST OF FIGURES.............................................. viii

ABSTRACT................................................... x

CHAPTER

1 INTRODUCTION ....................... ................................I

Characteristics of Children with Autism ................................I
Educational Services for Students with Autism 3 2 REVIEW OF LITERATURE 10

Functional Assessment ...........................................10
Review of Functional Assessment Studies ............................. 21
Conclusion --------------------- ----38

3 M ET H O D ....................................................................................................................... 40

Participants ......................................................................... ............................................ 40
Settings and Therapists .................................................................................................. 41
Materials .................................................... 42
Operational Definitions, Observation System, and Interobserver Agreement ...... 47 D esign and D ata A nalysis ............................................................................................. 57

4 RESULTS ................................................... 60

Comparison of Descriptive Assessments to Functional Analyses ........................62
Comparison of Functional Analyses across Environments ......................................... 75
Comparison of Perceived Function and Actual Consequence to the Functional
Analysis... 80.............. .................... ....................80
Sum m ary ............. .. ............................................................................................... 92







5 DISCUSSION............................................... 93

Interpretation of Results ...................... ................. .............. 94
Limitations and Extensions ........................................99
Summary ..................... .. .................. ........................ 103

APPENDIX

A SAMPLE FUNCTIONAL ASSESSMENT OBSERVATION FORM .........107 B SAMPLE FUNCTIONAL ASSESSMENT INTERVIEW FORM .......................... 109

C SAMPLE MOTIVATIONAL ASSESSMENT SCALE ..................... 119

D TARGET BEHAVIORS .........................................123

E SUMMARY STATEMENTS FROM THE FUNCTIONAL ASSESSMENT
INTERVIEW IN THE SCHOOL ENVIRONMENT .....................125

F SUMMARY STATEMENTS FROM THE FUNCTIONAL
ASSESSMENT INTERVIEW IN THE HOME ENVIRONMENT ...........129

G FUNCTIONAL ANALYSIS GRAPHS ..............................133

H FUNCTIONS OF BEHAVIOR IDENTIFIED BY INVESTIGATORS ........146 I GRAPHS OF SCHOOL INTERVENTION WITH JOSH 150 REFERENCES ..................................................152

BIOGRAPHICAL SKETCH 159













LIST OF TABLES

Table page

4-1 Summary of the Percent Agreement between the Hypothesized
Functions of Behaviors Generated by Instruments within the
School Environment and the Functions of Behaviors Determined by
the Functional A nalysis ............................................................................................ 64

4-2 Summary of the Hypothesized Function of Behaviors across
Instruments in the School Environment for Josh - ------...... 66

4-3 Summary of the Hypothesized Function of Behaviors across
Instruments in the School Environment for Brad - --------- 67

4-4 Summary of the Hypothesized Function of Behaviors across
Instruments in the School Environment for Drake - ---------68

4-5 Summary of the Percent Agreement between the Hypothesized Functions of
Behaviors Generated by Instruments within the Home Environment and the
Functions of Behaviors Determined by the Functional Analysis ....................... 70

4-6 Summary of the Hypothesized Function of Behaviors across
Instruments in the Home Environment for Josh - ----------72

4-7 Summary of the Hypothesized Function of Behaviors across
Instruments in the Home Environment for Brad - ----------73

4-8 Summary of the Hypothesized Function of Behaviors across
Instruments in the Home Environment for Drake - --------- 74

4-9 Summary of the Results of Functional Analyses across Environments
for Josh ---------------------- ---77

4-10 Summary of the Results of Functional Analyses across Environments
for Brad ---------- ---....... . .........78

4-11 Summary of the Results of Functional Analyses across Environments
for Drake ----------------- -----------79








4-12 Summary of the Percent Agreement between the FAO Categories
and the Functional Analysis within the School Environment for Josh ................ 82

4-13 Summary of the Percent Agreement between the FAO Categories
and the Functional Analysis within the School Environment for Brad ................ 83

4-14 Summary of the Percent Agreement between the FAO Categories
and the Functional Analysis within the School Environment for Drake .............. 83

4-15 Summary of the Hypothesized Function of Behaviors Generated
by FAO Categories and the Functional Analysis in the School Environment
for Josh ........ ..... . ...................................- .. 84

4-16 Summary of the Hypothesized Function of Behaviors Generated
by FAO Categories and the Functional Analysis in the School Environment
for Brad ---------------------- -- 85

4-17 Summary of the Hypothesized Function of Behaviors Generated
by FAO Categories and the Functional Analysis in the School Environment
for D rake ................................................ ..............................--------------.................... 85

4-18 Summary of the Percent Agreement between the FAO Categories and the
Functional Analysis within the Home Environment for Josh ............................... 87

4-19 Summary of the Percent Agreement between the FAO Categories and the
Functional Analysis within the Home Environment for Brad .................-......... - 87

4-20 Summary of the Percent Agreement between the FAO Categories and the
Functional Analysis within the Home Environment for Drake ............................. 88

4-21 Summary of the Hypothesized Function of Behaviors Generated by
FAO Categories and the Functional Analysis in the Home Environment
for Josh ---------------------- ---90

4-22 Summary of the Hypothesized Function of Behaviors Generated by
FAO Categories and the Functional Analysis in the Home Environment
for Brad 90

4-23 Summary of the Hypothesized Function of Behaviors Generated by
FAO Categories and the Functional Analysis in the Home Environment
for Drake --------------------- ---91

E-1 Summary Statements from the Functional Assessment Interview in the
School Environment for Josh ----- ------------126








E-2 Summary Statements from the Functional Assessment Interview in the
School Environment for Brad - --...... .............. 127

E-3 Summary Statements from the Functional Assessment Interview in the
School Environment for Drake .....-.-...... . . . . . . . ..... 128

F-1 Summary Statements from the Functional Assessment Interview in the
Home Environment for Josh ------ -----------130

F-2 Summary Statements from the Functional Assessment Interview in the
Home Environment for Brad ................. .1... 31

F-3 Summary Statements from the Functional Assessment Interview in the
Home Environment for Drake .-- ------------. 132

H-1 Reliability Data on Functional Analysis Graphs for Josh .................................... 147

H-2 Reliability Data on Functional Analysis Graphs for Brad .................................... 148

H-3 Reliability Data on Functional Analysis Graphs for Drake ....... ...................... 149













LIST OF FIGURES


Table pag

G-1 Josh vocalization behavior across conditions at school ................................. 134

G-2 Josh disruptive behavior across conditions at school ...................................-------134

G-3 Josh noncompliant behavior across conditions at school ..................................... 135

G-4 Josh disruptive and noncompliant behavior across conditions at school ............. 135

G-5 Josh vocalization behavior across conditions at home 136 G-6 Josh noncompliant behavior across conditions at home ...................................... 136

G-7 Josh disruptive behavior across conditions at home ............................................. 137

G-8 Josh disruptive and noncompliant behavior across conditions at home ......-....... 137

G-9 Brad disruptive behavior across conditions at school ............................................ 138

G-10 Brad noncompliant behavior across conditions at school ................................... 138

G-11 Brad aggressive behavior across conditions at school ......................................... 139

G-12 Brad disruptive, noncompliant, and aggressive behavior across conditions at
school ----------------------- --139

G-13 Brad disruptive behavior across conditions at home .......................................... 140

G-14 Brad noncompliant behavior across conditions at home ...................................... 140

G-15 Brad aggressive behavior across conditions at home .......................................... 141

G-16 Brad disruptive, noncompliant, and aggressive behavior across conditions at
home ------------------------- 141

G-17 Drake noncompliant behavior across conditions at school .................................. 142

G-18 Drake tantrum behavior across conditions at school 142








G-19 Drake vocalization behavior across conditions at school ............. 143

G-20 Drake disruptive behavior across conditions at school ....................................... 143

G-21 Drake noncompliant behavior across conditions at home .................................. 144

G-22 Drake tantrum behavior across conditions at home ....... ......144

G-23 Drake vocalization behavior across conditions at home - -------145 G-24 Drake disruptive behavior across conditions at home .................... ... .... 145

I-1 Josh noncompliant and disruptive behaviors at school during
treatm ent phases. ..................................................................................................... 151

I-2 Josh task accuracy at school during treatment phases ...................................... 151













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

COMPARISON OF DESCRIPTIVE FUNCTIONAL ASSESSMENT INSTRUMENTS TO EXPERIMENTAL FUNCTIONAL ANALYSES FOR CHILDREN WITH AUTISM By

Andrea Chait

August 2002

Chair: Dr. Jennifer Asmus
Major Department: Educational Psychology

Individuals with autism often have problematic behaviors that hinder their success in the classroom. Successfully including students with autism who display problematic behavior in the general education setting requires appropriate interventions and support services. The Individuals with Disabilities Education Act Amendments of 1997 require the use of a functional assessment for students with disabilities who display significant problematic behavior. However, no consensus exists on the most appropriate functional assessment instrument to use in schools today. In addition, few studies have compared the results of different functional assessment instruments to functional analyses to determine their consistency. This study was conducted to determine the consistency of three commonly used descriptive functional assessment instruments: the Functional Assessment Observation form, the Functional Assessment Interview, and the Motivation Assessment Scale. Results from the three instruments were compared with results of an








experimental functional analysis within settings. Functional analysis results were compared across settings to determine whether functions of behaviors differed across settings. In addition, the Functional Assessment Observation form was further explored to determine whether the "perceived functions" or the "actual consequences" category provided more useful information. Comparisons were conducted separately at school and at home to determine whether one or more of these descriptive assessment procedures can accurately identify the function of behavior within settings. Results suggest that these instruments have a low level of consistency when compared to the functional analysis. The Functional Assessment Observation form demonstrated a moderate level of consistency and is a better instrument than either the Functional Assessment Interview or the Motivation Assessment Scale for identifying the function of behavior. The perceived function category on the Functional Assessment Observation form more accurately identified the function of behavior than did the actual consequences category. In addition, functions of behavior identified by the functional analyses could not be generalized across settings approximately half of the time. Functions across settings were usually the same for one function, but not for all of the functions identified. The implications for research and practice along with the limitations of the study are discussed.













CHAPTER 1
INTRODUCTION

Characteristics of Children with Autism

Autism is a developmental disorder that is behaviorally defined. According to the most recent Diagnostic and Statistical Manual of Mental Disorders (DSM-IV; American Psychiatric Association, 1994), six broad categories of impairment must be met for the diagnosis of autism. First, the individual must display significant impairment in social interactions such as an inability to develop relationships with peers and an impairment in the ability to use nonverbal communication to assist in social interactions. Second, the individual must have impairments in communication such as a delay or lack of development of verbal language. Third, evidence of a markedly restricted, repetitive, and stereotyped pattern of behaviors, interests, and activities should be observed such as body movem.nts or hand flapping. Fourth, a total of at least six of the twelve characteristics in the above categories must be met. Fifth, the onset of symptoms must exist before the age of three. Sixth, the symptoms must not be better accounted for by Rett's Disorder, a condition occurring only in females that causes mental retardation, or Childhood Disintegrative Disorder, a condition in which there is a persistent and progressive loss of skills (American Psychiatric Association, 1994).

Individuals with autism often have problematic behaviors that hinder their success in the classroom. There is a tendency for individuals with autism to engage in stereot) ped, repetitive patterns of behavior (Bailey, Phillips. & Rutter, 1996). They often








display abnormal preoccupations, adherence to nonfunctional routines or rituals, restricted interest patterns, stereotyped body movements, abnormal attachments to objects, and unusual responses to sensory stimuli (Bailey et al., 1996). Matson, Benavidez, Compton, Paclawskyj, and Baglio (1996) noted that the most commonly addressed problematic behaviors in individuals with autism include stereotypy, aggression, and self-injurious behaviors. Nevertheless, the problematic behaviors of children with autism vary widely from harmless deviations to severe destructive behaviors (Schopler & Mesibov, 1994). Schopler and Mesibov suggested that a variety of behavioral problems emerge from frustrations due to difficulties understanding, communicating, and relating with others.

Bailey et al. (1996) pointed out that individuals with autism often have an

impaired ability to develop relationships through social interactions. Many individuals with autism lack social reciprocity and are deficient in many social skills. For example, they may act indifferent to people and respond in the same manner to strangers as they would to family members (Volmer, 1995). As a result, social relationships within the classroom are often impaired. Their speech and language skills may be severely delayed or absent. Even when language has developed, individuals with autism often have difficulty communicating (Bailey et al., 1996). They are often unable to initiate or sustain a conversation with others. They may exhibit repetitive, stereotyped, or idiosyncratic language (American Psychiatric Association, 1994). Furthermore, many lack the ability to spontaneously engage in creative or pretend play (Bailey et al., 1996). Problematic behaviors in the classroom impose a tremendous challenge on the teacher, consume instructional time, and decrease the academic time for all students (Scott, DeSimone, Fowler, & Webb, 2000).








Educational Services for Students with Autism Prevalence

The American Psychiatric Association (1994) estimates that there are two to five cases of autism per 10,000 individuals. Other studies, using various definitions of autism, have found prevalence rates ranging from 33 to 160 per 10,000 individuals (Wing, 1997). Other prevalence estimates have indicated that autism may occur in as many as 1 per 1,000 individuals (Gresham & MacMillan, 1997). Many professionals believe that the incidence of autistic disorders is rising (Wing, 1997). However, this impression may be because of a greater awareness of the disorder or changes in the diagnostic criteria rather than an actual increase in the incidence of autism (Wing, 1997). Further studies are needed to evaluate this phenomenon. In either case, it is clear that the number of students identified as having autism has increased nationally. According to the U.S. Department of Education (1998), in 1992 there were 15,580 students classified with autism who were served under the Individuals with Disabilities Education Act (IDEA) and in 1997 there were 34,101.

Several factors are believed to be responsible for the increase in the number of individuals identified with autism in the schools. First, advocacy groups lobbied for improvements in the identification of and services provided to students with autism. This resulted in the development of better assessment tools and support services for individuals with autism and their families. Secondly, in the last 20 years, many laws have been passed requiring public schools to provide educational services for students with severe disabilities (Heward, 1996). The Education for All Handicapped Children Act of 1975 was implemented and since has been revised and renamed the Individuals








with Disabilities Education Act (IDEA) (Boomer & Garrison-Harrell, 1995). Initially, IDEA enabled students with autism to receive services under the category "Other Health Impaired." However, in 1990, Congress reauthorized IDEA and included autism as a separate category of individuals eligible for services. The unique needs of students with autism were recognized and warranted a separate categorization. Finally, there has been an increase in the number of students identified with autism or autistic-like characteristics. This may be because of the broadening of diagnostic categories such as Pervasive Developmental Disorder (PDD) to include Pervasive Developmental Disorder Not Otherwise Specified (PDD-NOS) as a label to describe individuals with symptoms of autism who do not meet the criteria for a diagnosis of autism or another type of disability (Simpson & Myles, 1998). Terminology has also changed. The five diagnostic categories that fall under the term PDD in the DSM-IV (Autistic Disorder, Asperger's Disorder, PDD-NOS, Child Disintegrative Disorder, and Rett's Syndrome) are now often referred to as Autism Spectrum Disorders (ASD). These disorders share clinical features but are distinct and separate from one another. In many instances, students classified with an Autism Spectrum Disorder are able to receive services under a variety of IDEA categories including autism. In the remainder of this paper, the term autism is used to include those individuals identified with ASD. Least Restrictive Environment

IDEA (1997) mandates that special education and related services are provided to children with disabilities. There are ten diagnostic categories that allow for children to be eligible for services: mental retardation, hearing impairments, speech or language impairments, visual impairments, emotional disturbance, orthopedic impairments, autism,








traumatic brain injury, specific learning disabilities, and other health impairments. In accordance with the federal law, the trend has been to serve students with disabilities in the least restrictive environment. The least restrictive environment refers to educating these students, to the maximum extent possible, with students who are not disabled (IDEA, 1997). This includes providing special services or instructional aids as necessary to educate students in the general education environment (IDEA, 1997). A student should only be removed from the general education environment when the student cannot benefit educationally from the services provided by the general education classroom (IDEA, 1997).

Most students identified with autism are educated in separate classrooms or separate schools (U.S. Department of Education, 1998). The behavioral problems displayed are often so severe that they interfere with the student's ability to learn in the general education environment (Eaves & Ho, 1997; Trevarthen, Aitken, Papoudi, & Robarts, 1996). Various professionals have different views on including students with autism in general education. Opponents of inclusion argue that there is a lack of scientific evidence documenting the benefits of serving students with autism in general education (Eaves & Ho, 1997; Simpson & Myles, 1998). However, proponents of inclusion reason that the segregation of students with disabilities into separate classrooms and schools lowers their self-esteem, hampers the development of their social skills, reduces their knowledge and skill attainment, and limits their instructional choices (Simpson & Myles, 1998). The reauthorization of IDEA has increased the emphasis on including students with autism in general education classrooms (Simpson & Sasso, 1992). Congress stated that it is beneficial to educate students with disabilities in the general education








environment (Community Alliance for Special Education (CASE) & Protection and Advocacy Inc. (PAl), 2000). The court system also has encouraged the integration of students with disabilities in the general classroom environment. Many federal court decisions regarding the placement of special education students have set the trend toward inclusion (CASE & PAl, 2000).

The movement toward inclusion was advocated without much empirical evidence to support the benefits of inclusion (Eaves & Ho, 1997; Simpson & Sasso, 1992). One aspect most researchers agree on is that children with autism have special needs and are capable of making significant gains in their abilities with appropriate interventions (Eaves & Ho, 1997; Simpson & Sasso, 1992). Successfully including students with autism in the general education environment requires appropriate interventions and support services (Simpson & Myles, 1998). Structured educational programs have been found to be effective in furthering the skill development of individuals with autism (Happe & Frith, 1996). In addition, behavior management programs have been found to be effective in modifying problem behaviors (Happe & Frith, 1996). Because of the nature of autism and the wide range of complex behaviors displayed by individuals with autism, techniques for more individualized interventions are necessary and have been discovered through applied behavioral analysis research (Mace, Lalli, & Lalli, 1991). Behavioral Interventions

Researchers have discovered that behavioral interventions are effective in

reducing the problematic behaviors of individuals with autism (Mace et al., 1991; Matson et al., 1996). Over the past 30 years, researchers in the field of applied behavior analysis have focused on the use of behavioral assessments and treatments to ameliorate








problematic behavior and to increase the academic and social skills of individuals with autism (Matson et al., 1996). Much of the applied behavioral research before the 1980s focused on the effectiveness of interventions without consideration of the variables responsible for behavior (Mace et al., 1991). Over the last 20 years, the behavioral literature has expanded to include methods for determining the variables promoting or maintaining behavior (Mace et al., 1991; Matson et al., 1996; Wehby, Symons, & Hollo, 1997).

Functional assessment is a technique that has been proven useful in identifying the variables influencing problem behavior (Wehby et al., 1997). Consequently, the IDEA amendments of 1997 require school districts to conduct a functional assessment when behavior is a problem (McConnell, Hilvitz, & Cox, 1998). As a result, school professionals frequently conduct functional assessments. The two main types of functional assessment methods used are descriptive assessments and experimental functional analyses (referred to throughout the remainder of the paper as functional analysis). Descriptive assessments include methods and instruments used to gather information regarding behavior such as behavioral observations, checklists, and interviews that are based on direct observations or information gathering in a natural context (no experimental manipulation). Information identified by these instruments is correlational in nature. Functional analyses involve the experimental manipulation of variables to validate hypotheses regarding the variables responsible for maintaining problem behavior. Information generated from a functional analysis leads to the identification of causal relationships that exist between environmental events and target behaviors.







The most commonly used functional assessment instruments in schools are

descriptive assessments (Heckaman, Conroy, Fox, & Chait, 2000). Surprisingly, little research has been conducted on most of these instruments to determine their reliability and validity. The results of functional analyses have been proven to be useful in developing effective treatment interventions (Mace et al., 1991; Matson et al., 1996; Wehby et al., 1997). For functional assessment information to be useful and lead to effective intervention, it must be reliable and accurate. Reliability refers to the consistency of the measurement. When the instrument is administered repeatedly to the sample participant under the same conditions, if the instrument is reliable, the results should be the same. Validity refers to the degree to which the instrument measures the theoretical construct or trait it is designed to measure. Cook and Campbell (1979) describe validity as the best approximation of the truth. There is a great need for studies that assess the reliability and validity of functional assessment instruments.

Chait (2001) compared the results of three functional assessment instruments (the Functional Assessment Observation form [FAO], the Functional Assessment Interview [FAI], and the Motivation Assessment Scale [MAS]) to each other to determine their reliability and validity. The results of that study indicated that the FAO and the FAI had a high level of interrater reliability (averaging 93% and 78% respectively). The results obtained by each instrument were not very stable across environments (i.e., home versus school). Agreement on the hypothesized functions of behaviors across environments was moderate on the FAO (averaging 62%) and low on the FAI (averaging 33%) and the MAS (averaging 22%). The relatively low to moderate level of stability across settings indicates that the hypotheses of the functions of behaviors generated in one environment







(school) were different from the hypotheses generated in another environment (home). There was a moderate to low level of agreement on the hypothesized functions of behavior across instruments when examining them within and across settings, averaging 57% within, and 44% across settings, indicating a moderate level of construct validity. A major limitation of this study was that the descriptive instruments were compared to each other. A comparison of the descriptive assessments to a functional analysis was not conducted; therefore, there was no way to know which instrument provided the most valid results.

A functional analysis allows for the identification of causal relationships between a behavior and its function (Arndorfer, Miltenberger, Woster, Rortvedt, & Gaffaney, 1994; Lerman & Iwata, 1993; Shriver, Anderson, & Procter, 2001). By comparing the hypothesized functions of behavior generated by descriptive assessments to the functions of behavior generated by functional analyses, the validity of the instruments can be assessed. There is a need for studies comparing these instruments to a functional analysis (Chait, 2001). The present study addresses this need by comparing the results of these three commonly used instruments to a functional analysis to determine the consistency of the functional assessment instruments. The following questions were addressed:

* Within settings, how consistent are each of these instruments in identifying a
hypothesized function of the target behavior in comparison to the function
identified in the functional analysis? Are the functional analysis findings the same
across settings?

* Within settings, which category of the Functional Assessment Observation form
(perceived functions vs. actual consequences) was most consistent with the results
of the functional analysis and thereby provided the most useful information?

* Based on current findings, is the use of a descriptive assessment appropriate to
identify the function of behavior in the general education and home environment?
Which instrument or category of an instrument provides the most consistent
findings in comparison to the functional analysis?













CHAPTER 2
REVIEW OF LITERATURE

Functional Assessment

Functional assessment methodology is used to determine the function of behavior (Carlson, Hagiwara, & Quinn, 1998; Fox, Dunlap, & Philbrick, 1997; Wehby et al., 1997). The term functional assessment refers to the procedures used to identify potential antecedents and consequences influencing behavior (Homer, 2000). Functional assessments attempt to identify relationships between problematic behaviors and environmental variables. Functional assessment techniques have been found to be powerful tools in determining appropriate interventions for individuals with autism (Matson et al., 1996). Functional based interventions directly address the purpose of the problem. The underlying assumption is that behavior is communicative and serves a function (Fox, Conroy, & Heckaman, 1998). Individuals engage in problematic behaviors because they enable the individual to gain attention or social contact to receive something that they need or want or to escape a task. Interventions are more effective if they relate to the function the problem behavior serves (Gable, 1996). With the knowledge of the function of behavior and the contingencies maintaining it, appropriate interventions can be developed that replace the problem behavior with an alternatively socially useful behavior, which serves the same function (Carr, Langdon, & Yarbrough, 1999; Fox et al., 1997).

Four functions of behavior commonly studied in the literature are identified

through functional analysis: escape, attention, tangible, and sensory reinforcement (Carr 10







et al., 1999). When an individual exhibits increased problematic behavior in response the presentation of demands, the function of that behavior is to escape. If an individual exhibits increased problematic behavior in response to the removal of attention, the function of that behavior is to gain attention. If an individual exhibits increased problematic behavior in response to restricted access to a preferred item or activity, the function of that behavior is to access tangibles. If an individual exhibits increased problematic behavior in the absence of social reinforcement (e.g., automatic reinforcement), the function of that behavior is automatic or sensory reinforcement. Functional analysis involves experimental manipulation of behavioral consequences to determine the function a behavior serves. For example, to determine if the function of behavior serves to escape a task (negative reinforcement) then, during a functional analysis, researchers would present a demand for a task and then remove the task contingent on the occurrence of problem behavior (Carr et al., 1999).

Linking the function of problematic behavior to the intervention involves disrupting the maintaining contingency while increasing an alternative appropriate response (Mace & Roberts, 1993). According to Mace and Roberts, the way to achieve this is through schedules of reinforcement. Reinforcement is the contingency occurring after a behavior that increases the probability of the behavior occurring again in the future. Individuals have several response options or choices as to how they will respond in situations. The choice they make can be greatly influenced by changing the rate, quality, and timing of reinforcement. Interventions involve manipulating the schedules of reinforcement to encourage the individual to choose an alternative appropriate response to attain the same goal, while at the same time weakening the maintaining contingency of







the inappropriate behavior. For example, if attention was found to be the function of a problematic behavior, a possible intervention would include reducing the amount of attention received after engaging in the problematic behavior, while providing increased attention for time intervals with no occurrence of the problematic behavior. If escape was found to be the function of a problematic behavior, a possible intervention would include preventing escape after the problematic behavior, while enabling the individual to request a break from the task after a period of engagement. If accessing tangibles was found to be the function of a problematic behavior, a possible intervention would include preventing access to the tangible after the problematic behavior, while providing access to the tangible for engagement in appropriate behavior. If sensory reinforcement was found to be the function of a problematic behavior, a possible intervention would include reducing the response reinforcer relationship, while increasing engagement in alternative appropriate behaviors that provide sensory stimulation. Interventions may require multiple strategies and there are a variety of ways they can be approached. The key is to weaken the contingencies maintaining behavior while reinforcing alternative appropriate behaviors by first matching the treatment to the function (Mace & Roberts, 1993).

There are several advantages for using functional assessments. Blakeslee, Sugai, and Gruba (1994) noted that functional assessment decreases the need for the use of punishment procedures and promotes skill building, promotes treatment to be derived based on hypotheses regarding the function the behavior serves, increases the likelihood of generalization and maintenance of intervention effects, and contributes to our scientific knowledgebase regarding interventions.







Clearly, functional assessments have proven useful for assessment and treatment. However, there is still ambiguity about how best to conduct a functional assessment. Generally, functional assessments fall under two broad categories: descriptive assessments and functional analyses.

Descriptive Assessments

Descriptive assessments include informal interviews, structured interviews,

behavior ratings, and direct observations used to develop hypotheses about the variables related to the occurrence and maintaining function(s) of problematic behavior. Descriptive assessments fall under two categories: indirect and direct assessments. Indirect assessment involves the assessment of behavior at a different time or place from when the behavior actually occurred, such as rating scales or interviews (Shapiro & Kratochwill, 2000). Direct assessment includes assessments that occur (often in the natural environment) at the same time as the behavior, such as direct observation (Shapiro & Kratochwill, 2000). Each descriptive assessment method has its advantages and disadvantages as described below.

Indirect assessments. Behavioral interviews include asking parents, teachers, or other informants about the problematic behavior and the variables associated with it. Interviews may be structured or unstructured but often include questions regarding the antecedents and the consequences of the problem behavior. Antecedents are the events occurring before a behavior that may serve as a stimulus that occasions the behavior given the consequences after the behavior (Schreibman, 1994). Consequences are the events occurring after a behavior that may serve to strengthen, maintain, or weaken a behavior (Mulick & Meinhold, 1994). The Functional Assessment Interview (O'Neill,








Homer, Albin, Sprague, & Storey, 1997) is an example of a structured interview used to assess the function of behavior. It is a comprehensive interview that includes questions regarding the antecedents, consequences, and setting events surrounding problematic behavior. Setting events are events that may change the nature of the stimulus-response relationship (Carr et al., 1999).

Interviews may be used alone or in combination with other assessment

techniques. Interviews are useful for gathering initial information but have limitations. They have many of the same problems as other indirect data collection methods, such as observer bias, observer expectations, and false recollection of events (Lennox & Miltenberger, 1989; Repp & Homer, 1999). Therefore, it is recommended that the interview serve as a starting point in the functional assessment process (Lennox & Miltenberger, 1989).

Rating scales, checklists, and other questionnaires provide a structured set of

questions that help in the formulation of the hypothesis of the function of behavior. The Motivation Assessment Scale (Durand & Crimmins, 1988) is an example of a questionnaire used to assess the function of behavior. The MAS is a behavioral rating scale administered to parents and teachers that assesses four possible functions (escape, attention, tangibles, and sensory stimulation) of problematic behavior. Questionnaires such as the MAS provide specific information related to the behavior and are convenient to administer (Mace, 1994). However, similar to the behavioral interview, information may be biased by the informant's memory, expectations, or interpretation of events (Lennox & Miltenberger, 1989).

Direct assessments. Direct observation of behavior, another type of assessment, reduces the biases associated with indirect data collection methods. However, these







methods are difficult to standardize and provide only correlational data that must be interpreted carefully (Mace, 1994; Taylor, 1994). One type of direct observation method is called the antecedent-behavior-consequence (A-B-C) assessment (Bijou, Peterson, & Ault, 1968). In this type of assessment, the observer observes the antecedents and consequences directly surrounding the target behavior. This information enables the observer to develop hypotheses regarding the variables influencing behavior. When conducting this type of assessment, the observer must be trained to describe events accurately. Lennox and Miltenberger (1989) recommend a method of conducting an A-B-C assessment that is more rigorous. First, information is gained through the A-B-C method described above. Then, a direct observation system is developed where the antecedents, behaviors, and consequences are recorded in the natural setting. This type of a system will enable analysis of the frequencies of environmental events and their relationship to the target behavior. The Functional Assessment Observation (FAO) form developed by O'Neill et al. (1997) is another type of direct observation instrument. Similar to Lennox and Miltenberger's method, it includes recording the antecedents and consequences associated with behavioral events. In addition, the FAO is designed to assess the possible functions of behavior, co-occurring behaviors, the number of events of problem behavior, and the times when the problem behavior is more or less likely to occur.

Functional Analyses

Functional analyses involve the manipulation of variables to gather information and validate hypotheses regarding the function of behavior. Functional analyses are the most reliable and valid measures of behavioral function (Mace, 1994). This experimental








methodology enables causal statements to be made about the function of behavior, whereas descriptive assessments only allow for descriptive or correlational statements (Gresham, Watson, & Skinner, 2001). Experimental methods are standardized procedures that isolate variables and control contingencies to determine effects on behavior. Functional analyses involve the examination of relationships between a person and environmental variables within a natural or analogue setting (Gable, 1996). An analogue setting refers to a setting outside of the natural environment that is designed to represent the natural environment (Carr et al., 1999).

Iwata, Dorsey, Slifer, Bauman, and Richman (1982/1994) designed the first standardized method for conducting a functional analysis. Treatment failures for self-injurious behavior (SIB) were thought to be because of a lack of understanding of the variables responsible for maintaining or producing the behavior (Iwata et al., 1982/1994). Therefore, Iwata et al. (1982/1994) built on previous research to develop a systematic method for identifying the function of SIB. Nine participants with moderate to high rates of SIB were exposed to four, 15 minute, experimental conditions: social disapproval (attention), academic demand, unstructured play, and alone time.

In the attention condition, the participant was placed in a room with toys. The

participant was told to play with the toys while the experimenter engaged in some work. The experimenter only provided the participant with attention contingent on SIB. This condition was designed to determine if the participant engaged in self-injury to gain attention (Iwata et al., 1982/1994). In the academic demand condition, the participant was given a difficult academic task to complete. The experimenter used a three-prompt procedure to help the participant. Contingent on the occurrence of SIB, the experimenter







would stop the academic session and turn away from the participant for 30 seconds. This condition was designed to determine if the participant engaged in SIB to escape or avoid difficult tasks (Iwata et al., 1982/1994). In the alone condition, the participant was placed in a room alone without any toys or materials. This condition was designed to determine if the participant engaged in SIB to access self-stimulation (Iwata et al., 1982/1994). In the unstructured play condition, the experimenter and participant were in a room with toys. The participant was allowed to play or move freely about the room. No demands were placed on the participant. When SIB did not occur, the experimenter provided the participant with positive attention and brief physical contact every 30 seconds. When SIB did occur, the participant was ignored unless the behavior became too severe, at which point the session was terminated. This condition served as the control condition. During this condition, the experimenter was present and provided attention, the participant had access to tangibles, there were no demands, and there were no consequences delivered for engaging in SIB (Iwata et al., 1982/1994).

The function of SIB was identified for seven of the nine participants. Self-injury was found to be at its highest level during the alone condition for four of the participants, suggesting that the function of their behavior was to access self-stimulation (automatic reinforcement) (Iwata et al., 1982/1994). Two participants exhibited high levels of SIB during the high-demand situation, suggesting that the function of their behavior was to escape the demand (negative reinforcement) (Iwata et al.). One participant engaged in high levels of SIB during the social disapproval condition, suggesting that the function of the behavior was to gain attention (positive reinforcement) (Iwata et al., 1982/1994).








Results from this study suggested that the function of SIB varies among and

within individuals (Iwata et al., 1982/1994). By conducting an experimental analysis, it is possible to identify variables associated with SIB before implementing a treatment and ultimately to develop more effective treatments (Iwata et al., 1982/1994). Since the development of the functional analysis methodology, it has been applied to a wide range of behavioral disorders (Mace, 1994).

Derby et al. (1992) provided a descriptive summary evaluating the use of a brief functional analysis technique on 79 individuals with developmental disabilities who displayed aberrant behavior. Individuals ranged from age 1 to 32 years with most diagnosed as profoundly mentally retarded. A change in the rate of aberrant behavior or appropriate behavior was evident in 77% of the cases. During the brief functional analyses, 63% of the individuals displayed the targeted aberrant behaviors. The function maintaining the behavior was identified 74% of the time. When contingencies were provided for appropriate behavior, there was a reduction in the aberrant target behavior in 54% of the cases. In 84% of the cases, the manipulation of contingencies after the aberrant target behavior resulted in behavioral control. When appropriate behavior was the focus of assessment, 65% of the cases increased their rates of appropriate behavior. Results of this descriptive summary showed that a brief functional analysis was effective in identifying the function of behavior for most clients with developmental delays who engaged in high-frequency behaviors within the clinic setting.

Iwata et al. (1994) conducted an epidemiological analysis of the results of functional analyses from 152 individuals with self-injurious behavior (SIB). The functional analyses were conducted when subjects were living in a residential facility or a








pediatric hospital. Subjects ranged from age 1 to 51 years. Each subject was repeatedly exposed to a series of three to eight 15-minute conditions intended to identify the function of SIB. Three experimental designs were used to assess the influence of antecedents and consequences on behavior: a multielement, reversal, or combined design. Results showed that social-negative reinforcement (escape) was the function of SIB in approximately 38% of the individuals. Social-positive reinforcement (attention or tangible) was the function of SIB in approximately 26% of the individuals. Automatic reinforcement (sensory stimulation or pain attenuation) was the function of SIB in approximately 26% of the individuals. In approximately 5% of the cases, multiple functions were identified. No function was clearly identified in approximately 5% of the cases.

Functional analysis methodology is not only useful in identifying the functions of behavior in individuals but may also allow the study of epidemiological data regarding the behavioral function of large groups of individuals (Iwata et al., 1994). For example, results of this study showed that social reinforcement was the function of SIB in approximately 67% of the sample. Therefore, SIB is primarily a learned disorder (Iwata et al., 1994). Individuals with SIB may not have the skills necessary to gain reinforcement through other means or their social environment may not respond to less severe forms of behavior (Iwata et al., 1994). Therefore, it is crucial that preventive strategies consist of providing individuals at risk for SIB with early language instruction (e.g., functional communication training) (Iwata et al., 1994). In addition, Iwata et al. emphasized the importance of replacing SIB with an alternative appropriate behavior serving the same function.








After the functional analysis, 121 subjects participated in a treatment program. The treatment program was designed to evaluate the effects of different interventions (Iwata et al., 1994). Overall, antecedent interventions such as extinction, differential reinforcement, and punishment, when designed for the function of the SIB, were effective in reducing SIB to below 10% of baseline in over 80% of the cases. When the function of SIB was to obtain social-positive reinforcement, noncontingent attention (or access to tangibles) was effective in reducing SIB. Attention extinction, differential reinforcement of alternative behavior (DRO), and time-out were also effective. However, providing verbal reprimands and response interruption was not effective. When the function of SIB was to obtain social-negative reinforcement, noncontingent reinforcement (e.g., removal of task demand) was effective in reducing SIB. Escape extinction, differential reinforcement, and reducing the frequency of task presentation were also effective. Attention extinction and time-out had no effect on the rates of SIB. The presentation of additional demands contingent on SIB and response interruption only modestly reduced SIB. When SIB was maintained by automatic reinforcement, noncontingent reinforcement effectively reduced the levels of SIB. Sensory extinction, differential reinforcement, and response interruption also reduced SIB. Attention extinction and timeout were not effective. When SIB was maintained by multiple functions or the function was not identified, noncontingent reinforcement was effective in reducing SIB. Attention extinction, verbal reprimands, and time-out were not effective.

The treatment data supported the validity and utility of functional analysis as a means to identify behavioral function (Iwata et al., 1994). Interventions designed to eliminate the maintaining contingency were more effective in reducing SIB than other







types of interventions (Iwata et al., 1994). This study lends further support to the research already available that has shown interventions related to the function of behavior are more effective than interventions selected at random (Iwata et al., 1994).

The main limitation of functional analyses is that, because of the nature of these types of assessments, they are often conducted in analogue settings. As a result, sometimes the findings may not be generalizable to the natural setting (Gresham et al., 2001; Repp & Homer, 1999; Gable, 1996; Mace, 1994). In addition, functional analyses often require a large amount of time to conduct and may not be practical for implementation in classrooms (Conroy, Fox, Crain, Jenkins, & Belcher, 1996; Gresham, et al., 2001; Mace, 1994).

Review of Functional Assessment Studies

Generally, the student's individual educational placement team is responsible for conducting the functional assessment and implementing the behavioral intervention plan (McConnell et al.,1998). There are many ways of conducting a functional assessment. Many researchers recommend that a multi-method approach be used when conducting a functional assessment (Crawford, Brockel, Schauss, & Miltenberger, 1992; Davis, 1998; Gable, 1996; Homer, 1994; Lennox & Miltenberger, 1989; Symons, McDonald, & Wehby, 1998; Wehby et al., 1997). A multi-method approach includes gathering information from many different assessment sources. The resulting information should then be systematically analyzed and hypotheses developed. Interventions should be designed based on the results of the functional assessment.

Wehby et al. (1997) suggest a three stage model as best practice in assessing problematic behaviors. First, information is gathered through verbal reports and








interviews. This information is used to generate hypotheses regarding the function of the behavior. Second, direct observations in the natural setting are conducted to either support the hypotheses or provide new additional information. Third, a functional analysis is conducted. A functional analysis includes the manipulation of environmental variables to determine their effect on the problematic behavior. Functional analyses are rigorous and therefore tend to be difficult to implement in the classroom setting (Harding et al, 1999). As a result, some suggest it may be sufficient to focus on interviews and other descriptive data to design interventions rather than conduct a functional analysis (Arndorfer, et al., 1994; Mace, 1994; Wehby et al., 1997). However, others note that it may be difficult to identify functional relationships using descriptive assessments (Harding et al., 1999). Furthermore, descriptive approaches may be just as time consuming and complicated to administer as functional analyses (Iwata, 1994).

Functional assessments can be conducted in a variety of ways using a variety of instruments. At present, there are no standardized procedures. Many of the functional assessment instruments in use today have yet to be studied in regards to their validity and reliability. Shriver et al., (2001) stated this is because of the differences between functional assessment and traditional psychological assessment. Traditional psychological assessment involves gathering information to determine if a particular pattern of behavior represents a hypothetical construct within the child. The construct (i.e., intelligence, attention, aggression) is hypothesized to be the cause of the relationship between the environment and behavior. In contrast, functional assessment is not concerned with evaluating whether or not a construct exists. Instead, functional assessment focuses on the behaviors that contribute to the construct. Functional








assessment seeks to explain the function of the behaviors that make up a construct. The outcomes of these two approaches differ, which inevitably effects their evaluation (Shriver et al., 2001).

When evaluating traditional psychological assessments, researchers look for evidence that the instrument consistently and accurately represents the construct of interest. In a functional assessment, hypotheses regarding the functional relationship are the outcome to be evaluated. These hypotheses are based on an interaction between the behavior of a child and the child's environment. In addition, these relationships can be directly observed, unlike psychological constructs. Establishing that a functional assessment is accurate lends evidence for its construct validity (Shriver et al., 2001). Construct validity, according to Messick (1995), includes all forms of reliability and validity evidence. Messick (1995) proposed that all forms of reliability and validity evidence contribute to the construct validity of an instrument. Therefore, the construct validation process may include the collection of many different types of reliability and validity evidence. The methods for evaluating the construct validity of traditional psychological assessments are applicable to the evaluation of functional assessment instruments (Shriver et al., 2001).

The need to determine the reliability and validity of functional assessments has been clearly documented (Conroy et al., 1996; Davis, 1998; Fox et al., 1998; Gable, 1996; Iwata, 1994; Shriver et al., 2001). In addition, few studies have compared the results of functional assessment instruments to the results of functional analyses. Functional analyses are considered the most reliable and valid measure of functional relationships (Arndorfer et al., 1994; Lerman & Iwata, 1993; Shriver et al., 2001). At







present, the best way to evaluate the accuracy of functional assessment instruments is to compare them to functional analyses (Shriver et al., 2001). Research on the Functional Assessment Observation Form, Functional Assessment
Interview, and the Motivational Assessment Scale

Functional Assessment Observation Form. Very little research overall has been conducted on the FAO. Arndorfer et al. (1994) found agreement among a functional analysis and different descriptive assessment measures including the FAO. However, their use of the FAO form was described as a way to collect A-B-C data. It was unclear as to whether they used the FAO form in the way it was intended. Cunningham and O'Neill (2000) conducted a study which compared a functional analysis to the FAO, the FAI, and the MAS. They reported overall agreement in the identified function of behavior across instruments and analyses. In addition, they found a high level of interobserver agreement on the FAO, the average ranging from 89% to 100%. Chait (2001) conducted a study that compared the FAO to the FAI and the MAS within and across environments. In the school and the home environment individually, there was a moderate level of agreement between the FAO and the findings of the FAI and the MAS (the FAO agreed with the findings of the FAI and the MAS 58% of the time with a range from 0% to 100%). There was moderate agreement (62%) on the hypothesized function of behavior generated by the FAO across environments. Across settings, the FAO at school agreed with the hypothesized function of behavior generated by the FAI and the MAS in the school and home environment 60% of the time with a range from 40% to 100%. The hypothesized function of behavior determined by the FAO in the home environment agreed with the hypothesized function of behavior generated by the FAI and the MAS in the school and home environment 53% of the time with a range from 0% to 100%. Interobserver agreement on the FAO was high (averaging 93% with a range from







83% to 99%). Further studies evaluating the reliability and validity of the FAO are recommended (Arndorfer et al.; Chait, 2001; Cunningham & O'Neill, 2000; Fox et al., 1998).

Functional Assessment Interview. The FAI generates a wealth of information regarding the variables influencing behavior. However, very little research has been conducted to determine the reliability and validity of this instrument (Arndorfer & Miltenberger, 1993; Fox et al., 1998; Sturmey, 1994). Arndorfer et al. (1994) conducted a study that included the comparison of the hypothesized functions generated from the FAI to a functional analysis. They found agreement on the function of behavior identified by the FAI and the functional analysis. In addition, inter-rater agreement on the hypothesized functions of behaviors generated by the FAI was 100%. Cunningham and O'Neill (2000) compared the hypothesized function of behavior generated from the FAI, MAS, and the FAO to a functional analysis. They found that the function of behavior identified by the FAI agreed with the results of the functional analysis. In addition, the FAI agreed with the function identified by the MAS and the FAO for two out of the three participants. They also found 100% inter-rater agreement on the FAI. Galensky, Miltenberger, Stricker, and Garlinghouse (2001) conducted a study that included a comparison of the hypothesized functions generated from the Functional Assessment Questionnaire of Mealtime Behaviors (FAQ), to the FAI, and an A-B-C checklist. These instruments were compared for three children with food refusal behaviors aged 2 to 6 years. They found agreement on one function (escape) for all children across instruments. The FAI inter-rater agreement was 100% for the escape function, 67% for the attention function, and 0% for the tangible function for a combined agreement of 43%. The A-B-C direct observation descriptive assessment yielded consistent findings of a second







function, attention, that was not identified by the other two indirect descriptive assessment instruments. Chait (2001) conducted a study that compared the FAI to the FAO and the MAS within and across environments. In the school environment, there was a low level of agreement between the FAI and the findings of the FAO and the MAS (the FAI agreed with the findings of the FAO and the MAS 44% of the time with a range from 0% to 100%). In the home environment, there was a low level of agreement between the FAI and the findings of the FAO and the MAS (the FAI agreed with the findings of the FAO and the MAS 35% of the time with a range from 0% to 50%). There was a low level of agreement (33%) on the hypothesized function of behavior generated by the FAI across environments. Across settings, the FAI at school agreed with the hypothesized function of behavior generated by the FAO and the MAS in the school and home environment 40% of the time with a range from 0% to 100%. The hypothesized function of behavior determined by the FAI in the home environment agreed with the hypothesized function of behavior generated by the FAO and the MAS in the school and home environment 42% of the time with a range from 20% to 80%. Interobserver agreement on the FAI was moderately high (averaging 78% with a range from 43% to 100%). The above studies on the FAI have shown moderately high to high levels of interrater agreement and some contradicting evidence regarding the construct validity of the instrument. Therefore, researchers strongly recommend that further studies be conducted to assess the reliability and validity of the FAI (Arndorfer & Miltenberger, 1993; Chait, 2001; Fox et al., 1998; Galensky et al., 2001; Sturmey, 1994).

Motivational Assessment Scale. Durand and Crimmins (1988) created the MAS to provide an alternative assessment procedure to functional analyses. The MAS was created after extensive interviews with teachers, parents, and clinicians of students with








developmental disabilities. To investigate the inter-rater and test-retest reliability of the instrument, Durand and Crimmins selected fifty students with developmental disabilities who engaged in self-injurious behaviors. To assess inter-rater reliability, the MAS was administered to the child's classroom teacher (primary rater) and the classroom paraprofessional (secondary rater). Results were compared from the primary and secondary raters through the calculation of a pearson correlation coefficient on the raw scores, mean scores, and ranked scores. The correlations for raw scores ranged from .66 to .92, mean scores ranged from .80 to .95, and ranked scores ranged from .66 to .81. To assess test-retest reliability, the MAS was administered to the classroom teacher a second time, 30 days after the initial administration. Pearson correlation coefficients were calculated and the correlation between raw scores ranged from .89 to .99, mean scores ranged from .92 to .98, and ranked scores ranged from .82 to .99. Results suggested the MAS was reliable among raters and stable over time (Durand & Crimmins, 1988). A second study was conducted to determine the validity of the MAS by comparing the primary rater's ranking to the results of a functional analysis. Eight participants were selected from the 50 subjects used in the reliability study. The subjects were exposed to five experimental conditions, three times, each lasting 10 minutes. The experimental conditions included a baseline, attention, escape, tangible, and unstructured condition. Correlations were calculated to compare the rankings on the MAS to the ranks generated from the functional analysis. Results suggest ratings on the MAS were highly predictive of the results from the functional analysis with a correlation of .99.

The MAS is the most extensively evaluated functional assessment instrument (Sturmey, 1994). Several researchers including the authors of the MAS conducted







reliability and validity studies and found the instrument to have high reliability and validity (e.g., Durand & Carr, 1991; Durand & Crimmins, 1988; Durand, Crimmins, Caulfield, & Taylor, 1989; Iwata et al., 1982/1994). Two studies focusing on the factor structure of the MAS scales found the scales to have high internal consistency (e.g., Bihm, Kienlen, Ness, & Poindexter, 1991; and Durand & Crimmins, 1988). However, five other studies found the MAS to have low inter-rater reliability, internal consistency, and validity (e.g., Crawford et al., 1992; Goza & Ricketts, 1993; Newton & Sturmey, 1991; Singh et al, 1993; Zarcone, Rodgers, Iwata, Rourke, & Sorsey, 1991). Several factors may be responsible for the lack of replication of the robustness of the MAS. The instrument was originally studied on a population of children with developmental delays who engaged in frequent self-injurious behaviors. The characteristics of the subjects, behaviors, or raters may have influenced the robustness of the instrument (Sturmey, 1994). However, the small number of items corresponding to each scale on the MAS, the limited number of consequences identified, or the lack of consideration of antecedents and setting events may have also contributed to the lack of replication (Sturmey, 1994).

Chait (2001) conducted a study that compared the MAS to the FAO and the FAI within and across environments. In the school environment, there was a low level of agreement between the MAS and the findings of the FAO and the FAI (the MAS agreed with the findings of the FAO and the FAI 29% of the time with a range from 0% to 100%). In the home environment, there was a moderate level of agreement between the MAS and the findings of the FAO and the FAI (the MAS agreed with the findings of the FAO and the FAI 50% of the time with a range from 0% to 100%). There was a low level of agreement (22%) on the hypothesized function of behavior generated by the MAS







across environments. Across settings, the MAS at school agreed with the hypothesized function of behavior generated by the FAO and the FAI in the school and home environment 34% of the time with a range from 0% to 100%. The hypothesized function of behavior determined by the MAS in the home environment agreed with the hypothesized function of behavior generated by the FAO and the FAI in the school and home environment 37% of the time with a range from 0% to 100%. Researchers recommend that this instrument be used with caution and urge future studies to continually assess the reliability and validity of this measure (Chait, 2001; Arndorfer & Miltenberger, 1993; Sturmey, 1994).

Arndorfer et al. (1994) conducted a study that compared the hypothesized

function of behavior generated by the MAS to a functional analysis. They found that the results of the MAS were not in agreement with the results of the functional analysis. Conversely, Cunningham & O'Neill (2000) found overall agreement on the function of behavior generated from the MAS and the functional analysis. In general, few studies have compared the results of specific functional assessment instruments to the results of functional analyses to determine their validity. Those studies that do exist have found inconsistent results supporting the need for additional research in this area. Comparisons Across Descriptive Assessments and Functional Analyses

Dunlap and his colleagues (1993) conducted a study on the use of functional

assessment and functional analysis procedures, in the special education classroom, with five elementary students with emotional and behavioral disorders. Functional assessments were conducted to gather data and develop hypotheses. Researchers gathered information through record reviews, interviews (based on the FAI) with teachers, students, and other







classroom staff, and direct observations. After hypotheses were developed, the researchers conducted functional analyses in the classroom setting to test the hypotheses. The functional analyses confirmed the hypotheses developed. Behavior varied consistently in the direction predicted for each condition implemented. Results demonstrated the effectiveness of using functional assessment procedures with students with emotional and behavioral disorders in the classroom setting (Dunlap et al., 1993). However, the study did not analyze the reliability and validity of the individual functional assessment instruments used. The accuracy of the functional assessment strategies implemented was verified through the functional analyses. Still, questions remain as to which functional assessment instrument provided the most reliable and valid information or if a combination of instruments was necessary.

Lalli, Browder, Mace, and Brown (1993) conducted two studies to assess the use of descriptive analysis in the classroom setting to decrease problematic behavior. In the first study, descriptive analyses were used to develop hypotheses regarding the function of students' problematic behavior. In the second study, the hypotheses were tested through a functional analysis and through analyzing the effects of treatment on the problematic behavior. Three students with severe and profound mental retardation, their teachers, and six instructional assistants participated in this study. The first study consisted of an assessment and the development of hypotheses regarding the function of problematic behaviors. Assessment consisted of a problem-identification interview, a scatter plot analysis (Touchette, Macdonald, & Langer, 1985), narrative recordings, and a descriptive analysis. The problem-identification interview assisted in gathering general information regarding the problematic behavior. The scatter plot analysis assisted in the








identification of times when the problematic behavior occurred most frequently. The narrative recordings assisted in identifying the topography of the problematic behavior and the environmental antecedents and consequences associated with the problematic behavior. The descriptive analysis consisted of five hours of direct observation and a continuous 10-second partial-interval recording procedure. The descriptive analysis assisted in the identification of relationships between the problematic behavior and the environment. Based on the results of the descriptive analyses, hypotheses regarding the function of participants' behaviors were developed. In the second study, interventions were designed to indirectly test the accuracy of the hypotheses. Teachers were trained to block the reinforcement of problematic behavior, reinforce appropriate alternative behaviors, and teach an adaptive behavior. To assess the effects of interventions based on the hypotheses, a multiple baseline across students was used to conduct a component analysis. Different components of the intervention were removed at different times to observe their effects on behavior. A reversal design was used to test the hypotheses. In the reversal, teachers provided consequences contingent on problematic behavior (functional analyses). All three students showed significantly lower levels of problematic behavior during the intervention phases and higher levels during the reversal phases. The effectiveness of the interventions and the results from the functional analyses supported the use of descriptive analyses to develop accurate hypotheses regarding the function of behavior (Lalli et al., 1993). This study provides evidence that descriptive analyses, when conducted in this fashion, can provide reliable and valid results. However, there is a need for continued research directly assessing the reliability and validity of this and other functional assessment procedures.








Arndorfer et al. (1994) conducted a study on the use of descriptive and functional analysis in the home environment. They found that with limited training parents were capable of taking part in the functional assessment process. Five children with behavioral problems and their mothers were included in this study. In the first phase of the study, the MAS, a behavioral interview (FAI), and a direct observation (FAO) of behavior were conducted. The data from phase one was used to generate hypotheses regarding the function of the behavior. In the second phase of their study, a functional analysis was conducted. The results from the descriptive assessments and the functional analysis were generally in agreement except for the MAS. The primary functions of behaviors identified by the MAS were generally not in agreement with the other assessment techniques. The general agreement found among instruments lead the authors to suggest that when descriptive assessments provide clear results a functional analysis may not be warranted (Arndorfer et al., 1994). Furthermore, parents' ability to partake in the assessment process encouraged the use of these methods in the natural environment for practicality and useful purposes (Arndorfer et al., 1994).

Sasso, et al. (1992) conducted a study to compare the results of a descriptive

assessment and a functional analysis on two children with autism who were educated in a self-contained classroom. Each child's primary teacher participated with the assessment. Teachers were trained to conduct the assessment procedures. Before the teachers conducted the assessments, the investigator conducted the same assessments. Teachers conducted an A-B-C assessment and a classroom functional analysis. The hypothesized functions of behavior generated by the A-B-C assessment were in agreement with the function of behavior identified by the functional analysis for both children. The







interventions developed based on the assessment results were effective in reducing targeted behaviors. Overall, these results suggest that descriptive assessments, such as the A-B-C assessment, may be used to accurately identify consequences maintaining behavior. Descriptive assessments may be a viable alternative when functional analyses are not possible (Sasso et al., 1992).

Freeman, Anderson, and Scotti (2000) compared a structured descriptive

assessment to an unstructured descriptive assessment and a functional analysis on two children with mental retardation. During the structured descriptive assessment, the environment was structured to increase the likelihood of environmental variables of interest occurring and thereby obtain a better sampling of behavior. The structured descriptive assessment occurred in the child's classroom and consisted of four different antecedent conditions (attention, demand, tangible, control). The structured descriptive assessment resulted in more frequent environmental events than the unstructured descriptive assessment. The results from the structured descriptive assessment were in agreement with the results of the functional analyses for both children. Results suggest that a structured descriptive assessment may be used to identify the maintaining consequences of behavior when a functional analysis is not feasible.

Cunningham and O'Neill (2000) conducted a study to compare the effectiveness and efficiency of the MAS, FAI, and FAO. Subjects included three male children with autism who were educated in a self-contained classroom. The child's primary teacher and a paraprofessional were selected for participation with each child. Before any assessments, the teachers and assistants were asked why they thought the child engaged in challenging behavior. Next, the FAI was independently administered to the teachers








and paraprofessionals. Afterwards, they were asked to generate hypotheses regarding the function of the child's behavior. If there were multiple hypotheses, they were asked to rank them according to primary and secondary functions. Later, responses documented by the interviewer were coded by two independent raters to determine if they were able to generate the same hypotheses regarding the function of behavior. Results were compared to determine inter-rater reliability of the instrument. Next, the MAS was administered to the teachers and paraprofessionals. The category scores on the MAS were averaged for each teacher/paraprofessional pair to determine the rank ordering of the functions of behavior. Teachers were then asked to collect direct observational data by using the FAO form. Rank orderings of the function of behavior were generated from this information. Finally, the authors conducted a functional analysis in analogue conditions. Interobserver agreement was collected on the descriptive observation sessions (averages ranged from 89% to 100%) and the analogue sessions (averages ranged from 98% to 100%) during 20% of the sessions. Results demonstrated overall agreement across methods in identifying behavioral functions.

Mace and Lalli (1991) conducted a descriptive and functional analysis on an adult male with mental retardation. The descriptive assessment suggested two possible functions of his behavior (attention and escape). The functional analysis identified only one function of his behavior (attention). Mace and Lalli (1991) suggested that the results of the descriptive assessment were useful in designing and interpreting the functional analysis. However, the descriptive assessment results by themselves may have lead to unnecessary treatment components and possibly even counterproductive treatment







components (Mace & Lalli, 1991). Treatments were designed based on the results from the functional analysis and found to be effective.

Umbreit (1995) conducted a study on the use of functional assessments in the regular education setting on an eight-year-old boy with Attention Deficit Hyperactivity Disorder. After conducting an analogue functional analysis, the functions of the child's disruptive behaviors were identified. Negative reinforcement was found to be the function of the child's disruptive behavior evidenced by the high rates of disruptive behavior occurring during the escape condition. After the functional analysis, the teachers, aides, and the student were given functional assessment interviews. In addition, A-B-C data was collected through direct observation. Based on the information obtained through the functional assessments, two hypotheses were developed and tested. Results confirmed the hypotheses and an intervention was developed. The intervention was found to be successful in reducing disruptive behavior and increasing appropriate behavior. Each of the functional assessment methods conducted assisted in providing information regarding the variables functionally related to the child's behavior (Umbreit, 1995). Interestingly, the functional analysis clearly identified escape as the function of the child's disruptive behavior. However, the other functional assessments conducted identified escape and social attention as functions of the child's disruptive behavior. Although this difference may have been due to the fact that the functional analysis was conducted in an analogue setting, it indicates the need to observe the child in the natural environment before designing interventions based on analyses done in an analogue environment (Umbreit, 1995). It was not clear what element each functional assessment method contributed or if all assessments were necessary (Umbreit, 1995). In addition, it








was unclear if any of the assessment methods by themselves would have lead to a successful intervention (Umbreit, 1995).

Conroy et al. (1996) compared the use of analogue functional analyses to direct observations in predicting the function of behavior. Their study was conducted on four male children with developmental disabilities and problematic behaviors. To determine which analogue probes to conduct, the authors conducted preliminary functional assessments including the FAI (O'Neill et al., 1997) and the MAS (Durand & Crimmins, 1992). After the administration of these instruments, direct observations occurred in the classroom and the analogue probes were conducted. The analogue probes were successful in identifying conditions that were motivating behavior in two of the four subjects. However, the results from the analogue probes on the two subjects were found to match the results from the direct observation on only one subject. The direct observation did not concur with the analogue probes. This raises questions as to the validity of analogue probes verses the validity of direct observations (Conroy et al., 1996). As a result, it is suggested that analogue probes be used with a battery of functional assessment techniques (Conroy et al., 1996). The authors point out the need to assess the reliability and validity of analogue probes and recommend that future research consider the best way to combine functional assessment techniques (Conroy et al., 1996).

Crawford et al. (1992) compared the results of three functional assessment methods on four adults with severe to profound mental retardation who engaged in stereotypic behavior. They administered the MAS, followed by a functional analysis, and then engaged in direct observations using an A-B-C assessment. The A-B-C assessment and the MAS produced the clearest results and were in agreement with each other.







However, the functional analysis produced more ambiguous results. For two out of the four participants, the functional analysis was not in agreement with the other methods. The authors suggested that the MAS and A-B-C assessments provide useful functional assessment data and that the relative ease of administration and interpretation of these assessment tools make them practical to use in an applied setting. However, it was noted that the inter-rater reliability on the MAS was not as high as expected. Also, there was no evidence as to the validity of the findings from these instruments (Crawford et al., 1992). Crawford et al. (1992) stated that best practices would include the use of multiple functional assessment measures because of the lack of research on the reliability and validity of different functional assessment methods.

The research assessing the validity of functional assessment instruments is limited and has produced mixed results. Fox and his colleagues (1998) conducted a review of the research on the use of functional assessments on students with or at-risk for emotional and behavioral disorders. They found that the most frequently used functional assessment techniques were direct observations and interviews. Most of the studies in this area used more than one functional assessment technique (Fox et al., 1998). Surprisingly, there was little data on the reliability and validity of the functional assessment instruments/techniques in use (Fox et al., 1998). Many researchers question the reliability and validity of indirect functional assessment procedures (e.g., Fox et al., 1998; Davis, 1998; Symons, et al., 1998; Wehby & Symons, 1996; Gable, 1996; Iwata, 1994). Indirect functional assessment measures are likely to increase in popularity because they are less intrusive (Fox et al., 1998). Carr (1994) emphasized the need for researchers to develop descriptive procedures that provide results congruent to functional analyses for practical







purposes. The need for reliability and validity studies on functional assessment techniques and instruments has been clearly documented (Shriver et al., 2001; Carr, 1994; Fox et al., 1998; Gable, 1996; Iwata, 1994).

Conclusion

There is a need for more studies comparing functional assessment instruments (Arndorfer & Miltenberger, 1993; Fox et al., 1998; Sturmey, 1994). It may be that each of these functional assessment instruments are useful in developing effective treatments or that a combination approach is necessary. Research needs to be conducted to determine the utility of different procedures to reliably and accurately assess the function of behavior (Arndorfer & Miltenberger, 1993; Cunningham & O'Neill, 2000).

The research conducted on the MAS has provided inconsistent findings with most of the research documenting the need for future studies in this area. Clearly, there is a lack of research on different functional assessment instruments such as the FAO and the FAI developed by O'Neill et al. (1997). The study conducted by Chait (2001) was the most comprehensive analysis done on these instruments to date. Although the findings were useful, they were a comparison across descriptive assessments only. The best measure of an instruments ability to identify functions of behavior would be to compare the instruments results with a functional analysis. Few studies have compared the results of different functional assessment instruments to a functional analysis and several authors have documented the need for future studies in this area (Arndorfer & Miltenberger, 1993; Chait, 2001; Cunningham & O'Neill, 2000; Fox et al., 1998; Gable, 1996; Iwata, 1994; Sturmey, 1994). The present study sought to advance the scientific knowledgebase regarding functional assessments by comparing the results of three commonly used








instruments, the FAO, FAI, and MAS, to a functional analysis to determine their consistency. The present study expanded the research conducted by Cunningham and O'Neill. Specifically, the study was conducted on three children with autism in an inclusive setting. Functional assessments were conducted in both the school and the home environment with teacher and parent participation. Results from the individual instruments were then compared to a functional analysis. In addition, the FAO form was further analyzed to determine which category (predicted function vs. actual consequences) lead to the most consistent results. The following questions were addressed:

* Within settings, how consistent are each of these instruments in identifying a
hypothesized function of the target behavior in comparison to the function
identified in the functional analysis? Are the functional analysis findings the same
across settings?

* Within settings, which category of the Functional Assessment Observation form
(perceived functions vs. actual consequences) was most consistent with the results
of the functional analysis and thereby provided the most useful information?

* Based on current findings, is the use of a descriptive assessment appropriate to
identify the function of behavior in the general education and home environment?
Which instrument or category of an instrument provides the most consistent
findings in comparison to the functional analysis?













CHAPTER 3
METHOD

Participants

Three 1st grade children who were classified with autism and exhibited

challenging behaviors participated in this study. To be included each child had to meet four criteria. First, the child had to be diagnosed with autism/autism spectrum disorder. Second, the child must have demonstrated challenging behaviors. Third, the child must have been included in general education at least 50% of the time. Fourth, consent to participate in this study was sought from the school principal, teacher, and parents. The following is a brief description of each child.

Josh was a 7-year-old Caucasian male with a diagnosis of autism/autism spectrum disorder. His problematic behaviors in the classroom included inappropriate vocalizations, disruption, noncompliance, and off-task behaviors. In the home environment, Josh's parents reported concern with his inappropriate vocalizations and noncompliant behavior. Intellectual assessments determined Josh was functioning within the below average range with a composite score of 78 on the Kaufman Assessment Battery for Children. Josh was enrolled in a 1st grade general education classroom for most of the day. He received special education in reading and writing 1 to 5 hours a week. Josh also received speech and language services three times a week for 30 minutes and occupational therapy once a month for 30 minutes.







Brad was a seven-year-old Caucasian male with a diagnosis of autism and a

speech and language impairment. Brad's problematic behaviors in the classroom included disruptions, elopement, noncompliance, off-task behavior, and aggression. In the home environment, Brad's parents reported the same behavioral concerns except for his off-task behavior. Intellectual assessments determined Brad was functioning within the below average range with a composite score of 78 on the Comprehensive Test of Nonverbal Intelligence. Brad was enrolled in a 1st grade general education classroom for most of the day and had a full time classroom aide. He received special education in reading and writing one to five hours a week. Brad also received speech and language services three times a week for 30 minutes and occupational therapy once a month for 30 minutes.

Drake was a 7-year-old Caucasian male with diagnoses of autism and language

impairment. Drake's problematic behaviors in the school and home environment included noncompliance, vocalizations, stereotypy, disruptions, aggression, and tantrums. Intellectual assessments determined Drake was functioning within the below average range with a composite score of 77 on the Stanford-Binet IV. Drake was enrolled in a 1 st grade general education classroom for most of the day. A general education and special education teacher were present in the classroom full time. He received speech and language therapy and occupational therapy on a weekly basis. Drake was on several medications for behavior problems and seizures including: Depakote, Resperdol, and Zoloft.

Settings and Therapists

The investigation was completed in two phases. During Phase 1, a series of

descriptive assessments were conducted in each child's general education classroom and







various areas in the home (e.g., living room, bedroom, kitchen, etc.). During Phase 2, a functional analysis was completed in an empty classroom in each child's school setting and in the living room of each child's home. All instruments and analyses were completed across two settings, the school and home. The FAO was conducted by a trained investigator through direct observations in the school and home setting. The FAI and MAS were administered to the general education teacher in the school setting and to the parent in the home setting by a trained investigator. The functional analyses were conducted by a trained investigator in an empty classroom in the school during the school day and by a parent in the home setting.

Materials

Descriptive Assessment Measures

The Functional Assessment Observation form (FAO) developed by O'Neill et al. (1997) is a structured method of direct observation of behavior over time/days. It was designed to assist an observer in the identification of the antecedents, consequences, possible functions, co-occurring behaviors, number of events of the problem behavior, and times in which the problem behavior is more or less likely to occur (O'Neill et al., 1997). The FAO form consists of eight sections that are to be completed while directly observing an individual. In the first section, identification information is recorded, such as the name of the individual being observed and the observation date. The second section is separated into time blocks to record the activity, setting, and time interval of the observation. The third section is composed of the target behaviors that have been identified for observation. The fourth section is made up of events that are considered potential predictors of the target behavior. The fifth section contains perceived functions of the target behavior. The sixth section is used to record the consequences of the target







behavior. The seventh section provides a commentary section for events occurring during a given block of time. The eighth section allows for the tracking of behavioral events daily and in total. The FAO was completed according to the instructions provided by the FAO manual developed by O'Neill et al. (1997). See Appendix A for a sample of this form.

The Functional Analysis Interview Scale (FAI) (O'Neill et al., 1997) is a structured interview that takes approximately 45 to 90 minutes to administer. It is designed to collect information about the individual and the events influencing his/her problem behavior. The FAI is divided into 11 sections. In the first section, problem behaviors are clearly described. For each target behavior, the interviewee is asked to describe its topography (how the behavior was performed), frequency (how often the behavior was performed), duration (how long the behavior lasted), and intensity (how damaging was the behavior). In the second section, potential setting events are identified. The interviewee is asked questions regarding the child's medications, medical conditions, sleep patterns, diet, and other similar types of questions. In the third section, questions are designed to identify antecedent events that do or do not occasion problem behavior. For example, the interviewee may be asked the time of day in which the behaviors are most and least likely to happen. The fourth section consists of questions aimed at identifying the consequences of the problematic behavior. The interviewee is asked to identify particular situations in which the behavior occurs and what the child gains or avoids by engaging in the behavior. The fifth section contains questions designed to uncover the efficiency of the problematic behavior. Interviewees are asked to rate the efficiency of the child's behaviors (how effective they are in obtaining desired outcomes) on a 5-point Likert scale (1 = low efficiency and 5 = high efficiency). The sixth section helps to







identify alternative behaviors in the individual's repertoire. Interviewees are asked "What socially appropriate behaviors or skills can the person already perform that may generate the same outcomes or reinforcers produced by the problem behaviors?" (O'Neill et al., 1997, Appendix B, p. 6). The seventh section contains questions regarding the individual's communicative abilities. For example, the interviewee is asked "What are the general expressive communication strategies used by or available to the person?" (O'Neill et al., 1997, Appendix B, p. 6). The eighth section explores general information about the types of teaching strategies and activities that may or may not work well with the individual. For example, "What things can you do to improve the likelihood that a teaching session or other activity will go well with this person?" (O'Neill et al., 1997, Appendix B, p. 7). The ninth section consists of questions to identify objects, activities, and events that are reinforcing to the individual. The interviewee is asked to list particular food items, toys, activities, and any other things that are reinforcing to the child. The tenth section is designed to gather information on programs or interventions previously attempted. The interviewee is asked how long the behavior has been a problem, what programs have been attempted to decrease the behavior, and the resulting effects of those programs. Section eleven is a form for the interviewer to summarize data by listing the major antecedents, consequences, and setting events surrounding behavior identified through the interview process. See Appendix B for a sample of this form.

The Motivation Assessment Scale (MAS) (Durand & Crimmins, 1988) is a 16item questionnaire designed to provide information regarding four hypothesized function(s) of the target behavior (Durand & Crimmins, 1988). The questions on the MAS describe a variety of situations in which the target behavior may occur. The likelihood of the target behavior occurring is then rated on a 7-point Likert scale (0 =







Never and 6 = Always). An example of an item on the MAS is, "Does the behavior occur following a command to perform a difficult task?" (Durand & Crimmins, 1988, p. 102). Each question on the MAS assesses the influence of one of four potential functions of the behavior: escape, attention, tangibles, and/or sensory stimulation. There are four questions to assess each of the four variables. After the scale is completed, the four questions relating to each of the four variables are averaged. High scores indicate that the variable may be maintaining the behavior. Refer to Appendix C for a sample of the MAS form.

Training. Investigators were trained on the administration and scoring of the instruments to ensure reliable data. The two primary investigators had extensive experience in using the FAO, FAI, and the MAS. Additional raters were trained to conduct the FAO and the FAI. In addition to reading the directions from the handbook developed by O'Neill et al. (1997), raters attended a short workshop conducted by a professional experienced in using the FAO and FAI. Practice sessions were conducted for the FAO in the general education setting. Two raters independently observed a participant at the same time and collected FAO data. Practice sessions were conducted twice a week for a total of 10 hours until the raters reached at least 80% agreement. Raters were trained to administer and score the MAS. Training involved reading the manual and practicing scoring until reaching 100% agreement which occurred in one session.

Functional Analysis

The functional analysis consisted of five conditions: attention, escape, tangible, free play, and ignore. During the attention, tangible, and free play conditions, preferred items were used. Preferred items were identified by parent and teacher reports as well as








a preference assessment. A preference assessment was conducted at the beginning of each session using procedures similar to those used by DeLeon and Iwata (1996) and involved providing the child with several items for a period of 5 minutes. The item the participant played with the most during the allotted time period was identified as the preferred item.

For Josh, the most highly preferred items at school included books, puzzles, handheld electronic games, and toy cars. At home, the most highly preferred items were video games, books, and toy soldiers. For Brad, the most highly preferred items at school included books, toy cars, and toy soldiers. At home, the most highly preferred items included books, videos, and action figure dolls. For Drake, the most highly preferred items at school and home included books, toy trains, and balls.

During the escape condition, nonpreferred tasks were used. Nonpreferred tasks

were identified by parent and teacher reports and included tasks that the child was able to perform but that were difficult for the child or occasioned problematic behavior. For Josh, the task was to write his spelling words three times each. For Brad, single-digit math problems were used. For Drake, the task at school was to read words and the task at home was to write letters.

Training. For each child, the functional analysis was conducted by an investigator experienced in conducting functional analyses. Additional investigators were trained through a combination of direct instruction and assigned readings. After training was complete, investigators practiced coding behaviors using videotaped examples of the participants' behaviors. Investigators practiced two hours a day twice a week for a total of 10 hours, until reaching 80% reliability using a 10 second agreement window.








Operational Definitions, Observation System, and Interobserver Agreement Operational Definitions

Seven behaviors were targeted across both phases of assessment during the study. As previously described, each child engaged in one or more of these behaviors. Not all children engaged in each of the behaviors defined and not all behaviors were assessed by each instrument across both settings. Disruption was defined as verbal talk which was out of context (e.g., above speaking volume) or without permission, making noise by pounding on the tables, walls, or floors with a closed fist or object, throwing objects (not at another person), ripping or tearing objects, writing on inappropriate objects, screaming or yelling, spitting (not at another person), playing with materials or toys inappropriately, trying to escape work situations, throwing work materials, yelling at others, verbal threats, and inappropriate manding (Josh, Brad, and Drake). If disruption occurred, it was recorded singly or in combination with other behaviors. Vocalizations were defined as noises, words, or phrases made by the participant that did not have to do with the specific task at hand (Josh and Drake). Noises included nonsense vocalizations, singing, humming, or talking to self. Words and phrases were in the form of echolalia, which is the verbal repeating of what had been said by others. Echolalia was scored if it occurred immediately or delayed. Delayed echolalia involved children repeating words or phrases previously heard. Noncompliance was defined as the failure to complete an instruction or to begin following instructions five seconds after the request was given (Josh, Brad, and Drake). Off-task behavior was defined as the participant facing away from the task/material/instructor or not engaging in the instructional activity when directed (Josh and Brad). Aggression was defined as hitting, biting, pinching, kicking, pulling another person, throwing objects, spitting, or pushing one's body-part or body into another person







(Brad and Drake). Stereotypy was defined as repetitive behaviors such as rocking, hand flapping, arm flicking, spinning (object or body), covering ears, finger wringing, and staring at fingers while wiggling them (Drake). Tantrums were defined as screaming and crying or multiple topographies of behavior occurring at once (Drake). In addition, tantrum included dropping to the ground (either on knees, bottom, or back) and remaining there. Refer to Appendix D for an outline of the targeted behaviors. Observation System

Descriptive assessments. Before administering the descriptive assessments, target behaviors were selected and operational definitions were developed by the investigators. The three descriptive assessment measures were conducted independently of each other. During the FAO, participants were directly observed in the classroom and home environment by a trained investigator using the FAO form for five hours in each setting across a 4 week period. Hypotheses regarding the function of the participant's behavior were developed by comparing the number of behavioral events out of the total that were coded in each of the perceived function categories.

Two trained investigators were present for the administration of the FAI. In the school environment, the primary interviewer was the same for all three children. The second interviewer was the primary investigator for each of the individual children. In the home environment, a different primary interviewer conducted the interview for all three children. The second interviewer remained the same for all three children. The FAI was administered to parents and teachers separately but within the same week. Questions were delivered orally by the primary interviewer. Responses to the questions were recorded in vivo directly on the FAI form separately and independently by the two investigators. After the interview, the investigators developed separate and independent







summary statements about the problem behaviors and generated one or more hypothesis(es) regarding the functions of behavior. These hypotheses were then compared across raters to determine the inter-rater reliability of the instrument. Before completing the MAS, parents and teachers were given a brief explanation of the purpose of the MAS and the operational definition of the targeted behaviors. For all children, more than one behavior was identified and the MAS was completed for each of those behaviors individually. For Josh, three behaviors were identified at school (vocalization, disruption, and noncompliance) and two behaviors at home (vocalization and noncompliance). For Brad, three behaviors were identified at school and at home (disruption, noncompliance, and aggression). For Drake, three behaviors were identified at school (vocalization, noncompliance, and tantrum) and two at home (noncompliance and tantrum). The MAS was administered separately to the teacher and parent of each participant by a trained investigator within a one week time period. Scores were averaged for each of the four variable categories. The highest scoring variables were hypothesized to be maintaining the behaviors.

Functional analyses. After the completion of the three descriptive assessments,

functional analyses were conducted. All of the children participated in a series of sessions using procedures similar to those described by Iwata, et al. (1982/1994). Sessions were five minutes in length and involved the following conditions: contingent attention, contingent escape, contingent tangibles, free play, and ignore. For two of the children (Josh and Brad), additional sessions were conducted to analyze an additional condition of ignore with access to tangibles. These sessions were conducted to determine if problematic behavior during the ignore condition was attenuated when allowed access to tangibles. It was a test to further determine if behavior would be maintained without








social consequences. If the behavior ceased with the introduction of tangibles during the ignore condition, the behavior was determined to have a social function (e.g., not identified as an automatic function).

During the attention condition, the child was instructed to play alone while the adult was present in the room (e.g., engaged in paperwork). The child had access to preferred toys. The adult diverted his/her attention away from the child. Attention was provided to the child for 15 seconds for each occurrence of problematic behavior. Attention was in the form of statements of disapproval (e.g., "Don't do that"). This condition was designed to determine if problem behavior was maintained by positive reinforcement in the form of social attention. In the tangible condition, a preferred tangible item was provided to the child for 1 minute before beginning the condition. Next, the adult removed access to the tangible item by stating "my turn" and the item was then visible to the child but out of reach. The child was provided with attention from the adult at least every 15 seconds and no task demands were present. The tangible item was returned to the child for 15 seconds for each occurrence of problematic behavior. After 15 seconds the tangible item was removed. This condition was designed to determine if problem behavior was maintained by positive reinforcement in the form of tangibles. In the escape condition, the child was given an academic task to complete and told "It's time to work". The task was removed for 15 seconds for each occurrence of problematic behavior. This condition was designed to determine if problematic behavior was maintained by negative reinforcement. During the free play condition, the child had access to preferred items and attention was provided by the adult at least every 15 seconds in the form of neutral statements (e.g., "You are playing with a car."). Problematic behavior was ignored. It simulated an enriched environment and no demands







were presented. This condition was a control condition and problem behavior was not expected to occur. During the ignore condition, the room was cleared of all toys and materials and the child was left with "nothing" to do. All behavior was ignored. This condition was designed to determine if the child's problem behavior was maintained in the absence of social consequences (automatic reinforcement). The ignore with tangibles condition was included if high rates of problematic behaviors occurred during the ignore condition. For Josh, this condition was added in the school environment only. For Brad, this condition was added in the school and home environment. The ignore with tangibles condition was the same as the ignore condition previously described with one exception, the addition of one preferred item. This condition was designed to determine if high rates of problem behavior in the ignore condition occurred because the behaviors were maintained by automatic reinforcement or because there was nothing else to do and behaviors escalated to gain access to items.

The order of the sessions was randomly selected for each child and began in the school environment. The sessions were videotaped and later coded on palm-top computers to identify the occurrence of problematic behaviors. A computerized program entitled the Multiple Option Observation System for Experimental Studies2 (MOOSES2) (Tapp, Wehby, & Ellis, 1995) was used to record the problematic behaviors in a direct, continuous, and sequential manner. MOOSES2 was used to calculate the frequency of problematic behaviors in each condition. Frequency data were converted to rate per minute. The rate of problematic behaviors in each condition was graphically displayed and visual inspection was used to determine when stability in the data was achieved. Behaviors were graphed individually and some were combined. Behaviors were only combined when it was hypothesized that they were in the same response class and they







served the same function(s). Therefore, not all children's behaviors in each setting were combined. To determine the function of behavior, two independent observers visually inspected the graphical display of data. The function(s) of behavior determined by the primary investigator were used for comparison. Only behaviors for which a function(s) was clearly determined were used for comparison. Interobserver Agreement

Functional Assessment Observation Form. To determine the reliability of the

FAO across raters in each setting, a second trained investigator was present during 30% of the FAO observation sessions. The data collected by the observers was compared to determine if each observer recorded the same number of specific events, behaviors, antecedents, functions, and consequences of the behavior. Agreement was calculated by dividing the number of agreements by the total number of agreements plus disagreements and multiplying by 100 (Kazdin, 1982). Interobserver agreement on the FAO was high. Total agreement between the two observers averaged 92% with a range from 74% to 99%. In the school environment, agreement averaged 90% with a range from 74% to 99%. In the home environment, agreement averaged 94% with a range from 86% to 98%.

For Josh, agreement between the two observers in the school environment

averaged 96% and 98% in the home environment. In the school environment, agreement on behaviors, predictors, and perceived functions averaged 97% with a range from 95% to 98%. Agreement on actual consequences averaged 93% with a range from 88% to 98%. In the home environment, reliability across raters was typically conducted in one observation period lasting 90 minutes in length. Agreement was 100% on behaviors, 95% on predictors, 100% on perceived functions, and 95% on actual consequences.







For Brad, agreement between the two observers averaged 74% in the school

environment and 86% in the home environment. In the school environment, agreement averaged 79% on behaviors (range = 65% to 95%), 72% on predictors (range = 62% to 91%), 75% on perceived functions (range = 59% to 91%), and 74% on actual consequences (range = 66% to 80%). In the home environment, agreement averaged 83% on behaviors, 92% on predictors, 83% on perceived functions, and 88% on actual consequences.

For Drake, agreement between the two observers averaged 99% in the school

environment and 97% in the home environment. In the school environment, agreement on behaviors, predictors, and perceived functions was 100%. Agreement on actual consequences averaged 96% (range = 91% to 100%). In the home environment, agreement averaged 98% on behaviors (range = 89% to 100%), 80% on predictors (range = 66% to 93%), 87% on perceived functions (range = 78% to 95%), and 100% on actual consequences.

Functional Assessment Interview. Inter-rater agreement in each setting on the FAI was calculated by comparing the hypotheses generated by each investigator. The resulting hypothesized function of each child's behavior was compared within settings by comparing the primary interviewer's hypothesized function of behavior for each child to the secondary interviewer's hypothesized function of behavior for each child. The primary interviewers were the standard to which comparisons were made on the FAI. For example, if the primary interviewer listed attention and escape as the consequence of behavior and the secondary interviewer only listed escape, it was considered one agreement and one disagreement. Agreement was calculated by dividing the number of agreements by the total number of agreements plus disagreements and multiplying by 100








(Kazdin, 1982). The average agreement on the consequences hypothesized to be maintaining behavior generated from the FAI, across settings, was 68% with a range from 50% to 100%. In the school environment, agreement on the consequences hypothesized to be maintaining behavior averaged 72% with a range from 60% to 100%. Refer to Appendix E for a detailed description of the summary statements made for each child by the primary and secondary interviewer in the school environment.

For Josh, the interviewers agreed 100% of the time on the consequences

hypothesized to be maintaining behavior (attention and escape for disruption, attention for vocalization, and escape for noncompliance and off-task behavior). For Brad, the interviewers agreed 66% of the time on the consequences hypothesized to be maintaining behavior (attention for noncompliance, stereotypy, and off-task behavior, attention and escape for aggression, and escape as the hypothesized function for elopement). For Drake, interviewers agreed 50% of the time on the consequences hypothesized to be maintaining behavior (automatic for stereotypy, attention and escape for aggression, and escape as the hypothesized function for noncompliance).

In the home environment, agreement on the consequences hypothesized to be

maintaining behavior averaged 89% with a range from 43% to 100%. Refer to Appendix F for a detailed description of the summary statements made for each child by the primary and secondary interviewer in the home environment. For Josh, the interviewers agreed 50% of the time on the consequences hypothesized to be maintaining behavior (attention for noncompliance and automatic for vocalizations). For Brad, the interviewers agreed 100% of the time on the consequences hypothesized to be maintaining behavior (escape). For Drake, the interviewers agreed 43% of the time on the consequences







hypothesized to be maintaining behavior (escape for noncompliance and automatic for stereotypy and vocalizations).

Functional analysis. To determine the interobserver agreement during the

functional analyses, two independent observers coded the videotapes of approximately 30% (with a range of 27% to 52%) of all the functional analysis sessions for each child. Data files created by MOOSES were printed and agreement was calculated manually. Agreement was calculated by dividing the total number of agreements by the total number of agreements plus disagreements and multiplying by 100 (Kazdin, 1982). An agreement was recorded when the primary and secondary observer recorded the same response within a 10-second interval. A disagreement was recorded when the secondary observer did not record the same response or recorded the same response outside of the 10-second interval. For Josh, interobserver agreement was collected on 30% of the functional analysis sessions in the school (averaged 95% with a range from 84% to 100%) and home (averaged 94% with a range from 80% to 100%). For Brad, interobserver agreement was collected on 32% of the sessions at school (averaged 81% with a range from 39% to 100%) and 52% of the sessions at home (averaged 87% with a range from 63% to 100%). For Drake, interobserver agreement was collected on 27% of the sessions at school (averaged 88% with a range from 62% to 97%) and 30% of the sessions at home (averaged 92% with a range from 77% to 100%). Across participants, in the school environment, agreement averaged 88% with a range from 39% to 100% and in the home environment, agreement averaged 91% with a range from 63% to 100%.

To determine the function of behaviors, two observers independently rated each graphical display of data. Agreement was calculated by dividing the number of agreements by the total number of agreements plus disagreements and multiplying by 100







(Kazdin, 1982). Refer to Appendix G to view functional analysis graphs and Appendix H for a detailed summary of the function(s) identified by each investigator. Across environments, agreement on the functions of behavior for individual and combined graphs averaged 87% with a range from 67% to 100%. Across environments, agreement on the function of behavior for individual graphs averaged 83% with a range from 50% to 100%. For Josh, three behaviors were analyzed in the school and home environment (vocalization, noncompliance, and disruption). Agreement among raters on the function of the individual graphs of behaviors was 100%. For Brad, three behaviors were analyzed in the school and home environment (disruption, noncompliance, and aggression). Agreement among raters on the function of the individual graphs of behaviors was 60%. For Drake, four behaviors were analyzed in the school and home environment (noncompliance, tantrums, vocalizations, and disruption). Agreement among raters on the function of the individual graphs of behaviors was 100%. Across environments, agreement of the function of behavior for combined graphs averaged 100%. For Josh, agreement among raters on the function of the combined graph of behaviors was 100%. For Brad, agreement among raters on the function of the combined graph of behavior was 100%. In the school environment, agreement on the function of behavior for individual and combined graphs averaged 88% with a range from 75% to 100%. Agreement on the function of behavior for individual graphs averaged 86% with a range from 75% to 100%. Agreement on the function of behavior for combined graphs averaged 100%. In the home environment, agreement on the function of behavior for individual and combined graphs averaged 85% with a range from 67% to 100%. Agreement on the function of behavior for individual graphs averaged 80% with a range from 50% to 100%. Agreement on the function of behavior for combined graphs averaged 100%.







Design and Data Analysis

Design

Descriptive assessment. For two of the three students (Josh and Brad) data

collection on the FAI and MAS began in the school environment first then in the home. For Drake, data collection began in the home. For all participants, the FAO was conducted in the school environment first then in the home. No experimental design was used for the descriptive assessment phase. The setting in which the FAI and the MAS were initially administered was counter-balanced to minimize order effects.

Functional analysis. During the functional analysis five conditions were randomly presented to each participant using a multi element design. The design was used to determine the maintaining contingency for problematic behavior across children and across settings.

Data Analysis

Descriptive assessments. The data generated from each instrument was analyzed to determine the hypothesized function of behavior. For the FAO, hypotheses regarding the function of the child's behavior were developed by comparing the number of behavioral events out of the total that were coded in each of the perceived function categories. The perceived function category with the highest occurrence of the problem behavior was hypothesized to be the primary function of that behavior. For the FAI, hypotheses regarding the function of the child's behavior were developed by analyzing the summary statements made by the primary interviewer. The maintaining consequence statements made by the primary interviewer were hypothesized to be the primary function of that behavior. For the MAS, results from the teacher and parent were individually tallied and recorded on the MAS form. This information generated rankings of







hypothesized functions of each behavior. The hypothesized function scoring the highest mean score was determined to be the primary function of each behavior.

Functional analysis. The data was analyzed across participants and settings using a single-subject design. The differences in problematic behavior across conditions and settings were then examined for each individual. Data were analyzed using line graphs that visually displayed the rate of problematic behavior across a 5-minute condition period. The rate indicated the number of times in which problem behavior occurred divided by five. The line graphs presented a graphical display of the rate of problem behavior in each experimental condition across the settings (Parsonson & Baer, 1992). This allowed for a visual evaluation of the differences in rate of problem behavior by each condition and by setting (Kazdin, 1982). The primary function was determined by visual display of the condition(s) with the most consistent and highest rates of problem behavior across one or more conditions. The function(s) identified by the primary investigator were used for comparison.

Comparing descriptive assessments to functional analyses. The results obtained on all three descriptive assessment instruments were individually compared to the functional analysis within settings for each participant. Comparisons were made on behaviors assessed by all three instruments and a functional analysis. Agreement within settings was calculated by determining the number of agreements on the hypothesized function of the behavior between the descriptive assessment instruments and the functional analyses and dividing that by the number of agreements plus disagreements and multiplying by 100 (Kazdin, 1982). High agreement was defined as 80% to 100% (Kazdin, 1982), moderate agreement was defined as 40% to 79%, and low agreement was defined as 0% to 39%.







To determine if functions of behavior differed across settings, the results of the functional analysis for each behavior were compared across the home and school. Agreement across settings was calculated by determining the number of agreements on the function of behavior between the home and school setting and dividing that by the number of agreements plus disagreements and multiplying by 100 (Kazdin, 1982). Behaviors that were assessed in both environments were compared.

Finally, to determine which category of the FAO was most useful in determining the function of behavior, the "actual consequence" category on the FAO and the "perceived function" of behavior category were individually compared to the functional analysis findings for each behavior to determine their agreement within settings. Agreement was calculated by determining the number of agreements on the hypothesized function of the behavior between the FAO categories and the functional analysis and dividing that by the number of agreements plus disagreements and multiplying by 100 (Kazdin, 1982).













CHAPTER 4
RESULTS

The purpose of this investigation was to determine the consistency of three descriptive assessment instruments by comparing their results within settings to the findings of a functional analysis, determine if functions of behavior are the same across settings by comparing the results of functional analyses across settings, determine the category of the FAO that was most likely to predict the same function of behavior as the functional analysis, and determine if the use of descriptive assessments are supported and if so which instrument is more consistent with the results of a functional analysis? These questions were addressed by conducting three descriptive assessments and a functional analysis across the home and school setting for three children identified with autism who were included in the general education classroom for most of the school day. Results were analyzed for individual and combined target behaviors.

In general, the results of this study indicate that the FAO, FAI, and MAS have a low level of agreement with the results of functional analyses (averaging 39%). This indicates that these instruments generally will provide different functions of behavior than those identified by a functional analysis. Individually, the FAO had a moderate level of agreement (averaging 45%). The FAI and the MAS had low levels of agreement (averaging 36% and 29%, respectively). This means that of the three descriptive assessment instruments evaluated, the FAO is a better instrument and likely to accurately







identify the function of behavior approximately half of the time. Overall agreement between the three instruments and the functional analyses was higher in the home setting (averaging 56%) than at school (averaging 28%). The instruments were more likely to agree with the functional analysis when escape or automatic reinforcement was identified as the function of behavior.

When examining functional analyses across settings, there was a moderate level of agreement on the function of behaviors across settings (averaging 59%). This means that the function of a behavior may actually differ depending on the setting. Graphs of combined behaviors were more likely to find the same function across settings (averaging 75%) than graphs of individual behaviors (averaging 54%). However, only two graphs of combined behaviors were used in this analysis compared to nine graphs of individual behavior. Overall, tangible and automatic functions were more likely to be the same across settings.

The perceived function and actual consequence categories on the FAO combined had a low level of agreement when compared to the results of functional analyses (averaging 27%). Individually, the perceived function category had a moderate level of agreement (averaging 45%). The actual consequences category had a low level of agreement (averaging 10%). This indicates that the perceived function category is more likely than the actual consequences category to accurately identify the function of behavior. Agreement between the FAO categories and the functional analyses were generally higher in the home setting (averaging 36%) than at school (averaging 21%). The perceived function and actual consequence category were more likely to be in agreement with the functional analysis when automatic reinforcement was the identified







function. In addition, the perceived function category was more likely to be in agreement with the functional analysis when escape was the identified function.

Comparison of Descriptive Assessments to Functional Analyses

Descriptive functional assessment instruments were compared to the functional analysis within settings. Behaviors were compared individually and then in some combinations depending on the hypothesized response class and function(s). The total number of behaviors compared differed across environments since only some of the behaviors occurred in both the school and home settings. In addition, only behaviors analyzed by both an instrument and a functional analysis were used for comparison. A total of ten individual behaviors across children were compared in the school setting (three for Josh, three for Brad, and four for Drake). Two combined groups of behaviors were compared in the school setting (one for Josh and one for Brad). In the home environment, a total of eight individual behaviors across children were compared (three for Josh, two for Brad, and three for Drake). Two combined groups of behaviors were compared in the home setting (one for Josh and one for Brad). The number of comparisons was limited for three reasons. First, fewer behaviors were assessed by the MAS than were observed during the functional analysis. Second, more behaviors were assessed by the FAI than were directly observed during the functional analysis and the FAO. Third, the results of two functional analyses were inconclusive and a function could not be clearly identified. Data will be discussed by summarizing all participants' data and then reporting individual participant data by setting. Comparisons within the School Environment

For all three participants, within the school environment, the FAO, FAI, and MAS agreed with the results of the functional analyses 28% of the time with agreement ranging








from 0% to 75%. The FAO agreed with the results of the functional analyses 6 out of 17 times (35%). The FAI agreed with the results of the functional analyses 3 out of 14 times (21%). The MAS agreed with the results of the functional analyses 3 out of 12 times (25%). Table 4-1 presents a summary of the percent agreement between the hypothesized function of behaviors generated across instruments and participants and the functional analysis within the school setting.

For Josh, the hypothesized functions of behavior generated from the FAO, FAI, and MAS in the school setting agreed with the results of the functional analysis 0 out of 12 times (0%). Individually, the FAO, FAI, and MAS did not agree with the results of the functional analysis. The tangible function identified in the functional analyses was not identified by any of the descriptive instruments as a hypothesized function for any of the target behaviors. For Brad, the hypothesized functions of behavior generated from the FAO, FAI, and MAS in the school setting agreed with the results of the functional analysis 6 out of 18 times (33%). Individually, the FAO agreed with the results of the functional analysis four out of eight times (50%). The FAO correctly hypothesized the escape function of behavior (functional analyses identified escape and tangible functions) for four of the target behaviors (noncompliance, disruption, aggression, and a combination of all 3 of these behaviors). The FAI agreed with the results of the functional analysis two out of six times (33%) by correctly hypothesizing the escape function for two of the target behaviors (aggression and a combination of noncompliance, disruption, and aggression). The MAS agreed with the results of the functional analysis zero out of four times (0%). The MAS hypothesized that the four target behaviors were maintained by automatic reinforcement. For Drake, the hypothesized functions of behavior generated







Table 4-1

Summary of the Percent Agreement between the Hypothesized Functions of Behaviors Generated by Instruments within the School Environment and the Functions ofBehaviors Determined by the Functional Analysis

School


Participants

Josh Brad Drake Total agreement


FAO

0% (0/4) 50% (4/8) 40% (2/5) 35% (6/17)


FAI

0% (0/4) 33% (2/6) 25%(1/4) 21% (3/14)


MAS

0% (0/4) 0% (0/4) 75% (3/4)

25% (3/12)


Total agreement

0% (0/12) 33% (6/18) 46% (6/13)

28% (12/43)







from the FAO, FAI, and MAS in the school setting agreed with the results of the functional analysis 6 out of 13 times (46%). Individually, the FAO agreed with the results of the functional analysis two out of five times (40%) by correctly hypothesizing the escape function of noncompliance and the automatic reinforcement function of vocalizations. The FAI agreed with the results of the functional analysis one out of four times (25%) by correctly hypothesizing the escape function of noncompliance. The MAS agreed with the results of the functional analysis three out of four times (75%) by correctly hypothesizing the escape function of noncompliance, the automatic reinforcement function of vocalizations, and the tangible function of tantrums.

When escape was identified as the function of the target behavior across all three participants, the FAO always matched this function with one exception (Drake's disruptive behavior), the FAI correctly identified it half of the time (3/6), and the MAS correctly identified it once (1/6). The MAS was the only instrument to correctly hypothesize a tangible function for Drake (1/2). However, none of the instruments identified the tangible function for Josh or Brad. The FAO and MAS correctly hypothesized the automatic function for Drake (1/1). In summary, these results indicate that there was limited to no correct matching to the functions of behavior identified in the functional analysis in the school setting. Tables 4-2, 4-3, and 4-4 present detailed summaries of the hypothesized function of behaviors generated across instruments and participants within the school setting.

Comparisons within the Home Environment

For all participants within the home setting, the FAO, FAI, and MAS agreed with the results of the functional analysis 56% of the time with agreement ranging from








Table 4-2

Summary of the Hypothesized Function of Behaviors across Instruments in the School Environment for Josh

Behaviors FAO FAI MAS Functional Total


Vocalization Disruption





Noncompliance Disruption and noncompliance


Attention Attention





Escape Attention and

escape


Aggression


Attention Attention and escape Escape Attention and escape Attention and escape


Escape Escape





Attention Attention and escape


analysis Tangible Tangible


Tangible Tangible


agreement

0/3 (0%) 0/3 (0%)


0/3 (0%) 0/3 (0%)


Off-Task - Escape - Total 0/4 0/4 0/4 0/12 agreement (0%) (0%) (0%) (0%)








Table 4-3

Summary of the Hypothesized Function ofBehaviors across Instruments in the School Environment for Brad

Behaviors FAO FAI MAS Functional Total


Noncompliance Disruption Aggression


Escape Escape Escape


Noncompliance, Escape disruption, and aggression Stereotypy Elopement



Off-Task Tantrum Aggression -


Attention Automatic Attention Automatic


Attention and

escape Attention and

escape Attention Attention and escape Attention Attention Attention and escape


Automatic Automatic


analysis

Escape and tangible Escape and tangible Escape and tangible



Escape and tangible


Flopping - Attention Total 4/8 2/6 0/4 6/18 agreement (50%) (33%) (0%) (33%)


agreement

1/4 (25%)


1/4 (25%) 2/5 (40%) 2/5 (40%)








Table 4-4

Summary of the Hypothesized. Environment for Drake

Behaviors FAO


Noncompliance Vocalization Tantrum





Disruption


Function of Behaviors across Instruments in the School


FAI


Escape Escape and

attention

Automatic Attention Attention Attention

and

escape

Attention -


MAS


Escape and tangible Automatic Tangible


Stereotypy Aggression


Automatic Attention and escape


Total 2/5 1/4 3/4 6/13 agreement (40%) (25%) (75%) (46%)


Total agreement

3/5 (60%)



2/3 (67%) 1/3 (33%)





0/2 (0%)


Functional analysis Escape



Automatic Tangible





Tangible and

escape







0% to 100%. The FAO agreed with the results of the functional analysis 8 out of 14 times (57%). The FAI agreed with the results of the functional analysis five out of eight times (63%). The MAS agreed with the results of the functional analysis two out of five times (40%). Table 4-5 presents a summary of the percent agreement between the hypothesized function of behaviors generated across instruments and participants and the functional analysis within the home setting. In comparison to the school setting, there was a larger percentage of function matches across instruments in the home setting. However, there were approximately one-half the number of individual or combined behaviors assessed in this home setting (27) in comparison to the school setting (43).

For Josh, the hypothesized functions of behaviors generated from the FAO, FAI, and MAS in the home environment agreed with the results of the functional analysis 7 out of 14 times (50%). Individually, the FAO agreed with the results of the functional analysis four out of seven times (57%) by correctly hypothesizing the automatic function of vocalizations and the escape function of noncompliance, disruption, and combined behaviors. The FAI agreed with the results of the functional analysis two out of four times (50%) by correctly hypothesizing the automatic function of vocalizations and the escape function of noncompliance. The MAS agreed with the results of the functional analysis one out of three times (33%) by correctly hypothesizing the tangible function of noncompliance. For Brad, the hypothesized functions of behaviors generated from the FAO, FAI, and MAS in the home environment agreed with the results of the functional analysis four out of six times (67%). Individually, the FAO agreed with the results of the functional analysis two out of four times (50%) by correctly hypothesizing the tangible function of disruption and combined behaviors. The FAI and MAS each agreed with the







Table 4-5

Summary of the Percent Agreement between the Hypothesized Functions of Behaviors Generated by Instruments within the Home Environment and the Functions of Behaviors Determined by the Functional Analysis

Home

Participants FAO FAI MAS Total agreement


Josh 57%(4/7) 50%(2/4) 33%(1/3) 50%(7/14) Brad 50%(2/4) 100%(1/1) 100%(1/1) 67%(4/6) Drake 67% (2/3) 67% (2/3) 0% (0/1) 57% (4/7) Total

agreement 57%(8/14) 63%(5/8) 40%(2/5) 56%(15/27)








results of the functional analysis one out of one time (100%) by correctly hypothesizing the escape function for noncompliance. For Drake, the hypothesized functions of behaviors generated from the FAO, FAI, and MAS in the home environment agreed with the results of the functional analysis four out of seven times (57%). Individually, the FAO and FAI each agreed with the results of the functional analysis two out of three times (67%) by correctly hypothesizing the escape function of noncompliance and the automatic function of vocalizations. The MAS agreed with the results of the functional analysis zero out of one time (0%). The MAS hypothesized automatic and attention as the function of noncompliance whereas the functional analysis determined escape to be the function. Tables 4-6, 4-7, and 4-8 present detailed summaries of the hypothesized function of behaviors generated across instruments and participants within the home setting.

When escape was identified as the function of the target behavior across all three participants, the FAO matched this function 67% of the time (4/6), the FAI correctly identified it 60% of the time (3/5), and the MAS correctly identified it 20% of the time (1/5). The FAO correctly hypothesized a tangible function 33% of the time (2/6). The MAS correctly hypothesized a tangible function once for Josh. The FAI never correctly identified a tangible function (0/6). The FAO and FAI correctly hypothesized the automatic function for Josh and Drake (2/2). In summary, these results are similar to the school setting and indicate that there was limited to no correct matching to the functions of behavior identified in the functional analysis in the home setting.

Across environments and participants, when escape was identified as the function of the target behavior, the FAO matched this function 75% of the time (9/12). The FAI matched the escape function 50% of the time (6/12) and the MAS matched the escape








Table 4-6

Summary of the Hypothesized Function of Behaviors across Instruments in the Home Environment for Josh

Behaviors FAO FAI MAS Functional Total


Vocalization Noncompliance Disruption





Noncompliance and disruption


Automatic Attention and Escape

automatic

Escape Attention and Tangible

escape


Attention and

escape Escape and attention


analysis

Automatic


Tangible and escape Tangible and escape


Tangible and escape


Total 4/7 2/4 1/3 7/14 agreement (57%) (50%) (33%) (50%)


agreement

2/4 (50%) 3/6 (50%) 1/2 (50%)





1/2 (50%)








Table 4-7

Summary of the Hypothesized Function of Behaviors across Instruments in the Home Environment for Brad

Behaviors FAO FAI MAS Functional Total


Noncompliance Disruption Aggression


Tangible Tangible Attention and

tangible


Escape


Escape


Automatic


analysis Escape Tangible Inconclusive


agreement

2/3 (67%)

1/1 (100%)


Off-Task Flopping Noncompliance, disruption, and aggression


Escape Escape


Tangible and

attention


Tangible and escape


Total 2/4 1/1 1/1 4/6 agreement (50%) (100%) (100%) (67%)


1/2 (50%)








Table 4-8

Summary of the Hypothesized Function of Behaviors across Instruments in the Home Environment for Drake

Behaviors FAO FAI MAS Functional Total


Noncompliance


Escape


Escape


Automatic and

attention


analysis Escape


agreement

2/3 (67%)


Vocalization Disruption Tantrum


Aggression


Automatic Attention Tangible


Automatic



Attention and

automatic


Tangible


Automatic Tangible Inconclusive


Attention and

tangible


Stereotypy


Automatic and

attention


Total 2/3 2/3 0/1 4/7 agreement (67%) (67%) (0%) (57%)


2/3 (67%) 0/1 (0%)







function 17% of the time (2/12). When tangible was identified as the function of the target behavior, the FAO and MAS matched this function 13% of the time (2/16). The FAI matched the tangible function 0% of the time (0/16). When automatic reinforcement was identified as the function of the target behavior, the FAO matched this function 100% of the time (3/3). The FAI matched the automatic function 67% of the time (2/3) and the MAS matched the tangible function 33% of the time (1/3).

In sum, there was a low level of agreement between the results of the functional assessment instruments and the functional analyses. Agreement between the functional analyses and the hypothesized functions of behaviors generated from all three instruments across settings averaged 39% (27 out of 70 times). Individually, agreement between the FAO and the functional analyses across settings was moderate, averaging 45% (14 out of 31 times). In general, functions of escape and automatic reinforcement were most often correctly identified by the FAO (75% and 100% respectively) across environments and participants. Agreement between the FAI and the functional analyses across settings was low, averaging 36% (8 out of 22 times). Agreement between the MAS and the functional analyses across settings was also low, averaging 29% (5 out of 17 times). On average, agreement among instruments and the functional analyses was higher in the home setting where there were fewer behaviors identified to evaluate.

Comparison of Functional Analyses across Environments

Functional analyses were compared across settings to determine if the function of behaviors differed across settings. The behaviors for which a functional analysis was conducted in both settings were compared. Only functional analyses in which the primary investigator was able to clearly identify a function were used. Tables 4-9, 4-10, and 4-11 present detailed summaries of the results of functional analyses across settings. A total of







nine individual behaviors across children were compared (three for Josh, two for Brad, and four for Drake). Two graphs of combined behaviors were also compared across settings. Results are presented individually and then summarized across participants.

For Josh, agreement on the function(s) of behavior across settings averaged 43% (3/7), all for the tangible function. At least one function of behavior across settings was identified for three of four target behaviors (not for vocalizations). The function(s) of behavior identified by individual graphs of behavior agreed across settings two out of five times (40%). The function(s) of behavior identified on combined graphs of behaviors agreed across settings one out of two times (50%). For Brad, agreement on the function(s) of behavior across settings averaged 67% (4/6), two times each for escape and tangible functions. The function(s) of behavior identified by individual graphs of behavior agreed across settings two out of four times (50%). The function(s) of behavior identified on combined graphs of behaviors agreed across settings two out of two times (100%). At least one function of behavior across settings was identified for three out of four target behaviors (not for aggression individually). For Drake, agreement on the function(s) of behavior identified by individual graphs of behavior across settings averaged 75% (3/4), one each for escape, tangible, and automatic functions. No combined graphs were analyzed. At least one function of behavior across settings was identified for three out of four target behaviors (not for tantrums).

In sum, agreement on the function(s) of individual and combined behaviors across settings averaged 59% (10/17). This indicates that functions of behavior differed across environments approximately 40% of the time. There was a higher level of agreement between functional analyses across settings when combined behavior graphs were







Table 4-9

Summary of the Results of Functional Analyses across Environments for Josh

Behaviors School Home Agreement Vocalization Tangible Automatic 0% (0/1) Disruption Tangible Tangible and 50% (1/2) escape

Noncompliance Tangible Tangible and escape 50% (1/2) Disruption and Tangible Tangible and escape 50% (1/2) noncompliance

Total individual and

combined behaviors 43% (3/7) Total individual

behaviors 40% (2/5) Total combined

behaviors 50%(1/2)








Table 4-10

Summary ofthe Results ofFunctional Analyses across Environments for Brad

Behaviors School Home Agreement Noncompliance Tangible and escape Escape 50%(1/2) Disruption Tangible and escape Tangible 50% (1/2) Aggression Tangible Inconclusive Noncompliance, Tangible and escape Tangible and escape 100% (2/2) disruption, and

aggression

Total individual and

combined behaviors 67% (4/6) Total individual

behaviors 50% (2/4) Total combined

behaviors 100% (2/2)







Table 4-11

Summary of the Results ofFunctional Analyses across Environments for Drake

Behaviors School Home Agreement Noncompliance Escape Escape 100% (1/1) Vocalization Automatic Automatic 100% (1/1) Tantrum Tangible Inconclusive Disruption Tangible and Tangible 50% (1/2) escape

Total individual

behaviors 75% (3/4)







compared. Agreement on the functions of individual behaviors across settings averaged 54% (7/13). Agreement on the functions of combined behaviors across settings averaged 75% (3/4). There was a combined total of 77% (10/13) of the time where at least one function identified was confirmed in each setting, indicating that most of the time at least one function of behavior was identified across both settings using the functional analysis. Across participants, the tangible function agreed across settings 60% of the time (6/10), the escape function 38% of the time (3/8), and the automatic function agreed across settings 50% of the time (1/2).

Comparison of Perceived Function and Actual Consequence to the Functional Analysis

The perceived function and actual consequence categories of the FAO were compared to the functional analyses within settings. Behaviors where a function was identified by the functional analysis and the FAO were used for comparison. A total of twelve behaviors across children were compared in the school setting (four for each child) and ten behaviors in the home setting (four for Josh, three for Brad, and three for Drake). Data will be discussed by summarizing all participants' data and then reporting individual participants' data by setting.

Comparison Between FAO Categories and the Functional Analyses within the School
Environment

For all participants within the school setting, the perceived function and actual

consequence categories agreed with the results of the functional analysis 21% of the time with agreement ranging from 0% to 50%. The perceived function category agreed with the functional analysis 6 out of 17 times (35%). The actual consequence category agreed with the functional analysis 1 out of 17 times (6%). The function of behavior was most often determined to be maintained by a tangible function (ten times) but neither category ever identified this function across participants. Escape was the next most frequent







function (six times) correctly identified five times by the perceived function category. Automatic reinforcement was identified once for Drake and identified by both categories correctly. Tables 4-12, 4-13, and 4-14 present detailed summaries of the percent agreement between the perceived function and actual consequences category on the FAO and the functional analysis within the school setting. For Josh, the perceived function and actual consequence categories did not agree with the results of the functional analysis. The tangible function identified by the functional analysis was not hypothesized as the function for any of the target behaviors on either category of the FAO. For Brad, the perceived function category agreed with the results of the functional analysis four out of eight times (50%) by correctly hypothesizing the escape function for four of the target behaviors (noncompliance, disruption, aggression, and combined behaviors). The actual consequence category agreed with the results of the functional analysis zero out of eight times (0%). The actual consequences category incorrectly hypothesized attention as the function of all four target behaviors. The functional analysis identified escape and tangible as the function of all four target behaviors. For Drake, the perceived function category agreed with the results of the functional analysis two out of five times (40%) by correctly hypothesizing the escape function of noncompliance and the automatic function of vocalizations. The actual consequence categories agreed with the results of the functional analysis one out of five times (20%) by correctly hypothesizing the automatic function of vocalizations. Tables 4-15, 4-16, and 4-17 present detailed summaries of the hypothesized function of behaviors generated by FAO categories and the functional analysis within the school setting.







Table 4-12

Summary of the Percent Agreement between the FAO Categories and the Functional Analysis within the School Environment for Josh


FAO


Behaviors Vocalization Disruption Noncompliance Vocalization, disruption,


Perceived function 0%(0/1) 0%(0/1) 0%(0/1)


Actual consequence

0% (0/1) 0% (0/1) 0%(0/1)


and noncompliance


Total


Combined

0% (0/2) 0% (0/2) 0% (0/2)


0% (0/1)

0% (0/4)


0%(0/1)

0% (0/4)


0%(0/2)

0% (0/8)







Table 4-13

Summary of the Percent Agreement between the FA 0 Categories and the Functional Analysis within the School Environment for Brad


FAO


Behaviors Noncompliance Disruption Aggression Noncompliance, disruption, and aggression


Perceived function

50% (1/2) 50%(1/2)

50%(1/2)

50%(1/2)


Actual consequence

0% (0/2) 0% (0/2) 0% (0/2) 0% (0/2)


Total 50%(4/8) 0%(0/8) 25%(4/16) Table 4-14

Summary of the Percent Agreement between the FAO Categories and the Functional Analysis within the School Environment for Drake

FAO

Behaviors Perceived function Actual consequence Combined Noncompliance 100% (1/1) 0% (0/1) 50% (1/2) Vocalization 100% (1/1) 100%(1/1) 100%(2/2) Tantrum 0%(0/1) 0%(0/1) 0%(0/2) Disruption 0%(0/2) 0%(0/2) 0%(0/4) Total 40%(2/5) 20%(1/5) 30%(3/10) Total across

participants 35%(6/17) 6%(1/17) 21%(7/34)


Combined 25% (1/4) 25% (1/4) 25% (1/4) 25% (1/4)







Table 4-15

Summary of the Hypothesized Function of Behaviors Generated by FAO Categories and the Functional Analysis in the School Environment for Josh

FAO


Behaviors Vocalization



Disruption





Noncompliance Disruption and noncompliance


Perceived function Attention



Attention





Escape Attention and escape


Actual consequence Functional analysis

Ignore Tangible (automatic)

Attention and Tangible ignore

(automatic)

Attention Tangible Attention and Tangible automatic








Table 4-16

Summary of the Hypothesized Function of Behaviors Generated by FAO Categories and the Functional Analysis in the School Environment for Brad

FAO


Behaviors Noncompliance Disruption Aggression


Perceived function

Escape Escape Escape


Actual consequence

Attention Attention Attention


Functional analysis

Escape and tangible Escape and tangible Escape and tangible


Noncompliance, Escape Attention Escape and disruption, and tangible aggression

Table 4-17

Summary of the Hypothesized Function of Behaviors Generated by FAO Categories and the Functional Analysis in the School Environment for Drake

FAO


Behaviors

Noncompliance Vocalization Tantrum Disruption


Perceived Function Actual Consequence

Escape Attention Automatic Automatic Attention Attention Attention Attention and ignore


Functional Analysis

Escape Automatic Tangible Tangible and escape







Comparison Among FAO Categories and the Functional Analysis within the Home
Environment

For all participants within the home setting, the perceived function and actual

consequence categories agreed with the results of the functional analysis 36% of the time with agreement ranging from 0% to 100%. The perceived function category agreed with the functional analysis 8 out of 14 times (57%). The actual consequence category agreed with the functional analysis 2 out of 14 times (14%). Escape and tangible functions were each identified five times with the perceived function category identifying the escape function four times and the tangible function once. Automatic reinforcement was identified twice and both categories correctly identified this function. Tables 4-18, 4-19, and 4-20 present summaries of the percent agreement among the perceived function and actual consequence categories on the FAO and the functional analysis within the home setting.

For Josh, the perceived function category agreed with the results of the functional analysis four out of seven times (57%) by correctly identifying the automatic function of vocalizations and the escape function of noncompliance, disruption, and combined behaviors. The actual consequence category agreed with the results of the functional analysis one out of seven times (14%) by correctly identifying the automatic function of vocalizations. For Brad, the perceived function category agreed with the results of the functional analysis two out of four times (50%) by correctly identifying the tangible function of disruption. The actual consequence category agreed with the results of the functional analysis zero out of four times (0%). The actual consequence category incorrectly identified attention as the hypothesized function of noncompliant and disruptive behavior. The functional analysis identified escape as the function of







Table 4-18

Summary of the Percent Agreement between the FAO Categories and the Functional Analysis within the Home Environment for Josh

FAO

Behaviors Perceived function Actual consequence Combined Vocalization 100% (1/1) 100% (1/1) 100% (2/2) Disruption 50%(1/2) 0%(0/2) 25%(1/4) Noncompliance 50% (1/2) 0% (0/2) 25% (1/4) Disruption and

noncompliance 50% (1/2) 0% (0/2) 25% (1/4) Total 57%(4/7) 14%(1/7) 36%(5/14) Table 4-19

Summary of the Percent Agreement between the FAO Categories and the Functional Analysis within the Home Environment for Brad

FAO

Behaviors Perceived function Actual consequence Combined Noncompliance 0% (0/1) 0% (0/1) 0% (0/2) Disruption 100%(1/1) 0%(0/1) 50%(1/2) Noncompliance,

disruption,

and aggression 50%(1/2) 0%(0/2) 25%(1/4) Total 50% (2/4) 0% (0/4) 25% (2/8)








Table 4-20

Summary of the Percent Agreement between the FA 0 Categories and the Functional Analysis within the Home Environment for Drake


Participant Noncompliance Vocalization Disruption Total


Perceived function

100% (1/1) 100%(1/1)

0% (0/1) 67% (2/3)


FAO

Actual consequence

0% (0/1) 100% (1/1)

0% (0/1) 33% (1/3)


Combined

50% (1/2) 100% (2/2)

0% (0/2) 50% (3/6)


Total across

participants 57% (8/14) 14% (2/14) 36% (10/28) Total across categories

and environments 45% (14/31) 10% (3/31) 27% (17/62)







noncompliance and tangible as the function of disruption. For Drake, the perceived function category agreed with the results of the functional analysis two out of three times (67%) by correctly identifying the escape function of noncompliance and the automatic function of vocalizations. The actual consequence category agreed with the results of the functional analysis one out of three times (33%) by correctly identifying the automatic function of vocalizations. Table 4-21, 4-22, and 4-23 present detailed summaries of the hypothesized function of behaviors generated by FAO categories and the functional analyses within the home setting.

Across settings, when escape was identified as the function of the target behavior across all three participants, the perceived function category matched this function 75% of the time (9/12). The actual consequence category never correctly identified the escape function of the target behaviors. When tangible was identified as the function of the target behavior across all three participants, the perceived function category matched this function 13% of the time (2/16). The actual consequence category never correctly identified the tangible function of the target behavior. When automatic reinforcement was identified as the function of the target behavior across all three participants, the perceived function category and actual consequence category matched this function 100% of the time (6/6).

Across environments, there was a low level of agreement between the FAO categories and the functional analyses (averaging 27%). Agreement between the perceived function category and the functional analysis was moderate, averaging 45%. Agreement between the actual consequences category and the functional analysis was low, averaging 10%. Agreement among the FAO categories and the functional analyses were higher in the home setting where there were fewer behaviors identified to evaluate




Full Text

PAGE 1

COMPARISON OF DESCRIPTIVE FUNCTIONAL ASSESSMENT INSTRUMENTS TO EXPERIMENTAL FUNCTIONAL ANALYSES FOR CHILDREN WITH AUTISM ANDREA CHAIT A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE I^EQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2002

PAGE 2

ACKNOWLEDGMENTS First and foremost, I would like to thank my dissertation chair, Dr. Jennifer Asmus, for her endless patience and unflagging support. I would also like to thank Drs. Maureen Conroy, Tina Smith-Bonahue, and Greg Valcante for their expertise and assistance on this research project. I thank the students and their families who donated their time to participate in this study. A special thank you goes to all of the members of the Autism Inclusion Project at the University of Florida who assisted in the data collection process. Also, I would like to thank my friends for believing in me and providing me with continuous encouragement throughout this process. Finally, I would like to thank my parents, Marvin and Rochelle Chait, for their love and support. This dissertation is dedicated to them. ii

PAGE 3

TABLE OF CONTENTS Eage ACKNOWLEDGMENTS ii LIST OF TABLES v LIST OF FIGURES viii ABSTRACT x CHAPTER 1 INTRODUCTION 1 Characteristics of Children with Autism 1 Educational Services for Students with Autism 3 2 REVIEW OF LITERATURE 10 Functional Assessment 10 Review of Functional Assessment Studies 21 Conclusion 38 3 METHOD 40 Participants 40 Settings and Therapists 41 Materials 42 Operational Definitions, Observation System, and Interobserver Agreement 47 Design and Data Analysis 57 4 RESULTS 60 Comparison of Descriptive Assessments to Functional Analyses 62 Comparison of Functional Analyses across Environments 75 Comparison of Perceived Function and Actual Consequence to the Functional Analysis 80 Summary 92 iii

PAGE 4

5 DISCUSSION Interpretation of Results 94 Limitations and Extensions 99 Summary APPENDIX A SAMPLE FUNCTIONAL ASSESSMENT OBSERVATION FORM 107 B SAMPLE FUNCTIONAL ASSESSMENT INTERVIEW FORM 109 C SAMPLE MOTIVATIONAL ASSESSMENT SCALE 119 D TARGET BEHAVIORS 123 E SUMMARY STATEMENTS FROM THE FUNCTIONAL ASSESSMENT INTERVIEW IN THE SCHOOL ENVIRONMENT 125 F SUMMARY STATEMENTS FROM THE FUNCTIONAL ASSESSMENT INTERVIEW IN THE HOME ENVIRONMENT 129 G FUNCTIONAL ANALYSIS GRAPHS 133 H FUNCTIONS OF BEHAVIOR IDENTIFIED BY INVESTIGATORS 146 I GRAPHS OF SCHOOL INTERVENTION WITH JOSH 150 REFERENCES 152 BIOGRAPHICAL SKETCH 159 iv

PAGE 5

LIST OF TABLES Table Eage 4-1 Summary of the Percent Agreement between the Hypothesized Functions of Behaviors Generated by Instruments within the School Environment and the Functions of Behaviors Determined by the Functional Analysis 64 4-2 Simimary of the Hypothesized Function of Behaviors across Instruments in the School Environment for Josh 66 4-3 Simimary of the Hypothesized Function of Behaviors across Instruments in the School Environment for Brad 67 4-4 Summary of the Hypothesized Function of Behaviors across Instruments in the School Environment for Drake 68 4-5 Summary of the Percent Agreement between the Hypothesized Functions of Behaviors Generated by Instruments within the Home Environment and the Functions of Behaviors Determined by the Functional Analysis 70 4-6 Summary of the Hypothesized Function of Behaviors across Instruments in the Home Environment for Josh 72 4-7 Summary of the Hypothesized Function of Behaviors across Instruments in the Home Environment for Brad 73 4-8 Summary of the Hypothesized Fvmction of Behaviors across Instruments in the Home Environment for Drake 74 4-9 Summary of the Results of Functional Analyses across Environments for Josh 77 41 0 Summary of the Results of Functional Analyses across Environments for Brad 78 4-11 Summary of the Results of Functional Analyses across Enviroimients for Drake 79 V

PAGE 6

4-12 Summary of the Percent Agreement between the FAO Categories and the Functional Analysis within the School Environment for Josh 82 4-13 Summary of the Percent Agreement between the FAO Categories and the Functional Analysis within the School Environment for Brad 83 4-14 Summary of the Percent Agreement between the FAO Categories and the Functional Analysis within the School Environment for Drake 83 41 5 Summary of the Hypothesized Function of Behaviors Generated by FAO Categories and the Functional Analysis in the School Environment for Josh 84 4-16 Summary of the Hypothesized Function of Behaviors Generated by FAO Categories and the Functional Analysis in the School Environment for Brad 85 41 7 Summary of the Hypothesized Function of Behaviors Generated by FAO Categories and the Functional Analysis in the School Environment for Drake 85 41 8 Summary of the Percent Agreement between the FAO Categories and the Functional Analysis within the Home Envirormient for Josh 87 4-19 Summary of the Percent Agreement between the FAO Categories and the Functional Analysis within the Home Environment for Brad 87 4-20 Summary of the Percent Agreement between the FAO Categories and the Functional Analysis within the Home Environment for Drake 88 4-21 Summary of the Hypothesized Function of Behaviors Generated by FAO Categories and the Functional Analysis in the Home Envirormient for Josh 90 4-22 Summary of the Hypothesized Function of Behaviors Generated by FAO Categories and the Fimctional Analysis in the Home Environment for Brad 90 4-23 Summary of the Hypothesized Function of Behaviors Generated by FAO Categories and the Functional Analysis in the Home Envirormient for Drake 91 E-1 Summary Statements from the Functional Assessment Interview in the School Environment for Josh 1 26 vi

PAGE 7

E-2 Summary Statements from the Functional Assessment Interview in the School Environment for Brad 127 E-3 Summary Statements from the Functional Assessment Interview in the School Environment for Drake 128 F-1 Summary Statements from the Functional Assessment Interview in the Home Environment for Josh 130 F-2 Summary Statements from the Functional Assessment Interview in the Home Environment for Brad 131 F-3 Summary Statements from the Functional Assessment Interview in the Home Environment for Drake 132 H-1 Reliability Data on Functional Analysis Graphs for Josh 147 H-2 Reliability Data on Functional Analysis Graphs for Brad 148 H-3 Reliability Data on Functional Analysis Graphs for Drake 149 ; « J.. : ' vii

PAGE 8

LIST OF FIGURES Table Eage G-1 Josh vocalization behavior across conditions at school 134 G-2 Josh disruptive behavior across conditions at school 1 34 G-3 Josh noncompliant behavior across conditions at school 135 G-4 Josh disruptive and noncompliant behavior across conditions at school 135 G-5 Josh vocalization behavior across conditions at home 136 G-6 Josh noncompliant behavior across conditions at home 136 G-7 Josh disruptive behavior across conditions at home 137 G-8 Josh disruptive and noncompliant behavior across conditions at home 137 G-9 Brad disruptive behavior across conditions at school 138 G-10 Brad noncompliant behavior across conditions at school 138 G-11 Brad aggressive behavior across conditions at school 139 G-1 2 Brad disruptive, noncompliant, and aggressive behavior across conditions at school 139 G-1 3 Brad disruptive behavior across conditions at home 140 G-1 4 Brad noncompliant behavior across conditions at home 140 G1 5 Brad aggressive behavior across conditions at home 1 4 1 G-1 6 Brad disruptive, noncompliant, and aggressive behavior across conditions at home 141 G1 7 Drake noncompliant behavior across conditions at school 1 42 G1 8 Drake tantrum behavior across conditions at school 1 42 viii

PAGE 9

1 G-19 Drake vocalization behavior across conditions at school 143 G-20 Drake disruptive behavior across conditions at school 143 G-21 Drake noncompliant behavior across conditions at home 144 G-22 Drake tantrum behavior across conditions at home 144 G-23 Drake vocalization behavior across conditions at home 145 G-24 Drake disruptive behavior across conditions at home 145 I-l Josh noncompliant and disruptive behaviors at school during treatment phases 151 1-2 Josh task accuracy at school during treatment phases 1 5 1 ix

PAGE 10

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 COMPARISON OF DESCRIPTIVE FUNCTIONAL ASSESSMENT INSTRUMENTS TO EXPERIMENTAL FUNCTIONAL ANALYSES FOR CHILDREN WITH AUTISM By ,1 Andrea Chait August 2002 Chair: Dr. Jennifer Asmus Major Department: Educational Psychology Individuals with autism often have problematic behaviors that hinder their success in the classroom. Successfully including students with autism who display problematic behavior in the general education setting requires appropriate interventions and support services. The Individuals with Disabilities Education Act Amendments of 1997 require the use of a functional assessment for students with disabilities who display significant problematic behavior. However, no consensus exists on the most appropriate functional assessment instrument to use in schools today. In addition, few studies have compared the results of different functional assessment instruments to functional analyses to determine their consistency. This study was conducted to determine the consistency of three commonly used descriptive fianctional assessment instruments: the Functional Assessment Observation form, the Functional Assessment Interview, and the Motivation Assessment Scale. Results from the three instruments were compared with results of an i X

PAGE 11

experimental functional analysis within settings. Functional analysis results were compared across settings to determine whether functions of behaviors differed across settings. In addition, the Functional Assessment Observation form was further explored to determine whether the "perceived functions" or the "actual consequences" category provided more useful information. Comparisons were conducted separately at school and at home to determine whether one or more of these descriptive assessment procedures can accurately identify the function of behavior within settings. Results suggest that these instruments have a low level of consistency when compared to the functional analysis. The Functional Assessment Observation form demonstrated a moderate level of consistency and is a better instrument than either the Functional Assessment Interview or the Motivation Assessment Scale for identifying the flinction of behavior. The perceived fiinction category on the Functional Assessment Observation form more accurately identified the function of behavior than did the actual consequences category. In addition, functions of behavior identified by the functional analyses could not be generalized across settings approximately half of the time. Functions across settings were usually the same for one function, but not for all of the functions identified. The implications for research and practice along with the limitations of the study are discussed. xi

PAGE 12

I CHAPTER 1 INTRODUCTION Characteristics of Children with Autism Autism is a developmemal disorder that is behaviorally defined. According to the most recent Diagnostic and Statistical Manual of Mental Disorders (DSM-IV; American Psychiatric Association, 1994), six broad categories of impairment must be met for the diagnosis of autism. First, the individual must display significant impairment in social interactions such as an inability to develop relationships with peers and an impairment in the ability to use nonverbal communication to assist in social interactions. Second, the individual must have impairments in communication such as a delay or lack of development of verbal language. Third, evidence of a markedly restricted, repetitive, and stereotyped pattern of behaviors, interests, and activities should be observed such as body movem mts or hand flapping. Fourth, a total of at least six of the twelve characteristics in the abo' e categories must be met. Fifth, the onset of symptoms must exist before the age of three. Sixth, the symptoms must not be better accounted for by Rett's Disorder, a condition occurring only in females that causes mental retardation, or Childhood Disintegrative Disorder, a condition in which there is a persistent and progressive loss of skills (American Psychiatric Association, 1994). Individuals with autism often have problematic behaviors that hinder their success in the classroom. There is a tendency for individuals with autism to engage in stereotyped, repetitive patterns of behavior (Bailey, Phillips. & Rutter, 1996). They often 1

PAGE 13

display abnormal preoccupations, adherence to nonfunctional routines or rituals, restricted interest patterns, stereotyped body movements, abnormal attachments to objects, and unusual responses to sensory stimuli (Bailey et al., 1996). Matson, Benavidez, Compton, Paclawskyj, and Baglio (1996) noted that the most commonly addressed problematic behaviors in individuals with autism include stereotypy, aggression, and self-injurious behaviors. Nevertheless, the problematic behaviors of children with autism vary widely from harmless deviations to severe destructive behaviors (Schopler &. Mesibov, 1994). Schopler and Mesibov suggested that a variety of behavioral problems emerge from frustrations due to difficulties understanding, communicating, and relating with others. Bailey et al. (1996) pointed out that individuals with autism often have an impaired ability to develop relationships through social interactions. Many individuals with autism lack social reciprocity and are deficient in many social skills. For example, they may act indifferent to people and respond in the same manner to strangers as they would to family members (Volmer, 1995). As a resuh, social relationships within the classroom are often impaired. Their speech and language skills may be severely delayed or absent. Even when language has developed, individuals with autism often have difficulty commimicating (Bailey et al., 1996). They are often unable to initiate or sustain a conversation with others. They may exhibit repetitive, stereotyped, or idiosyncratic language (American Psychiatric Association, 1994). Furthermore, many lack the ability to spontaneously engage in creative or pretend play (Bailey et al., 1996). Problematic behaviors in the classroom impose a tremendous challenge on the teacher, consume instructional time, and decrease the academic time for all students (Scott, DeSimone, Fowler, & Webb, 2000).

PAGE 14

s • Educational Services for Students with Autism Prevalence The American Psychiatric Association (1994) estimates that there are two to five cases of autism per 10,000 individuals. Other studies, using various definitions of autism, have found prevalence rates ranging from 33 to 160 per 10,000 individuals (Wing, 1997). Other prevalence estimates have indicated that autism may occur in as many as 1 per 1,000 individuals (Gresham & MacMillan, 1997). Many professionals believe that the incidence of autistic disorders is rising (Wing, 1997). However, this impression may be because of a greater awareness of the disorder or changes in the diagnostic criteria rather than an actual increase in the incidence of autism (Wing, 1997). Further studies are needed to evaluate this phenomenon. In either case, it is clear that the number of students identified as having autism has increased nationally. According to the U.S. Department of Education (1998), in 1992 there were 15,580 students classified with autism who were served under the Individuals with Disabilities Education Act (IDEA) and in 1997 there were 34,101. Several factors are believed to be responsible for the increase in the number of individuals identified with autism in the schools. First, advocacy groups lobbied for improvements in the identification of and services provided to students with autism. This resulted in the development of better assessment tools and support services for individuals with autism and their families. Secondly, in the last 20 years, many laws have been passed requiring public schools to provide educational services for students with severe disabilities (Reward, 1996). The Education for All Handicapped Children Act of 1975 was implemented and since has been revised and renamed the Individuals

PAGE 15

with Disabilities Education Act (IDEA) (Boomer & Garrison-Harrell, 1995). Initially, IDEA enabled students with autism to receive services under the category "Other Health Impaired." However, in 1990, Congress reauthorized IDEA and included autism as a separate category of individuals eligible for services. The unique needs of students with autism were recognized and warranted a separate categorization. Finally, there has been an increase in the number of students identified with autism or autistic-like characteristics. This may be because of the broadening of diagnostic categories such as Pervasive Developmental Disorder (PDD) to include Pervasive Developmental Disorder Not Otherwise Specified (PDD-NOS) as a label to describe individuals with symptoms of autism who do not meet the criteria for a diagnosis of autism or another type of disability (Simpson & Myles, 1998). Terminology has also changed. The five diagnostic categories that fall under the term PDD in the DSM-IV (Autistic Disorder, Asperger's Disorder, PDD-NOS, Child Disintegrative Disorder, and Rett's Syndrome) are now often referred to as Autism Spectrum Disorders (ASD). These disorders share clinical features but are distinct and separate from one another. In many instances, students classified with an Autism Spectrum Disorder are able to receive services under a variety of IDEA categories including autism. In the remainder of this paper, the term autism is used to include those individuals identified with ASD. Least Restrictive Environment IDEA (1997) mandates that special education and related services are provided to children with disabilities. There are ten diagnostic categories that allow for children to be eligible for services: mental retardation, hearing impairments, speech or language impairments, visual impairments, emotional disturbance, orthopedic impairments, autism.

PAGE 16

traumatic brain injury, specific learning disabilities, and other health impairments. In accordance with the federal law, the trend has been to serve students with disabilities in the least restrictive environment. The least restrictive environment refers to educating these students, to the maximum extent possible, with students who are not disabled (IDEA, 1997). This includes providing special services or instructional aids as necessary to educate students in the general education environment (IDEA, 1997). A student should only be removed from the general education environment when the student cannot benefit educationally from the services provided by the general education classroom (IDEA, 1997) . Most students identified with autism are educated in separate classrooms or separate schools (U.S. Department of Education, 1998). The behavioral problems displayed are often so severe that they interfere with the student's ability to learn in the general education environment (Eaves & Ho, 1997; Trevarthen, Aitken, Papoudi, & Robarts, 1996). Various professionals have different views on including students with autism in general education. Opponents of inclusion argue that there is a lack of scientific evidence documenting the benefits of serving students with autism in general education (Eaves & Ho, 1997; Simpson & Myles, 1998). However, proponents of inclusion reason that the segregation of students with disabilities into separate classrooms and schools lowers their self-esteem, hampers the development of their social skills, reduces their knowledge and skill attainment, and limits their instructional choices (Simpson & Myles, 1998) . The reauthorization of IDEA has increased the emphasis on including students with autism in general education classrooms (Simpson & Sasso, 1992). Congress stated that it is beneficial to educate students with disabilities in the general education

PAGE 17

6 environment (Community Alliance for Special Education (CASE) & Protection and Advocacy Inc. (PAI), 2000). The court system also has encouraged the integration of students with disabilities in the general classroom environment. Many federal court decisions regarding the placement of special education students have set the trend toward inclusion (CASE & PAI, 2000). The movement toward inclusion was advocated without much empirical evidence to support the benefits of inclusion (Eaves & Ho, 1997; Simpson & Sasso, 1992). One aspect most researchers agree on is that children with autism have special needs and are capable of making significant gains in their abilities with appropriate interventions (Eaves & Ho, 1997; Simpson & Sasso, 1992). Successfully including students with autism in the general education environment requires appropriate interventions and support services (Simpson & Myles, 1998). Structured educational programs have been found to be effective in furthering the skill development of individuals with autism (Happe & Frith, 1996). In addition, behavior management programs have been found to be effective in modifying problem behaviors (Happe & Frith, 1996). Because of the nature of autism and the wide range of complex behaviors displayed by individuals with autism, techniques for more individualized interventions are necessary and have been discovered through applied behavioral analysis research (Mace, Lalli, & Lalli, 1991). Behavioral Interventions Researchers have discovered that behavioral interventions are effective in reducing the problematic behaviors of individuals with autism (Mace et al., 1991; Matson et al., 1996). Over the past 30 years, researchers in the field of applied behavior analysis have focused on the use of behavioral assessments and treatments to ameliorate

PAGE 18

7 problematic behavior and to increase the academic and social skills of individuals with autism (Matson et al., 1996). Much of the applied behavioral research before the 1980s focused on the effectiveness of interventions without consideration of the variables responsible for behavior (Mace et al., 1991). Over the last 20 years, the behavioral literature has expanded to include methods for determining the variables promoting or maintaining behavior (Mace et al., 1991; Matson et al., 1996; Webby, Symons, & Hollo, 1997). Functional assessment is a technique that has been proven useful in identifying the variables influencing problem behavior (Wehby et al., 1997). Consequently, the IDEA amendments of 1 997 require school districts to conduct a functional assessment when behavior is a problem (McConnell, Hilvitz, & Cox, 1 998). As a result, school professionals frequently conduct functional assessments. The two main types of functional assessment methods used are descriptive assessments and experimental functional analyses (referred to throughout the remainder of the paper as functional analysis). Descriptive assessments include methods and instruments used to gather information regarding behavior such as behavioral observations, checklists, and interviews that are based on direct observations or information gathering in a natural context (no experimental manipulation). Information identified by these instruments is correlational in nature. Functional analyses involve the experimental manipulation of variables to validate hypotheses regarding the variables responsible for maintaining problem behavior. Information generated from a functional analysis leads to the identification of causal relationships that exist between environmental events and target behaviors.

PAGE 19

8 The most commonly used functional assessment instruments in schools are descriptive assessments (Heckaman, Conroy, Fox, &. Chait, 2000). Surprisingly, little research has been conducted on most of these instruments to determine their reliability and validity. The results of functional analyses have been proven to be usefiil in developing effective treatment interventions (Mace et al., 1991; Matson et al., 1996; Wehby et al., 1997). For functional assessment information to be useful and lead to effective intervention, it must be reliable and accurate. Reliability refers to the consistency of the measurement. When the instrument is administered repeatedly to the sample participant under the same conditions, if the instrument is reliable, the results should be the same. Validity refers to the degree to which the instrument measures the theoretical construct or trait it is designed to measure. Cook and Campbell (1979) describe validity as the best approximation of the truth. There is a great need for studies that assess the reliability and validity of functional assessment instruments. Chait (2001) compared the results of three functional assessment instruments (the Functional Assessment Observation form [FAO], the Functional Assessment Interview [FAI], and the Motivation Assessment Scale [MAS]) to each other to determine their reliability and validity. The results of that study indicated that the FAO and the FAI had a high level of interrater reliability (averaging 93% and 78% respectively). The results obtained by each instrument were not very stable across environments (i.e., home versus school). Agreement on the hypothesized functions of behaviors across environments was moderate on the FAO (averaging 62%) and low on the FAI (averaging 33%) and the MAS (averaging 22%). The relatively low to moderate level of stability across settings indicates that the hypotheses of the functions of behaviors generated in one environment

PAGE 20

(school) were different from the hypotheses generated in another environment (home). There was a moderate to low level of agreement on the hypothesized functions of behavior across instruments when examining them within and across settings, averaging 57% within, and 44% across settings, indicating a moderate level of construct validity. A major limitation of this study was that the descriptive instruments were compared to each other. A comparison of the descriptive assessments to a functional analysis was not conducted; therefore, there was no way to know which instrument provided the most valid results. A functional analysis allows for the identification of causal relationships between a behavior and its function (Amdorfer, Miltenberger, Woster, Rortvedt, & Gaffaney, 1994; Lerman & Iwata, 1993; Shriver, Anderson, & Procter, 2001). By comparing the hypothesized functions of behavior generated by descriptive assessments to the functions of behavior generated by functional analyses, the validity of the instruments can be assessed. There is a need for studies comparing these instruments to a functional analysis (Chait, 2001). The present study addresses this need by comparing the results of these three commonly used instruments to a functional analysis to determine the consistency of the functional assessment instruments. The following questions were addressed: • Within settings, how consistent are each of these instruments in identifying a hypothesized function of the target behavior in comparison to the function identified in the functional analysis? Are the functional analysis findings the same across settings? • Within settings, which category of the Functional Assessment Observation form (perceived functions vs. actual consequences) was most consistent with the results of the functional analysis and thereby provided the most useful information? • Based on current findings, is the use of a descriptive assessment appropriate to identify the ftmction of behavior in the general education and home environment? Which instrument or category of an instrument provides the most consistent findings in comparison to the functional analysis?

PAGE 21

CHAPTER 2 REVIEW OF LITERATURE Functional Assessment ^ ' Functional assessment methodology is used to determine the function of behavior (Carlson, Hagiwara, & Quinn, 1998; Fox, Dunlap, & Philbrick, 1997; Wehby et al., 1997). The term functional assessment refers to the procedures used to identify potential antecedents and consequences influencing behavior (Homer, 2000). Functional assessments attempt to identify relationships between problematic behaviors and envirormiental variables. Functional assessment techniques have been found to be powerful tools in determining appropriate interventions for individuals with autism (Matson et al., 1996). Functional based interventions directly address the purpose of the problem. The underlying assumption is that behavior is communicative and serves a function (Fox, Conroy, & Heckaman, 1998). Individuals engage in problematic behaviors because they enable the individual to gain attention or social contact to receive something that they need or want or to escape a task. Interventions are more effective if they relate to the function the problem behavior serves (Gable, 1996). With the knowledge of the function of behavior and the contingencies maintaining it, appropriate interventions can be developed that replace the problem behavior with an alternatively socially useful behavior, which serves the same function (Carr, Langdon, & Yarbrough, 1999; Fox et al., 1997). Four functions of behavior commonly studied in the literature are identified through fiinctional analysis: escape, attention, tangible, and sensory reinforcement (Carr 10

PAGE 22

11 et al., 1999). When an individual exhibits increased problematic behavior in response the presentation of demands, the function of that behavior is to escape. If an individual exhibits increased problematic behavior in response to the removal of attention, the function of that behavior is to gain attention. If an individual exhibits increased problematic behavior in response to restricted access to a preferred item or activity, the function of that behavior is to access tangibles. If an individual exhibits increased problematic behavior in the absence of social reinforcement (e.g., automatic reinforcement), the fimction of that behavior is automatic or sensory reinforcement. Functional analysis involves experimental manipulation of behavioral consequences to determine the function a behavior serves. For example, to determine if the fiinction of behavior serves to escape a task (negative reinforcement) then, during a functional analysis, researchers would present a demand for a task and then remove the task contingent on the occurrence of problem behavior (Carr et al., 1 999). Linking the function of problematic behavior to the intervention involves disrupting the maintaining contingency while increasing an alternative appropriate response (Mace & Roberts, 1993). According to Mace and Roberts, the way to achieve this is through schedules of reinforcement. Reinforcement is the contingency occurring after a behavior that increases the probability of the behavior occurring again in the future. Individuals have several response options or choices as to how they will respond in situations. The choice they make can be greatly influenced by changing the rate, quality, and timing of reinforcement. Interventions involve manipulating the schedules of reinforcement to encourage the individual to choose an alternative appropriate response to attain the same goal, while at the same time weakening the maintaining contingency of

PAGE 23

12 the inappropriate behavior. For example, if attention was found to be the function of a problematic behavior, a possible intervention would include reducing the amount of attention received after engaging in the problematic behavior, while providing increased attention for time intervals with no occurrence of the problematic behavior. If escape was found to be the function of a problematic behavior, a possible intervention would include preventing escape after the problematic behavior, while enabling the individual to request a break fi-om the task after a period of engagement. If accessing tangibles was found to be the ftmction of a problematic behavior, a possible intervention would include preventing access to the tangible after the problematic behavior, while providing access to the tangible for engagement in appropriate behavior. If sensory reinforcement was foimd to be the ftmction of a problematic behavior, a possible intervention would include reducing the response reinforcer relationship, while increasing engagement in alternative appropriate behaviors that provide sensory stimulation. Interventions may require multiple strategies and there are a variety of ways they can be approached. The key is to weaken the contingencies maintaining behavior while reinforcing alternative appropriate behaviors by first matching the treatment to the ftmction (Mace & Roberts, 1993). There are several advantages for using ftinctional assessments. Blakeslee, Sugai, and Gruba (1994) noted that functional assessment decreases the need for the use of punishment procedures and promotes skill building, promotes treatment to be derived based on hypotheses regarding the ftanction the behavior serves, increases the likelihood of generalization and maintenance of intervention effects, and contributes to our scientific knowledgebase regarding interventions.

PAGE 24

13 Clearly, functional assessments have proven useful for assessment and treatment. However, there is still ambiguity about how best to conduct a functional assessment. Generally, functional assessments fall under two broad categories: descriptive assessments and functional analyses. Descriptive Assessments Descriptive assessments include informal interviews, structured interviews, behavior ratings, and direct observations used to develop hypotheses about the variables related to the occurrence and maintaining function(s) of problematic behavior. Descriptive assessments fall under two categories: indirect and direct assessments. Indirect assessment involves the assessment of behavior at a different time or place from when the behavior actually occurred, such as rating scales or interviews (Shapiro & Kratochwill, 2000). Direct assessment includes assessments that occur (often in the natural environment) at the same time as the behavior, such as direct observation (Shapiro & Kratochwill, 2000). Each descriptive assessment method has its advantages and disadvantages as described below. Indirect assessments. Behavioral interviews include asking parents, teachers, or other informants about the problematic behavior and the variables associated with it. Interviews may be structured or unstructured but often include questions regarding the antecedents and the consequences of the problem behavior. Antecedents are the events occurring before a behavior that may serve as a stimulus that occasions the behavior given the consequences after the behavior (Schreibman, 1994). Consequences are the events occurring after a behavior that may serve to strengthen, maintain, or weaken a behavior (Mulick & Meinhold, 1994). The Functional Assessment Interview (O'Neill,

PAGE 25

.14 * ^. ^ Homer, Albin, Sprague, & Storey, 1997) is an example of a structured interview used to assess the function of behavior. It is a comprehensive interview that includes questions regarding the antecedents, consequences, and setting events surrounding problematic behavior. Setting events are events that may change the nature of the stimulus-response relationship (Carr et al., 1999). Interviews may be used alone or in combination with other assessment techniques. Interviews are usefiil for gathering initial information but have limitations. They have many of the same problems as other indirect data collection methods, such as observer bias, observer expectations, and false recollection of events (Lennox & Miltenberger, 1989; Repp & Homer, 1999). Therefore, it is recommended that the interview serve as a starting point in the functional assessment process (Lennox & Miltenberger, 1989). Rating scales, checklists, and other questionnaires provide a stmctured set of questions that help in the formulation of the hypothesis of the function of behavior. The Motivation Assessment Scale (Durand & Crimmins, 1988) is an example of a questionnaire used to assess the function of behavior. The MAS is a behavioral rating scale administered to parents and teachers that assesses four possible functions (escape, attention, tangibles, and sensory stimulation) of problematic behavior. Questionnaires such as the MAS provide specific information related to the behavior and are convenient to administer (Mace, 1994). However, similar to the behavioral interview, information may be biased by the informant's memory, expectations, or interpretation of events (Lennox & Miltenberger, 1989). Direct assessments. Direct observation of behavior, another type of assessment, reduces the biases associated with indirect data collection methods. However, these

PAGE 26

15 methods are difficult to standardize and provide only correlational data that must be interpreted carefully (Mace, 1994; Taylor, 1994). One type of direct observation method is called the antecedent-behavior-consequence (A-B-C) assessment (Bijou, Peterson, & Ault, 1968). In this type of assessment, the observer observes the antecedents and consequences directly surrounding the target behavior. This information enables the observer to develop hypotheses regarding the variables influencing behavior. When conducting this type of assessment, the observer must be trained to describe events accurately. Lermox and Miltenberger (1989) recommend a method of conducting an A-B-C assessment that is more rigorous. First, information is gained through the A-B-C method described above. Then, a direct observation system is developed where the antecedents, behaviors, and consequences are recorded in the natural setting. This type of a system will enable analysis of the frequencies of environmental events and their relationship to the target behavior. The Functional Assessment Observation (FAO) form developed by O'Neill et al. (1997) is another type of direct observation instrument. Similar to Lennox and Miltenberger' s method, it includes recording the antecedents and consequences associated with behavioral events. In addition, the FAO is designed to assess the possible fiinctions of behavior, co-occurring behaviors, the number of events of problem behavior, and the times when the problem behavior is more or less likely to occur. Functional Analyses Functional analyses involve the manipulation of variables to gather information and validate hypotheses regarding the function of behavior. Functional analyses are the most reliable and valid measures of behavioral function (Mace, 1994). This experimental

PAGE 27

methodology enables causal statements to be made about the function of behavior, whereas descriptive assessments only allow for descriptive or correlational statements (Gresham, Watson, & Skinner, 2001). Experimental methods are standardized procedures that isolate variables and control contingencies to determine effects on behavior. Functional analyses involve the examination of relationships between a person and environmental variables within a natural or analogue setting (Gable, 1996). An analogue setting refers to a setting outside of the natural environment that is designed to represent the natural environment (Carr et al., 1999). Iwata, Dorsey, Slifer, Bauman, and Richman (1982/1994) designed the first standardized method for conducting a functional analysis. Treatment failures for self-injurious behavior (SIB) were thought to be because of a lack of understanding of the variables responsible for maintaining or producing the behavior (Iwata et al., 1982/1994). Therefore, Iwata et al. (1982/1994) built on previous research to develop a systematic method for identifying the function of SIB. Nine participants with moderate to high rates of SIB were exposed to four, 15 minute, experimental conditions: social disapproval (attention), academic demand, unstructured play, and alone time. In the attention condition, the participant was placed in a room with toys. The participant was told to play with the toys while the experimenter engaged in some work. The experimenter only provided the participant with attention contingent on SIB. This condition was designed to determine if the participant engaged in self-injury to gain attention (Iwata et al., 1982/1994). In the academic demand condition, the participant was given a difficult academic task to complete. The experimenter used a three-prompt procedure to help the participant. Contingent on the occurrence of SIB, the experimenter

PAGE 28

17 would stop the academic session and turn away from the participant for 30 seconds. This condition was designed to determine if the participant engaged in SIB to escape or avoid difficult tasks (Iwata et al., 1982/1994). In the alone condition, the participant was placed in a room alone without any toys or materials. This condition was designed to determine if the participant engaged in SIB to access self-stimulation (Iwata et al., 1982/1994). In the unstructured play condition, the experimenter and participant were in a room with toys. The participant was allowed to play or move freely about the room. No demands were placed on the participant. When SIB did not occur, the experimenter provided the participant with positive attention and brief physical contact every 30 seconds. When SIB did occur, the participant was ignored unless the behavior became too severe, at which point the session was terminated. This condition served as the control condition. During this condition, the experimenter was present and provided attention, the participant had access to tangibles, there were no demands, and there were no consequences delivered for engaging in SIB (Iwata et al., 1982/1 994). The function of SIB was identified for seven of the nine participants. Self-injury was found to be at its highest level during the alone condition for four of the participants, suggesting that the fijnction of their behavior was to access self-stimulation (automatic reinforcement) (Iwata et al., 1982/1994). Two participants exhibited high levels of SIB during the high-demand situation, suggesting that the ftmction of their behavior was to escape the demand (negative reinforcement) (Iwata et al.). One participant engaged in high levels of SIB during the social disapproval condition, suggesting that the fimction of the behavior was to gain attention (positive reinforcement) (Iwata et al., 1982/1994).

PAGE 29

18 .A Results from this study suggested that the function of SIB varies among and within individuals (Iwata et al., 1982/1994). By conducting an experimental analysis, it is possible to identify variables associated with SIB before implementing a treatment and ultimately to develop more effective treatments (Iwata et al., 1982/1994). Since the development of the functional analysis methodology, it has been applied to a wide range of behavioral disorders (Mace, 1994). Derby et al. (1992) provided a descriptive summary evaluating the use of a brief functional analysis technique on 79 individuals with developmental disabilities who displayed aberrant behavior. Individuals ranged from age 1 to 32 years with most diagnosed as profoundly mentally retarded. A change in the rate of aberrant behavior or appropriate behavior was evident in 77% of the cases. During the brief functional analyses, 63% of the individuals displayed the targeted aberrant behaviors. The function maintaining the behavior was identified 74% of the time. When contingencies were provided for appropriate behavior, there was a reduction in the aberrant target behavior in 54% of the cases. In 84% of the cases, the manipulation of contingencies after the aberrant target behavior resulted in behavioral control. When appropriate behavior was the focus of assessment, 65% of the cases increased their rates of appropriate behavior. Results of this descriptive summary showed that a brief functional analysis was effective in identifying the function of behavior for most clients with developmental delays who engaged in high-frequency behaviors within the clinic setting. Iwata et al. (1994) conducted an epidemiological analysis of the results of functional analyses from 152 individuals with self-injurious behavior (SIB). The functional analyses were conducted when subjects were living in a residential facility or a

PAGE 30

19 pediatric hospital. Subjects ranged from age 1 to 51 years. Each subject was repeatedly exposed to a series of three to eight 15-minute conditions intended to identify the function of SIB. Three experimental designs were used to assess the influence of antecedents and consequences on behavior: a multielement, reversal, or combined design. Results showed that social-negative reinforcement (escape) was the function of SIB in approximately 38% of the individuals. Social-positive reinforcement (attention or tangible) was the function of SIB in approximately 26% of the individuals. Automatic reinforcement (sensory stimulation or pain attenuation) was the function of SIB in approximately 26% of the individuals. In approximately 5% of the cases, multiple functions were identified. No function was clearly identified in approximately 5% of the cases. Functional analysis methodology is not only useful in identifying the functions of behavior in individuals but may also allow the study of epidemiological data regarding the behavioral function of large groups of individuals (Iwata et al., 1994). For example, results of this study showed that social reinforcement was the function of SIB in approximately 67% of the sample. Therefore, SIB is primarily a learned disorder (Iwata et al, 1994). Individuals with SIB may not have the skills necessary to gain reinforcement through other means or their social environment may not respond to less severe forms of behavior (Iwata et al., 1994). Therefore, it is crucial that preventive strategies consist of providing individuals at risk for SIB with early language instruction (e.g., functional communication training) (Iwata et al., 1994). In addition, Iwata et al. emphasized the importance of replacing SIB with an alternative appropriate behavior serving the same function.

PAGE 31

20 After the ftmctional analysis, 121 subjects participated in a treatment program. The treatment program was designed to evaluate the effects of different interventions (Iwata et al., 1994). Overall, antecedent interventions such as extinction, differential reinforcement, and punishment, when designed for the fimction of the SIB, were effective in reducing SIB to below 10% of baseline in over 80% of the cases. When the fiinction of SIB was to obtain social-positive reinforcement, noncontingent attention (or access to tangibles) was effective in reducing SIB. Attention extinction, differential reinforcement of alternative behavior (DRO), and time-out were also effective. However, providing verbal reprimands and response interruption was not effective. When the function of SIB was to obtain social-negative reinforcement, noncontingent reinforcement (e.g., removal of task demand) was effective in reducing SIB. Escape extinction, differential reinforcement, and reducing the frequency of task presentation were also effective. Attention extinction and time-out had no effect on the rates of SIB. The presentation of additional demands contingent on SIB and response interruption only modestly reduced SIB. When SIB was maintained by automatic reinforcement, noncontingent reinforcement effectively reduced the levels of SIB. Sensory extinction, differential reinforcement, and response interruption also reduced SIB. Attention extinction and timeout were not effective. When SIB was maintained by multiple ftinctions or the function was not identified, noncontingent reinforcement was effective in reducing SIB. Attention extinction, verbal reprimands, and time-out were not effective. The treatment data supported the validity and utility of ftmctional analysis as a means to identify behavioral function (Iwata et al., 1994). Interventions designed to eliminate the maintaining contingency were more effective in reducing SIB than other

PAGE 32

21 types of interventions (Iwata et al., 1994). This study lends further support to the research already available that has shown interventions related to the function of behavior are more effective than interventions selected at random (Iwata et al., 1994). The main limitation of functional analyses is that, because of the nature of these types of assessments, they are often conducted in analogue settings. As a result, sometimes the findings may not be generalizable to the natural setting (Gresham et al., 2001; Repp & Homer, 1999; Gable, 1996; Mace, 1994). In addition, ftinctional analyses often require a large amoimt of time to conduct and may not be practical for implementation in classrooms (Conroy, Fox, Grain, Jenkins, «& Belcher, 1996; Gresham, etal., 2001; Mace, 1994). Review of Functional Assessment Studies Generally, the student's individual educational placement team is responsible for conducting the fiinctional assessment and implementing the behavioral intervention plan (McConnell et al.,1998). There are many ways of conducting a fiinctional assessment. Many researchers recommend that a multi-method approach be used when conducting a functional assessment (Crav^ord, Brockel, Schauss, &. Miltenberger, 1992; Davis, 1998; Gable, 1996; Homer, 1994; Lennox & Miltenberger, 1989; Symons, McDonald, & Wehby, 1998; Wehby et al., 1997). A multi-method approach includes gathering information from many different assessment sources. The resulting information should then be systematically analyzed and hypotheses developed. Interventions should be designed based on the results of the functional assessment. Wehby et al. (1997) suggest a three stage model as best practice in assessing problematic behaviors. First, information is gathered through verbal reports and

PAGE 33

22 interviews. This information is used to generate hypotheses regarding the function of the behavior. Second, direct observations in the natural setting are conducted to either support the hypotheses or provide new additional information. Third, a functional analysis is conducted. A functional analysis includes the manipulation of environmental variables to determine their effect on the problematic behavior. Functional analyses are rigorous and therefore tend to be difficult to implement in the classroom setting (Harding et al, 1999). As a result, some suggest it may be sufficient to focus on interviews and other descriptive data to design interventions rather than conduct a fimctional analysis (Amdorfer, et al., 1994; Mace, 1994; Wehby et al., 1997). However, others note that it may be difficult to identify fimctional relationships using descriptive assessments (Harding et al., 1999). Furthermore, descriptive approaches may be just as time consuming and complicated to administer as fimctional analyses (Iwata, 1994). Functional assessments can be conducted in a variety of ways using a variety of instruments. At present, there are no standardized procedures. Many of the functional assessment instruments in use today have yet to be studied in regards to their validity and reliability. Shriver et al., (2001) stated this is because of the differences between functional assessment and traditional psychological assessment. Traditional psychological assessment involves gathering information to determine if a particular pattern of behavior represents a hypothetical construct within the child. The construct (i.e., intelligence, attention, aggression) is hypothesized to be the cause of the relationship between the environment and behavior. In contrast, fimctional assessment is not concerned with evaluating whether or not a construct exists. Instead, functional assessment focuses on the behaviors that contribute to the construct. Functional

PAGE 34

23 assessment seeks to explain the function of the behaviors that make up a construct. The outcomes of these two approaches differ, which inevitably effects their evaluation (Shriveretal.,2001). When evaluating traditional psychological assessments, researchers look for •: evidence that the instrument consistently and accurately represents the construct of interest. In a functional assessment, hypotheses regarding the functional relationship are the outcome to be evaluated. These hypotheses are based on an interaction between the behavior of a child and the child's environment. In addition, these relationships can be directly observed, unlike psychological constructs. Establishing that a functional assessment is accurate lends evidence for its construct validity (Shriver et al., 2001). Construct validity, according to Messick (1995), includes all forms of reliability and validity evidence. Messick (1995) proposed that all forms of reliability and validity evidence contribute to the construct validity of an instrument. Therefore, the construct validation process may include the collection of many different types of reliability and validity evidence. The methods for evaluating the construct validity of traditional psychological assessments are applicable to the evaluation of functional assessment instruments (Shriver et al., 2001 ). The need to determine the reliability and validity of fimctional assessments has been clearly documented (Conroy et al., 1996; Davis, 1998; Fox et al., 1998; Gable, 1996; Iwata, 1994; Shriver et al., 2001). In addition, few studies have compared the results of functional assessment instruments to the results of functional analyses. Functional analyses are considered the most reliable and valid measure of functional relationships (Amdorfer et al., 1994; Lerman «& Iwata, 1993; Shriver et al., 2001). At

PAGE 35

24 present, the best way to evaluate the accuracy of functional assessment instruments is to compare them to functional analyses (Shriver et al., 2001). Research on the Functional Assessment Observation Form, Functional Assessment Interview, and the Motivational Assessment Scale Functional Assessment Observation Form . Very little research overall has been conducted on the FAO. Amdorfer et al. (1994) found agreement among a functional analysis and different descriptive assessment measures including the FAO. However, their use of the FAO form was described as a way to collect A-B-C data. It was unclear as to whether they used the FAO form in the way it was intended. Cuimingham and O'Neill (2000) conducted a study which compared a functional analysis to the FAO, the FAI, and the MAS. They reported overall agreement in the identified function of behavior across instruments and analyses. In addition, they found a high level of interobserver agreement on the FAO, the average ranging from 89% to 100%. Chait (2001) conducted a study that compared the FAO to the FAI and the MAS within and across envirormients. In the school and the home environment individually, there was a moderate level of agreement between the FAO and the findings of the FAI and the MAS (the FAO agreed with the findings of the FAI and the MAS 58% of the time with a range from 0% to 100%). There was moderate agreement (62%) on the hypothesized function of behavior generated by the FAO across environments. Across settings, the FAO at school agreed with the hypothesized function of behavior generated by the FAI and the MAS in the school and home environment 60% of the time with a range from 40% to 100%. The hypothesized function of behavior determined by the FAO in the home environment agreed with the hypothesized function of behavior generated by the FAI and the MAS in the school and home environment 53% of the time with a range from 0% to 100%. Interobserver agreement on the FAO was high (averaging 93% with a range from

PAGE 36

25 83% to 99%). Further studies evaluating the reliability and validity of the FAO are recommended (Amdorfer et al.; Chait, 2001; Cunningham & O'Neill, 2000; Fox et al., 1998). Functional Assessment Interview . The FAI generates a wealth of information regarding the variables influencing behavior. However, very little research has been conducted to determine the reliability and validity of this instrument (Amdorfer & Miltenberger, 1993; Fox et al., 1998; Sturmey, 1994). Amdorfer et al. (1994) conducted a study that included the comparison of the hypothesized functions generated from the FAI to a functional analysis. They found agreement on the function of behavior identified by the FAI and the functional analysis. In addition, inter-rater agreement on the hypothesized functions of behaviors generated by the FAI was 100%. Cunningham and O'Neill (2000) compared the hypothesized function of behavior generated from the FAI, MAS, and the FAO to a functional analysis. They found that the function of behavior identified by the FAI agreed with the results of the functional analysis. In addition, the FAI agreed with the function identified by the MAS and the FAO for two out of the three participants. They also found 100% inter-rater agreement on the FAI. Galensky, Miltenberger, Strieker, and Garlinghouse (2001) conducted a study that included a comparison of the hypothesized functions generated from the Functional Assessment Questionnaire of Mealtime Behaviors (FAQ), to the FAI, and an A-B-C checklist. These instruments were compared for three children with food refiisal behaviors aged 2 to 6 years. They found agreement on one function (escape) for all children across instruments. The FAI inter-rater agreement was 100% for the escape function, 67% for the attention function, and 0% for the tangible function for a combined agreement of 43%. The A-B-C direct observation descriptive assessment yielded consistent findings of a second

PAGE 37

26 function, attention, that was not identified by the other two indirect descriptive assessment instruments. Chait (2001) conducted a study that compared the FAI to the FAO and the MAS within and across environments, hi the school environment, there was a low level of agreement between the FAI and the findings of the FAO and the MAS (the FAI agreed with the findings of the FAO and the MAS 44% of the time with a range firom 0% to 100%). In the home environment, there was a low level of agreement between the FAI and the findings of the FAO and the MAS (the FAI agreed with the findings of the FAO and the MAS 35% of the time with a range from 0% to 50%). There was a low level of agreement (33%) on the hypothesized fimction of behavior generated by the FAI across environments. Across settings, the FAI at school agreed with the hypothesized function of behavior generated by the FAO and the MAS in the school and home environment 40% of the time with a range from 0% to 100%. The hypothesized function of behavior determined by the FAI in the home environment agreed with the hypothesized function of behavior generated by the FAO and the MAS in the school and home environment 42% of the time with a range from 20% to 80%. Interobserver agreement on the FAI was moderately high (averaging 78% with a range from 43% to 100%). The above studies on the FAI have shown moderately high to high levels of interrater agreement and some contradicting evidence regarding the construct validity of the instrument. Therefore, researchers strongly recommend that further studies be conducted to assess the reliability and validity of the FAI (Amdorfer & Miltenberger, 1993; Chait, 2001; Fox et al., 1998; Galensky et al., 2001; Sturmey, 1994). Motivational Assessment Scale . Durand and Crimmins (1988) created the MAS to provide an alternative assessment procedure to functional analyses. The MAS was created after extensive interviews with teachers, parents, and clinicians of students with

PAGE 38

27 developmental disabilities. To investigate the inter-rater and test-retest reliability of the instrument, Durand and Crimmins selected fifty students with developmental disabilities who engaged in self-injurious behaviors. To assess inter-rater reliability, the MAS was administered to the child's classroom teacher (primary rater) and the classroom paraprofessional (secondary rater). Results were compared fi-om the primary and secondary raters through the calculation of a pearson correlation coefficient on the raw scores, mean scores, and ranked scores. The correlations for raw scores ranged from .66 to .92, mean scores ranged from .80 to .95, and ranked scores ranged from .66 to .81. To assess test-retest reliability, the MAS was administered to the classroom teacher a second time, 30 days after the initial administration. Pearson correlation coefficients were calculated and the correlation between raw scores ranged fi-om .89 to .99, mean scores ranged from .92 to .98, and ranked scores ranged fi-om .82 to .99. Results suggested the MAS was reliable among raters and stable over time (Durand & Crimmins, 1988). A second study was conducted to determine the validity of the MAS by comparing the primary rater's ranking to the results of a fiinctional analysis. Eight participants were selected from the 50 subjects used in the reliability study. The subjects were exposed to five experimental conditions, three times, each lasting 10 minutes. The experimental conditions included a baseline, attention, escape, tangible, and unstructured condition. Correlations were calculated to compare the rankings on the MAS to the ranks generated from the functional analysis. Results suggest ratings on the MAS were highly predictive of the results from the fiinctional analysis with a correlation of .99. The MAS is the most extensively evaluated fiinctional assessment instrument (Sturmey, 1994). Several researchers including the authors of the MAS conducted

PAGE 39

28 reliability and validity studies and found the instrument to have high reliability and validity (e.g., Durand &. Carr, 1991; Durand & Crimmins, 1988; Durand, Crimmins, Caulfield, & Taylor, 1989; Iwata et ai., 1982/1994). Two studies focusing on the factor structure of the MAS scales found the scales to have high internal consistency (e.g., Bihm, Kienlen, Ness, & Poindexter, 1991; and Durand & Crimmins, 1988). However, five other studies found the MAS to have low inter-rater reliability, internal consistency, and validity (e.g., Crawford et al., 1992; Goza & Ricketts, 1993; Newton & Sturmey, 1991; Singh et al, 1993; Zarcone, Rodgers, Iwata, Rourke, & Sorsey, 1991). Several factors may be responsible for the lack of replication of the robustness of the MAS. The instrument was originally studied on a population of children with developmental delays who engaged in frequent self-injurious behaviors. The characteristics of the subjects, behaviors, or raters may have influenced the robustness of the instrument (Sturmey, 1994). However, the small number of items corresponding to each scale on the MAS, the limited number of consequences identified, or the lack of consideration of antecedents and setting events may have also contributed to the lack of replication (Sturmey, 1994). Chait (2001) conducted a study that compared the MAS to the FAO and the FAI within and across environments. In the school environment, there was a low level of agreement between the MAS and the findings of the FAO and the FAI (the MAS agreed with the findings of the FAO and the FAI 29% of the time with a range fi-om 0% to 100%). In the home environment, there was a moderate level of agreement between the MAS and the findings of the FAO and the FAI (the MAS agreed with the findings of the FAO and the FAI 50% of the time with a range fi-om 0% to 100%)). There was a low level of agreement (22%) on the hypothesized fimction of behavior generated by the MAS

PAGE 40

29 across environments. Across settings, the MAS at school agreed with the hypothesized function of behavior generated by the FAO and the FAI in the school and home environment 34% of the time with a range from 0% to 100%. The hypothesized function of behavior determined by the MAS in the home environment agreed with the hypothesized function of behavior generated by the FAO and the FAI in the school and home environment 37% of the time with a range from 0% to 100%. Researchers recommend that this instrument be used with caution and urge fiiture studies to continually assess the reliability and validity of this measure (Chait, 2001 ; Amdorfer & Miltenberger, 1993; Sturmey, 1994). Amdorfer et al. (1994) conducted a study that compared the hypothesized function of behavior generated by the MAS to a functional analysis. They found that the results of the MAS were not in agreement with the results of the functional analysis. Conversely, Curmingham & O'Neill (2000) found overall agreement on the function of behavior generated firom the MAS and the functional analysis. In general, few studies have compared the results of specific functional assessment instruments to the results of functional analyses to determine their validity. Those studies that do exist have found inconsistent results supporting the need for additional research in this area. Comparisons Across Descriptive Assessments and Functional Analvses Dunlap and his colleagues (1993) conducted a study on the use of functional assessment and functional analysis procedures, in the special education classroom, with five elementary students with emotional and behavioral disorders. Functional assessments were conducted to gather data and develop hypotheses. Researchers gathered information through record reviews, interviews (based on the FAI) with teachers, students, and other

PAGE 41

30 classroom staff, and direct observations. After hypotheses were developed, the researchers conducted ftinctional analyses in the classroom setting to test the hypotheses. The fimctional analyses confirmed the hypotheses developed. Behavior varied consistently in the direction predicted for each condition implemented. Results demonstrated the effectiveness of using functional assessment procedures with students with emotional and behavioral disorders in the classroom setting (Dunlap et al., 1993). However, the study did not analyze the reliability and validity of the individual ftmctional assessment instruments used. The accuracy of the functional assessment strategies implemented was verified through the ftmctional analyses. Still, questions remain as to which functional assessment instrument provided the most reliable and valid information or if a combination of instruments was necessary. Lalli, Browder, Mace, and Brown (1993) conducted two studies to assess the use of descriptive analysis in the classroom setting to decrease problematic behavior. In the first study, descriptive analyses were used to develop hypotheses regarding the function of students' problematic behavior. In the second study, the hypotheses were tested through a functional analysis and through analyzing the effects of treatment on the problematic behavior. Three students with severe and profound mental retardation, their teachers, and six instructional assistants participated in this study. The first study consisted of an assessment and the development of hypotheses regarding the function of problematic behaviors. Assessment consisted of a problem-identification interview, a scatter plot analysis (Touchette, Macdonald, & Langer, 1985), narrative recordings, and a descriptive analysis. The problem-identification interview assisted in gathering general information regarding the problematic behavior. The scatter plot analysis assisted in the

PAGE 42

31 identification of times when the problematic behavior occurred most frequently. The narrative recordings assisted in identifying the topography of the problematic behavior and the environmental antecedents and consequences associated with the problematic behavior. The descriptive analysis consisted of five hours of direct observation and a continuous 10-second partial-interval recording procedure. The descriptive analysis assisted in the identification of relationships between the problematic behavior and the environment. Based on the results of the descriptive analyses, hypotheses regarding the function of participants' behaviors were developed. In the second study, interventions were designed to indirectly test the accuracy of the hypotheses. Teachers were trained to block the reinforcement of problematic behavior, reinforce appropriate alternative behaviors, and teach an adaptive behavior. To assess the effects of interventions based on the hypotheses, a multiple baseline across students was used to conduct a component analysis. Different components of the intervention were removed at different times to observe their effects on behavior. A reversal design was used to test the hypotheses. In the reversal, teachers provided consequences contingent on problematic behavior (functional analyses). All three students showed significantly lower levels of problematic behavior during the intervention phases and higher levels during the reversal phases. The effectiveness of the interventions and the results from the functional analyses supported the use of descriptive analyses to develop accurate hypotheses regarding the function of behavior (Lalli et al., 1993). This study provides evidence that descriptive analyses, when conducted in this fashion, can provide reliable and valid results. However, there is a need for continued research directly assessing the reliability and validity of this and other functional assessment procedures.

PAGE 43

32 Amdorfer et al. (1994) conducted a study on the use of descriptive and functional analysis in the home environment. They found that with limited training parents were capable of taking part in the functional assessment process. Five children with behavioral problems and their mothers were included in this study. In the first phase of the study, the MAS, a behavioral interview (FAI), and a direct observation (FAO) of behavior were conducted. The data from phase one was used to generate hypotheses regarding the fiinction of the behavior. In the second phase of their study, a functional analysis was conducted. The results from the descriptive assessments and the functional analysis were generally in agreement except for the MAS. The primary functions of behaviors identified by the MAS were generally not in agreement with the other assessment techniques. The general agreement found among instruments lead the authors to suggest that when descriptive assessments provide clear results a functional analysis may not be warranted (Amdorfer et al., 1994). Furthermore, parents' ability to partake in the assessment process encouraged the use of these methods in the natural environment for practicality and useful purposes (Amdorfer et al., 1994). Sasso, et al. (1992) conducted a study to compare the results of a descriptive assessment and a functional analysis on two children with autism who were educated in a self-contained classroom. Each child's primary teacher participated with the assessment. Teachers were trained to conduct the assessment procedures. Before the teachers conducted the assessments, the investigator conducted the same assessments. Teachers conducted an A-B-C assessment and a classroom functional analysis. The hypothesized functions of behavior generated by the A-B-C assessment were in agreement with the function of behavior identified by the functional analysis for both children. The

PAGE 44

33 interventions developed based on the assessment results were effective in reducing targeted behaviors. Overall, these results suggest that descriptive assessments, such as the A-B-C assessment, may be used to accurately identify consequences maintaining behavior. Descriptive assessments may be a viable alternative when functional analyses are not possible (Sasso et al., 1992). Freeman, Anderson, and Scotti (2000) compared a structured descriptive assessment to an unstructured descriptive assessment and a functional analysis on two children with mental retardation. During the structured descriptive assessment, the environment was structured to increase the likelihood of environmental variables of interest occurring and thereby obtain a better sampling of behavior. The structured descriptive assessment occurred in the child's classroom and consisted of four different antecedent conditions (attention, demand, tangible, control). The structured descriptive assessment resulted in more frequent environmental events than the unstructured descriptive assessment. The results from the structured descriptive assessment were in agreement with the results of the functional analyses for both children. Results suggest that a structured descriptive assessment may be used to identify the maintaining consequences of behavior when a functional analysis is not feasible. Cunningham and O'Neill (2000) conducted a study to compare the effectiveness and efficiency of the MAS, FAI, and FAO. Subjects included three male children with autism who were educated in a self-contained classroom. The child's primary teacher and a paraprofessional were selected for participation with each child. Before any assessments, the teachers and assistants were asked why they thought the child engaged in challenging behavior. Next, the FAI was independently administered to the teachers

PAGE 45

34 and paraprofessionals. Afterwards, they were asked to generate hypotheses regarding the function of the child's behavior. If there were muhiple hypotheses, they were asked to rank them according to primary and secondary ftanctions. Later, responses documented by the interviewer were coded by two independent raters to determine if they were able to generate the same hypotheses regarding the function of behavior. Results were compared to determine inter-rater reliability of the instrument. Next, the MAS was administered to the teachers and paraprofessionals. The category scores on the MAS were averaged for each teacher/paraprofessional pair to determine the rank ordering of the functions of behavior. Teachers were then asked to collect direct observational data by using the FAO form. Rank orderings of the function of behavior were generated from this information. Finally, the authors conducted a functional analysis in analogue conditions. Interobserver agreement was collected on the descriptive observation sessions (averages ranged from 89% to 100%) and the analogue sessions (averages ranged from 98% to 100%) during 20% of the sessions. Results demonstrated overall agreement across methods in identifying behavioral functions. Mace and Lalli (1991) conducted a descriptive and functional analysis on an adult male with mental retardation. The descriptive assessment suggested two possible functions of his behavior (attention and escape). The functional analysis identified only one function of his behavior (attention). Mace and Lalli (1991) suggested that the results of the descriptive assessment were useful in designing and interpreting the fiinctional analysis. However, the descriptive assessment results by themselves may have lead to unnecessary treatment components and possibly even counterproductive treatment

PAGE 46

35 components (Mace & Lalli, 1991). Treatments were designed based on the results from the fiinctional analysis and found to be effective. Umbreit (1995) conducted a study on the use of functional assessments in the regular education setting on an eight-year-old boy with Attention Deficit Hyperactivity Disorder. After conducting an analogue ftinctional analysis, the fiinctions of the child's disruptive behaviors were identified. Negative reinforcement was found to be the fiinction of the child's disruptive behavior evidenced by the high rates of disruptive behavior occurring during the escape condition. After the functional analysis, the teachers, aides, and the student were given functional assessment interviews. In addition, A-B-C data was collected through direct observation. Based on the information obtained through the fiinctional assessments, two hypotheses were developed and tested. Results confirmed the hypotheses and an intervention was developed. The intervention was found to be successfiil in reducing disruptive behavior and increasing appropriate behavior. Each of the fiinctional assessment methods conducted assisted in providing information regarding the variables ftmctionally related to the child's behavior (Umbreit, 1995). Interestingly, the fiinctional analysis clearly identified escape as the fimction of the child's disruptive behavior. However, the other fiinctional assessments conducted identified escape and social attention as fiinctions of the child's disruptive behavior. Although this difference may have been due to the fact that the fianctional analysis was conducted in an analogue setting, it indicates the need to observe the child in the natural environment before designing interventions based on analyses done in an analogue environment (Umbreit, 1995). ft was not clear what element each fiinctional assessment method contributed or if all assessments were necessary (Umbreit, 1995). In addition, it

PAGE 47

36 was unclear if any of the assessment methods by themselves would have lead to a successful intervention (Umbreit, 1 995). Conroy et al. (1996) compared the use of analogue functional analyses to direct observations in predicting the fiinction of behavior. Their study was conducted on four male children with developmental disabilities and problematic behaviors. To determine which analogue probes to conduct, the authors conducted preliminary functional assessments including the FAI (O'Neill et al., 1997) and the MAS (Durand &. Crimmins, 1992). After the administration of these instruments, direct observations occurred in the classroom and the analogue probes were conducted. The analogue probes were successftil in identifying conditions that were motivating behavior in two of the four subjects. However, the results from the analogue probes on the two subjects were found to match the results from the direct observation on only one subject. The direct observation did not concur with the analogue probes. This raises questions as to the validity of analogue probes verses the validity of direct observations (Conroy et al., 1996). As a result, it is suggested that analogue probes be used with a battery of functional assessment techniques (Conroy et al., 1996). The authors point out the need to assess the reliability and validity of analogue probes and recommend that future research consider the best way to combine functional assessment techniques (Conroy et al., 1996). Crawford et al. (1992) compared the results of three functional assessment methods on four adults with severe to profound mental retardation who engaged in stereotypic behavior. They administered the MAS, followed by a functional analysis, and then engaged in direct observations using an A-B-C assessment. The A-B-C assessment and the MAS produced the clearest results and were in agreement with each other.

PAGE 48

37 However, the functional analysis produced more ambiguous results. For two out of the four participants, the functional analysis was not in agreement with the other methods. The authors suggested that the MAS and A-B-C assessments provide useful functional assessment data and that the relative ease of administration and interpretation of these assessment tools make them practical to use in an applied setting. However, it was noted that the inter-rater reliability on the MAS was not as high as expected. Also, there was no evidence as to the validity of the fmdings from these instnunents (Crawford et al., 1992). Crawford et al. (1992) stated that best practices would include the use of multiple functional assessment measures because of the lack of research on the reliability and validity of different functional assessment methods. The research assessing the validity of functional assessment instruments is limited and has produced mixed results. Fox and his colleagues (1998) conducted a review of the research on the use of functional assessments on students with or at-risk for emotional and behavioral disorders. They found that the most frequently used functional assessment techniques were direct observations and interviews. Most of the studies in this area used more than one functional assessment technique (Fox et al., 1998). Surprisingly, there was little data on the reliability and validity of the functional assessment instruments/techniques in use (Fox et al., 1998). Many researchers question the reliability and validity of indirect functional assessment procedures (e.g.. Fox et al., 1998; Davis, 1998; Symons, et al., 1998; Wehby & Symons, 1996; Gable, 1996; Iwata, 1994). Indirect functional assessment measures are likely to increase in popularity because they are less intrusive (Fox et al., 1998). Carr (1994) emphasized the need for researchers to develop descriptive procedures that provide results congruent to functional analyses for practical

PAGE 49

38 purposes. The need for reliability and validity studies on functional assessment techniques and instruments has been clearly documented (Shriver et al., 2001; Carr, 1994; Fox et al., 1998; Gable, 1996; Iwata, 1994). Conclusion There is a need for more studies comparing functional assessment instruments (Amdorfer & Miltenberger, 1993; Fox et al., 1998; Sturmey, 1994). It may be that each of these functional assessment instruments are usefiil in developing effective treatments or that a combination approach is necessary. Research needs to be conducted to determine the utility of different procedures to reliably and accurately assess the function of behavior (Amdorfer & Miltenberger, 1993; Cunningham & O'Neill, 2000). The research conducted on the MAS has provided inconsistent findings with most of the research documenting the need for future studies in this area. Clearly, there is a lack of research on different functional assessment instruments such as the FAO and the FAI developed by O'Neill et al. (1997). The study conducted by Chait (2001) was the most comprehensive analysis done on these instruments to date. Although the findings were useful, they were a comparison across descriptive assessments only. The best measure of an instruments ability to identify ftinctions of behavior would be to compare the instruments results with a fiinctional analysis. Few studies have compared the results of different functional assessment instruments to a functional analysis and several authors have documented the need for fiiture studies in this area (Amdorfer & Miltenberger, 1993; Chait, 2001; Cunningham & O'Neill, 2000; Fox et al., 1998; Gable, 1996; Iwata, 1994; Sturmey, 1994). The present study sought to advance the scientific knowledgebase regarding functional assessments by comparing the results of three commonly used

PAGE 50

39 instruments, the FAO, FAI, and MAS, to a functional analysis to determine their consistency. The present study expanded the research conducted by Cunningham and O'Neill. Specifically, the study was conducted on three children with autism in an inclusive setting. Functional assessments were conducted in both the school and the home environment with teacher and parent participation. Results from the individual instruments were then compared to a functional analysis. In addition, the FAO form was further analyzed to determine which category (predicted function vs. actual consequences) lead to the most consistent results. The following questions were addressed: • Within settings, how consistent are each of these instruments in identifying a hypothesized function of the target behavior in comparison to the function identified in the functional analysis? Are the functional analysis findings the same across settings? • Within settings, which category of the Functional Assessment Observation form (perceived functions vs. actual consequences) was most consistent with the results of the functional analysis and thereby provided the most useful information? • Based on current findings, is the use of a descriptive assessment appropriate to identify the function of behavior in the general education and home environment? Which instniment or category of an instrument provides the most consistent findings in comparison to the functional analysis?

PAGE 51

CHAPTER 3 METHOD Participants Three 1 st grade children who were classified with autism and exhibited challenging behaviors participated in this study. To be included each child had to meet four criteria. First, the child had to be diagnosed with autism/autism spectrum disorder. Second, the child must have demonstrated challenging behaviors. Third, the child must have been included in general education at least 50% of the time. Fourth, consent to participate in this study was sought from the school principal, teacher, and parents. The following is a brief description of each child. Josh was a 7-year-old Caucasian male with a diagnosis of autism/autism spectrum disorder. His problematic behaviors in the classroom included inappropriate vocalizations, disruption, noncompliance, and off-task behaviors. In the home environment. Josh's parents reported concern with his inappropriate vocalizations and noncompliant behavior. Intellectual assessments determined Josh was functioning within the below average range with a composite score of 78 on the Kaufman Assessment Battery for Children. Josh was enrolled in a 1st grade general education classroom for most of the day. He received special education in reading and writing 1 to 5 hours a week. Josh also received speech and language services three times a week for 30 minutes and occupational therapy once a month for 30 minutes. 40

PAGE 52

41 Brad was a seven-year-old Caucasian male with a diagnosis of autism and a speech and language impairment. Brad's problematic behaviors in the classroom included disruptions, elopement, noncompliance, off-task behavior, and aggression. In the home environment. Brad's parents reported the same behavioral concerns except for his off-task behavior. Intellectual assessments determined Brad was functioning within the below average range with a composite score of 78 on the Comprehensive Test of Nonverbal Intelligence. Brad was enrolled in a 1 st grade general education classroom for most of the day and had a full time classroom aide. He received special education in reading and writing one to five hours a week. Brad also received speech and language services three times a week for 30 minutes and occupational therapy once a month for 30 minutes. Drake was a 7-year-old Caucasian male with diagnoses of autism and language impairment. Drake's problematic behaviors in the school and home environment included noncompliance, vocalizations, stereotypy, disruptions, aggression, and tantrums. Intellectual assessments determined Drake was functioning within the below average range with a composite score of 77 on the Stanford-Binet IV. Drake was enrolled in a 1 st grade general education classroom for most of the day. A general education and special education teacher were present in the classroom full time. He received speech and language therapy and occupational therapy on a weekly basis. Drake was on several medications for behavior problems and seizures including: Depakote, Resperdol, and Zoloft. Settings and Therapists The investigation was completed in two phases. During Phase I, a series of descriptive assessments were conducted in each child's general education classroom and

PAGE 53

42 various areas in the home (e.g., living room, bedroom, kitchen, etc.). During Phase 2, a functional analysis was completed in an empty classroom in each child's school setting and in the living room of each child's home. All instruments and analyses were completed across two settings, the school and home. The FAO was conducted by a trained investigator through direct observations in the school and home setting. The FAI and MAS were administered to the general education teacher in the school setting and to the parent in the home setting by a trained investigator. The functional analyses were conducted by a trained investigator in an empty classroom in the school during the school day and by a parent in the home setting. Materials Descriptive Assessment Measures The Functional Assessment Observation form (FAO) developed by O'Neill et al. (1997) is a structured method of direct observation of behavior over time/days. It was designed to assist an observer in the identification of the antecedents, consequences, possible fimctions, co-occurring behaviors, number of events of the problem behavior, and times in which the problem behavior is more or less likely to occur (O'Neill et al., 1997). The FAO form consists of eight sections that are to be completed while directly observing an individual. In the first section, identification information is recorded, such as the name of the individual being observed and the observation date. The second section is separated into time blocks to record the activity, setting, and time interval of the observation. The third section is composed of the target behaviors that have been identified for observation. The fourth section is made up of events that are considered potential predictors of the target behavior. The fifth section contains perceived functions of the target behavior. The sixth section is used to record the consequences of the target

PAGE 54

43 behavior. The seventh section provides a commentary section for events occurring during a given block of time. The eighth section allows for the tracking of behavioral events daily and in total. The FAO was completed according to the instructions provided by the ^ FAO manual developed by O'Neill et al. (1997). See Appendix A for a sample of this form. : ' The Functional Analysis Interview Scale (FAI) (O'Neill et al., 1997) is a structured interview that takes approximately 45 to 90 minutes to administer. It is designed to collect information about the individual and the events influencing his/her problem behavior. The FAI is divided into 1 1 sections. In the first section, problem behaviors are clearly described. For each target behavior, the interviewee is asked to describe its topography (how the behavior was performed), frequency (how often the behavior was performed), duration (how long the behavior lasted), and intensity (how damaging was the behavior). In the second section, potential setting events are identified. The interviewee is asked questions regarding the child's medications, medical conditions, sleep patterns, diet, and other similar types of questions. In the third section, questions are designed to identify antecedent events that do or do not occasion problem behavior. For example, the interviewee may be asked the time of day in which the behaviors are most and least likely to happen. The fourth section consists of questions aimed at identifying the consequences of the problematic behavior. The interviewee is asked to identify particular situations in which the behavior occurs and what the child gains or avoids by engaging in the behavior. The fifth section contains questions designed to uncover the efficiency of the problematic behavior. Interviewees are asked to rate the efficiency of the child's behaviors (how effective they are in obtaining desired outcomes) on a 5-point Likert scale (1 = low efficiency and 5 = high efficiency). The sixth section helps to

PAGE 55

44 identify alternative behaviors in the individual's repertoire. Interviewees are asked "What socially appropriate behaviors or skills can the person already perform that may generate the same outcomes or reinforcers produced by the problem behaviors?" (O'Neill et al., 1997, Appendix B, p. 6). The seventh section contains questions regarding the individual's communicative abilities. For example, the interviewee is asked "What are the general expressive communication strategies used by or available to the person?" (O'Neill et al., 1997, Appendix B, p. 6). The eighth section explores general information about the types of teaching strategies and activities that may or may not work well with the individual. For example, "What things can you do to improve the likelihood that a teaching session or other activity will go well with this person?" (O'Neill et al., 1997, Appendix B, p. 7). The ninth section consists of questions to identify objects, activities, and events that are reinforcing to the individual. The interviewee is asked to list particular food items, toys, activities, and any other things that are reinforcing to the child. The tenth section is designed to gather information on programs or interventions previously attempted. The interviewee is asked how long the behavior has been a problem, what programs have been attempted to decrease the behavior, and the resulting effects of those programs. Section eleven is a form for the interviewer to summarize data by listing the major antecedents, consequences, and setting events surroimding behavior identified through the interview process. See Appendix B for a sample of this form. The Motivation Assessment Scale (MAS) (Durand & Crimmins, 1988) is a 16item questionnaire designed to provide information regarding four hypothesized function(s) of the target behavior (Durand & Crimmins, 1988). The questions on the MAS describe a variety of situations in which the target behavior may occur. The likelihood of the target behavior occurring is then rated on a 7-point Likert scale (0 =

PAGE 56

45 Never and 6 = Always). An example of an item on the MAS is, "Does the behavior occur following a command to perform a difficuh task?" (Durand & Crimmins, 1988, p. 102). Each question on the MAS assesses the influence of one of four potential functions of the behavior: escape, attention, tangibles, and/or sensory stimulation. There are four questions to assess each of the four variables. After the scale is completed, the four questions relating to each of the four variables are averaged. High scores indicate that the variable may be maintaining the behavior. Refer to Appendix C for a sample of the MAS form. Training. Investigators were trained on the administration and scoring of the instruments to ensure reliable data. The two primary investigators had extensive experience in using the FAO, FAI, and the MAS. Additional raters were trained to conduct the FAO and the FAI. In addition to reading the directions from the handbook developed by O'Neill et al. (1997), raters attended a short workshop conducted by a professional experienced in using the FAO and FAI. Practice sessions were conducted for the FAO in the general education setting. Two raters independently observed a participant at the same time and collected FAO data. Practice sessions were conducted twice a week for a total of 10 hours until the raters reached at least 80% agreement. Raters were trained to administer and score the MAS. Training involved reading the manual and practicing scoring until reaching 100% agreement which occurred in one session. Functional Analysis The fimctional analysis consisted of five conditions: attention, escape, tangible, free play, and ignore. During the attention, tangible, and free play conditions, preferred items were used. Preferred items were identified by parent and teacher reports as well as

PAGE 57

46 a preference assessment. A preference assessment was conducted at the beginning of each session using procedures similar to those used by DeLeon and Iwata (1996) and involved providing the child with several items for a period of 5 minutes. The item the participant played with the most during the allotted time period was identified as the preferred item. For Josh, the most highly preferred items at school included books, puzzles, handheld electronic games, and toy cars. At home, the most highly preferred items were video games, books, and toy soldiers. For Brad, the most highly preferred items at school included books, toy cars, and toy soldiers. At home, the most highly preferred items included books, videos, and action figure dolls. For Drake, the most highly preferred items at school and home included books, toy trains, and balls. During the escape condition, nonpreferred tasks were used. Nonpreferred tasks were identified by parent and teacher reports and included tasks that the child was able to perform but that were difficult for the child or occasioned problematic behavior. For Josh, the task was to write his spelling words three times each. For Brad, single-digit math problems were used. For Drake, the task at school was to read words and the task at home was to write letters. Training. For each child, the functional analysis was conducted by an investigator experienced in conducting functional analyses. Additional investigators were trained through a combination of direct instruction and assigned readings. After training was complete, investigators practiced coding behaviors using videotaped examples of the participants' behaviors. Investigators practiced two hours a day twice a week for a total of 10 hours, until reaching 80% reliability using a 10 second agreement window.

PAGE 58

47 Operational Definitions. Observation System, and Interobserver Agreement Operational Definitions Seven behaviors were targeted across both phases of assessment during the study. As previously described, each child engaged in one or more of these behaviors. Not all children engaged in each of the behaviors defined and not all behaviors were assessed by each instrument across both settings. Disruption was defined as verbal talk which was out of context (e.g., above speaking volume) or without permission, making noise by pounding on the tables, walls, or floors with a closed fist or object, throwing objects (not at another person), ripping or tearing objects, writing on inappropriate objects, screaming or yelling, spitting (not at another person), playing with materials or toys inappropriately, trying to escape work situations, throwing work materials, yelling at others, verbal threats, and inappropriate manding (Josh, Brad, and Drake). If disruption occurred, it was recorded singly or in combination with other behaviors. Vocalizations were defined as noises, words, or phrases made by the participant that did not have to do with the specific task at hand (Josh and Drake). Noises included nonsense vocalizations, singing, humming, or talking to self Words and phrases were in the form of echolalia, which is the verbal repeating of what had been said by others. Echolalia was scored if it occurred immediately or delayed. Delayed echolalia involved children repeating words or phrases previously heard. Noncompliance was defined as the failure to complete an instruction or to begin followdng instructions five seconds after the request was given (Josh, Brad, and Drake). Off-task behavior was defined as the participant facing away from the task/material/instructor or not engaging in the instructional activity when directed (Josh and Brad). Aggression was defined as hitting, biting, pinching, kicking, pulling another person, throwing objects, spitting, or pushing one's body-part or body into another person

PAGE 59

48 (Brad and Drake). Stereotypy was defined as repetitive behaviors such as rocking, hand flapping, arm flicking, spinning (object or body), covering ears, finger wringing, and staring at fingers while wiggling them (Drake). Tantrums were defined as screaming and crying or multiple topographies of behavior occurring at once (Drake). In addition, tantrum included dropping to the ground (either on knees, bottom, or back) and remaining there. Refer to Appendix D for an outline of the targeted behaviors. Observation System Descriptive assessments. Before administering the descriptive assessments, target behaviors were selected and operational definitions were developed by the investigators. The three descriptive assessment measures were conducted independently of each other. During the FAO, participants were directly observed in the classroom and home environment by a trained investigator using the FAO form for five hours in each setting across a 4 week period. Hypotheses regarding the fimction of the participant's behavior were developed by comparing the number of behavioral events out of the total that were coded in each of the perceived fimction categories. Two trained investigators were present for the administration of the FAI. In the school environment, the primary interviewer was the same for all three children. The second interviewer was the primary investigator for each of the individual children. In the home environment, a different primary interviewer conducted the interview for all three children. The second interviewer remained the same for all three children. The FAI was administered to parents and teachers separately but within the same week. Questions were delivered orally by the primary interviewer. Responses to the questions were recorded in vivo directly on the FAI form separately and independently by the two investigators. After the interview, the investigators developed separate and independent

PAGE 60

49 summary statements about the problem behaviors and generated one or more hypothesis(es) regarding the functions of behavior. These hypotheses were then compared across raters to determine the inter-rater reliabihty of the instrument. Before completing the MAS, parents and teachers were given a brief explanation of the purpose of the MAS and the operational definition of the targeted behaviors. For all children, more than one behavior was identified and the MAS was completed for each of those behaviors individually. For Josh, three behaviors were identified at school (vocalization, disruption, and noncompliance) and two behaviors at home (vocalization and noncompliance). For Brad, three behaviors were identified at school and at home (disruption, noncompliance, and aggression). For Drake, three behaviors were identified at school (vocalization, noncompliance, and tantrum) and two at home (noncompliance and tantrum). The MAS was administered separately to the teacher and parent of each participant by a trained investigator within a one week time period. Scores were averaged for each of the four variable categories. The highest scoring variables were hypothesized to be maintaining the behaviors. Functional analyses. After the completion of the three descriptive assessments, fimctional analyses were conducted. All of the children participated in a series of sessions using procedures similar to those described by Iwata, et al. (1982/1994). Sessions were five minutes in length and involved the following conditions: contingent attention, contingent escape, contingent tangibles, free play, and ignore. For two of the children (Josh and Brad), additional sessions were conducted to analyze an additional condition of ignore with access to tangibles. These sessions were conducted to determine if problematic behavior during the ignore condition was attenuated when allowed access to tangibles. It was a test to further determine if behavior would be maintained without

PAGE 61

50 social consequences. If the behavior ceased with the introduction of tangibles during the ignore condition, the behavior was determined to have a social function (e.g., not identified as an automatic function). During the attention condition, the child was instructed to play alone while the adult was present in the room (e.g., engaged in paperwork). The child had access to preferred toys. The adult diverted his/her attention away from the child. Attention was provided to the child for 1 5 seconds for each occurrence of problematic behavior. Attention was in the form of statements of disapproval (e.g., "Don't do that"). This condition was designed to determine if problem behavior was maintained by positive reinforcement in the form of social attention. In the tangible condition, a preferred tangible item was provided to the child for 1 minute before beginning the condition. Next, the adult removed access to the tangible item by stating "my turn" and the item was then visible to the child but out of reach. The child was provided with attention from the adult at least every 15 seconds and no task demands were present. The tangible item was returned to the child for 1 5 seconds for each occurrence of problematic behavior. After 15 seconds the tangible item was removed. This condition was designed to determine if problem behavior was maintained by positive reinforcement in the form of tangibles. In the escape condition, the child was given an academic task to complete and told "It's time to work". The task was removed for 15 seconds for each occurrence of problematic behavior. This condition was designed to determine if problematic behavior was maintained by negative reinforcement. During the free play condition, the child had access to preferred items and attention was provided by the adult at least every 1 5 seconds in the form of neutral statements (e.g., "You are playing with a car."). Problematic behavior was ignored. It simulated an enriched envirormient and no demands

PAGE 62

51 were presented. This condition was a control condition and problem behavior was not expected to occur. During the ignore condition, the room was cleared of all toys and materials and the child was left with "nothing" to do. All behavior was ignored. This condition was designed to determine if the child's problem behavior was maintained in the absence of social consequences (automatic reinforcement). The ignore with tangibles condition was included if high rates of problematic behaviors occurred during the ignore condition. For Josh, this condition was added in the school environment only. For Brad, this condition was added in the school and home environment. The ignore with tangibles condition was the same as the ignore condition previously described with one exception, the addition of one preferred item. This condition was designed to determine if high rates of problem behavior in the ignore condition occurred because the behaviors were maintained by automatic reinforcement or because there was nothing else to do and behaviors escalated to gain access to items. The order of the sessions was randomly selected for each child and began in the school environment. The sessions were videotaped and later coded on palm-top computers to identify the occurrence of problematic behaviors. A computerized program entitled the Multiple Option Observation System for Experimental Studies2 (M00SES2) (Tapp, Wehby, & Ellis, 1995) was used to record the problematic behaviors in a direct, continuous, and sequential manner. M00SES2 was used to calculate the frequency of problematic behaviors in each condition. Frequency data were converted to rate per minute. The rate of problematic behaviors in each condition was graphically displayed and visual inspection was used to determine when stability in the data was achieved. Behaviors were graphed individually and some were combined. Behaviors were only combined when it was hypothesized that they were in the same response class and they

PAGE 63

52 served the same flinction(s). Therefore, not all children's behaviors in each setting were combined. To determine the function of behavior, two independent observers visually inspected the graphical display of data. The flinction(s) of behavior determined by the primary investigator were used for comparison. Only behaviors for which a function(s) was clearly determined were used for comparison. Interobserver Agreement Functional Assessment Observation Form . To determine the reliability of the FAO across raters in each setting, a second trained investigator was present during 30% of the FAO observation sessions. The data collected by the observers was compared to determine if each observer recorded the same number of specific events, behaviors, antecedents, functions, and consequences of the behavior. Agreement was calculated by dividing the number of agreements by the total number of agreements plus disagreements and multiplying by 100 (Kazdin, 1982). Interobserver agreement on the FAO was high. Total agreement between the two observers averaged 92% with a range from 74% to 99%. In the school environment, agreement averaged 90% with a range from 74% to 99%. In the home environment, agreement averaged 94% with a range from 86% to 98%. For Josh, agreement between the two observers in the school environment averaged 96% and 98% in the home environment. In the school environment, agreement on behaviors, predictors, and perceived functions averaged 97% with a range from 95% to 98%. Agreement on actual consequences averaged 93% with a range from 88% to 98%. In the home environment, reliability across raters was typically conducted in one observation period lasting 90 minutes in length. Agreement was 100% on behaviors, 95% on predictors, 100% on perceived functions, and 95% on actual consequences.

PAGE 64

53 For Brad, agreement between the two observers averaged 74% in the school environment and 86% in the home environment. In the school environment, agreement averaged 79% on behaviors (range = 65% to 95%), 72% on predictors (range = 62% to 91%)), 75% on perceived functions (range ^ 59% to 91%), and 74%) on actual consequences (range = 66% to 80%). In the home environment, agreement averaged 83% on behaviors, 92%) on predictors, 83% on perceived functions, and 88%) on actual consequences. For Drake, agreement between the two observers averaged 99% in the school environment and 97% in the home environment. In the school environment, agreement on behaviors, predictors, and perceived functions was 100%). Agreement on actual consequences averaged 96% (range = 91% to 100%). In the home environment, agreement averaged 98% on behaviors (range = 89%) to 100%)), 80% on predictors (range = 66% to 93%), 87% on perceived functions (range = 78% to 95%), and 100% on actual consequences. Functional Assessment Interview . Inter-rater agreement in each setting on the FAI was calculated by comparing the hypotheses generated by each investigator. The resulting hypothesized function of each child's behavior was compared within settings by comparing the primary interviewer's hypothesized function of behavior for each child to the secondary interviewer's hypothesized function of behavior for each child. The primary interviewers were the standard to which comparisons were made on the FAI. For example, if the primary interviewer listed attention and escape as the consequence of behavior and the secondary interviewer only listed escape, it was considered one agreement and one disagreement. Agreement was calculated by dividing the number of agreements by the total number of agreements plus disagreements and multiplying by 100

PAGE 65

54 (Kazdin, 1982). The average agreement on the consequences hypothesized to be maintaining behavior generated from the FAI, across settings, was 68% with a range from 50% to 100%. In the school environment, agreement on the consequences hypothesized to be maintaining behavior averaged 72% with a range from 60% to 100%. Refer to Appendix E for a detailed description of the summary statements made for each child by the primary and secondary interviewer in the school environment. For Josh, the interviewers agreed 100% of the time on the consequences hypothesized to be maintaining behavior (attention and escape for disruption, attention for vocalization, and escape for noncompliance and off-task behavior). For Brad, the interviewers agreed 66%) of the time on the consequences hypothesized to be maintaining behavior (attention for noncompliance, stereotypy, and off-task behavior, attention and escape for aggression, and escape as the hypothesized function for elopement). For Drake, interviewers agreed 50% of the time on the consequences hypothesized to be maintaining behavior (automatic for stereotypy, attention and escape for aggression, and escape as the hypothesized function for noncompliance). In the home environment, agreement on the consequences hypothesized to be maintaining behavior averaged 89% with a range from 43% to 100%. Refer to Appendix F for a detailed description of the summary statements made for each child by the primary and secondary interviewer in the home envirorunent. For Josh, the interviewers agreed 50% of the time on the consequences hypothesized to be maintaining behavior (attention for noncompliance and automatic for vocalizations). For Brad, the interviewers agreed 100%) of the time on the consequences hypothesized to be maintaining behavior (escape). For Drake, the interviewers agreed 43% of the time on the consequences

PAGE 66

55 hypothesized to be maintaining behavior (escape for noncompliance and automatic for stereotypy and vocalizations). Functional analysis . To determine the interobserver agreement during the functional analyses, two independent observers coded the videotapes of approximately 30% (with a range of 27% to 52%) of all the functional analysis sessions for each child. Data files created by MOOSES were printed and agreement was calculated manually. Agreement was calculated by dividing the total number of agreements by the total number of agreements plus disagreements and multiplying by 100 (Kazdin, 1982). An agreement was recorded when the primary and secondary observer recorded the same response within a 10-second interval. A disagreement was recorded when the secondary observer did not record the same response or recorded the same response outside of the 10-second interval. For Josh, interobserver agreement was collected on 30% of the functional analysis sessions in the school (averaged 95% with a range from 84% to 100%) and home (averaged 94% with a range from 80% to 100%). For Brad, interobserver agreement was collected on 32% of the sessions at school (averaged 81% with a range from 39% to 100%) and 52% of the sessions at home (averaged 87% with a range from 63% to 100%). For Drake, interobserver agreement was collected on 27% of the sessions at school (averaged 88% with a range from 62% to 97%) and 30% of the sessions at home (averaged 92% with a range from 77% to 100%). Across participants, in the school environment, agreement averaged 88% with a range from 39% to 100% and in the home environment, agreement averaged 91% with a range from 63% to 100%. To determine the function of behaviors, two observers independently rated each graphical display of data. Agreement was calculated by dividing the number of agreements by the total number of agreements plus disagreements and multiplying by 100

PAGE 67

56 (Kazdin, 1982). Refer to Appendix G to view functional analysis graphs and Appendix H for a detailed summary of the function(s) identified by each investigator. Across environments, agreement on the functions of behavior for individual and combined graphs averaged 87% with a range from 67% to 1 00%. Across envirormients, agreement on the function of behavior for individual graphs averaged 83% with a range from 50% to 100%. For Josh, three behaviors were analyzed in the school and home environment (vocalization, noncompliance, and disruption). Agreement among raters on the function of the individual graphs of behaviors was 100%. For Brad, three behaviors were analyzed in the school and home environment (disruption, noncompliance, and aggression). Agreement among raters on the fimction of the individual graphs of behaviors was 60%. For Drake, four behaviors were analyzed in the school and home environment (noncompliance, tantrums, vocalizations, and disruption). Agreement among raters on the function of the individual graphs of behaviors was 100%. Across envirormients, agreement of the function of behavior for combined graphs averaged 100%. For Josh, agreement among raters on the ftinction of the combined graph of behaviors was 100%. For Brad, agreement among raters on the fimction of the combined graph of behavior was 100%. In the school envirormient, agreement on the fimction of behavior for individual and combined graphs averaged 88% with a range from 75% to 100%. Agreement on the function of behavior for individual graphs averaged 86% with a range from 75% to 100%. Agreement on the function of behavior for combined graphs averaged 100%. In the home environment, agreement on the fimction of behavior for individual and combined graphs averaged 85% with a range from 67% to 100%. Agreement on the function of behavior for individual graphs averaged 80% with a range from 50% to 100%. Agreement on the function of behavior for combined graphs averaged 100%.

PAGE 68

57 Design and Data Analysis Design Descriptive assessment. For two of the three students (Josh and Brad) data collection on the FAI and MAS began in the school environment first then in the home. For Drake, data collection began in the home. For all participants, the FAO was conducted in the school environment first then in the home. No experimental design was used for the descriptive assessment phase. The setting in which the FAI and the MAS were initially administered was counter-balanced to minimize order effects. Functional analysis. During the flinctional analysis five conditions were randomly presented to each participant using a multi element design. The design was used to determine the maintaining contingency for problematic behavior across children and across settings. Data Analysis Descriptive assessments. The data generated from each instrument was analyzed to determine the hypothesized function of behavior. For the FAO, hypotheses regarding the function of the child's behavior were developed by comparing the number of behavioral events out of the total that were coded in each of the perceived function categories. The perceived function category with the highest occurrence of the problem behavior was hypothesized to be the primary function of that behavior. For the FAI, hypotheses regarding the function of the child's behavior were developed by analyzing the summary statements made by the primary interviewer. The maintaining consequence statements made by the primary interviewer were hypothesized to be the primary function of that behavior. For the MAS, results fi-om the teacher and parent were individually tallied and recorded on the MAS form. This information generated rankings of

PAGE 69

58 hypothesized functions of each behavior. The hypothesized function scoring the highest mean score was determined to be the primary function of each behavior. Functional analysis. The data was analyzed across participants and settings using a single-subject design. The differences in problematic behavior across conditions and settings were then examined for each individual. Data were analyzed using line graphs that visually displayed the rate of problematic behavior across a 5-minute condition period. The rate indicated the number of times in which problem behavior occurred divided by five. The line graphs presented a graphical display of the rate of problem behavior in each experimental condition across the settings (Parsonson & Baer, 1992). This allowed for a visual evaluation of the differences in rate of problem behavior by each condition and by setting (Kazdin, 1982). The primary function was determined by visual display of the condition(s) with the most consistent and highest rates of problem behavior across one or more conditions. The function(s) identified by the primary investigator were used for comparison. Comparing descriptive assessments to functional analyses. The resuhs obtained on all three descriptive assessment instruments were individually compared to the functional analysis within settings for each participant. Comparisons were made on behaviors assessed by all three instruments and a functional analysis. Agreement within settings was calculated by determining the number of agreements on the hypothesized function of the behavior between the descriptive assessment instruments and the functional analyses and dividing that by the number of agreements plus disagreements and muhiplying by 100 (Kazdin, 1982). High agreement was defined as 80% to 100% (Kazdin, 1982), moderate agreement was defined as 40% to 79%, and low agreement was defined as 0% to 39%.

PAGE 70

59 To determine if functions of behavior differed across settings, the results of the functional analysis for each behavior were compared across the home and school. Agreement across settings was calculated by determining the number of agreements on the function of behavior between the home and school setting and dividing that by the number of agreements plus disagreements and multiplying by 100 (Kazdin, 1982). Behaviors that were assessed in both envirormients were compared. Finally, to determine which category of the FAO was most usefiil in determining the function of behavior, the "actual consequence" category on the FAO and the "perceived function" of behavior category were individually compared to the functional analysis findings for each behavior to determine their agreement within settings. Agreement was calculated by determining the number of agreements on the hypothesized fianction of the behavior between the FAO categories and the functional analysis and dividing that by the number of agreements plus disagreements and multiplying by 100 (Kazdin, 1982).

PAGE 71

CHAPTER 4 RESULTS The purpose of this investigation was to determine the consistency of three descriptive assessment instruments by comparing their results within settings to the findings of a functional analysis, determine if fiinctions of behavior are the same across settings by comparing the results of functional analyses across settings, determine the category of the FAO that was most likely to predict the same function of behavior as the functional analysis, and determine if the use of descriptive assessments are supported and if so which instrument is more consistent with the results of a functional analysis? These questions were addressed by conducting three descriptive assessments and a functional analysis across the home and school setting for three children identified with autism who were included in the general education classroom for most of the school day. Results were analyzed for individual and combined target behaviors. In general, the results of this study indicate that the FAO, FAI, and MAS have a low level of agreement with the results of functional analyses (averaging 39%). This indicates that these instruments generally will provide different functions of behavior than those identified by a functional analysis. Individually, the FAO had a moderate level of agreement (averaging 45%). The FAI and the MAS had low levels of agreement (averaging 36% and 29%, respectively). This means that of the three descriptive assessment instruments evaluated, the FAO is a better instrument and likely to accurately

PAGE 72

61 identify the function of behavior approximately half of the time. Overall agreement between the three instruments and the functional analyses was higher in the home setting (averaging 56%) than at school (averaging 28%). The instruments were more likely to agree with the fiinctional analysis when escape or automatic reinforcement was identified as the function of behavior. When examining functional analyses across settings, there was a moderate level of agreement on the fimction of behaviors across settings (averaging 59%). This means that the function of a behavior may actually differ depending on the setting. Graphs of combined behaviors were more likely to find the same function across settings (averaging 75%) than graphs of individual behaviors (averaging 54%). However, only two graphs of combined behaviors were used in this analysis compared to nine graphs of individual behavior. Overall, tangible and automatic functions were more likely to be the same across settings. The perceived function and actual consequence categories on the FAO combined had a low level of agreement when compared to the results of functional analyses (averaging 27%). Individually, the perceived function category had a moderate level of agreement (averaging 45%). The actual consequences category had a low level of agreement (averaging 1 0%). This indicates that the perceived function category is more likely than the actual consequences category to accurately identify the function of behavior. Agreement between the FAO categories and the functional analyses were generally higher in the home setting (averaging 36%) than at school (averaging 21%). The perceived function and actual consequence category were more likely to be in agreement with the functional analysis when automatic reinforcement was the identified

PAGE 73

62 function. In addition, the perceived function category was more likely to be in agreement with the functional analysis when escape was the identified function. Comparison of Descriptive Assessments to Functional Analyses Descriptive functional assessment instruments were compared to the functional analysis within settings. Behaviors were compared individually and then in some combinations depending on the hypothesized response class and function(s). The total nvmiber of behaviors compared differed across enviroimients since only some of the behaviors occurred in both the school and home settings. In addition, only behaviors analyzed by both an instrument and a functional analysis were used for comparison. A total of ten individual behaviors across children were compared in the school setting (three for Josh, three for Brad, and four for Drake). Two combined groups of behaviors were compared in the school setting (one for Josh and one for Brad). In the home environment, a total of eight individual behaviors across children were compared (three for Josh, two for Brad, and three for Drake). Two combined groups of behaviors were compared in the home setting (one for Josh and one for Brad). The number of comparisons was limited for three reasons. First, fewer behaviors were assessed by the MAS than were observed during the fiinctional analysis. Second, more behaviors were assessed by the FAI than were directly observed during the functional analysis and the FAO. Third, the results of two functional analyses were inconclusive and a function could not be clearly identified. Data will be discussed by summarizing all participants' data and then reporting individual participant data by setting. Comparisons within the School Environment For all three participants, within the school environment, the FAO, FAI, and MAS agreed with the results of the functional analyses 28% of the time with agreement ranging

PAGE 74

63 from 0% to 75%. The FAO agreed with the results of the functional analyses 6 out of 17 times (35%). The FAI agreed with the results of the functional analyses 3 out of 14 times (21%). The MAS agreed with the results of the functional analyses 3 out of 12 times (25%). Table 4-1 presents a summary of the percent agreement between the hypothesized function of behaviors generated across instruments and participants and the functional analysis within the school setting. For Josh, the hypothesized functions of behavior generated from the FAO, FAI, and MAS in the school setting agreed with the results of the functional analysis 0 out of 12 times (0%). Individually, the FAO, FAI, and MAS did not agree with the results of the functional analysis. The tangible function identified in the fimctional analyses was not identified by any of the descriptive instruments as a hypothesized function for any of the target behaviors. For Brad, the hypothesized functions of behavior generated from the FAO, FAI, and MAS in the school setting agreed with the results of the functional analysis 6 out of 18 times (33%). Individually, the FAO agreed with the results of the functional analysis four out of eight times (50%). The FAO correctly hypothesized the escape function of behavior (functional analyses identified escape and tangible functions) for four of the target behaviors (noncompliance, disruption, aggression, and a combination of all 3 of these behaviors). The FAI agreed with the results of the functional analysis two out of six times (33%) by correctly hypothesizing the escape function for two of the target behaviors (aggression and a combination of noncompliance, disruption, and aggression). The MAS agreed with the results of the functional analysis zero out of four times (0%). The MAS hypothesized that the four target behaviors were maintained by automatic reinforcement. For Drake, the hypothesized functions of behavior generated

PAGE 75

64 Table 4-1 Summary of the Percent Agreement between the Hypothesized Functions of Behaviors Generated by Instruments within the School Environment and the Functions of Behaviors Determined by the Functional Analysis School Participants FAQ FAI MAS Total agreement Josh 0%(0/4) 0%(0/4) 0%(0/4) 0%(0/12) Brad 50% (4/8) 33% (2/6) 0%(0/4) 33% (6/18) Drake 40% (2/5) 25% (1/4) 75% (3/4) 46% (6/13) Total 35% (6/17) 21% (3/14) 25% (3/12) 28% (12/43) agreement

PAGE 76

65 from the FAO, FAI, and MAS in the school setting agreed with the resuUs of the functional analysis 6 out of 13 times (46%). Individually, the FAO agreed with the results of the functional analysis two out of five times (40%) by correctly hypothesizing the escape function of noncompliance and the automatic reinforcement function of vocalizations. The FAI agreed with the results of the functional analysis one out of four times (25%) by correctly hypothesizing the escape function of noncompliance. The MAS agreed with the results of the functional analysis three out of four times (75%) by correctly hypothesizing the escape function of noncompliance, the automatic reinforcement function of vocalizations, and the tangible function of tantrums. When escape was identified as the function of the target behavior across all three participants, the FAO always matched this function with one exception (Drake's disruptive behavior), the FAI correctly identified it half of the time (3/6), and the MAS correctly identified it once (1/6). The MAS was the only instrument to correctly hypothesize a tangible function for Drake (1/2). However, none of the instruments identified the tangible function for Josh or Brad. The FAO and MAS correctly hypothesized the automatic function for Drake (1/1). In summary, these results indicate that there was limited to no correct matching to the functions of behavior identified in the functional analysis in the school setting. Tables 4-2, 4-3, and 4-4 present detailed summaries of the hypothesized function of behaviors generated across instruments and participants within the school setting. Comparisons within the Home Environment For all participants within the home setting, the FAO, FAI, and MAS agreed with the results of the functional analysis 56% of the time with agreement ranging from

PAGE 77

66 Table 4-2 Summary of the Hypothesized Function of Behaviors across Instruments in the School Environment for Josh Behaviors FAO FAI MAS Functional analysis Total agreement Vocalization Attention Attention Escape Tangible 0/3 (0%) Disruption Attention Attention and escape Escape Tangible 0/3 (0%) Noncompliance Escape Escape Attention Tangible 0/3 (0%) Disruption and noncompliance Attention and escape Attention and escape Attention and escape Tangible 0/3 (0%) Aggression — Attention and escape — — — Off-Task Escape Total 0/4 0/4 0/4 0/12 agreement (0%) (0%) (0%) (0%)

PAGE 78

67 Table 4-3 Summary of the Hypothesized Function of Behaviors across Instruments in the School Environment for Brad Behaviors FAO FAI MAS Functional analysis Total agreement Noncompliance Escape Attention Automatic Escape and tangible 1/4 (25%) Disruption Escape Attention Automatic Escape and tangible 1/4 (25%) Aggression Escape Attention and escape Automatic Escape and tangible 2/5 (40%) Noncompliance, Escape Attention Automatic Escape and 2/5 (40%) disruption, and and tangible aggression escape Stereotypy — Attention — — — Elopement — Attention and escape — — — Off-Task Attention Tantrum — Attention — — — Aggression — Attention and escape Flopping Attention Total 4/8 2/6 0/4 6/18 agreement (50%) (33%) (0%) (33%)

PAGE 79

68 Table 4-4 Summary of the Hypothesized Function of Behaviors across Instruments in the School Environment for Drake Behaviors FAO FAl MAS Functional Total analysis agreement Noncompliance Escape Escape and Escape and Escape attention tangible Vocalization Automatic Attention Automatic Automatic Tantrum Disruption Attention Attention Tangible and escape Attention — — Stereotypy — Aggression — Automatic Attention and escape Tangible 3/5 (60%) 2/3 (67%) 1/3 (33%) Tangible 0/2 (0%) and escape Total agreement 2/5 (40%) 1/4 (25%) 3/4 (75%) 6/13 (46%)

PAGE 80

69 0% to 100%. The FAO agreed with the resuUs of the functional analysis 8 out of 14 times (57%). The FAl agreed with the results of the functional analysis five out of eight times (63%). The MAS agreed with the results of the functional analysis two out of five times (40%). Table 4-5 presents a summary of the percent agreement between the hypothesized function of behaviors generated across instruments and participants and the functional analysis within the home setting. In comparison to the school setting, there was a larger percentage of function matches across instruments in the home setting. However, there were approximately one-half the number of individual or combined behaviors assessed in this home setting (27) in comparison to the school setting (43). For Josh, the hypothesized functions of behaviors generated from the FAO, FAI, and MAS in the home environment agreed with the results of the functional analysis 7 out of 14 times (50%). Individually, the FAO agreed with the results of the functional analysis four out of seven times (57%) by correctly hypothesizing the automatic function of vocalizations and the escape function of noncompliance, disruption, and combined behaviors. The FAI agreed with the results of the functional analysis two out of four times (50%) by correctly hypothesizing the automatic function of vocalizations and the escape function of noncompliance. The MAS agreed with the results of the functional analysis one out of three times (33%) by correctly hypothesizing the tangible function of noncompliance. For Brad, the hypothesized functions of behaviors generated from the FAO, FAI, and MAS in the home environment agreed with the results of the fionctional analysis four out of six times (67%). Individually, the FAO agreed with the results of the functional analysis two out of four times (50%) by correctly hypothesizing the tangible function of disruption and combined behaviors. The FAI and MAS each agreed with the

PAGE 81

70 Table 4-5 Summary of the Percent Agreement between the Hypothesized Functions of Behaviors Generated by Instruments within the Home Environment and the Functions of Behaviors Determined by the Functional Analysis Home Participants FAQ FAI MAS Total agreement Josh Brad Drake 57% (4/7) 50% (2/4) 67% (2/3) 50% (2/4) 100% (1/1) 67% (2/3) 33% (1/3) 100% (1/1) 0%(0/l) 50% (7/14) 67% (4/6) 57% (4/7) Total agreement 57% (8/14) 63% (5/8) 40% (2/5) 56% (15/27)

PAGE 82

71 results of the ftinctional analysis one out of one time (100%) by correctly hypothesizing the escape function for noncompliance. For Drake, the hypothesized functions of behaviors generated from the FAO, FAI, and MAS in the home environment agreed with the results of the functional analysis four out of seven times (57%). Individually, the FAO and FAI each agreed with the results of the functional analysis two out of three times (67%) by correctly hypothesizing the escape function of noncompliance and the automatic function of vocalizations. The MAS agreed with the results of the functional analysis zero out of one time (0%). The MAS hypothesized automatic and attention as the function of noncompliance whereas the functional analysis determined escape to be the function. Tables 4-6, 4-7, and 4-8 present detailed summaries of the hypothesized function of behaviors generated across instruments and participants within the home setting. When escape was identified as the fimction of the target behavior across all three participants, the FAO matched this function 67% of the time (4/6), the FAI correctly identified it 60% of the time (3/5), and the MAS correctly identified it 20% of the time (1/5). The FAO correctly hypothesized a tangible function 33% of the time (2/6). The MAS correctly hypothesized a tangible function once for Josh. The FAI never correctly identified a tangible function (0/6). The FAO and FAI correctly hypothesized the automatic function for Josh and Drake (2/2). In summary, these results are similar to the school setting and indicate that there was limited to no correct matching to the functions of behavior identified in the functional analysis in the home setting. Across environments and participants, when escape was identified as the function of the target behavior, the FAO matched this function 75% of the time (9/12). The FAI matched the escape function 50% of the time (6/12) and the MAS matched the escape

PAGE 83

72 Table 4-6 Summary of the Hypothesized Function of Behaviors across Instruments in the Home Environment for Josh Behaviors FAO FAI MAS Functional analysis Total agreement Vocalization Automatic Attention and automatic Escape Automatic 2/4 (50%) Noncompliance Escape Attention and escape Tangible Tangible and escape 3/6 (50%) Disruption Attention and escape — — Tangible and escape 1/2 (50%) Noncompliance Escape and Tangible and 1/2 (50%) and disruption attention escape Total 4/7 2/4 1/3 7/14 agreement (57%) (50%) (33%) (50%)

PAGE 84

73 1 Table 4-7 Summary of the Hypothesized Function of Behaviors across Instruments in the Home Environment for Brad Behaviors FAO FAI MAS Functional analysis Total agreement Noncompliance Tangible Escape Escape Escape 2/3 (67%) Disruption Tangible — — Tangible 1/1 (100%) Aggression Attention and tangible — Automatic Inconclusive — Off-Task — Escape — — — Flopping — Escape — — — Noncompliance, Tangible — — Tangible 1/2 (50%) disruption, and and and aggression attention escape Total 2/4 1/1 1/1 4/6 agreement (50%) (100%) (100%) (67%) i

PAGE 85

74 Table 4-8 Summary of the Hypothesized Function of Behaviors across Instruments in the Home Environment for Drake Behaviors rAU T? A T r Al \ /f A C MAo Functional analysis 1 otai agreement Noncompliance Escape Escape Automatic and attention Escape 2/3 (67%) Vocalization Automatic Automatic Automatic 2/3 (67%) Disruption Attention Tangible 0/1 (0%) Tantrum Tangible Attention and automatic Tangible Inconclusive Aggression Attention and tangible — — — Stereotypy — Automatic and attention — — — Total 2/3 2/3 0/1 4/7 agreement (67%) (67%) (0%) (57%)

PAGE 86

75 function 17% of the time (2/12). When tangible was identified as the function of the target behavior, the FAO and MAS matched this function 13% of the time (2/16). The FAX matched the tangible function 0% of the time (0/16). When automatic reinforcement was identified as the function of the target behavior, the FAO matched this function 100% of the time (3/3). The FAI matched the automatic function 67% of the time (2/3) and the MAS matched the tangible function 33% of the time (1/3). In sum, there was a low level of agreement between the results of the functional assessment instruments and the functional analyses. Agreement between the functional analyses and the hypothesized functions of behaviors generated from all three instruments across settings averaged 39% (27 out of 70 times). Individually, agreement between the FAO and the functional analyses across settings was moderate, averaging 45% (14 out of 31 times). In general, functions of escape and automatic reinforcement were most often correctly identified by the FAO (75% and 100% respectively) across environments and participants. Agreement between the FAI and the functional analyses across settings was low, averaging 36% (8 out of 22 times). Agreement between the MAS and the functional analyses across settings was also low, averaging 29% (5 out of 17 times). On average, agreement among instruments and the functional analyses was higher in the home setting where there were fewer behaviors identified to evaliiate. Comparison of Functional Analvses across Environments Functional analyses were compared across settings to determine if the fiinction of behaviors differed across settings. The behaviors for which a functional analysis was conducted in both settings were compared. Only functional analyses in which the primary investigator was able to clearly identify a function were used. Tables 4-9, 4-10, and 4-1 1 present detailed summaries of the results of functional analyses across settings. A total of

PAGE 87

76 nine individual behaviors across children were compared (three for Josh, two for Brad, and four for Drake). Two graphs of combined behaviors were also compared across settings. Results are presented individually and then summarized across participants. For Josh, agreement on the fimction(s) of behavior across settings averaged 43% (3/7), all for the tangible function. At least one function of behavior across settings was identified for three of four target behaviors (not for vocalizations). The ftinction(s) of behavior identified by individual graphs of behavior agreed across settings two out of five times (40%). The function(s) of behavior identified on combined graphs of behaviors agreed across settings one out of two times (50%). For Brad, agreement on the function(s) of behavior across settings averaged 67% (4/6), two times each for escape and tangible functions. The function(s) of behavior identified by individual graphs of behavior agreed across settings two out of four times (50%). The function(s) of behavior identified on combined graphs of behaviors agreed across settings two out of two times (100%). At least one function of behavior across settings was identified for three out of four target behaviors (not for aggression individually). For Drake, agreement on the function(s) of behavior identified by individual graphs of behavior across settings averaged 75% (3/4), one each for escape, tangible, and automatic functions. No combined graphs were analyzed. At least one function of behavior across settings was identified for three out of four target behaviors (not for tantrums). In sum, agreement on the function(s) of individual and combined behaviors across settings averaged 59% (10/17). This indicates that functions of behavior differed across environments approximately 40% of the time. There was a higher level of agreement between functional analyses across settings when combined behavior graphs were

PAGE 88

77 Table 4-9 nuiiic V ULallZallUn 1 angiDie /\UlUmalll/ yj /o yyj/ 1 ) Til ctn it^fi UlalUpilOIl 1 all^lOlC OIIU »/v/ /O yll escape Noncompliance Tangible Tangible and escape 50% (1/2) Disruption and Tangible Tangible and escape 50% (1/2) noncompliance Total individual and combined behaviors 43% (3/7) Total individual behaviors 40% (2/5) Total combined behaviors 50% (1/2)

PAGE 89

78 Table 4-10 Summary of the Results of Functional Analyses across Environments for Brad OCliCtVlUl o Aprpement INUIlCUiliUiiCuiCC DisruDtion Taneible and escaoe Taneible 50% (1/2) Aggression Tangible Inconclusive 1 00% (2IT\ 1 \J\J / U ^ I disruption, and Total individual and Total individual behaviors 50% (2/4) Total combined behaviors 100% (2/2)

PAGE 90

I Table 4-11 Summary of the Results of Functional Analyses across Environments for Drake DCIlaVlUrb OLnooi Home /\^rcciiiciii Noncomnliance Fscane Fscane 100% (1/1) Vocalization Automatic Automatic 100% (1/1) Tantrum Tangible Inconclusive Disruption Tangible and Tangible 50% (1/2) escape Total individual behaviors 75% (3/4)

PAGE 91

80 compared. Agreement on the functions of individual behaviors across settings averaged 54% (7/13). Agreement on the functions of combined behaviors across settings averaged 75% (3/4). There was a combined total of 77% (10/13) of the time where at least one function identified was confirmed in each setting, indicating that most of the time at least one function of behavior was identified across both settings using the functional analysis. Across participants, the tangible function agreed across settings 60% of the time (6/10), the escape function 38% of the time (3/8), and the automatic function agreed across settings 50% of the time (1/2). Comparison of Perceived Function and Actual Consequence to the Fimctional Analvsis The perceived function and actual consequence categories of the FAO were compared to the functional analyses within settings. Behaviors where a function was identified by the functional analysis and the FAO were used for comparison. A total of twelve behaviors across children were compared in the school setting (four for each child) and ten behaviors in the home setting (four for Josh, three for Brad, and three for Drake). Data will be discussed by summarizing all participants' data and then reporting individual participants' data by setting. Comparison Between FAO Categories and the Functional Analyses within the School Environment For all participants within the school setting, the perceived function and actual consequence categories agreed with the resuhs of the functional analysis 21% of the time with agreement ranging from 0% to 50%. The perceived function category agreed with the functional analysis 6 out of 17 times (35%). The actual consequence category agreed with the functional analysis 1 out of 17 times (6%). The function of behavior was most often determined to be maintained by a tangible function (ten times) but neither category ever identified this function across participants. Escape was the next most frequent

PAGE 92

81 function (six times) correctly identified five times by the perceived function category. Automatic reinforcement was identified once for Drake and identified by both categories correctly. Tables 4-12, 4-13, and 4-14 present detailed summaries of the percent agreement between the perceived fiinction and actual consequences category on the FAO and the fimctional analysis within the school setting. For Josh, the perceived function and actual consequence categories did not agree with the results of the functional analysis. The tangible function identified by the functional analysis was not hypothesized as the function for any of the target behaviors on either category of the FAO. For Brad, the perceived fiinction category agreed with the results of the fimctional analysis four out of eight times (50%) by correctly hypothesizing the escape function for four of the target behaviors (noncompliance, disruption, aggression, and combined behaviors). The actual consequence category agreed with the results of the functional analysis zero out of eight times (0%). The actual consequences category incorrectly hypothesized attention as the function of all four target behaviors. The functional analysis identified escape and tangible as the function of all four target behaviors. For Drake, the perceived function category agreed with the results of the fimctional analysis two out of five times (40%) by correctly hypothesizing the escape function of noncompliance and the automatic function of vocalizations. The actual consequence categories agreed with the results of the functional analysis one out of five times (20%) by correctly hypothesizing the automatic fimction of vocalizations. Tables 4-15, 4-16, and 4-17 present detailed summaries of the hypothesized function of behaviors generated by FAO categories and the functional analysis within the school setting.

PAGE 93

82 Table 4-12 Summary of the Percent Agreement between the FAO Categories and the Functional Analysis within the School Environment for Josh FAO Behaviors Perceived function Actual consequence Combined Vocalization 0%(0/l) 0% (0/1) 0% (0/2) Disruption 0%(0/l) 0%(0/l) 0% (0/2) Noncompliance 0%(0/l) 0% (0/1) 0% (0/2) Vocalization, disruption, and noncompliance 0%(0/l) 0% (0/1) 0% (0/2) Total 0% (0/4) 0% (0/4) 0% (0/8)

PAGE 94

83 i Table 4-13 Summary of the Percent Agreement between the FAO Categories and the Functional Analysis within the School Environment for Brad FAO Behaviors Perceived function Actual consequence Combined Noncompliance 50% (1/2) 0% (0/2) 25% (1/4) Disruption 50% (1/2) 0% (0/2) 25% (1/4) Aggression 50% (1/2) 0%(0/2) 25% (1/4) Noncompliance, 50% (1/2) 0% (0/2) 25% (1/4) disruption. and aggression Total 50% (4/8) 0% (0/8) 25% (4/16) Table 4-14 Summary of the Percent Agreement between the FAO Categories and the Functional Analysis within the School Environment for Drake FAO Behaviors Perceived function Actual consequence Combined Noncompliance 100% (1/1) 0%(0/l) 50% (1/2) Vocalization 100% (1/1) 100% (1/1) 100% (2/2) Tantrum 0%(0/l) 0% (0/1) 0% (0/2) Disruption 0% (0/2) 0% (0/2) 0% (0/4) Total 40% (2/5) 20% (1/5) 30% (3/10) Total across participants 35% (6/17) 6% (1/17) 21% (7/34)

PAGE 95

84 Table 4-15 Summary of the Hypothesized Function of Behaviors Generated by FAO Categories and the Functional Analysis in the School Environment for Josh Behaviors Perceived function FAO Actual consequence Functional analysis Vocalization Attention Ignore Tangible (automatic) Disruption Attention Attention and Tangible ignore (automatic) Noncompliance Escape Attention Tangible Disruption and Attention and Attention and Tangible noncompliance escape automatic

PAGE 96

85 Table 4-16 Summary of the Hypothesized Function of Behaviors Generated by FAO Categories and Behaviors FAO Functional analysis Perceived function Actual consequence Noncompliance Escape Attention Escape and tangible Disruption Escape Attention Escape and tangible Aggression Escape Attention Escape and tangible Noncompliance, Escape Attention Escape and disruption, and tangible aggression Table 4-17 Summary of the Hypothesized Function of Behaviors Generated by FAO Categories and the Functional Analysis in the School Environment for Drake FAO Behaviors Noncompliance Vocalization Tantrum Disruption Perceived Function Actual Consequence Functional Analysis Escape Attention Escape Automatic Automatic Automatic Attention Attention Attention Attention and Tangible Tangible and

PAGE 97

86 Comparison Among FAQ Categories and the Functional Analysis within the Home Environment For all participants within the home setting, the perceived function and actual consequence categories agreed with the results of the functional analysis 36% of the time with agreement ranging from 0% to 100%. The perceived function category agreed with the functional analysis 8 out of 14 times (57%). The actual consequence category agreed with the functional analysis 2 out of 14 times (14%). Escape and tangible functions were each identified five times with the perceived function category identifying the escape function four times and the tangible function once. Automatic reinforcement was identified twice and both categories correctly identified this function. Tables 4-18, 4-19, and 4-20 present summaries of the percent agreement among the perceived function and actual consequence categories on the FAO and the functional analysis within the home setting. For Josh, the perceived function category agreed with the results of the functional analysis four out of seven times (57%) by correctly identifying the automatic function of vocalizations and the escape function of noncompliance, disruption, and combined behaviors. The actual consequence category agreed with the results of the functional analysis one out of seven times (14%) by correctly identifying the automatic fimction of vocalizations. For Brad, the perceived function category agreed with the results of the functional analysis two out of four times (50%) by correctly identifying the tangible function of disruption. The actual consequence category agreed with the results of the functional analysis zero out of four times (0%). The actual consequence category incorrectly identified attention as the hypothesized function of noncompliant and disruptive behavior. The functional analysis identified escape as the function of

PAGE 98

87 Table 4-18 Summary of the Percent Agreement between the FAO Categories and the Functional Analysis within the Home Environment for Josh FAO Behaviors Perceived function Actual consequence Combined Vocalization 100% (1/1) 100% (1/1) 100% (2/2) Disruption 50% (1/2) 0% (0/2) 25% (1/4) Noncompliance 50% (1/2) 0% (0/2) 25% (1/4) Disruption and noncompliance 50% (1/2) 0% (0/2) 25% (1/4) Total 57% (4/7) 14% (1/7) 36% (5/14) Table 4-19 Summary of the Percent Agreement between the FAO Categories and the Functional Analysis within the Home Environment for Brad FAO Behaviors Perceived function Actual consequence Combined Noncompliance 0% (0/1) 0%(0/l) 0% (0/2) Disruption 100% (1/1) 0% (0/1) 50% (1/2) Noncompliance, disruption. and aggression 50% (1/2) 0% (0/2) 25% (1/4) Total 50% (2/4) 0% (0/4) 25% (2/8)

PAGE 99

88 Table 4-20 Summary of the Percent Agreement between the FAO Categories and the Functional Analysis within the Home Environment for Drake FAO Participant Perceived function Actual consequence Combined Noncompliance 100% (1/1) 0%(0/l) 50% (1/2) Vocalization 100% (1/1) 100% (1/1) 100% (2/2) Disruption 0%(0/l) 0%(0/l) 0% (0/2) Total 67% (2/3) 33% (1/3) 50% (3/6) Total across participants 57% (8/14) 14% (2/14) 36% (10/28) Total across categories and environments 45% (14/31) 10% (3/31) 27% (17/62) 1

PAGE 100

89 noncompliance and tangible as the function of disruption. For Drake, the perceived function category agreed with the results of the functional analysis two out of three times (67%) by correctly identifying the escape function of noncompliance and the automatic function of vocalizations. The actual consequence category agreed with the results of the functional analysis one out of three times (33%) by correctly identifying the automatic function of vocalizations. Table 4-21, 4-22, and 4-23 present detailed summaries of the hypothesized function of behaviors generated by FAO categories and the functional analyses within the home setting. Across settings, when escape was identified as the function of the target behavior across all three participants, the perceived function category matched this function 75% of the time (9/12). The actual consequence category never correctly identified the escape function of the target behaviors. When tangible was identified as the function of the target behavior across all three participants, the perceived function category matched this function 13% of the time (2/16). The actual consequence category never correctly identified the tangible function of the target behavior. When automatic reinforcement was identified as the function of the target behavior across all three participants, the perceived function category and actual consequence category matched this function 1 00% of the time (6/6). Across environments, there was a low level of agreement between the FAO categories and the functional analyses (averaging 27%). Agreement between the perceived function category and the functional analysis was moderate, averaging 45%. Agreement between the actual consequences category and the functional analysis was i low, averaging 10%. Agreement among the FAO categories and the functional analyses were higher in the home setting where there were fewer behaviors identified to evaluate

PAGE 101

90 Table 4-21 Summary of the Hypothesized Function of Behaviors Generated by FAO Categories and Behaviors Perceived function FAO Actual consequence Functional analysis Vocalization Automatic Automatic Automatic Disruption " ' Attention and Attention Escape and escape tangible Noncompliance Escape Attention Escape and tangible Disruption and Attention and Attention Escape and noncompliance escape tangible Table 4-22 Summary of the Hypothesized Function of Behaviors Generated by FAO Categories and the Functional Analysis in the Home Environment for Brad FAO Behaviors Perceived function Actual consequence Fimctional analysis Noncompliance Disruption Noncompliance, disruption, and Tangible Tangible Tangible and attention Attention and Escape ignore Attention Tangible Attention and ignore Tangible and escape

PAGE 102

91 Table 4-23 Summary of the Hypothesized Function of Behaviors Generated by FAO Categories and the Functional Analysis in the Home Environment for Drake FAO Behaviors Perceived function Actual consequence Functional analysis Noncompliance Escape Attention Escape Vocalization Automatic Automatic Automatic Disruption Attention Attention and Tangible ignore

PAGE 103

' \ v., . ' 92 (10 in the home, 12 in the school setting). Although the perceived functions category more often correctly identified the fimction determined by the ftmctional analysis, the 45% agreement is still considered only a moderate level. Simimary The hypothesized ftmctions of behaviors generated by the FAO, FAX, and MAS have a low level of agreement with the results of a functional analysis. The FAO displayed a moderate level of agreement with the results of functional analyses. The FAI and MAS displayed low levels of agreement with the results of functional analyses. All three instruments were more likely to agree with the functional analysis when automatic reinforcement or escape was the identified function of behavior. Across settings, there was a moderate level of agreement between the functions of behavior identified by the functional analyses. Agreement was generally higher when combined behavior graphs were compared across settings than when individual behavior graphs were compared. Also, tangible and automatic functions were more likely to be in agreement across settings. There was a low level of agreement between the hypothesized functions of behaviors identified by the FAO categories and the functional analyses. However, agreement between the functional analyses and the hypothesized functions of behaviors identified by the perceived function category was higher than the level of agreement between the functional analyses and the hypothesized functions generated from the actual consequences category. Both categories on the FAO were more likely to be in agreement with the fimctional analysis when the function of behavior was identified to be automatic reinforcement. The perceived fiinction category was more likely to be in agreement with the functional analysis when escape was the identified function of behavior.

PAGE 104

CHAPTER 5 DISCUSSION There has been a renewed interest in the use of functional assessments since the 1997 Amendments to the Individuals with Disabilities Act (IDEA, 1997). However, the amendments do not provide guidelines as to how functional assessment should be conducted or the specific components that must be included to appropriately develop an intervention plan (Ervin et al., 2001). The result has been that individual states and local districts determine on their own what constitutes appropriate and legally defensible functional assessment. There is a wide range of functional assessment methods available from indirect descriptive assessments (e.g., FAI and MAS) and direct descriptive assessments (e.g., FAO) to experimental functional analyses. Functional assessment has the potential to be a powerful tool in the determination of effective treatments based on the function of behavior (Gresham et al., 2001). However, the reliability and validity of different functional assessment methods has not been adequately addressed (Gresham et al.,2001). This study was conducted to determine the consistency of three commonly used functional assessment instruments: the FAO, FAI, and the MAS. These three instnmients were compared to the results of a functional analysis within settings. Fimctional analysis results were compared across settings to determine if functions of behaviors differed across settings (external validity). In addition, the FAO was further explored to determine which category, "perceived functions" or "actual consequences", provided the most 93

PAGE 105

94 useful information. The comparisons were made to ultimately determine if these descriptive assessment procedures were able to accurately identify the fimction of behavior in the general education and home setting. In general, the results of this study indicated that the FAO, FAI, and MAS have a low level of agreement with the results of a functional analysis. Across settings and participants, the instruments agreed with the results of the functional analyses 39% of the time. This indicates that these instruments are able to accurately identify the function of behavior less than half of the time. These results disagree with the results from other researchers who found high levels of agreement between descriptive assessments and functional analyses (Amdorfer et al., 1994; Cunningham & O'Neill, 2000). The difference in results may be because of the way in which the studies were conducted. Other researchers have compared functional analyses to the primary and secondary functions identified by descriptive assessments. Agreement was determined if the primary or secondary function was the same as the results of the functional analysis. In this study, primary and secondary functions were not differentiated; all behavioral functions identified were treated as primary functions and compared to the results of the functional analysis. Interpretation of Results The hypothesized functions of behaviors generated by the FAO had the highest level of agreement with functions identified by fiinctional analyses (45%). This may have occurred because the FAO is an observation system that allows for the direct observation of behavior, thereby reducing biases and providing the observer with more objective information. The FAI and MAS are indirect assessments and the information gained may

PAGE 106

95 be influenced by biases and some relevant information may be left out. Other researchers (e.g., Galensky, et al., 2001) have found more consistency between direct descriptive assessment measures of behavior than indirect methods. Across settings and participants, the FAO had a moderate level of agreement with the functional analysis, averaging 45%. It should be noted that investigators received a high level of training before data collection which may have inflated the results. Investigators practiced using the FAO until reaching 80% agreement. The FAI and the MAS had low levels of agreement with the ftinctional analysis. Across settings and participants, the FAI agreed with the ftmctional analysis 36% of the time. This level of agreement is lower than those reported in previous studies (Amdorfer et al., 1994; Cunningham & O'Neill, 2000). It should be noted that the FAI generated the largest number of behaviors to evaluate (28 across all participants and settings compared to 21 for the FAO, 20 for the ftinctional analysis, and 15 for the MAS). The MAS agreed with the results of the ftmctional analysis 29% of the time. This level of agreement is lower than has been found in previous studies (Amdorfer et al.; Cunningham & O'Neill, 2000). This may be the result of comparing only primary functions of behavior to the ftinctional analysis. Previous studies have compared the primary and secondary ftmctions to determine agreement. Interestingly, agreement between the instruments and the ftmctional analyses were higher in the home than in the school. On average, instruments agreed with the ftmctional analysis 28% of the time at school and 56% of the time at home. Specifically, the FAO agreed with the results of the ftinctional analysis 35% of the time at school and 57% of the time at home. The FAI agreed with the results of the ftmctional analysis 21% of the

PAGE 107

time at school and 63% of the time at home. The MAS agreed with the results of the fimctional analysis 25% of the time at school and 40% of the time at home. The FAI and MAS are indirect assessments where the parent is the informant at home and the teacher is the informant at school. Therefore, higher rates of agreement in the home setting may be because of the parent's knowledge of his/her child. The FAO is an observation system that was completed by an investigator. The higher level of agreement between the FAO and the functional analyses in the home setting may have implications for practice. Often, children are assessed in schools or clinics and the resulting information is used to plan interventions in the home (Carr et al., 1999). If observations are more effective at identifying the function of behavior in the home setting, practitioners may want to consider conducting these types of assessments in the home. It is important to note that differences in the settings in which the functional analyses were conducted may have influenced the results. In the home envirormient, both the descriptive assessments and the functional analyses were conducted in the natural setting. In the school environment, the descriptive assessments were conducted in the natural setting but the functional analyses were conducted in an analog setting. Future research should explore the influence of setting on the accuracy of assessment data. One final consideration to make in interpreting the difference in agreements across settings is that there were a total of 27 behaviors compared in the home setting and 43 behaviors compared in the school setting. Having close to twice the number of behaviors to compare and evaluate appeared to reduce the consistency with which functions would match in the school setting. Since the FAI appeared to produce the largest number of behaviors for comparison, fiiture research

PAGE 108

97 may want to make sure that all behaviors evaluated by the FAI are evaluated by the other instruments. Across settings, the functional analysis demonstrated a moderate level of agreement. The function identified in the school setting agreed with the function identified in the home setting 59% of the time. This indicates that functions of behavior differed across settings less than half of the time. When individual behaviors were compared, agreement across settings averaged 54%. When combined behavior graphs were compared, agreement across settings averaged 75%. It is possible that when behaviors are combined, fiinctional analysis results are more robust across settings. However, only two combined graphs of behavior were compared in this study. Future studies should compare the results of functional analyses using individual and combined behavior graphs to determine if combined behavior graphs are in fact more robust or if functions of behavior actually differ across settings. If functions of behavior are different across settings, caution must be used when applying information from one setting for application in another. However, 77% of the time (10/13) one of the functions of behavior identified in one setting was identified in the other. This appears to indicate that most of the time the same function of behavior will be present across settings and that additional functions of behavior are missed by not conducting separate analyses in each setting of interest. Clearly, this is an area of research that needs to be investigated further. Across settings, the FAO categories demonstrated a low level of agreement with the functional analysis, averaging 27%. Higher levels of agreement were found between the perceived function category and the functional analysis than the actual consequence category. Agreement between the functional analysis and the perceived function category

PAGE 109

98 was moderate, averaging 45%. Agreement between the functional analysis and the actual consequences category was much lower, averaging 10%. This indicates that the perceived function category provided more accurate information than the actual consequence category. The perceived function category is a subjective measure based primarily on clinical impression. This implies that results are more likely to be accurate when the observer has a high level of clinical training. However, the FAO is still unable to match the function of behavior identified in a functional analysis approximately half of the time. Agreement between the functional analysis and the FAO categories was higher in the home than school setting. Within the school setting, agreement between the perceived function category and the functional analysis was 35%. Agreement between the actual consequences category and the functional analysis was 6%. Within the home setting, agreement between the perceived function category and the functional analysis was 57%. Agreement between the actual consequences category and the functional analysis was 14%. This data implies observations are more accurate in identifying the function of behavior in the home setting. However, this may have also occurred because of the differences in the settings in which the functional analyses were conducted. This further supports the need to consider the setting when collecting assessment data. There were fewer behaviors (10) identified in the home setting in comparison to the school setting (12). This small difference does not appear to play a role in the explanation of the differences observed; however, future research should consider the number of behaviors in an empirical manner when making comparisons across instruments and categories of instruments.

PAGE 110

99 Throughout this study, the descriptive assessment instruments were more likely to agree with the results of the functional analyses when escape or automatic reinforcement was identified as the function of behavior. This indicates that descriptive assessment instruments may be more accurate when identifying these particular functions of behavior. Future research needs to be conducted to determine if descriptive assessment instruments are more accurate in identifying certain functions of behavior. Also, across environments, the functions of behaviors identified by the functional analyses were more likely to agree when the function identified was escape or automatic reinforcement. This may imply that certain fiinctions of behavior are more likely to be the same across environments. Further research is necessary to determine if specific functions of behavior are more stable across environments. Limitations and Extensions Limitations The results of this investigation contribute to the functional assessment literature. However, some limitations of this study must be noted. First, the investigators were trained in the administration and scoring of the FAO, FAI, and MAS. Training may have inflated the results. Therefore, if teachers were to use these instruments with little or no training, the resulting data may be less valid than the current findings. Future research should examine the effects of training on the use and interpretation of these instruments. Second, this study was conducted with only three students. Future studies are needed using a larger number of participants to determine if the findings here have external validity. Third, consistent rules were not established in the functional analysis phase to ensure that all individual behaviors assessed during the functional analysis were

PAGE 111

continued until each demonstrated a specific function. For example, aggression occurred so infrequently across individuals that to continue to conduct the functional analysis until a steady state of behavioral function occurred would have been impractical. Future research should address determination of primary target behaviors and then conduct functional analyses on those behaviors until clear functions are identified. Fourth, the setting in which the functional analyses were conducted differed. In the school setting, functional analyses were conducted in an analog setting. However, in the home setting, functional analyses were conducted in the natural environment. Descriptive assessments were conducted in the natural environment at school and home. Differences in settings may have influenced the results. Future studies should compare descriptive assessments conducted in the natural environment to fimctional analyses conducted in the natural environment. Fifth, treatment was developed based on the results of the functional analysis. To determine if the functions identified by the descriptive assessments were useful, treatment should have been developed and implemented based on the descriptive assessment results. Future studies should compare treatments based on functional analysis results to those based on descriptive assessment results. Sixth, raters were not completely reliable when identifying the function of behavior determined by the functional analyses. Therefore, caution should be used when drawing conclusions based on functional analyses that were not reliable. Finally, this study was conducted on individuals with autism. It is important for future studies to be conducted on other populations of students to determine the validity of these instruments. However, it should be noted that recent literature reviews have indicated that integrated settings were rarely

PAGE 112

101 the focus of functional assessment for students identified with disabilities, so this research adds to the sparse literature (Ervin et al., 2001). Extensions There are several suggestions for future research which have been discussed throughout this section such as the replication of data interpretation methods for the FAO, examination of the effects of the number of behaviors on the reliability and validity of results, examination of the effects of training, examination of the effects of the assessment environment, examination of similar types of instruments to the functional analysis results (e.g., multiple indirect or direct assessments but not combining the two types), additional evaluation of functional analyses across settings to determine the external validity of functional analyses, and the examination of descriptive assessments based on functions identified. Since the purpose of functional assessment is to utilize the information to develop and implement function based treatments (Gresham et al., 2001), a summary of Josh's treatment data at school is discussed to demonstrate how the data collected from these instruments and assessments was used to extend the assessment findings. After a review of functional assessment data and consulting with his parent and teacher, it was determined that treatment should be aimed at reducing Josh's disruptive and noncompliant behavior and increasing his task accuracy. The function of his noncompliant behavior as identified by the FAO and FAI was escape. The MAS identified attention as the function of his noncompliant behavior. The function of his disruptive behavior identified by the FAO was attention. The FAI identified attention and escape as the functions of his disruptive behavior. The MAS identified escape as the

PAGE 113

102 function of his disruptive behavior. None of the descriptive assessments agreed with the results of the functional analysis. The fimctional analysis identified tangible as the function of his noncompliant and disruptive behavior at school. Additional analyses conducted after the functional analysis demonstrated that problem behaviors were more frequent during high-demand, low-preference tasks. There was also evidence that adult proximity decreased these two problem behaviors for Josh. Therefore, Phase 1 of treatment involved providing Josh with specific directions to complete a high demand, low-preference task with a tangible item in proximity. Josh was provided with noncontingent adult proximity every 1 0 seconds. The purpose of this intervention was to determine if adult proximity alone would decrease his disruptive and noncompliant behaviors and increase task accuracy. The intervention was not successful (See Appendix I, Phase 1). Phase 2 of treatment was designed to determine if the tangible function of Josh's behavior was responsible for exerting control over problem behaviors. Similar to Phase 1, Josh was given directions to complete a high demand, low-preference task with a tangible item in proximity. An adult was present providing Josh with non-contingent proximity every 10 seconds. However, in this phase. Josh was notified in advance that he had to complete a certain amount of work (1 worksheet) to receive access to the tangible item. Josh was provided with 1 minute of access to the tangible item contingent on task completion. This intervention successfully decreased Josh's inappropriate behaviors (See Appendix 1, Phase 2). A third phase was conducted to determine if tangible reinforcement without non-contingent adult proximity was effective in reducing his target behaviors. Phase 3 was the same as Phase 2 with the removal of non-contingent adult proximity.

PAGE 114

103 This intervention successfully decreased Josh's target behaviors (See Appendix I, Phase 3). It also provided evidence that tangible reinforcement was the variable responsible for reducing behavior and not adult proximity. A fourth phase was conducted to determine if increasing the amount of access to the tangible reinforcer would successfully increase his task accuracy. Phase 4 was the same as Phase 3 with the addition of doubling the amount of access to the tangible reinforcer (2 minutes) for increased task accuracy (80% or higher). The generalization phase demonstrated successful application of the treatment into Josh's classroom setting (for three data points) and then the treatment was transferred and successfully implemented by his classroom aide for the final 2 data points of the phase. This intervention was effective in decreasing targeted behavior and increasing task accuracy (See Appendix 1, Phase 4). Before this intervention, Josh displayed relatively frequent rates of disruptive and noncompliant behavior (averaging 2 per minute). During the last phase of treatment, Josh's disruptive and noncompliant behavior was almost nonexistent, hi addition, before treatment, Josh's task accuracy averaged 43%. During the last phase of treatment. Josh's task accuracy doubled (averaging 86%). Refer to Appendix I to view a graphical display of Josh's intervention. The success of this intervention lends further support to the use of functional analyses in identifying the variable responsible for maintaining behavior and providing valuable information in the development of a successful treatment. Summary Functional assessment is becoming common practice in schools. New methods for conducting functional assessments are constantly being developed and need validation (Cunningham & O'Neill, 2000). This study demonstrated a method for comparing data

PAGE 115

104 generated from different functional assessment instruments to functional analyses. The results suggest the need for future studies to determine the reliability and validity of these and other functional assessment instruments. Overall, the results of this study suggest that these descriptive assessments have low levels of agreement with functional analyses and should not be used in isolation to identify the fianction of behavior in the general education or home setting. If descriptive assessments are used in isolation, a direct descriptive assessment appears to provide more reliable results, especially when conducted by a practitioner with extensive clinical experience and training. The FAO demonstrated a moderate level of agreement with the functional analysis and is a better instrument than the FAI and the MAS. The perceived fimction category on the FAO more accurately identified the function of behavior than the actual consequences category. These findings have important implications for practitioners and researchers. Based on these findings, practitioners should use caution when using and interpreting the results from these instruments to develop treatment plans. One piupose of functional assessment is to identify the function of problematic behaviors so that interventions can be generated. Results from this study suggest that practitioners may hypothesize different functions of behaviors depending on which instrument they use. Clearly, this will influence treatment planning and therefore the utility and effectiveness of those treatments. Future studies are needed to determine the reliability and validity of these and other functional assessment instruments and the efficacy of these instruments in the development of treatment plans. The present findings, while different from previous studies (e.g., Amdorfer et al., 1994; Cunningham & O'Neill, 2000), adds to the limited research investigating the validity of functional

PAGE 116

105 assessment instruments. However, additional studies are needed to provide evidence for one set of findings. It is imperative that the reliability and validity of functional assessment instruments continue to be assessed to discover effective and efficient methods for conducting assessments (Cunningham & O'Neill, 2000). The functions of behaviors identified by functional analyses cannot consistently be generalized across settings. In this study, the fiinctions of behaviors determined by the functional analyses differed across settings approximately 40% of the time. This finding indicates that assessment may be necessary in both settings. The goal of conducting a functional analysis is to identify the function of behavior in order to select intervention procedures suited to address the identified function. When functions of behavior differ across settings, different interventions may be necessary. If practitioners conduct a functional analysis in only one setting (e.g., school), the recommended interventions may not generalize across settings (e.g., home). At the very least, not all functions of behaviors are identified in any one of the settings. The results of this study suggest that different instruments may provide different results. At present, no descriptive functional assessment instrument has been found to consistently provide accurate and reliable information. Future studies that assess the reliability and validity of descriptive functional assessment instruments are needed. Researchers should continue to refine and validate functional assessment instruments. Until then, to reliably and accurately identify the functions of behaviors, functional analyses should be conducted in the setting where intervention is to occur. When conducting a fiinctional analysis is not possible, multiple forms of assessments should be used within and across settings. The more evidence gathered to support the hypothesized

PAGE 117

106 functions, the more confidence there will be in the accuracy of the results (Crawford et al., 1992; Davis, 1998; Gable, 1996; Homer, 1994; Lennox & Miltenberger, 1989; Symons et al., 1998; Wehby et al., 1997). Accurate assessment will lead to the development of effective interventions. Future research assessing the reliability and validity of functional assessment instruments is necessary to determine their \ ' * . -' ?! . effectiveness.

PAGE 118

APPENDIX A SAMPLE FUNCTIONAL ASSESSMENT OBSERVATION FORM

PAGE 119

108

PAGE 120

APPENDIX B SAMPLE FUNCTIONAL ASSESSMENT INTERVIEW FORM

PAGE 121

110 FUNCTIONAL ASSESSMENT INTERVIEW (FAI) Person of concern Age Sex M F Date of interview Interviewer Respondents A. DESCRIBE THE BEHAVIORS. 1. For each of the behaviors of concern, define the topography (how it is performed), frequency (how often it occurs per day, week, or month), duration (how long it lasts when it occurs), and intensity (how damaging or destructive the behaviors are when they occur). Behavior Topography Frequency Duration Intensity a. c. d. h. 2. Which of the behaviors described above are likely to occur together in some way? Do they occur about the same time? In some kind of predictable sequence or 'chain7 In response to the same type of situation? 1

PAGE 122

Ill B. DEFINE ECOLOGICAL EVENTS (SETTING EVENTS) THAT PREDICT OR SET UP THE PROBLEM BEHAVIORS. 1. What medications is the person taking (if any), and how do you believe these may affect his or her behavior? 2. What medical or physical conditions (if any) does the person experience that may affect his or her behavior (e.g., asthma, allergies, rashes, sinus infections, seizures, problems related to menstruation)? 3. Describe the sleep patterns of the individual and the extent to which these patterns may affect his or her behavior. 4. Describe the eating routines and diet of the person and the extent to which these may aflfect his or her behavior. 5a. Briefly list below the person's typical daily schedule of activities. (Check the boxes by those activities the person enjoys and those activities most associated with problems.) Enjoys Problems Enjoys Problems 6:00 n 2:00 7:00 n 3:00 8:00 n 4:00 9:00 n 5:00 10:00 n 6:00 11:00 n 7:00 12:00 n 8:00 1:00 n 9:00 2

PAGE 123

112 5b. To what extent are the activities on the daily schedule predictable for the person, with regard to what will be happening, when it will occur, with whom, and for how long? 5c. To what extent does the person have the opportunity during the day to make choices about his or her activities and reinforcing events? (e.g., food, clothing, social companions, leisure activities) 6. How many other persons are typically around the individual at home, school, or work (including staff, classmates, and housemates)? Does the person typically seem bothered in situations that are more crowded and noisy? 7. What is the pattern of staffing support that the person receives in home, school, work, and other settings (e.g., 1:1, 2:1)? Do you beheve that the number of staff, the training of staff, or their social interactions with the person affect the problem behaviors? DEFINE SPECIFIC IMMEDIATE ANTECEDENT EVENTS THAT PREDICT WHEN THE BEHAVIORS ARE UKELY AND NOT LIKELY TO OCCUR. 1. Times of Day: When are the behaviors most and least hkely to happen? Most likely: Least likely: 3

PAGE 124

113 2. Settings: Where are the behaviors most and least likely to happen? Most likely: Least likely: 3. People: With whom are the behaviors most and least likely to happen? Most likely: Least likely: 4. Activity: What activities are most and least likely to produce the behaviors? Most likely: Least likely: 5. Are there particular or idiosyncratic situations or events not listed above that sometimes seem to "set ofiF" the behaviors, such as particular demands, noises, Ughts, clothing? 6. What one thing could you do that would most likely make the vmdesirable behaviors occur? 7. Briefly describe how the person's behavior would be affected if . . . a. You asked him or her to perform a difficult task. b. You interrupted a desired activity, such as eating ice cream or watching TV. c. You unexpectedly changed his or her typical routine or schedule of activities.

PAGE 125

114 d. She or he wanted something but wasn't able to get it (e.g., a food item up on a shelf). e. You didn't pay attention to the person or left her or him alone for a while (e.g., 15 minutes). D. IDENTIFY THE CONSEQUENCES OR OUTCOMES OF THE PROBLEM BEHAVIORS THAT MAY BE MAINTAINING THEM (I.E., THE FUNCTIONS THEY SERVE FOR THE PERSON IN PARTICULAR SITUATIONS). 1. Think of each of the behaviors listed in Section A, and try to identify the specific consequences or outcomes the person gets when the behaviors occur in different situations. What exactly What exactly Behavior Particular situations does he or she get? does she or he avoid? a. b. d. f gh. j. . E. CONSIDER THE OVERALL EFFICIENCY OF THE PROBLEM BEHAVIORS. EFFICIENCY IS THE COMBINED RESULT OF (A) HOW MUCH PHYSICAL EFFORT IS REQUIRED, (B) HOWOFTEN'mE BEHAVIOR IS PERFORMED BEFORE IT IS REWARDED, AND iOHOW LONG THE PERSON MUST WAIT TO GET THE REWARD. Low Hi^ Efficiency Efficiency 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 6

PAGE 126

115 WHAT FUNCTIONAL ALTERNATNE BEHAVIORS DOES THE PERSON ALREADY KNOW HOW TO DO? 1. What socially appropriate behaviors or skills can the person already perform that may generate the same outcomes or reinforcers produced by the problem behaviors? G. WHAT ARE THE PRIMARY WAYS THE PERSON COMMUNICATES WITH OTHER PEOPLE? 1. What are the general expressive communication strategies used by or available to the person? These might include vocjil speech, signs/gestures, communication boards/books, or electronic devices. How consistently are the strategies used? 2. On the following chart, indicate the behaviors the person uses to achieve the communicative outcomes listed: Communicative Functions Complex speech (sentences) Multiple-word phrases 1 n V u C « u •E O 1 0) a 0 Echolalia 1 Other vocalizing 1 Complex signing 1 09
PAGE 127

116 3. With regard to the person's receptive communication, or ability to understand other persons . . . a. Does the person follow spoken requests or instructions? If so, approximately how many? (List if only a few.) b. Does the person respond to signed or gestural requests or instructions? If so, approximately how many? (List if only a few.) c. Is the person able to imitate if you provide physical models for various tasks or activities? (Listifonly afew.) d. How does the person typically indicate yes or no when asked if she or he wants something, wants to go somewhere, and so on? H. WHAT ARE THINGS YOU SHOULD DO AND THINGS YOU SHOULD AVOID IN WORKING WITH AND SUPPORTING THIS PERSON? 1. What things can you do to improve the likelihood that a teaching session or other activity will go well with this person? 2. What things should you avoid that might interfere with or disrupt a teaching session or activity with this person? I. WHAT ARE THINGS THE PERSON LIKES AND ARE REINFORCING FOR HIM OR HER? 1. Food items: 7

PAGE 128

117 2. Toys and objects: 3. Activities at home: 4. Activities /outings in the community: 5. Other: J. WHAT DO YOU KNOW ABOUT THE HISTORY OF THE UNDESIRABLE BEHAVIORS THE PROGRAMS THAT HAVE BEEN ATTEMPTED TO DECREASE OR ELIMINATE THEM' AND THE EFFECTS OF THOSE PROGRAMS? How long has this Behavior been a problem? Programs Effects 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 8

PAGE 129

118 K DEVEIX)PSmiMARYSTATEMENTSF0REAOTMAJORPREDICroRAOTl«RCO^ Distant Setting Immediate Antecedent Problem Maintaining Event (Predictor) Behavior Consequence O'Neill etal. (1997)

PAGE 130

APPENDIX C SAMPLE MOTIVATIONAL ASSESSMENT SCALE

PAGE 131

120 MOTIVATION ASSESSMENT SCALE NRHE . ^"^^ BEHAVIOR DESCRiroiOH SETTING DESCRIPTION INSTRUCTIONS I The Motivation AeeeBenient Scale is a questionnaire designed to identify those situations in which an individual is likely to behave in certain ways. From this information, more informed decisions can be made concerning the selection of appropriate reinforcers and treatments. To complete the Motivation Assessment Scale, select one behavior that is of particular interest. It is important that you identify the behavior very specifically. Aggressive, for exainple, is not as good a description as "hits his sister". Once you have specified the behavior to be rated, read each question carefully and record the number that corresponds to the response that best describes your observations of the behavior in the appropriate column. QUESTIONS ANSWERS 0 NEVER 1 ALMOST NEVER 2 SELDOM 3 BALF THE TIME 4 USUALLY 5 ALMOST ALWAYS ( ALWAYS 1 1. Would the behavior occur continuouslv. over and ovi»r iftthis person was left alone for long periods of time? (For example, several hours) 2. Does the behavior occur following a request to perform a difficult task? 3. Does the behavior seem to occur in response to your talking to other persons in the room? 4. Does the behavior ever occur to get an item, food, or activity that this person has been told that he or she can't have? 5. Would the behavior occur repeatedly, in the same way, for long periods of time if noone was around? (For example, rocking back and forth for over an hour) 6. Does the behavior occur when any request is made of this person? 7. Does the behavior occur whenever you stop attending to this person?

PAGE 132

121 QUESTIONS ANSWERS n v NEVER I ALHOST NEVER 2 SELDOM 3 HALF THE TIME 4 USUAUY 5 ALMOST ALWAYS ==" 6 ALWAYS 6. Does the behavior occur when you take away a activity? t' • 9. Does it appear to you 1 that thia person enjoya 1 performing the behavior? 1 fT^ f an 1 ^ABt*ttB Innlra emella, and/or sounds pleasing) 10. Does this person seem to do the behavior to upset or annoy you when you are trying to get hits or her to do what you ask? 11. Does this person seem to do the behavior to upset or annoy you when you are not paying attention or him or her? (For example, if you are sitting in a separate room, interacting with another person) 12. Does the behavior stop occurring shortly after you give this person the item, food or activity he or she has requested? 13. When the behavior is occurring, does this person seem calm and unaware of everything else going on around him or her? 14. Does the behavior stop occurring shortly after (1 to 5 minutes) you stop working or making demands of this person? 15. Does this person seem to do thfi bphft V 1 nr* f'rt naitrf^.t **w */ciiciv^uL to ucw yOu to spend some time with him or her? 16. Does the behavior seem to occur when this person has been told that he or she can't do something he or she had wanted to do?

PAGE 133

122 SEM50RX ESCAPE ATTEirriON TAMOIBLE 12. 3.^ 4. S.^ 6. 7. 8. 9.. 10. 11. 12. 13.^ 14. 15. 16. Total scora Mean score Relatlva ranking Durand & Crimmins (1988)

PAGE 134

APPENDIX D TARGET BEHAVIORS

PAGE 135

124 Problematic Behaviors Definitions Disruption (Josh, Brad, & Drake) Vocalization (Josh & Drake) Disruption consisted of any of the following, singly or in combination with other behaviors: verbal talk which was out of context or without permission, making noise by pounding on the tables, walls, or floors with a closed fist or object, throwing objects (not at another person), ripping or tearing objects, writing on inappropriate objects, screaming or yelling, spitting (not at another person), playing with materials or toys inappropriately, trying to escape work situations, throwing work materials, yelling at others, verbal threats, and inappropriate manding. Vocalizations included any noises, words or phrases made by the participant that did not have to do with the specific task at hand. Noises included nonsense vocalizations, singing, humming, or talking to self Words and phrases were in the form of echolalia, which was the verbal repeating of what had been said by others. Noncompliance (Josh, Brad, & Drake) Off-task (Josh & Brad) Aggression (Brad and Drake) Stereotypy (Drake) Tantrums (Drake) Noncompliance consisted of the failure to complete an instruction, or to begin following instructions five seconds after the request was given. ~ Off-task behavior occurred when the participant faced away from the task/material/instructor or was not engaged in the instructional activity when directed. Aggression included hitting, biting, pinching, kicking, pulling another person, throwing objects, spitting, or pushing one's bodypart of body into another person. Stereotypy included repetitive behaviors such as rocking, hand flapping, arm flicking, spinning (object or body), covering ears, finger wringing, and staring at fingers while wiggling them. Tantrums included screaming and crying or multiple topographies of behavior occurring at once. In addition, tantrum included dropping to the ground (either on knees, bottom, or back) and remaining there.

PAGE 136

APPENDIX E SUMMARY STATEMENTS FROM THE FUNCTIONAL ASSESSMENT INTERVIEW IN THE SCHOOL ENVIRONMENT

PAGE 137

126 Table E-1 Summary Statements from the Functional Assessment Interview in the School Environment for Josh Problem behaviors Consequences Primary interviewer Second interviewer Primary interviewer Second interviewer Disruption * Vocalization * Disruption Vocalization Off-task Aggression Off-task Attention & escape Attention & escape Attention Attention & escape Non-compliance * Non-compliance Escape Escape Attention and escape Attention & escape Attention & escape 80% agreement 100% agreement

PAGE 138

127 Table E-2 Summary Statements from the Functional Assessment Interview in the School Environment for Brad Problem behaviors Consequences Primary Secondary Primary Secondary interviewer interviewer interviewer interviewer Disruption * Disruption Attention Escape Non-compliance * Non-compliance Attention Attention Stereotypy * Stereotypy Attention Attention Elope * Elope Attention & escape Escape Off-task * Off-task Attention Attention Tantrum Attention Flopping * Flopping Attention Escape Aggression * Aggression Attention & escape Attention & escape 83% agreement 66% agreement

PAGE 139

128 Table E-3 Summary Statements from the Functional Assessment Interview in the School Environment for Drake Problem behaviors Consequences Primary Secondary Primary Secondary interviewer interviewer interviewer interviewer Non-compliance * Non-comnlifinrp J^dCajpc OC dllCXllliJli Vocalizations * Vocalizations Attention Automatic Stereotypy * 1 . Stereotypy Automatic Automatic Aggression * Aggression Attention & escape Attention & Tantrum * Tantrum Attention & escape escape Tangible 100% agreement 50% agreement

PAGE 140

APPENDIX F SUMMARY STATEMENTS FROM THE FUNCTIONAL ASSESSMENT INTERVIEW IN THE HOME ENVIRONMENT

PAGE 141

130 Table F-1 Summary Statements from the Functional Assessment Interview in the Home Environment fi)r Josh Problem behaviors Antecedents Consequences Primary Secondary Primary Secondary Primary Secondary interviewer interviewer interviewer interviewer interviewer interviewer Non NonEngaged in Removal of Attention Attention & compliance * compliance preferred desired & escape continued activity or activity or instruction demand difficult task Vocalization VocEngaged in Playing by Intrinsic Automatic alization solitary self or reinplay or engaged forcement watching in passive & videotape activity attention 100% agreement 1 00% agreement 50% agreement y

PAGE 142

131 Table F-2 Summary Statements from the Functional Assessment Interview in the Home Environment for Brad Problem behaviors Antecedents Consequences Primar> Secondary Primary Secondary Primary Secondary interviewer interviewer interviewer interviewer interviewer interviewer NonNonPreferred Homework Escape Escape compliance * compliance activity or demand (demand ) Off-Task Preferred activity or demand Escape Flopping Flopping Preferred activity or demand Shopping (demand) Escape Escape Disruption Low attention or access Attention or access Pica High stress Attention 67% agreement 50% agreement 1 00% agreement

PAGE 143

132 Table F-3 Summary Statements from the Functional Assessment Interview in the Home Environment for Drake Problem behaviors Antecedents Consequences Primary int. Secondary int. Primary interviewer Secondary interviewer Primary interviewer Secondary int. NonNonDemand or Demand or Escape Escape compliance * compliance non-pref. task non-pref. task Tantrum * Tantrum Reinforced No access to Attention Tangible previously tangibles or a and and escape difficult task automatic oiercuiypy oiereotypy Passive Little overall Automatic Automatic activity & attention VocVocPassive All the time Automatic Automatic alization * alization activity & attention Aggression Toward Unknown brother 100% agreement 25% agreement 43% agreement Note: * = Agreement was calculated for antecedents and consequences only on the behaviors recorded by both observers.

PAGE 144

APPENDIX G FUNCTIONAL ANALYSIS GRAPHS

PAGE 145

134 14 Ml Conditions Figure G-2. Josh disruptive behavior across conditions at school

PAGE 146

135 4 -, 3.5 1 2 3 < 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 2 3 24 25 26 27 28 29 30 31 32 33 34 35 36 37 Conditions Figure G-3. Josh noncompliant behavior across conditions at school Figiire G-4. Josh disruptive and noncompliant behavior across conditions at school

PAGE 147

136 Figure G-6. Josh noncompliant behavior across conditions at home

PAGE 148

Figure G-8. Josh disruptive and noncompliant behavior across conditions at home

PAGE 149

I . . . ' ' ' ' « 138 5 I 2 > 5 6 7 8 9 10 11 12 13 14 t6 16 17 18 19 20 21 22 23 24 25 26 27 28 Condltiona Figure G-9. Brad disruptive behavior across conditions at school 3-1 Conditions Figure G-10. Brad noncompliant behavior across conditions at school

PAGE 150

139 Conditions Figure G-11 . Brad aggressive behavior across conditions at school 10 1 5 6 7 8 9 10 11 U 13 14 16 16 17 18 1« 20 21 22 23 24 25 26 27 28 Conditions Figure G-12. Brad disruptive, noncompliant, and aggressive behavior across conditions at school

PAGE 151

140 7 T 6 Conditions Figure G-13. Brad disruptive behavior across conditions at home 1.4 T Conditions Figure G-14. Brad noncompliant behavior across conditions at home

PAGE 152

141 8 -; Figure G-16. Brad disruptive, noncompliant, and aggressive behavior across conditions at home

PAGE 153

142 2.S 1.4 0 I x o — •-e — «-e — ^ *-e ^-e — xo 4 0 — x-e — •-e — •-e — *-e — x^ ^ — -e — ' 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Condition* Figure G-17. Drake noncompliant behavior across conditions at school Conditions Figure G-18. Drake tantrum behavior across conditions at school

PAGE 154

143 Figure G-20. Drake disruptive behavior across conditions at school

PAGE 155

144 3 T Conditions Figure G-21. Drake noncompliant behavior across conditions at home 1.8 T Conditions Figure G-22. Drake tantrum behavior across conditions at home

PAGE 156

145 9n Conditions Figure G-23. Drake vocalization behavior across conditions at home < 1.2 n Condltioni Figure G-24. Drake disruptive behavior across conditions at home

PAGE 157

APPENDIX H FUNCTIONS OF BEHAVIOR IDENTIFIED BY INVESTIGATORS

PAGE 158

147 Table H-1 Reliability Data on Functional Analysis Graphs for Josh Reliability at school Behavior^ Primarv i n vesti pator ill vC'Oii^aiLfi idle Vocalization Tangible Tangible 1/1 DisniDtion Tanpihip Toncti nip 1/ 1 Noncompliance Tangible Tangible 1/1 Disruption and noncompliance Tangible Tangible 1/1 Total agreement 4/4= 100% Reliability at home Behavior(s) Primary Secondary Agreement investigator investigator rate Vocalization Automatic Automatic 1/1 Noncompliance Tangible and escape Tangible and escape 1/1 Disruption Tangible and escape Tangible and escape 1/1 Disruption and Tangible and escape Tangible and escape 1/1 noncompliance Total agreement 4/4= 100%

PAGE 159

148 Table H-2 Reliability Data on Functional Analysis Graphs for Brad Reliability at school * xillllaiy lIlVcallgd.lOr Secondary investigator Agreement rate ioii^iuic ana escape Escape 1/2 Disruption Tangible and escape Tangible and escape 2/2 Aggression Tangible Tangible and escape 1/2 Noncompliance, Tangible and escape Tangible and escape 2/2 disruption, and aggression Total agreement 6/8 = 75% Reliability at home Behavior(s) Primary investigator Secondary investigator Agreement rate Noncompliance Escape Escape 1/1 Disruption Tangible Tangible and escape 1/2 Aggression Inconclusive Escape 0/1 Noncompliance, Tangible and escape Tangible and escape 2/2 disruption, and aggression Total agreement 4/6 = 67%

PAGE 160

149 Table H-3 Reliability Data on Functional Analysis Graphs for Drake Reliability at school JUClld V IVJi^iS J rllllloiy lllVcbllgalOr oeconuary Agreement investigator rate Noncompliance Escape Escape 1/1 Vocalization Automatic Automatic 1/1 Tantrum Tangible Tangible 1/1 Disruption Tangible and escape Tangible and escape 2/2 Total agreement 5/5 = 100% Reliability at home Behavior(s) Primary Investigator Secondary Agreement investigator rate Noncompliance Escape Escape 1/1 Vocalization Automatic Automatic 1/1 Tantrum Inconclusive Inconclusive Disruption Tangible Tangible 1/1 Total agreement 3/3 = 100%

PAGE 161

APPENDIX I GRAPHS OF SCHOOL INTERVENTION WITH JOSH

PAGE 162

151 Baseline Ph 1 Ph 2 Ph 1 Ph 2 Ph 3 Ph4 Gen. 10 0) Q. 0) w o o Q. 2 Problem Aide i TCO If) hO) 1-T-T-T-CNCMCMCgCMCOCO in CO CO Sessions Figure I-l. Josh noncompliant and disruptive behaviors at school during treatment phases Ph 1 Ph 2 Ph 1 Ph 2 Ph 3 Sessions Figure 1-2. Josh task accuracy at school during treatment phases

PAGE 163

REFERENCES American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4"^ ed.). Washington, DC: American Psychiatric Association. Amdorfer, R. E., &. Miltenberger, R. G. (1993). Functional assessment and treatment of challenging behavior: A review with implications for early childhood. Topics in Early Childhood Special Education. 13. 82-105. Amdorfer, R. E., Miltenberger, R. G., Woster, S.H., Rortvedt, A. K., & Gaffaney, T. (1994). Home-based descriptive and experimental analysis of problem behaviors in children. Topics in Early Childhood Special Education. 14. 64-87. Bailey, A., Phillips, W., & Rutter, M. (1996). Autism: Towards an integration of clinical, genetic, neuropsychological, and neurobiological perspectives. The Journal of Child Psychology and Psychiatry and Allied Disciplines. 37 . 89-126. Bihm, E. M., Kienlen, T. L., Ness, M. E., & Poindexter, A. R. (1991). Factor structure of the Motivational Assessment Scale for persons with mental retardation. Psychological Reports. 84. 1235-1238. Bijou, S. W., Peterson, R. F., & Ault, M. H. (1968). A method to integrate descriptive and experimental field studies at the level of data and empirical concepts. Journal of Applied Behavior Analysis. 1. 175-191. Blakeslee, T., Sugai, G., & Gruba, J. (1994). A review of functional assessment use in data-based intervention studies. Journal of Behavioral Education. 4. 397-413. Boomer, L. W., & Garrison-Harrell, L. (1995). Legal issues concerning children with autism and pervasive developmental disabilities. Behavioral Disorders. 21(1). 5361. Carlson, J. K., Hagiwara, T., & Quinn, C. (1998). Assessment of Students with Autism. In R.L. Simpson & B.S. Myles (Eds.), Educating children and youth with autism: Strategies for effective practice (pp. 1-20). Austin, TX: Pro-Ed. Cart, E. G., Langdon, N.A., 8c Yarbrough, S. C. (1999). Hypothesis-based intervention for severe problem behavior. In A.C. Repp & R.H. Homer (Eds.), Functional analysis of pro blem behavior: From effective assessment to effective support (pp. 9-31). Belmont, CA: Wadsworth Publishing Company. 152

PAGE 164

153 Chait, A (2001). A comparison of descriptive functional assessment instruments for children with autism. Unpublished master's thesis, University of Florida. Community Alliance For Special Education (CASE) and Protection and Advocacy, Inc. (PAl). (2000). Special Education Rights and Responsibilities . Sacramento, CA: PEERS Project. Conroy, M., Fox, J., Crain, L., Jenkins, A., & Belcher, K. (1996). Evaluating the social and ecological validity of analog assessment procedures for challenging behaviors in young children. Education and Treatment of Children, 19, 233-256. Cook, T. D. & Campbell, D. T. (1979). Quasi-experimentation: Design and analysis issues for field setting. Boston, MA: Houghton Mifflin Company. Crawford, J., Brockel, B., Schauss, S., & Miltenberger, R. G. (1992). A comparison of methods for the fiinctional assessment of stereotypic behavior. Journal of the Association for Persons with Severe Handicaps, 1 7, 77-86. Cunningham, E., & O'Neill, R. E. (2000). Comparison of results of functional assessment and analysis methods with young children with autism. Education and Training in Mental Retardation and Developmental Disabilities. 35. 406-4 14. Davis, C. A. (1998). Functional assessment: Issues in implementation and applied research. Preventing School Failure. 43. 34-36. DeLeon, 1. G., & Iwata, B. A. (1996). Evaluation of multiple-stimulus presentation format for assessing reinforcer preference. Journal of Applied Behavior Analysis. 29,519-533. Derby, M. K., Wacker, D. P., Sasso, G., Steege, M., Northup, J., Cigrand, K., & Asmus, J. (1992). Brief functional assessment techniques to evaluation aberrant behavior in an outpatient setting: A summary of 79 cases. Journal of Applied Behavior Analysis. 25. 713-721. Dunlap, G., Kern, L., dePerczel, M., Shelley, C, Wilson, D., Childs, K. E., White, R., & Falk, G. D. (1993). Functional analysis of classroom variables for students with emotional and behavioral disorders. Behavioral Disorders. 18. 275291. Durand, V. M., & Carr, E. G. (1991). Functional communication training to reduce challenging behavior: Maintenance and application in new settings. Journal of Applied Behavior Analvsis, 24, 25 1 -264. Durand, V. M., & Crimmins, D. B. (1988). Identifying the variables maintaining selfinjurious behavior. Journal of Autism and Developmental Disorders. 18. 99-1 17.

PAGE 165

154 Durand, V. M., & Crimmins, D. B., Caulfield, M., & Taylor, J. (1989). Reinforcer assessment I: Using problem behaviors to select reinforcers. Journal of the Association of Persons with Severe Handicaps, 14, 1 13-126. Eaves, L. C, & Ho., H. H. (1997). School placement and academic achievement in children with autistic spectrum disorders. Journal of Developmental «fe Physical Disabilities. 9 (4). 277-29 1 . Ervin, R. A., Radford, M. P., Bertsch, K., Piper, A. L., Ehrhardt, K. E., & Poling, A. (2001). A descriptive analysis and critique of the empirical literature on schoolbased functional assessment. School Psychology Review. 30 (2), 193-210. Fox, J., Conroy, M., & Heckaman, K. (1998). Research issues in functional assessment of challenging behaviors of students with emotional and behavioral disorders. Behavioral Disorders. 24. 26-33. Fox, L., Dunlap, G., & Philbrick, L. A. (1997). Providing individual supports to young children with autism and their families. Journal of Early Intervention. 21, 1-14. Freeman, K. A., Anderson, C. M., & Scotti, J. R. (2000). A structured descriptive methodology: Increasing agreement between descriptive and experimental analyses. Education and Training in Mental Retardation and Developmental Disabilities. 35. 55-66. Gable, R. A. (1996). A critical analysis of functional assessment: Issues for researchers and practitioners. Behavioral Disorders. 22. 36-40. Galensky, T. L., Miltenberger, R. G., Strieker, J. M., & Garlinghouse, M. A. (2001). Functional assessment and treatment of mealtime behavior problems. Journal of Positive Behavior Interventions. 3 (4) . 21 1-224. Gresham, F. M., & MacMillan, D. L. (1997). Autistic recovery? An analysis and critique of the empirical evidence on the early intervention project. Behavioral Disorders. 22 (4). 185-201. Gresham, F. M., Watson, T. S., & Skinner, C. H. (2001). Functional behavioral assessment: Principles, procedures, and future directions. School Psychology Review. 30. 156-172. Happe, F., & Frith, U. (1996). The neuropsychology of Autism. Brain. 119. 1377-1400. Harding, J., Wacker, D. P., Cooper, L. J., Asmus, J., Jensen-Kovalan, P., & Grisolano, L. (1999). Combining descriptive and experimental analysis of young children with behavior problems in preschool settings. Behavior Modification. 23. 316-333.

PAGE 166

155 Heckaman, K., Conroy, M., Fox, J., & Chait, A. (2000). Functional assessment-based intervention research on students with or at risk for emotional and behavioral disorders in school settings. Behavioral Disorders, 25, 196-210. Reward, W. L. (1996). Exceptional children. Englewood Cliffs, NJ: Merrill. Homer, R. H. (1994). Functional assessment: Contributions and future directions. Journal of Applied Behavior Analysis. 27, 401-404. Homer, R. H. (2000). Poshive behavior supports. Focus on Autism and Other Developmental Disabilities. 15. 97-105. (Reprinted from Mental Retardation in the2f' Centurv by M. L. Wehmeyer and J. R. Patton (Eds.), 2000, Austin, TX: PRO-ED. IDEA (1997). Individual with Disabilities Education Act Amendments of 1997, Public Law 105-17, 20 USC Chapter 33, Section 1415 et seq. (EDLAW, 1997). Iwata, B. A. (1994). Functional analysis methodology: Some closing comments. Joumal of Applied Behavior Analysis. 27. 413-418. Iwata, B. A., Dorsey, M. F., Slifer, K. J., Bauman, K. E., & Richman, G. S. (1982/1994). Toward a functional analysis of self-injury. Joumal of Applied Behavior Analysis, 27j 197-209. (Reprinted from Analysis and Intervention in Developmental Disabilities. 2. 3-20. 1982). Iwata, B. A., Pace, G. M., Dorsey, M. F., Zarcone, J. R., Vollmer, T. R., Smith, R. G., Rodgers, T. A., Lerman, D. C, Shore, B. A., Mazaleski, J. L., Goh, H. L., Cowdery, G. E., Kalsher, M. J., McCosh, K. C, & Willis, K. D. (1994). The functions of self-injurious behavior: An experimental-epidemiological analysis. Joumal of Applied Behavior Analysis, 27. 2 1 5-240. Kazdin, A. E. (1982). Single-case research designs: Methods for clinical and applied settings. New York: Oxford University Press. Lalli, J. S., Browder, D. M., Mace, F. C, & Brown, D. K. (1993). Teacher use of descriptive analysis data to implement interventions to decrease students' problem behaviors. Joumal of Applied Behavior Analysis. 26. 227-238. Lennox, D. B., & Miltenberger, R. G. (1989). Conducting a functional assessment of problem behavior in applied settings. Joumal of the Association for Persons with Severe Handicaps. 14. 304-3 1 1 . Lerman, D. C, & Iwata, B. A. (1993). Descriptive and experimental analyses of variables maintaining self-injurious behavior. Joumal of Applied Behavior Analysis. 26. 293-319.

PAGE 167

156 Mace, F. C. (1994). The significance and future of functional analysis methodologies. Journal of Applied Behavior Analysis, 27, 385-392. Mace, F. C, & Lalli, J. S. (1991). Linking descriptive and experimental analyses in the treatment of bizarre speech. Journal of Applied Behavior Analysis, 24, 553-562. Mace, F. C, Lalli, J. S., & Lalli, E. P. (1991). Functional analysis and treatment of aberrant behavior. Research in Developmental Disabilities, 12, 155-180. Mace, F. C, & Roberts, M. L. (1993). Factors affecting selection of behavioral interventions. In J. Reichle & D. P. Wacker (Eds.), Communicative alternatives to challenging behavior: Integrating functional assessment and intervention strategies (pp. 1 13-133). Baltimore, MD: Brookes Publishing Co., Inc. Matson, J. L., Benavidez, D. A., Compton, L. S., Paclawskyj, T., &. Baglio, C. (1996). Behavioral treatment of autistic persons: A review of research from 1 980 to the present. Research in Developmental Disabilities, 17. 433-465. McConnell, M. E., Hilvitz, P. B., & Cox, C. J. (1998). Functional assessment: A systematic process for assessment and intervention in general and special education classrooms. Intervention in School and Clinic. 34. 10-20. Messick, S. (1995). Validity of psychological assessment: Validation of inferences from persons' responses and performances as scientific inquiry into score meaning. ; ; American Psychologist, 50. 741-749. Mulick, J. A., & Meinhold, P. M. (1994). Developmental disorders and broad effects of the environment on learning and treatment effectiveness. In E. Schopler & G.B. Mesibov (Eds.) Behavioral issues in autism (pp. 99-128). Newton, T., & Sturmey, P. (1991). The Motivation Assessment Scale: Inter-rater reliability and internal consistency in a British sample. Journal of Intellectual Disabilities Research. 35. 372-374. O'Neill, R., Homer, R. H., Albin, R. W., Sprague, J. R., &. Storey, K. (1997). Functional assessment and program development for behavior problems (2"** edition!. CA: Brooks/Cole Publishing Co. Parsonson, B. S., & Baer, D. M. (1992). The visual analysis of data, and current research into the stimuli controlling it. In T. Kratochwill & J. Levin (Eds.), Single-case research design and analysis: New directions for psychology and education (pp. 15-40). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Repp, A. C, & Homer, R. H. (1999). Functional analysis of problem behavior: From effective ass essment to effective support. Belmont, CA: Wadsworth Publishing Company.

PAGE 168

157 Sasso, G. M., Reimers, T. M., Cooper, L. J., Wacker, D., Berg, W., Steege, M., Kelly, L., & Allaire, A. (1992). Use of descriptive and experimental analyses to identify the functional properties of aberrant behavior in school settings. Journal of Applied Behavior Analvsis. 25. 809-821. Schopler, E., & Mesibov, G. B. (1994). Introduction to behavioral issues in autism. In E. Schopler & G. B. Mesibov (Eds.), Behavioral issues in autism (pp. 3-10). New York, NY: Plenum Press. Schreibman, L. (1994). General principles of behavior management. In E. Schopler & G.B. Mesibov (Eds.) Behavioral issues in autism (pp. 1 1-38). Scott, T. M., DeSimone, C, Fowler, W., & Webb, E. (2000). Using functional assessment to develop interventions for challenging behaviors in the classroom: Three case studies. Preventing School Failure, 44. 5 1-56. Shapiro, E. S., & Kratochwill, T. R. (2000). Introduction: Conducting a multidimensional behavioral assessment. In E. S. Shapiro & T. R. Kratochwill (Eds.), Conducting school-based assessments of child and adolescent behavior (pp. 1 -20). New York, NY: The Guilford Press. Shriver, M. D., Anderson, C. M., & Proctor, B. (2001). Evaluating the validity of functional behavior assessment. School Psvcholoev Review. 30. 180-192. Simpson, R. L. & Myles, B. S. (1998). Understanding and responding to the needs of students with autism. In R. L. Simpson & B. S. Myles (Eds.), Educating children and youth with autism: Strategies for effective practice (pp. 1-20). Austin, TX: Pro-Ed. Simpson, R. L. & Sasso, G. M. (1992). Full inclusion of students with autism in general education settings: Values versus science. Focus on Autistic Behavior. 7(3). 1-13. Singh, N. N., Donatelli, L. S., Best, A., Williams, E. E., Barrera, F. J., Lenz, M.W., Landrum, T. J., Ellis, C. R., & Moe, T. L. (1993). Factor structure of the Motivation Assessment Scale. Journal of Intellectual Disabilities Research. 19. 56-90. Sturmey, P. (1994). Assessing the functions of aberrant behaviors: A review of psychometric instruments. Journal of Autism and Developmental Disorders. 24. 293-304. Symons, F. J., McDonald, L. M., & Wehby, J. (1998). Functional assessment and teacher collected data. Education and Treatment of Children. 21. 135-159. Tapp, J., Wehby, J., & Ellis, D. (1995). A mukiple option observation system for

PAGE 169

158 experimental studies: MOOSES. Behavioral Research Methods. Instruments, and Computers, 27, 25-3 1 . Taylor, J. (1994). Functional assessment and functionally-derived treatment for child behavior problems. Special Services in the Schools, 9, 39-67. Touchette, P. E., Macdonald, R. F., & Langer, S. N. (1985). A scatter plot for identifying stimulus control of problem behavior. Journal of Applied Behavior Analysis. 1 8, 343-351. Trevarthen, C, Aitken, K., Papoudi, D., & Robarts, J. (1996). Children vyith autism: Diagnosis and interventions to meet their needs . Bristol, PA: Jessica Kingsley Publishers. Umbreit, J. (1995). Functional assessment and intervention in a regular classroom setting for the disruptive behavior of a student with Attention Deficit Hyperactivity Disorder. Behavioral Disorders. 20. 267-278. U.S. Department of Education (1998). To assure the free and appropriate education of all children with disabilities: Twentieth annual report to congress on the implementation of the Individuals with Disabilities Education Act. Washington, DC: author. Volmer, L. (1995). Best practices in working with students with autism. In Thomas, A., & Grimes, J. (Eds.), Best practices in school psvchology. Washington, DC: National Association of School Psychologists. Wehby, J. H., & Symons, F. J. (1996). Revisiting conceptual issues in the measurement of aggressive behavior. Behavioral Disorders, 22. 29-35. Wehby, J. H., Symons, F. J., & Hollo, A. (1997). Promote appropriate assessment. Journal of Emotional and Behavioral Disorders. 5. 45-54. Wing, L. (1997). The autistic spectrum. The Lancet. 350 . 451-459. Zarcone, J. R., Rodgers, T. A., Iwata, B. A., Rourke, D. A., & Dorsey, M. F. (1991). Reliability analysis of the Motivation Assessment Scale. Research in Developmental Disabilities. 12. 349-360.

PAGE 170

BIOGRAPHICAL SKETCH Andrea Chait was bom in Buffalo, New York. She completed her undergraduate studies in health education at Ithaca College in 1992. After receiving her degree, she worked as a paraprofessional at Stanley G. Falk, in Buffalo, New York (a school for students with emotional and behavioral disorders). This fostered her desire to learn more about students with disabilities and to pursue her degree in special education. In 1995, she completed her Master of Education degree from the University of Florida in special education. After graduation, she went on to teach students with disabilities at Paul D. West Middle School in Atlanta, Georgia. With a continued interest in understanding and helping students with disabilities, Andrea returned to the University of Florida to study school psychology. In 2001, she completed her Master of Arts in Education degree from the University of Florida (in school psychology). Upon completion of her Doctor of Philosophy degree in school psychology, she plans to become gainfully employed. Andrea hopes to make a difference in the lives of children. 159

PAGE 171

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor o^^^ilosophy. ' ()y\ Jennifer M. Asmus, Chair Assistant Professor of Educational Psychology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is ftjlly adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Maureen A. Conroy (^j " ' \ Associate Professor of Special Education I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. V V> tliia Smith-Bonahue Associate Professor of Educational Psychology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Greg Valca Director of the Center for Autism and Related Disabilities, Gainesville, Florida This dissertation was submitted to the Graduate Faculty of the College of Education and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. August 2002 "Chairman, Educational Psychology Dean, Graduate School