Citation
Parent-child interaction therapy with behavior problem children

Material Information

Title:
Parent-child interaction therapy with behavior problem children generalization of treatment effects to the school setting
Creator:
Bodiford, Cheryl Ann, 1961-
Publication Date:
Language:
English
Physical Description:
vii, 133 leaves : ill. ; 29 cm.

Subjects

Subjects / Keywords:
Behavior problems ( jstor )
Child psychology ( jstor )
Children ( jstor )
Classrooms ( jstor )
Deviant behavior ( jstor )
Parent training ( jstor )
Parents ( jstor )
Schools ( jstor )
Tax noncompliance ( jstor )
Treatment compliance ( jstor )
Clinical and Health Psychology thesis Ph.D ( mesh )
Dissertations, Academic -- Clinical and Health Psychology -- UF ( mesh )
Family Therapy ( mesh )
Parent-Child Relations ( mesh )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph.D.)--University of Florida, 1989.
Bibliography:
Bibliography: leaves 126-132.
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Cheryl Ann Bodiford.

Record Information

Source Institution:
University of Florida
Holding Location:
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:
024264832 ( ALEPH )
20895969 ( OCLC )

Downloads

This item has the following downloads:


Full Text
PARENT-CHILD INTERACTION THERAPY WITH
BEHAVIOR PROBLEM CHILDREN: GENERALIZATION OF
TREATMENT EFFECTS TO THE SCHOOL SETTING
By
CHERYL ANN BODIFORD
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 1989




ACKNOWLEDGMENTS
I would like to thank each of my committee members, Dr. Sheila Eyberg, Dr. Randy Carter, Dr. Eileen Fennell, Dr. Jacque Goldman, and Dr. James Johnson, for providing conceptual guidance and critical reviews for this manuscript. I particularly would like to thank the chair of my committee, Dr. Eyberg, for the long evenings she put into this research, as well as for her knowledgeable contributions and support.
Toni Hembree Eisenstadt, M.S., deserves special
acknowledgment for her assistance and dedication to the study from its conception through completion. Other valued contributors to this work include Katharine Newcomb, Beverly Funderburk, Bill Eisenstadt, and Lori Rossi.
I also would like to thank those individuals who provided much needed moral support throughout this research. I appreciate motivating phone calls from Dr. Shari Rediess that assisted me in meeting unreasonable deadlines. Dr. Dan McNeil also is very appreciated for the patience, support, consultation, and equipped office that he provided during the manuscript preparation phase of this project. Finally, I very much would like to
ii




thank my parents, Otis and Donna Bodiford, for providing continued support of my educational goals and making the attainment of this degree possible.
iii




TABLE OF CONTENTS
page
ACKNOWLEDGMENTS ................................... ii
ABSTRACT .......................................... vi
CHAPTERS
1 INTRODUCTION ................................1
Review of Effectiveness of Treatment
Components ................................ 4
Review of Treatment Efficacy Studies........ 10
Maintenance Literature (Generalization
Over Time) ................................ 16
Generalization to Home, Untreated
Behavior, and Siblings .................... 19
Overview of Hanf Model Treatment Outcome
Literature ................................ 24
Review of Generalization to School............ 26
Statement of Problem and Hypotheses........... 36
2 METHOD ...................................... 40
Subjects .................................... 40
Treatment Outcome Measures..................... 44
School Generalization Measures................. 49
Procedures .................................. 59
3 RESULTS ..................................... 64
4 DISCUSSION .................................. 83
APPENDICES
A DSM-III-R CHECKLIST...................... 99
B EYBERG CHILD BEHAVIOR INVENTORY....... 104 C CLASSROOM CODING SYSTEM................ 107
D REVISED CONNERS TEACHER RATING SCALE.. 113
E SUTTER-EYBERG STUDENT BEHAVIOR
INVENTORY ........................... 115
iv




page
APPENDICES (continued)
F THE WALKER-McCONNELL SCALE OF SOCIAL
COMPETENCE AND SCHOOL ADJUSTMENT: A SOCIAL SKILLS RATING SCALE FOR
TEACHERS ............................ 118
G SAMPLE TREATMENT SESSION OUTLINE:
SECOND PDI COACHING SESSION.......... 124 REFERENCES......... ............................... 126
BIOGRAPHICAL SKETCH ............................... 133
v




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
PARENT-CHILD INTERACTION THERAPY WITH
BEHAVIOR PROBLEM CHILDREN: GENERALIZATION OF
TREATMENT EFFECTS TO THE SCHOOL SETTING By
Cheryl Ann Bodiford
May 1989
Chairman: Sheila Eyberg, Ph.D. Major Department: Clinical and Health Psychology
Generalization of treatment effects from the home to school setting was evaluated in ten children between the ages of two and seven years who were referred for treatment of severe conduct problem behaviors that were occurring both at home and in the classroom. Referred children received 14 weeks of Parent-Child Interaction Therapy, a treatment program directed towards improving the parent-child relationship and modifying noncompliant and disruptive child behaviors. No direct classroom interventions were conducted.
The referred children demonstrated clinically significant improvements to within normal limits on measures of home and clinic behavior problems upon completion of therapy. Post-treatment behavior changes
vi




in the school setting were evaluated through comparisons of the treatment children to ten untreated deviant classroom controls and ten normal classroom controls on multiple observational and teacher report measures. The treatment group was found to display significantly greater improvements than both control groups on all measures of conduct problem behavior in the classroom. These classroom behavioral gains were judged to be clinically significant in that the mean scores of treatment group children on these measures improved to within normal limits following therapy. Results in the areas of hyperactivity/distractibility and social behavior were less supportive of generalization. The positive school generalization results in this study contradict previous findings that children's behavior in the classroom either remains the same or worsens following parent training. Possible reasons for the discrepant findings are discussed, as well as the clinical and research implications of this investigation.
vii




CHAPTER 1
INTRODUCTION
Parent-Child Interaction Therapy is based on a model of treatment devised by Constance Hanf, Ph.D. (1969) at the Oregon Health Sciences University. The Hanf model is based on operant conditioning and was designed to modify the dysfunctional interaction patterns between multiply handicapped children and their mothers. Hanf outlined a two-stage model of treatment to be implemented using structured laboratory situations. In the first stage mothers were taught to let the children lead the play and to use differential attention for positive and negative child behaviors. Mothers were instructed to give attention to all appropriate behavior and to ignore all inappropriate behavior. The second stage was devoted to teaching mothers to lead the play. Mothers were taught to give clear directions, to praise compliance, and to punish noncompliance with time-out. One factor that particularly set the Hanf parent-child interaction model apart from other parent training models was the emphasis on working with the mother and child together in the therapy sessions and providing immediate feedback to the
1




2
mother to aid in the acquisition of the parenting skills. The Hanf model also differed from other parent training programs in that it emphasized the restructuring of family interaction patterns rather than the modification of specific targeted behaviors.
Several of Hanf's students and colleagues have
modified and extended the original model such that there currently are different versions of the Hanf parent-child interaction program (e.g., Barkley, 1987; Eyberg & Boggs, in press; Forehand & McMahon, 1981). Perhaps the most well-known modification of the Hanf program is the one proposed by Forehand and McMahon (1981) for the treatment of noncompliance in young children. In the book, Helping the Noncompliant Child (Forehand & McMahon, 1981), these authors describe their approach to treatment and present results from the programmatic research that they have conducted on the Hanf approach to therapy. Parent-Child Interaction Therapy (PCIT) is a specific treatment approach that also is based on the Hanf model (Eyberg, 1979). PCIT was designed by Sheila Eyberg, Ph.D., as a behavioral family therapy approach that integrates operant concepts with principles of behavior change used in traditional play therapy. While many of the Hanf techniques and skills for improving behavior problems in young children




3
are included in PCIT, there also are several unique features to the program and a slightly different theoretical framework. Compared to other Hanf model approaches, PCIT places a greater emphasis on establishing a strong, positive relationship between the parent and child by providing relatively extensive training to parents in the use of specific play therapy skills (Eyberg & Matarazzo, 1980). PCIT also can be differentiated from other parent-child interaction programs by the emphasis on problem-solving skills training as a means of helping parents to learn effective parenting skills (Eyberg & Boggs, in press).
The present study examines the issue of whether
improvements in the home behavior of children following Parent-Child Interaction Therapy generalize to the children's behavior at school. Before the methodology and findings of the current research are discussed, a review of the literature is presented to provide a framework for analyzing the study. Because school generalization cannot be examined without first documenting that treatment has been effective in the primary setting, the literature review first addresses the issue of whether Hanf model treatment programs are effective for improving parenting skills and child behavior in the home setting. This literature includes




4
studies of the effectiveness of individual components of the treatment package as well as studies of the overall efficacy of the treatment program as a whole. A review of the literature regarding treatment generalization is then presented. This literature includes generalization over time, to the home setting, to untreated behaviors, and to siblings. Finally, the school generalization literature is reviewed. Because the literature on school generalization with Hanf model programs is sparce, a review of school generalization in the general parent training literature also is provided.
Review of Effectiveness of Treatment Components
A number of studies have evaluated the
effectiveness of specific components within each of the two stages of Hanf model programs. Particular emphasis has been placed on evaluating the effectiveness of various types of commands, time-out procedures, and praise. In a study of command training, Forehand and Scarboro (1975) found that increases in the number of parental commands were related to decreases in child compliance. This finding was viewed as supporting the initial use of didactic sessions to teach parents to reduce the number of commands given to their children (Forehand & Scarboro, 1975). Similarly, research on




5
various time-out procedures has supported the time-out approach used in most Hanf model programs. Time-out has been found to be most effective when the child is removed from all potential sources of reinforcement (Scarboro & Forehand, 1975), when the length of time-out is between one minute and four minutes (Hobbs, Forehand, & Murray, 1978), and when release from time-out is under parental rather than child control (Bean & Roberts, 1981). In a study evaluating the relative contribution of time-out and commands, it was found that training parents to give specific, single commands led to an increase in child compliance when compared to a placebo control group, but that even greater improvements in child compliance were obtained when command training was combined with a time-out procedure (Roberts, McMahon, Forehand, & Humphreys, 1978). This finding supports the use of a combination of command and time-out training in the second stage of Hanf programs. With respect to the use of praise, Bernhardt and Forehand (1975) found that both labeled (specific) and unlabeled (nonspecific) praises led to increases in targeted child behaviors on a marble dropping task. However, labeled praise was found to result in a significantly greater increase in correct marble dropping than unlabeled praise (Bernhardt & Forehand, 1975). This finding supports the emphasis




6
in Hanf programs on teaching parents to use specific, labeled praise to increase appropriate child behaviors.
Walle, Hobbs, and Caldwell (1984) assessed the effectiveness of attention (in the form of praise following compliance) and time-out (in the form of standing in a corner following noncompliance) in various sequences for reducing noncompliance. Twenty-eight mother-child dyads were assigned to groups that received either no treatment or various sequences of attention and time-out training. Contrary to studies indicating that positive reinforcement techniques are more appealing to parents than punishment techniques (e.g., Calvert & McMahon, 1987; Kazdin, 1980, 1981), parents in this study rated praise and time-out as equally acceptable treatment components. The sequence in which the components were presented did not affect parental ratings of overall treatment acceptability. Attention alone was not found to produce significant reductions in noncompliance, and attention combined with time-out was not found to be superior to time-out alone. Analyses of sequencing effects, however, indicated that attention is an important variable. Receiving attention before time-out was found to enhance the efficacy of time-out for reducing noncompliance (i.e. the group that received attention before time-out was found to have greater




7
compliance in the time-out phase of the intervention than the group that received attention training after time-out). Interestingly, receiving attention after time-out also was found to have specific therapeutic benefits. Improvements in compliance were found to maintain better over time when the attention phase followed the time-out phase. As specific therapeutic benefits were obtained both when the attention stage preceded the time-out stage and when time-out was presented first, neither stage sequence was shown to be superior for the treatment of noncompliance. The authors concluded that teaching parents to use a combination of time-out for noncompliance and praise for compliance, regardless of stage sequence, is an effective strategy for producing immediate behavior change (Walle et al., 1984).
Another component of the Hanf program that has been evaluated is the relative effectiveness of various instructional techniques for teaching parenting skills. Flanagan, Adams, and Forehand (1979) compared the following four methods of teaching parents timb-out: written presentation, lecture presentation, videotaped modeling, and role-playing. After time-out training, parents in the four groups were not found to differ on a written test of knowledge of time-out principles.




8
However, role-playing was found to be superior to the other presentations when outcome was measured by parental responses to an audiotaped analogue task. When outcome was assessed through parental performance of time-out with their own children in the home, the videotaped modeling group was found to implement the time-out more effectively than the other groups (Flanagan et al., 1979).
Studies evaluating the relative effectiveness of group versus individual instruction and videotaped training versus live modeling have reported contradictory results. Webster-Stratton, Hollinsworth, and Kolpacoff (in press) compared the long-term effectiveness and clinical significance of the following three cost-effective presentations of a Hanf model program: a) group videotaped modeling with therapist-led discussion, b) an individually self-administered videotaped modeling presentation, and c) a group discussion presentation with no videotaped modeling. All presentations resulted in significant behavioral improvements as measured by parent report of behavior problems and observations in the home, and these improvements were found to be maintained at one year follow-up. The only significant finding differentiating the three presentations was that parents in the group




9
videotaped modeling presentation with therapist-led discussion reported greater satisfaction with treatment (Webster-Stratton et al., in press). In another study of the relative effectiveness of individual and group presentations, Webster-Stratton (1984) randomly assigned 35 families with conduct disordered children to the following groups: a) a waiting-list control group, b) a 9 week individual parent-child interaction program, and c) a 9 week standardized group videotape modeling program. Improvements in parenting skills and parental attitudes toward the children, as well as reductions in deviant and noncompliant behavior, were noticed equally in both groups. As no between group differences were found on the treatment outcome measures, it was concluded that individual treatment with live modeling and group treatment with videotape modeling are equally effective for treating the behavior problems of young children (Webster-Stratton, 1984).
Contradictory findings were evident in a study
comparing individual Parent-Child Interaction Therapy with group didactic training using 29 families with children who had speech and language disorders (Eyberg & Matarazzo, 1980). In the individual training group, parents demonstrated improved parenting skills, reported significantly fewer home problems, and a more positive




10
attitude toward their children. Additionally, child behaviors improved significantly with fewer children exhibiting inappropriate behavior and a mean compliance increase from 79% at pre-treatment to 91% at post-treatment. In the group didactic training, mothers reported significantly fewer home problems and a more positive attitude toward their children, but no pre- to post-treatment behavioral differences were noted in parenting skills or child behavior. Also, parents expressed more satisfaction with the individual presentation than the group didactic training. These authors concluded that parenting skills are improved most readily through direct observation and immediate feedback (Eyberg & Matarazzo, 1980). When studies of the relative effectiveness of different instructional methods are considered together, there are consistent data to support that the Hanf model is effective in both an individual and group training format. However, the issue of whether individual direct feedback training is more effective than group videotape/didactic training remains unclear.
Review of Treatment Efficacy Studies
In an initial study of the overall effectiveness of the Hanf program, Hanf and Kling (1974) treated 40 pairs




11
of mothers and their severely physically handicapped, noncompliant children. Comparisons of pre- and post-treatment data demonstrated that maternal verbalizations changed in the direction expected based on the skills taught in the program. Mothers significantly increased their use of verbal rewards and decreased their use of commands and questions. The program's focus on noncompliance also was found to be effective in that the children displayed significant increases in overall rate of compliance following treatment. Improvements in parenting skills and child compliance were found to maintain over a three month time interval (Hanf & Kling, 1974). Further evidence for the effectiveness of the Hanf program for treatment of behavior problems in handicapped children was provided in a case study of a deaf child. Consistent with the group findings, the compliance rate of this child was found to increase significantly following the parent-child interaction program (Forehand, Cheney, & Yoder, 1974).
Forehand and King (1974) empirically demonstrated the effectiveness of a modified version of Hanf's program for the treatment of noncompliant preschoolers without physical handicaps. After approximately six sessions, mothers of eight noncompliant preschool




12
children were found to increase significantly their use of rewards and decrease their use of commands and questions in the first stage of treatment (which Forehand and colleagues refer to as the Child's Game). Maternal verbal rewards also increased significantly in the second stage of treatment (i.e. Parent's Game), as did child compliance (Forehand & King, 1974). In a subsequent study Forehand and King (1977) again used the modified Hanf program for the treatment of noncompliant preschoolers. However, this study used parent perception measures as well as clinic observations of mother-child interactions to evaluate treatment outcome. As in prior studies, mothers significantly increased their use of rewards and decreased their use of commands and questions while children's compliance to parental commands increased. Results of the parent perception measures demonstrated that mothers perceived their children as significantly better adjusted following treatment. An additional comparison of the 11 treatment children to 11 nonclinic normal controls showed that the treated children were less compliant than the normals before treatment and more compliant following treatment. Although treatment mothers had more negative perceptions of their children than normal control mothers prior to treatment, scores on parent




13
perception measures were not found to differ between the two groups after the Hanf program. All improvements in parent and child behaviors were found to maintain over a three month time interval (Forehand & King, 1977).
In addition to behavioral observations and measures of parental perceptions of their children, researchers have examined consumer satisfaction and treatment acceptability as indicators of treatment efficacy. Several studies have used either the Parent's Consumer Satisfaction Questionnaire (Forehand & McMahon, 1981) or the Therapy Attitude Inventory (Eyberg, 1974) to evaluate parental satisfaction. Results indicate that Hanf parent-child interaction programs are typically rated as moderately to highly satifactory by parents (Eyberg & Matarazzo, 1980; Eyberg & Robinson, 1982; McMahon, Forehand, & Griest, 1981). While one study found that parents perceived praise and time-out as equally acceptable methods of modifying child behavior (Walle et al., 1984), most studies indicate that parents rate positive techniques for increasing deficit behavior (e.g., rewards, commands, attends) as more acceptable than strategies for reducing behavioral excesses (e.g., time-out, ignoring) (Calvert & McMahon, 1987; Kazdin, 1980, 1981). Results of studies evaluating the acceptability of various aspects of Hanf model programs




14
must be considered with caution, however, because subjects typically are not clinical subjects. Parents of conduct disordered children might view the acceptability of the Hanf treatment strategies differently than parents whose children are not in need of treatment.
Some researchers have suggested that drop-out rates can be viewed as an outcome measure or as an indicator of consumer satisfaction. In a review of 45 parent training studies in eight journals from 1972 to 1982, it was found that about half of the studies reported drop-out data (Forehand, Middlebrook, Rogers, & Steffe, 1983). Using the drop-out statistics from these studies, it was estimated that the drop-out rate in parent training is approximately 28% (Forehand et al., 1983). Although it is unknown how this rate compares to drop-out rates in other types of child psychotherapy because of the lack of research in the area, the 28% drop-out figure compares favorably to drop-out rates reported in the adult literature (Baekeland & Laundwall, 1975). Preliminary data reported by McMahon, Forehand, Griest, and Wells (1981) suggest that parents who are depressed or socioeconomically disadvantaged account for a substantial portion of the parent training drop-out




15
population. A meta-analysis of the overall drop-out rate specifically for Hanf parent-child interaction programs has not yet been conducted.
In addition to evaluating whether statistically
significant improvements in parenting skills and child behavior occur following treatment, studies have examined the issue of whether treatment gains are socially and clinically valid. Forehand, Wells, and Griest (1980) examined the social validity of a Hanf treatment program in the following ways: a) comparing referred families to nonclinic normal controls, b) having parents rate their children's progress in treatment, and c) having parents provide consumer satisfaction data. Results demonstrated that the clinic children were less compliant and more deviant than the normal controls prior to treatment but not after treatment or at the two month follow-up evaluation. Parental ratings indicated that parents perceived improvements in their children's behavior and were satisfied with the treatment program. Because the treatment gains represented not only statistical changes but normative improvements to within the normal range, the authors were able to conclude that the Hanf program resulted in a clinically significant (i.e. socially valid) outcome (Forehand et al., 1980).




16
Maintenance Literature (Generalization Over Time)
In a review of the generalization research in the parent training literature, Forehand and Atkeson (1977) discuss maintenance of treatment effects as a form of generalization termed "temporal generalization." In one study addressing the temporal generalization issue, home observations of ten parent-child pairs were conducted immediately after a Hanf parent-child interaction program and again at 6 and 12 month follow-up intervals (Forehand, Sturgis, McMahon, Aguar, Green, Wells, & Breiner, 1979). Post-treatment changes in parental attitudes toward the children, parental use of commands, attends, and questions, and child behavior were found to be maintained both at 6 and 12 months following termination of therapy (Forehand et al., 1979). However, maternal rewards (i.e. labeled and unlabeled praise) at the follow-up evaluations did not differ significantly from the pre-treatment level. This finding suggests that maternal use of praise, a central parenting skill of Hanf programs, did not generalize over time as expected.
Positive temporal generality results were obtained in several more recent maintenance studies. For example, Baum and Forehand (1981) conducted follow-up




17
assessments on 36 families at intervals ranging from
1 1/2 to 4 1/2 years following treatment. Although parental use of attends, rewards, and questions decreased from the post-treatment to follow-up evaluations, the frequency with which parents demonstrated these skills at follow-up remained significantly improved from the pre-treatment levels. Improvements in child compliance, maternal use of indirect commands, and maternal report of child behavior were found to be maintained at follow-up. Consumer satisfaction measures indicated that the parents continued to have a favorable view of treatment up to
4 1/2 years after termination of therapy. Additionally, measures of child deviant behavior taken both in the clinic and the home demonstrated that the behavior of the children further improved from the post-treatment to follow-up evaluation (Baum & Forehand, 1981). Similar results were obtained in a study by Webster-Stratton (1984) in which one year follow-up evaluations were conducted on families who had received either individual Parent-Child Interaction Therapy or a videotape modeling program. Both groups of treated mothers were found to be significantly more positive, less critical, and less negative in interactions with their children. As in the Baum and Forehand (1981)




18
study, treated children displayed significant reductions in noncompliant and deviant behaviors from the post-treatment to follow-up evaluation.
Forehand and Long (1988) conducted a long term follow-up study addressing the adjustment and functioning of 11 to 14 year old children and their families who had participated in a Hanf model treatment program for child noncompliance 4 1/2 to 10 1/2 years earlier. Of the 43 families who previously had been treated and were eligible for the study, 21 families agreed to participate. These families were compared to a control group of 21 families who had a child in early adolescence who had never received psychological treatment. The two subject groups were matched on age, gender of the adolescent, and socioeconomic status. Multiple measures of child and parent functioning were used that included observational measures, self-report, parent report, teacher report, and academic grades from school files.
With respect to externalizing behaviors, Forehand
and Long (1988) found no differences between the treated children and controls on parent report of conduct, attentional, and delinquency problems in the home. However, parents who had been in treatment reported a greater amount of conflict with their adolescents than




19
was reported by control parents. Teachers reported significantly more attentional problems and lower grades in the children who had been treated than the controls, but no differences were noted in the areas of delinquency and aggression. Observational, self-report, and parent report measures showed no differences between the groups on depression, anxiety, and social behaviors. The only internalizing or social behavior difference noted by teachers was that children who had been in treatment were viewed as being more anxious than the control children. No differences between the groups were found in the areas of parenting skills and parental adjustment. The authors concluded that these findings are generally positive and are indicative of "moderate" long-term effectiveness (Forehand & Long, 1988). Taken together, these maintenance studies support the presence of temporal generalization following the successful treatment of behavior problems in young children using a Hanf model treatment approach. Generalization to Home, Untreated Behavior, and Siblings
Most researchers agree that an adequate evaluation of parent training outcome requires an assessment of the functioning of the parent and child outside of the clinic setting. The two most common methods of




20
assessing home behavior are the use of parent questionnaires and direct observations in the home setting. An example of the use of parent questionnaires is a study by Eyberg and Robinson (1982) evaluating the effectiveness of Parent-Child Interaction Therapy for reducing child behavior problems at home. Parents reported a significant decline in the number and intensity of home behavior problems on the Eyberg Child Behavior Inventory (Robinson, Eyberg, & Ross, 1980) upon completion of treatment, indicating that behavioral improvements generalize from the clinic to home setting. With respect to direct observations, Peed, Roberts, and Forehand (1977) evaluated setting generality by conducting behavioral observations in both the clinic and home setting. Six mother-child pairs who received a Hanf model treatment program were compared to six mother-child pairs who were on a waiting-list to receive treatment. Clinic behavioral observations demonstrated improvements in maternal parenting skills and child behavior. Similarly, home behavioral observations demonstrated significant increases in child compliance and maternal skills following treatment. In contrast, no significant changes were found in the waiting list children or parents after completion of the waiting period. The authors concluded that the positive




21
treatment results were not attributable to the passage of time, nonspecific environmental events, or the effects of observation, and that generalization occurred from the clinic to home setting (Peed et al., 1977). Several more recent studies using behavioral observations in the home have provided additional evidence that improvements noted in the clinic generalize to the home setting (e.g., Baum & Forehand, 1981; Forehand, Sturgis, et al., 1979).
Treatment generalization also has been examined by assessing whether nontargeted deviant behavior improves following treatment of noncompliance using a Hanf parent-child interaction program. Wells, Forehand, and Griest (1980) compared 12 mother-child dyads who were referred for treatment of noncompliance to 12 mother-child normal controls on measures of noncompliance and deviant behavior. With respect to compliance (which was a behavior targeted for change in treatment), the clinic children were found to become significantly more compliant after treatment whereas no compliance differences were found in the nonclinic children after a comparable time period. Additionally, although the clinic and nonclinic groups differed significantly on rate of compliance before treatment, no between-group compliance differences were noted upon




22
completion of therapy. Once it was determined that treatment had been effective for the targeted behavior, a further analysis was conducted to examine generalization of improvements in child noncompliance to other deviant behaviors. The treated children were found to have significant decreases in such deviant behaviors as aggressiveness, yelling, and whining whereas the nonclinic group demonstrated no change in deviant behavior. The authors concluded that the successful treatment of noncompliance is sufficient in many cases to reduce other untreated deviant behaviors (Wells, Forehand, and Griest, 1980).
Several studies have examined the hypothesis that generalization of treatment effects from the target child to untreated siblings occurs following improvements in parental child management skills. Two case studies have been reported in which successful parent training was found to result not only in improvements in the behavior of the child targeted as having behavior problems but in the untargeted siblings as well (Laviqueur, Peterson, Sheese, & Peterson, 1973; Resnick, Forehand, & McWhorter, 1976). In a more controlled study of 8 clinic referred children, their parents, and untreated siblings, positive behavior changes in parental interactions with nontargeted




23
siblings were noted (Humphreys, Forehand, McMahon, & Roberts, 1978). Observations of the parent with the untreated sibling revealed that the parents used significantly more rewards and contingent attention and fewer indirect commands following a Hanf treatment program. These changes in parenting skills were found to be accompanied by compliance rate increases from 34% to 44% in the untreated sibling, suggesting the presence of generalization across children (Humphreys et al., 1978).
In a second controlled group study of sibling
generalization, Eyberg and Robinson (1982) evaluated the impact of treatment on nontargeted siblings in seven families. Prior to treatment, the children targeted for treatment of behavior problems were found to be 39% compliant to total parental commands and their nontargeted siblings demonstrated compliance rates of 53%. Following treatment, the rates of compliance in both groups of children improved significantly with post-treatment compliance rates of 89% and 73% respectively. Thus this study supports previous findings that improvements in parenting skills and the target child's behavior generalize to the behavior of siblings who are not directly participating in the parent-child interaction program.




24
Overview of Hanf Model Treatment Outcome Literature
Individual studies in the Hanf parent-child interaction literature can be criticized for methodological inadequacies that bring into question specific results. Examples of methodological problems in this literature include the following: a) failure to demonstrate that children had clinically significant behavior problems prior to treatment according to DSM-III criteria, b) use of improvements in child noncompliance and deviant behavior as primary outcome criteria with little attention given to other aspects of child functioning (e.g., activity level, attention span, affectionate behavior, self-esteem, attachment to parents, attitude, social skills, mood), and c) failure to use standardized measures to evaluate the social validity of findings (i.e. outside of normal limits at pre-treatment and within normal limits at post-treatment). While these criticisms are valid for individual studies, the parent-child interaction literature as a whole is considered to be relatively well-researched and composed primarily of methodologically sound, empirical investigations. The major results in the field have been consistently obtained across research groups and replicated in




25
multiple subsequent studies, supporting the validity of the positive treatment outcome findings reviewed above.
In summary, researchers have consistently
documented the effectiveness of the Hanf approach for treating behavior problems in young children. Studies of the individual components of the treatment support the use of single, direct commands (e.g., Forehand & Scarboro, 1975), time-out with release contingencies under parental control (e.g., Bean & Roberts, 1981), and specific, labeled praise (e.g., Bernhardt & Forehand, 1975). Many studies have demonstrated that the Hanf model is effective for reducing noncompliance and deviant behavior in the clinic setting (e.g., Eyberg & Robinson, 1982), and generalization of these positive behavior changes from the clinic to the home has been consistently documented (e.g., Peed et al., 1977). Treatment effects also have been shown to generalize over time (e.g., Webster-Stratton, 1984), with follow-up studies demonstrating maintenance of up to 10 and 1/2 years following treatment (Forehand & Long, 1988). Furthermore, positive gains resulting from treatment have been shown to generalize to nontargeted deviant behavior (e.g., Wells et al., 1980) and to the behavior of untreated siblings (e.g., Eyberg & Robinson, 1982).




26
Given the positive outcome results and
generalization findings, it would be expected that treatment gains also would result in positive behavior changes within the school setting. As the following literature review demonstrates, however, the research on generalization of treatment effects to the school setting has produced contradictory and confusing findings. Because few studies directly have addressed school generalization with Hanf approaches to treatment, school generalization in the general parent training literature will be reviewed along with the Hanf studies in the following section.
Review of Generalization to School
In the first published parent training study addressing whether improvements in home behavior generalize to the school setting, Wahler (1969) presented case studies of two boys (ages 5 and 8) referred by the school for treatment of behavior problems both at home and in the classroom. The stated hypothesis for this study was that generalization would not be expected because of the situational specificity of behavior. According to Wahler, "If a child's behavior is a principal function of its short-term environmental consequences and antecedents, one could argue that the behavior is situation-specific. That is,




27
the child's behavicr in various settings should conform to the contingencies present, regardless of between-setting contingency differences (Wahler, 1969, p. 239)." The treatment for the two subjects in this study involved training the parents in the home environment to use differential attention, verbal and physical rewards, and time-out to modify specific behaviors. Each of the families received only two to three one-hour sessions. Behavioral observations in the home demonstrated higher levels of cooperative behavior during treatment sessions than baseline sessions. Yet, behavioral observations in the schools of these boys revealed no improvements in the classroom until a direct behavioral intervention was conducted with the teachers. The study thus provided no evidence of setting generalization following this brief parent training intervention. Wahler (1969) concluded that the home and school settings were functionally independent such that the behavior of the child only improved in the setting in which the contingencies were operating.
Wahler (1975) conducted a second study to further investigate the question of whether the home and school settings are functionally independent. Subjects were two boys (ages 10 and 11) who were both referred by teachers for predelinquent behavior problems in the




28
classroom. One of the boys also displayed acting-out behavior problems in the home. A parent training program was used in which parents were taught a combination of reinforcement and time-out procedures. Baseline behavioral observations showed that the behavioral clusters evident at home were different from those being displayed in the school setting. As in the previous school generalization study (Wahler, 1969), the successful management of one of the subject's behavior in the home resulted in no significant changes in school behavior. For the second subject, improvements at home were found to be associated with increased oppositional behavior in the classroom, a result which Wahler referred to as covariation (later termed a "behavioral contrast effect"). Because this was a case study and one of the subjects did not demonstrate the covariation response, conclusions that home to school generalization either does not occur or that children become more deviant at school following successful parent management training were judged to be premature. Instead, Wahler (1975) emphasized the need for additional research on setting generality with parent training programs.
Johnson, Bolstad, and Lobitz (1976) also examined school generalization following a parent training program. Treatment subjects were eight children (mean




29
age = 8.5 years; range = 5.3 to 12.8) who were referred for treatment of home behavior problems. Control subjects were eight untreated children (mean age = 7.5 years; range = 7.2 to 8.3) who were exhibiting deviant behavior at school but were not classmates of the referred children. Initial school observations and teacher report measures showed that the school behavior of the nonreferred control subjects was significantly more deviant than the pre-treatment school behavior of children who were referred for treatment. Thus it is unclear whether the children referred for treatment of home behavior problems also were demonstrating significant behavior problems at school.
The referred children received a treatment program that involved teaching the parents to modify specific, targeted behaviors using behavioral principles. After three months of treatment, parents displayed more positive attitudes toward their children and improvement was evident in targeted behaviors. However, no overall changes in deviant behavior in the home were observed. Post-treatment classroom evaluations were conducted 3.5 months after the pre-treatment evaluation for referred children, whereas the deviant control children were reevaluated after a 1.8 month time interval. The post-treatment classroom assessment showed a




30
nonsignificant increase in observed deviant behavior and no change in teacher report of behavior problems for the referred subjects. The untreated control group demonstrated more classroom deviant behavior at pre-treatment and less at post-treatment than the referred children. These pre- and post-treatment between-group differences did not attain significance. However, the slight improvement in the control group's behavior concurrent with the slight deterioration in the behavior of the treated group represented a significant Group by Time interaction (Johnson et al., 1976).
Although Johnson et al. (1976) concluded that their data supported a "behavioral contrast effect," they cautioned against drawing definitive conclusions from their study due to the small sample size and small magnitude of improvement in home behavior. In addition to these problems, the method of selecting and assessing the treatment and control children makes between group comparisons of school behavior difficult to interpret. Because the treatment subjects were not referred for school problems, it is not clear whether their classroom behavior was sufficiently deviant before treatment to be affected by improvements in home behavior. Also, because the control children were not selected from the referred children's classrooms, between-classroom




31
differences in teaching and disciplining styles may have contributed to the behavioral changes. Another consideration in interpreting these findings is that the control group received their "post-treatment" assessment after a shorter time-interval than the treatment group. It is unknown how the control group's behavior would have compared to that of the treatment group after a comparable amount of time.
In a well-designed study of the overall
effectiveness of behavioral parent training and cognitive-behavioral self-control therapy for ADD-H children, school generalization was examined as one indicator of treatment outcome. Horn, Ialongo, Popovich, and Peradotto (1987) randomly assigned 24 families with ADD-H children between the ages of seven and eleven years to either group behavioral parent training only, group self-control training only, or a combination of group parent training and self-control training. The treatment groups each met for eight weekly 90-minute sessions. Pre-treatment assessment of classroom performance included achievement testing, teacher report of activity/attentional problems (Conners' Teacher Questionnaire), and a 30-minute behavioral observation of disruptiveness, demanding the teacher's attention, time off-task, and gross motor




32
movements. All three treatment groups evidenced significant behavior improvments in the home following treatment and at 1-month follow-up, but no treatment was found to be superior to the others. With respect to behavior in the classroom, no evidence for school generalization or a behavioral contrast effect was found for any of the three treatment groups. The authors concluded that because of the situationally specific nature of childhood disorders, "a successful treatment program for a child with behavioral problems at home and in school must develop coordinated treatment components for home and school settings" (Horn et al., 1987, p. 65).
Sayger and Horne (1987) obtained more favorable school generalization findings in a study of the long-range effectiveness of a social learning family therapy model. The sample consisted of 20 aggressive boys in the second through sixth grade whose parents chose to participate in treatment following referral by school personnel. All families participated in a 10-session treatment program which included training in the following five areas: setting-up for success, self-control, discipline, reinforcement, and communication. Children were assessed before and after treatment and at a nine month follow-up. A number of




33
improvements were found in the home behavior of the children, as well as in family relationships and problem-solving skills. School behavior was assessed using the Daily Behavior Checklist (Prinz, O'Connor, & Wilson, 1981) which allows teachers to rate the occurrence of classroom behavior problems over three consecutive days. The children's classroom behavior was found to improve significantly from pre-treatment to post-treatment (22.50 to 13.25 on the DBC) and from post-treatment to follow-up (13.25 to 7.75) (Sayger & Horn, 1987). These results are difficult to interpret, however, because classroom behavior was assessed using only a single teacher report measure, and a classroom control group was not included.
For the most part, the school generalization
studies reviewed thus far have provided little evidence to suggest that treatment effects resulting from Parent Child Interaction Therapy might generalize to classroom behavior. However, none of these studies directly examined the effectiveness of the Hanf model. Instead, they represent such diverse treatment approaches as behavioral parent training targeting single behaviors, social learning family therapy, and group parent management training. It is unknown whether the results of studies using these approaches are applicable to PCIT




34
because PCIT is based on a different treatment model and involves a different set of procedures.
More favorable classroom findings actually could be expected from a Hanf approach because behavior problems are managed through the implementation of general improvements in the parent-child relationship, rather than targeting a small number of specific behaviors. Positive school behavior changes also would be expected because the Hanf model is used primarily with very young children whose relatively short learning histories may make them particularly amenable to the types of behavioral and personality changes that are necessary for generalization. In fact, the treatment outcome literature on the Hanf model has provided strong evidence for generalization across time (e.g., Forehand et al., 1979), and behaviors (e.g., Wells et al., 1980), as well as from the clinic to home setting (e.g., Eyberg & Robinson, 1982; Webster-Stratton, 1984). Additionally, multiple studies have demonstrated that Hanf model programs lead to decreases in child noncompliance and disruptiveness (e.g., Eyberg & Ross, 1978), two areas of behavioral improvement that would be most beneficial to the children's adjustment to the behavioral demands of the classroom setting.




35
Despite the above evidence suggesting the
likelihood of school generalization, Hanf model studies to date have provided no evidence of improvement in classroom behavior following this clinic-based treatment approach. In the first of two controlled group studies using the Hanf model, the classroom behavior of eight children referred for treatment of noncompliance was compared to that of eight normal control children (Forehand et al., 1979). The treatment children were not specifically referred for school behavior problems, and the treatment and control groups were not found to differ on level of inappropriate school behavior before treatment. Following treatment, the referred children's rate of compliance in the home increased from 88% to 93%. In the classroom, nonsignificant increases in inappropriate behavior occurred for both groups of children over the eight week treatment period. Although there were no significant differences between the groups, five of the treated children, in contrast to only four of the control children, were found to demonstrate more inappropriate behavior after treatment. It was suggested that this finding could be indicative of a behavioral contrast effect (Forehand et al., 1979).
In the second school generalization study, Breiner and Forehand (1981) evaluated noncompliance and inappropriate behavior in the classroom using a larger




36
sample of subjects. As in the previous study (Forehand et al., 1979), the treatment children were not specifically referred for school problems, and pre-treatment school observations revealed no significant differences between the 16 treatment children and their normal classroom controls (Breiner & Forehand, 1981). After approximately eight weeks of treatment, 11 of the 16 treated children became less deviant and more compliant in the home. In the classroom, both the treatment and control children became more compliant and less oppositional, and there still were no significant differences between the groups. It was concluded that, "if school problems exist, they will not be reduced by treatment directed toward home problems. School problems will have to be directly programmed into an overall treatment package in order to successfully reduce them" (Breiner & Forehand, 1981, p. 41).
Statement of Problem and Hypotheses
Because Hanf model programs differ from other
parent training programs in that they have a different set of procedures, the Hanf model generalization literature is considered most applicable to the question of whether home behavioral improvements generalize to




37
the school setting following Parent-Child Interaction Therapy. The two Hanf model generalization studies both provided no evidence of school generalization and concluded that direct classroom interventions are necessary for school behavior change. However, two methodological issues of concern in the previous Hanf model studies indicate that this negative conclusion regarding school generalization may be premature. First, to evaluate cross setting generalization, it is necessary to document that treatment resulted in clinically significant improvements in the primary setting. In the previous Hanf model investigations of school generalization, child compliance in the home increased by only 5% (Forehand et al., 88% to 93%; Breiner & Forehand, 30% to 35%) and home deviant behavior decreased only from 9% to 6% after treatment (Breiner & Forehand, 1981). Because of the apparently limited magnitude of change in home behavior, it may be that the children's overall adjustment was not affected to such a degree that improvements could be noted across settings. Second, to evaluate school generalization, it is necessary to document that the treatment children have conductproblems in the classroom to which the positive changes in home behavior can generalize. Children in both of these studies displayed normal




38
classroom behavior before treatment, making it difficult to demonstrate significant improvements in the school setting. Therefore, the issue of school generalization with the Hanf model, and more specifically Parent-Child Interaction Therapy, is considered unanswered and in need of further investigation.
The purpose of the present study was to evaluate school generalization in referred children who demonstrated severe conduct problem behaviors in both the home and school settings before treatment and who demonstrated clinically significant improvements in home behavior after treatment. Comparisons of the treatment children to both normal and deviant classroom controls were conducted on multiple measures to evaluate the clinical significance of school changes. It was hypothesized that a) PCIT would result in home behavior improvements to within normal limits following treatment, b) the treated children would demonstrate greater classroom behavior improvements in pre-treatment problem behaviors than either the normal or untreated deviant classroom controls, and c) the treated children would demonstrate clinically significant improvements in their school behavior problems upon completion of therapy (i.e., outside of normal limits before treatment and within normal limits after treatment). Social




39
behavior was an exploratory variable in this study that also was hypothesized to improve following treatment, but as a result of second order, rather than first order, generalization. That is, the generalization of behavioral improvements to the school setting was hypothesized to lead to improved social skills because of the increased opportunity for the treated children to display prosocial, rather than deviant, behaviors in interactions with peers.




CHAPTER 2
METHOD
Subjects
Subjects were 30 children between the ages of 2
years, 0 months and 7 years, 0 months. Twenty-four of the children attended preschool, three were in kindergarten, and three were in first grade. There were three subject groups: Treatment Group (TG) (n=10), Normal Classroom Controls (NC) (n=10), and Untreated Deviant Classroom Controls (DC) (n=10).
The TG consisted of 10 parent-child dyads
consecutively referred to the Shands Hospital Psychology Clinic by physicians, mental health professionals, and school personnel for treatment of child behavior problems both at home and at school. All TG children met the following criteria: a) received parent ratings on the Eyberg Child Behavior Inventory (ECBI) above the published cut-off scores for child deviancy (i.e., Intensity Score > 127; Problem Score > 11) (Eyberg & Ross, 1978); b) demonstrated a compliance ratio (averaged across three semi-structured situations) less than that typically shown by nonreferred children (i.e., < 62%) (Robinson & Eyberg, 1981) on at least one of the two baseline clinic observations using the Dyadic
40




41
Parent-Child Interaction Coding System (DPICS) (Robinson & Eyberg, 1981); c) received a DSM-III-R diagnosis (via structured interview) of either Oppositional Defiant Disorder, Attention-deficit Hyperactivity Disorder, or Conduct Disorder (American Psychiatric Association, 1987); and d) received teacher ratings on the Revised Conners Teacher Rating Scale at least one standard deviation above the mean for nonreferred children on either the Conduct Problem factor (i.e., for 3 to 5 year olds, > 1.23; for 6 to 8 year olds, > .71) or Hyperactivity Index (i.e., for 3 to 5 year olds, > 1.60; for 6 to 8 year olds, > 1.05) (Goyette, Conners, & Ulrich, 1978). Families were excluded from participation if either the child or the primary caretaker had a history of moderate to profound mental retardation (n = 2) or the child was taking medication for hyperactivity (n = 2).
To determine whether children met DSM-III-R
criteria for Disruptive Behavior Disorders (APA, 1987), a structured clinical interview was conducted with the parents (see Appendix A). The interviewer enquired specifically about each of the symptoms that comprise the diagnoses of Attention-deficit Hyperactivity Disorder, Conduct Disorder, and Oppositional Defiant Disorder. Parents were asked the duration of the




42
child's particular problems and whether the child demonstrated each behavior rarely, occasionally, pretty often, or very often (Campbell, Ewing, Breaux, & Szumowski, 1986). A child was considered to display a particular symptom only if the parent indicated that it occurred pretty often or very often (Campbell et al., 1986). Diagnoses were made according to whether the child demonstrated the minimum number of symptoms specified in the DSM-III-R for each disorder. Interrater agreement was evaluated by comparing the interview checklist data collected by the primary interviewer with that generated by an independent observer (the co-therapist). Interrater agreement was calculated by dividing the number of agreements by the number of agreements plus disagreements. The percentages of agreement were found to be 100% for the duration of the behavior problems and 99% for the frequency of specific diagnostic symptoms.
Structured parental interviews addressing the DSM-III-R Disruptive Behavior Disorders (APA, 1987) resulted in the following diagnostic breakdown for the TG children: a) five children met the criteria for both Oppositional Defiant Disorder (ODD) and Attention-deficit Hyperactivity Disorder (ADHD), b) three children met the criteria for ODD, ADHD, and




43
Conduct Disorder, c) one child met only the criteria for ODD, and d) one child met only the criteria for ADHD. The mean age of TG children was 4.5 years (range = 31 to 79 months; SD = 14.8 months). All of the TG children were male, and nine were White (one child was Oriental). The mean income for TG families was $21,360 (SD = $25,370), with a median income of $12,000. Welfare funds were the sole source of income for four of the TG families, and four of the TG children were being raised by single mothers. Of the six two parent families, four of the fathers participated in treatment.
A Normal Classroom Control (NC) subject and an
Untreated Deviant Classroom Control (DC) subject were selected from each of the classrooms of the TG children. To reduce the possibility of expectancy effects, the two classroom controls were selected by an individual who was familiar with the students in the classroom (e.g., classroom aide, counselor, daycare director) but was not the main classroom teacher who completed the teacher report measures. This individual was asked to select a child in the TG child's classroom who demonstrated average behavior as the NC subject and a child who demonstrated behavior problems in that classroom as the DC subject. Teachers who completed questionnaires were reimbursed $20 for their time.




44
The mean age of NC subjects was 4.9 years (range = 30 to 73 months; SD = 13.9 months). Six of the NC subjects were male, and eight were White (two were Black). The mean age of DC subjects was 4.4 years (range = 31 to 86 months; SD = 13.1 months). Seven of the DC children were male, and six were White (three were Black, one was Spanish).
Treatment Outcome Measures
Dyadic Parent-Child Interaction Coding System (DPICS)
The DPICS is a measure of parent-child social
interaction that specifically was designed to assess treatment progress and to be used as a pre- and post-treatment observational measure (Eyberg & Robinson, 1983a; Robinson & Eyberg, 1981). Parent-child interactions are coded during three standard five-minute situations: Child-Directed Interaction (CDI), Parent-Directed Interaction (PDI), and Clean-Up. The following instructions were given to parents for CDI: "In this situation, tell (child's name) that he/she may play whatever he/she chooses. Let him/her choose any activity he/she wishes. You just follow his/her lead and play along with him/her." For PDI, the parents were given the following instructions: "That was fine. Now we'll switch to another situation. Tell (child's name) that it is your turn to choose the game. You may choose




45
any activity. Keep him/her playing with you according to your rules." For the Clean-Up situation, the parents were given the following instructions (slightly modified from those provided in the manual): "That was fine. Now I'd like you to tell (child's name) that it is time to put the toys away. Get him/her to put all the toys away without your help. Make sure that you have him/her put them away all by himself/herself. You should get (child's name) to put each toy in its container and then put all the containers into the big toy box (Robinson & Eyberg, 1981)."
The DPICS was standardized for the CDI and PDI
clinic situations with a sample of 22 normal families and 20 families referred for treatment of a conduct problem child. All children in the standardization samples were between the ages of 2.0 and 7.0 years (Robinson & Eyberg, 1981). Norms are reported in the DPICS manual for both parent and child behaviors (Eyberg & Robinson, 1983a). The mean compliance ratio for children in the normal control families during ten minutes of PDI was 62 percent. As norms have not yet been reported for the mean compliance ratio in Clean-up, 62 percent was used in this study as the best available estimate of normative compliance across the three DPICS situations. The mean inter-rater reliability for the




46
child behaviors was found to be .92 (range = .76 to
1.0).
Several studies provide supportive evidence for the validity of DPICS. The coding system has been found to distinguish the behavior of conduct problem children from their siblings and normal children (Robinson & Eyberg, 1981). In a study by Robinson, Eyberg, and Ross (1980), the coding system was found to predict 61% of the variance in parent report of home behavior problems on the ECBI. The validity of DPICS is further supported by data demonstrating its sensitivity to treatment effects. Child compliance, as measured by DPICS, has been shown to increase from 39% at pre-treatment to 89% at post-treatment (Eyberg & Robinson, 1982).
All DPICS coders for the present study were blind to the stage of intervention and coded all observations from videotapes. Coders received approximately 30 hours of training and met a 75% level of agreement with the criterion coding of a prerecorded videotape. DPICS reliability checks (using all three of the structured situations) were conducted on five of the ten TG subjects for a total of 11 of the 40 observations (i.e., 27.5%).. Five of the reliability checks were conducted on pre-treatment data, while the remaining six of the checks were conducted on post-treatment data.




47
Intercoder reliability coefficients were as follows: total r = .982, compliance r = .929, and deviant behavior r = .979.
Although a number of parent behaviors are included
in DPICS, only the child compliance (total # of complies across three DPICS situations divided by total # of commands) and deviant behavior (total # of deviant behaviors across three DPICS situations) were analyzed in this study. The DPICS observations were conducted twice before treatment and twice after treatment (with a one week interval both pre- and post-treatment). Pearson stability correlation coefficients were as follows: a) pre-treatment compliance r = .537 (p = .21), b) pre-treatment deviance r = .978 (p < .01), c) post-treatment compliance r = .438 (p = .24), d) post-treatment deviance r = .572 (p = .11), e) total (pre- and post-treatment combined) compliance r = .664 (p < .01), f) total deviance r = .888 (p < .01). The compliance and deviant behavior scores used in the statistical analyses were mean scores that had been averaged across the two DPICS observations. Eyberg Child Behavior Inventory (ECBI)
The ECBI is a 36-item parent report inventory
measuring a variety of behavior problems of children between the ages of 2 and 16 years (Eyberg & Robinson,




48
1983b; Eyberg & Ross, 1978; Robinson, Eyberg, & Ross, 1980) (see Appendix B). The ECBI includes such items as "refuses to go to bed on time," "teases or provokes other children," "acts defiant when told to do something," and "is overactive or restless." This inventory allows the parent to indicate both the current frequency of occurence of particular behaviors and whether they are considered to be problems. Frequency ratings are scored on a scale of 1 ("never" occurs) to 7 ("always" occurs) and are summed to yield an overall problem behavior Intensity Score. For each item, the parent also indicates whether the behavior is currently a problem by circling either "yes" or "no," and an overall Problem Score is computed by summing the number of "yes" responses.
The psychometric properties of the ECBI originally were established on samples of 42 normal preschool children and 43 behavior problem preschoolers (age range for both samples = 2.0 to 7.0 years) (Eyberg & Ross, 1978). The mean Intensity Score for the normal preschoolers was 102.6 (SD = 25.5), while the mean Intensity Score for behavior problem preschoolers was 158.3 (SD = 31.7). The mean Problem Score for the normal subjects was 4.62 (SD = 5.7), while the mean for behavior problem subjects was 18.6 (SD = 7.2). Several




49
studies have demonstrated that the ECBI has satisfactory test-retest reliability (r = .86 at an interval of 3 weeks) and internal consistency (r = .98 for both scales) (Eyberg & Ross, 1978; Robinson, Eyberg, & Ross, 1980). Findings of significant inter-item correlations, high internal consistency coefficients, and homogeneity of each of the ECBI scales provide good evidence for internal validity (Robinson, Eyberg, & Ross, 1980). External validity is supported by findings that the ECBI correlates well with independent observations of children's behavior and discriminates between normals and children referred for treatment of conduct problems (Eyberg & Ross, 1978; Robinson, Eyberg, & Ross, 1980).
School Generalization Measures Classroom Coding System
This system was developed from coding systems used by Forehand and colleagues (Breiner & Forehand, 1981; Forehand et al., 1979) and those used by Walker, Shinn, O'Neill, and Ramsey (1987). Three behavior categories were coded for each child: a) Appropriate Behavior vs. Oppositional Behavior, b) Comply vs. Noncomply vs. Unsure/ No Command Given, and c) On Task vs. Off Task vs. Not Applicable. The Appropriate Behavior category and the Comply category were used to code classroom behavior in two previous studies of school




50
generalization (Breiner & Forehand, 1981; Forehand et al., 1979). The On Task category is based on the definition of academically engaged time proposed by Walker et al. (1987). (See Appendix C for definitions of each of the coding categories.)
The classroom coding system yields three scores for each child: percent appropriate behavior (# of intervals child was appropriate divided by total # of intervals), percent compliance (# of commands obeyed divided by total # of commands), and percent on task (# of intervals on task divided by the total # of intervals in which the child was expected to be on task). Data were coded using an interval sampling procedure similar to that used by Forehand et al. (1979) and Breiner and Forehand (1981). Coders used a tape recorder and earphones to listen to a prerecorded audiotape that cued them to the beginning and end of each interval. The TG child and the two control children were observed sequentially. Each target child was observed during a 10-second period followed by a 10-second marking period. Two 45 interval classroom observations were conducted prior to treatment and two 45 interval sessions were conducted at the end of treatment. The 45 interval sessions typically were conducted two days apart and during the morning hours. Pearson stability




51
coefficients (first 45 interval observation compared to second 45 interval observation) for the three behavioral categories were as follows: Compliance r = .392 (p < .01), Appropriate r = .734 (p < .01), and On Task r = .745 (r < .01).
Breiner and Forehand (1981) obtained high
interrater reliability scores for both the Appropriate vs. Oppositional category (84%) and the Comply vs. Noncomply behavioral category (88%). Forehand et al. (1979) did not report reliability for individual categories but obtained an overall interrater reliability of 93%. High interrater reliability has also been obtained using the On Task vs. Off Task category. Walker et al. (1987) obtained interrater reliability coefficients ranging from 91% to 100% with an average agreement of 97% for the On Task category. This study also provided evidence for the discriminant validity of the On Task category. Walker et al. (1987) found that the sixteen fifth grade antisocial subjects obtained a mean On Task time of 68% (SD = 16.8%) whereas the non-antisocial control children achieved a mean On Task time of 85% (SD = 12.7%).
Coders received approximately 25 hours of training and met an 80% interrater agreement criterion with the principal investigator before being permitted to code in




52
the classroom. School coders were blind to group assignment of individual children and stage of intervention. Reliability checks were conducted for 25 of the 40 observations (62.5%) by using a dual-jack earplug that cued both coders to the beginning and end of the 10-second intervals. Reliability checks were conducted on 81% or pre-treatment observations and 40% of post-treatment observations. Intermittent training sessions were conducted throughout the data collection to minimize the possibility of observer drift. Interrater reliability percentages (# of Agreements divided by # of Agreements plus Disagreements were as follows: Total = 92.3% (range 85 to 96%), Comply = 93.2% (range 83 to 99%), Appropriate = 92.6% (range 87 to 97%), and On Task = 90.9% (range 80 to 96). Pre-treatment total agreement was 90.8% (SD = 4.2), and post-treatment total agreement was 93.9% (SD = 2.2). Revised Conners Teacher Rating Scale
The Revised Conners is a 28 item teacher report inventory that is widely used for the assessment of hyperactivity in children ages 3 through 17 (Goyette, Conners, & Ulrich, 1978) (see Appendix D). This inventory is based on an earlier 39 item scale (Conners, 1969). The Revised Conners includes such items as "temper outbursts and unpredictable behavior," "mood




53
changes quickly and drastically," "excitable, impulsive," "excessive demands for teacher attention," and "uncooperative with classmates." For each item the teacher marks whether the behavior is present "not at all," "just a little," "pretty much," or "very much." These ratings are given weights of 0, 1, 2, and 3 respectively, with the higher score representing greater symptomatology. A total score is obtained by summing the weights of the scored items and dividing the sum by 28. Scores also are obtained for the Hyperactivity (i.e. Hyperkinesis) Index as well as the following three factors: Conduct Problems, Hyperactivity, and Inattentive-Passive. Only the Conduct Problems and Hyperactivity Index factor scores were analyzed in this study. These scores are obtained by summing the points for each of the items assigned to the factor and dividing by the number of factor items.
The Revised Conners was standardized on a sample of 570 children ranging in age from 3 to 17 (Goyette, Conners, & Ulrich, 1978). Three to five year old children (n = 24) in the normative sample obtained the following mean factor scores: Conduct Problem = .49 (SD = .74); Hyperactivity = .74 (SD = .74); Inattentive-Passive = .83 (SD = .87); Hyperactivity Index = .78 (SD = .82). Six to eight year old children




54
(n = 102) obtained the following mean factor scores: Conduct Problem = .30 (SD = .41); Hyperactivity = .46 (SD = .57); Inattentive-Passive = .64 (SD = .71); Hyperactivity Index = .49 (SD = .56).
The Revised Conners has been shown to have adequate reliability. Test-retest reliability scores for a one week interval have ranged from .88 to .98 for the various factors (Edelbrock, Greenbaum, & Conover, 1985; Edelbrock & Reed, 1984). The 39-item version was found to have adequate test-retest reliability for all three factors over a one month interval (ranging from r = .72 to .91) (Conners, 1969). Inter-rater reliability gathered from parents and teachers on the Hyperkinesis Index (a 10 item subscale that is identical for parent and teacher forms) of the Revised Conners was found to be r = .49 (Goyette et al., 1978). Acceptable correlations between parent and teacher factor scores also have been obtained providing evidence of convergent validity (Conduct Problem = .33; Inattentive-Passive/ Learning Problem = .45; Hyperactivity/ Impulsive-Hyperactive = .36) (Goyette et al., 1978). Numerous studies have shown that the original form of this rating scale (Conners, 1969) discriminates between normal and hyperactive children (e.g., Conners, 1970), and both the original and the Revised Conners have been




55
shown to be sensitive to drug treatment effects (e.g., Barkley, 1977; Barkley, Fischer, Newby, & Breen, 1988). Sutter-Eyberg Student Behavior Inventory (SESBI)
The SESBI is a 36-item teacher report inventory measuring a variety of conduct problem behaviors of young children (Sutter & Eyberg, 1984) (see Appendix E). The SESBI includes such items as "argues with teachers about rules or instructions," "sasses teacher," "teases or provokes other students," "has difficulty staying on task," and "demands teacher attention." The teacher indicates both the frequency of occurrence of particular behaviors and whether they currently are considered to be problems. As is the case for the ECBI, frequency ratings are scored on a scale of 1 ("never" occurs) to 7 ("always" occurs) and are summed to yield an overall Intensity Score. The teacher indicates whether each behavior is a problem by circling either "yes" or "no," and the overall Problem Score is the sum of the number of "yes" responses.
The SESBI was standardized on 55 three to five year old children (Funderburk & Eyberg, in press). The mean Intensity Score was 100.9 (range: 36 to 228) with a standard deviation of 47.6. The Problem Scores ranged from 0 to 33 with a mean of 6.0 (SD = 8.8). The Intensity Score and the Problem Score were significantly




56
correlated (r = .65, p < .0001), indicating that the scales measure similar but not identical dimensions. The teacher completed SESBI ratings and the parent completed ECBI ratings were not found to be significantly correlated (Intensity Scores r = -.03). However, the SESBI has been shown to correlate significantly with two other teacher measures of behavior problems. The correlation between the SESBI and the Preschool Behavior Questionnaire (Behar & Stringfield, 1974) was .73 (Funderburk & Eyberg, in press). Correlations between the SESBI and the Revised Conners Teacher Rating Scale were as follows: Conduct Factor with Intensity Score = .90 (Problem Score = .88), Hyperactivity Factor with Intensity Score = .89 (P.S. = .88), and Inattentive/Passive Factor with Intensity Score = .75 (P.S. = .67) (Sosna, Ladish, Warner, & Burns, 1989).
Walker-McConnell Test of Children's Social Skills
The Walker-McConnell consists of 43 positively
stated items describing social-behavioral competencies of children in their interactions with peers and teachers (Walker & McConnell, 1987) (see Appendix F). This scale includes such items as "makes friends easily with other children," "expresses anger appropriately,"




57
"listens carefully to teacher instructions and directions for assignments," "listens while others are speaking," "controls temper," "can accept not getting his or her own way," and "gains peer attention in an appropriate manner" (Walker & McConnell, 1987). Teachers rate the occurrence of each behavior on a Likert-type scale ranging from 1 ("never") to 5 ("frequently"). The scale produces a total score and the following three subscale scores: 1) Teacher-preferred social behavior, which is defined as "peer-related behavioral competencies which teachers value as appropriate to an academic setting" (e.g., sharing, assisting others, taking turns), 2) Peer-preferred social behavior, which is defined as "peer-related behavioral competencies that facilitate the development of friendships and social acceptance" (e.g., complimenting others, having extended conversations, playing games skillfully), and 3) School adjustment behavior, which is defined as "behavioral competencies that determine a positive teacher-pupil adjustment within instructional contexts" (e.g., making assistance needs known in an appropriate manner, following classroom rules, compliance).
The Walker-McConnell was standardized on a sample of 762 children in grades kindergarten through sixth.




58
This scale was chosen as the measure of social behavior for this study despite that fact that preschoolers were not included in the norm group for three reasons: a) the lack of available social behavior measures designed for young children, b) the individual items on the Walker-McConnell appear appropriate for preschoolers, and c) the measure has been shown to have good psychometric properties for children as young as five years of age.
The norm sample for the Walker-McConnell obtained the following scores: mean Total Adjustment score = 159.78 (SD = 33.16), mean Teacher-preferred social behavior = 58.66 (SD = 13.31), mean Peer-preferred social behavior = 66.98 (SD = 12.92), and mean School Adjustment behavior = 38.12 (SD = 10.27). Test-retest reliability over intervals of two weeks to one month have been found to be at or above .80 (Walker et al., 1987). Estimates of internal consistency yield alpha coefficients exceeding .90 for the total score as well as the three subscale scores. The Walker et al. (1987) study supported the discriminant validity of the scale. Antisocial subjects were rated by teachers as substantially less competent/adjusted than non-antisocial subjects on all scales except for the peer-preferred subscale. Mean scores for the 16




59
antisocial subjects were as follows: Total Adjustment score = 134.94 (SD = 33.74), Teacher-preferred social behavior = 47.25 (SD = 11.88), Peer-preferred social behavior = 57.25 (SD = 14.55), and School Adjustment behavior = 30.44 (SD= 9.38) (Walker et al., 1987).
Procedures
Preparation for Classroom Evaluation
Parents of children referred for treatment of both home and school behavior problems initially completed the ECBI and the DSM-III-R Structured Interview to determine eligibility for inclusion in the TG. Then, the parent-child dyads were observed interacting in the three standard DPICS situations. These clinic behavioral observations were repeated one week later to increase the likelihood of obtaining a representative sample of behavior. DPICS data from the three structured situations were combined, and the resulting data for the two observation sessions were averaged. Families who met the inclusion criteria for clinically significant behavior problems in the home were then given the Revised Conners to be completed by the teacher. If the scores on the Revised Conners met the inclusion criteria for clinically significant behavior problems in the classroom, the family was invited to participate in the study. All families asked to participate agreed to be included in the research.




60
Pre-Treatment Classroom Evaluation
The Treatment Group child and the two control children were observed in the classroom for approximately 45 minutes on two separate occasions. To minimize the children's reactivity to observers in the classroom, the children were told by their teacher that the data collectors were "people interested in finding out what we do in a preschool class." Because previous studies on school generalization have found that teachers sometimes give too few commands to code noncompliance reliably (e.g., Forehand et al., 1979), teachers were asked to give as many group commands as possible to the classroom as a whole during the observation period. The classroom observations were conducted as described in the section on the Classroom Coding System with the observers seated in an unobtrusive location in the room. Teachers completed the SESBI, Revised Conners, and Walker-McConnell for the Treatment Group child, the Deviant Control Child, and the Normal Control Child. Each teacher was reimbursed $20 for the time that was required to complete the questionnaires.
Description -of Treatment
Parent-Child Interaction Therapy (Boggs & Eyberg, in press; Eyberg & Robinson, 1982) is conducted in two




61
treatment phases: Child-Directed Interaction (CDI) and Parent-Directed Interaction (PDI) (Eyberg & Robinson, 1982). Both phases involve didactic and coaching sessions. During didactic sessions, therapists first model and then have the parents role play skills. In subsequent sessions, parents are actively coached with their child using positive feedback through a bug-in-the-ear device. In CDI, the parents are taught to allow their child to lead the play activity. Parents are instructed to describe and praise the child's appropriate behavior, reflect appropriate child speech, and ignore inappropriate behavior. They learn not to criticize the child and not to use commands and leading questions which make it difficult for the child to lead the play. The major goal of CDI is to "create or strengthen a positive and mutually rewarding relationship" between the parent and child (Eyberg & Robinson, 1982). In PDI, the parents are taught how to direct their child's activity. They are instructed in the use of clear, positively stated direct commands and consistent consequences for behavior (i.e., praise for compliance, time-out in a chair for noncompliance). Parents learn to establish and enforce "house-rules" and to manage their child's behavior both at home and in




62
public places. The major goal of PDI is to decrease problematic behavior while increasing lowrate prosocial behaviors. For a more complete description of Parent-Child Interaction Therapy, see Eyberg and Boggs (in press) and Eyberg (1988).
All TG families received both the CDI and the PDI treatment phases. Because these subjects also were participating in a study comparing the effects of the two treatment phases, six of the subjects received CDI before PDI and the remainder received PDI before CDI. Thus, by the end of therapy all families had been taught the same child management skills, but there were two different phase order presentations. All TG families received 14 weekly one-hour sessions of Parent-Child Interaction Therapy (7 in CDI and 7 in PDI). No direct intervention was conducted in the classrooms. Session outlines were followed to avoid divergence from the treatment protocol (see Appendix G for a sample outline). Clinical psychology graduate students trained in the therapeutic approach served as the therapists for the study. Post-Treatment Clinic and Classroom Evaluations
Upon completion of treatment (approximately four months after the initial intake evaluation) parents of TG children again completed the ECBI and were observed




63
interacting with their children in the three standard DPICS situations. As in the pre-treatment clinic evaluation, DPICS observations were repeated one week later and the scores were averaged. A second classroom evaluation was conducted approximately one week after treatment. The TG children and the two controls were observed again on two separate days approximately two days apart. Teachers again completed the SESBI, Revised Conners, and Walker-McConnell for each of the three subjects and were reimbursed $20 for their time. Data Coding and Reliability Checks
The primary coders were graduate students blind to group assignment and stage of intervention. To keep coders blind to stage of intervention, different coders were used for a particular TG subject's pre- and post-treatment observations. For each classroom observation session, there was one primary coder whose data were used in the study. For 25 of the 40 observations, the primary coder was accompanied by a reliability coder. The reliability coder was an advanced graduate student skilled in the use of the coding system. This reliability coder was not always blind to group assignment and stage of intervention. Thus, in all cases it was the data of the primary coder, not the reliability coder, that were analyzed for the study.




CHAPTER 3
RESULTS
To be included in the analyses, TG children were required to improve to within normal limits following treatment on either the DPICS compliance ratio or the ECBI Intensity score (see Method for information regarding normative data). All ten of the TG children met this criterion, indicating that each TG child demonstrated clinically significant improvements in behavior with their parents (see Table 1). Statistical significance was tested using paired-comparison t-tests evaluating pre- and post-treatment TG mean scores on DPICS compliance and deviance and ECBI Intensity and Problem scores (see Table 2). Significant pre- to post-treatment improvements were found on all variables except DPICS deviant behavior. Although mean deviant behavior change was in the positive direction, this change failed to reach statistical significance using a two-tailed test (v < .10). The TG mean scores for measures of home and clinic behavior (see Table 2) improved to within normal limits following treatment (Eyberg & Ross, 1978; Robinson & Eyberg, 1981).
As mentioned previously in this paper, in addition to this generalization study TG subjects participated in a study evaluating the relative effectiveness of the 64




65
Table 1. Pre- and Post-Treatment Scores of Individual Treatment Group Children on Measures of Home and Clinic Behavior Problems
ECBI ECBI DPICS
Subj. Intensity Problem Compliance'
Pre / Post Pre / Post Pre / Post
#1 175 / 118* 24 / 8* 44 / 85* #2 189 / 85* 22 / 2* 19 / 47 #3 220 / 133 30 / 1* 51 / 73* #4 221 / 105* 24 / 6* 58 / 100* #5 185 / 159 33 / 26 52 / 82* #6 154 / 117* 25 / 8* 18 / 50 #7 171 / 56* 22 / 0* 52 / 75* #8 194 / 106* 28 / 3* 46 / 68* #9 170 / 76* 12 / 0* 9 / 59 #10 128 / 104* 13 / 7* 58 / 65*
*Within normal limits
ITotal # of complies in the three DPICS situations divided by the total # of commands (averaged across two observations)
two phases of PCIT and the impact of treatment phase order on overall outcome. As such, six of the TG subjects received CDI before PDI and the remainder received PDI before CDI. An analysis of possible differences between these two groups was conducted to




66
Table 2. Pre- and Post-Treatment Comparisons of Treatment Group Means on Measures of Home and Clinic Behavior Problems
Pre-Treatment Post-Treatment
Measures M (SD) M (SD) t(df = 9)
DPICS
Comply, 40.7 (18.2) 70.4 (16.3) 7.59* Deviance2 23.8 (27.6) 9.8 (12.6) 1.42 ECBI
Intensity 180.7 (28.2) 105.9 (29.2) 6.60* Problem 23.3 (6.7) 6.1 (7.7) 7.37*
*p < .01
'Total # of complies in the three DPICS situations divided by total # of commands (averaged across two observations)
2Total # of deviant behaviors across three DPICS situations (averaged across two observations)
determine whether treatment phase order influenced the degree of improvement in home and clinic behavior problems. No significant differences were found between the CDI-PDI group and the PDI-CDI group on pre- to post-treatment difference scores for the following measures of home and clinic behavior: ECBI Intensity score, ECBI Problem score, DPICS Compliance, DPICS Deviant Behavior (all p's > .05). Because no between




67
group differences were found on magnitude of behavior change using measures of home and clinic behavior problems, there was no evidence to suggest that treatment phase order would affect degree of school generalization. Thus, the CDI first group (n = 6) was combined with the PDI first group (n = 4) for the analyses of the school generalization measures.
By the time that the TG children completed the PCIT program (approximately four months after the pre-treatment school evaluation), one NC child and two DC children were no longer members of the classrooms in which they originally had been observed. Because inclusion of the incomplete data of these children could skew the findings, all data collected on subjects who were unavailable for the post-treatment school evaluation were excluded from analyses.
To determine whether the improvements in home behavior generalized to the school setting, a multivariate analysis of variance was conducted that included the following measures of classroom behavior in the statistical package: percent appropriate behavior, percent compliance, percent of time on task, SESBI Intensity score, SESBI Problem score, Revised Conners Conduct Problem factor, Revised Conners Hyperactivity Index, and the Walker-McConnell scales (i.e., Total




68
score, Teacher-preferred behavior, Peer-preferred behavior, and School Adjustment). This analysis yielded a significant result (p < .01), supporting the use of separate analyses of variance for individual measures. Thus, a 3 X 2 analysis of variance with one nonrepeated factor (group) and one repeated factor (time) was conducted for each of the dependent school variables listed above. For those scales displaying significant group by time interaction effects, pairwise group comparisons of difference scores (pre-treatment minus post-treatment) were done using the Duncan's Multiple Range Test.
Results of analyses comparing behavioral changes of TG children with the DC and NC groups on observational measures of classroom behavior are presented in Table 3. TG subjects were found to demonstrate significantly greater improvements in a) percent of time behaving appropriately in the classroom (p < .01) and b) compliance to teacher commands (p < .05) than either the DC or NC groups. The average percentage of time that the TG was found to display appropriate behavior during school observations after treatment was within one-half of one standard deviation of that obtained by the NC group, suggesting that the behavior improved to within normal limits. Similarly, although the TG obtained a




69
Table 3. Comparisons of Treatment Group to Untreated Deviant Controls and Normal Controls on Observational Measures of Classroom Behavior
TG8 DCb NCc
Pre Post Pre Post Pre Post
Measures M M M M M M F(2,24)
(SD) (SD) (SD) (SD) (SD) (SD)
% Appropriate 65 87 78 89 89 90 8.09**1
(14) (14) (12) (5) (8) (9)
% Compliance 54 87 64 75 73 80 4.24*1
(23) (16) (18) (15) (14) (13)
% On Task 69 82 80 80 86 84 1.75
(14) (16) (16) (10) (6) (16)
an = 10. bn = 8. cn = 9.
*~ < .05, **2 < .01
'Results of pairwise comparisons: TG difference score greater than both NC and DC (all p's ( .05). No difference between NC and DC (p > .05).
significantly lower mean compliance ratio (54%) than the NC group (73%) or the DC group (64%) prior to treatment (both p values ( .05), the post-treatment compliance ratio (87%) was not significantly different from those obtained by the NC (80%) and the DC (75%) groups. These findings suggest that in addition to demonstrating




70
statistically greater improvements in appropriate classroom behavior and compliance than the control groups, the TG displayed clinically significant gains to within normal limits on these two measures of school behavior. In contrast, significant group by time interaction effects were not found for the observational measure of on task behavior. Although the TG obtained significant pre- to post-treatment increases in percentage of time on task (p < .05), the TG was not found to display significantly greater improvements on this measure than the two control groups.
Qualitative analyses of the individual TG children were conducted to evaluate whether a subgroup of children displayed either no improvement or behavioral decrements that could not be detected in the group comparisons. As Eyberg and Johnson (1974) found a 30 percent change in pre- to post-treatment scores to be an effective cut-off value for determining the clinical significance of behavioral change in individual families, the 30 percent criterion was used in the present study for interpreting the classroom behavioral changes of individual TG children. The percent change values were obtained by dividing the pre- to post-treatment difference score by the pre-treatment score. A 30 percent improvement was interpreted as




71
behavioral gain, whereas a 30 percent decrement was interpreted as behavioral deterioration.
Pre- to post-treatment changes in the scores of individual TG children on the classroom observational measures are presented in Table 4. While four children improved in the area of appropriate classroom behavior and four children improved in the area of on-task time, the greatest number of behavioral improvements (n = 8) was evidenced in the area of compliance with teacher commands. Eight of the ten children demonstrated behavioral improvements on at least one of the observational measures (i.e., appropriate behavior, compliance, time on-task), and one child displayed a behavioral decrement on one of the measures (i.e., time on-task). In general, these data support the presence of behavioral improvements in the majority of TG children at post-treatment, while providing only minimal evidence to suggest that individual TG children became more deviant in the classroom after therapy.
Results of analyses comparing the TG to the two control groups on teacher report measures of school behavior problems are presented in Table 5. Teachers rated TG children as demonstrating significantly greater improvements than both the DC and NC children on three measures of oppositional and disruptive behavior (i.e.,




72
Table 4. Pre- and Post-Treatment Scores of Individual Treatment Group Children on Observational Measures of School Behavior
Subj. % Appropriate % Compliance % On Task
Pre / Post Pre / Post Pre / Post
#1 40 / 91+ 45 / 100+ 41 / 93+ #2 46 / 89+ 31 / 85+ 61 / 86+ #3 74 / 93 100 / 83 82 / 70 #4 70 / 96+ 53 / 91+ 62 / 95+ #5 50 / 72+ 55/ 82+ 58 / 71 #6 80 / 97 62 / 86+ 85 / 80 #7 76 / 95 83 / 100 78 / 86 #8 76 / 96 37 / 100+ 78 / 96 #9 60 / 55 23 / 47+ 75 / 46#10 73 / 86 54 / 100+ 73 / 96+
4Greater than 30% improvement relative to pre-treatment
-Greater than 30% decrement relative to pre-treatment
Revised Conners Conduct Problem factor, SESBI Intensity score, SESBI Problem score) (all p values < .05). The TG's mean scores on the Conduct Problem factor of the Revised Conners and the Intensity and Problem scores of the SESBI did not fall within one standard deviation of the NC group at post-treatment. However, the TG's mean




73
Table 5. Comparisons of Treatment Group to Untreated Deviant Controls and Normal Controls on Teacher Report Measures of Classroom Behavior
TG DC NC
Pre Post Pre Post Pre Post
Measures M M M M M M F(2,24)
(SD) (SD) (SD) (SD) (SD) (SD)
Revised Conners
Conduct Probs. 1.6 1.0 1.3 1.2 .35 .33 13.90**1
(.4) (.6) (.7) (.6) (.2) (.2)
Hyper. Index 1.9 1.4 1.5 1.3 .28 .36 4.51*2
(.6) (.8) (.5) (.7) (.3) (.4)
SESBI
Intensity 155 116 126 115 57 67 9.88***
(21) (34) (36) (35) (15) (21)
Problem 20 10 13 10 .67 1.1 14.00**1
(6) (7) (8) (9) (1) (2)
*p < .05, **p < .01
*Results of pairwise comparisons: TG difference score greater than both NC and DC (all ,p's < .05). No difference between NC and DC (p > .05). 2Results of pairwise comparisons: TG difference score greater than NC (p < .05). DC difference score greater than NC (p < .05). No difference between TG and DC (p > .05).




74
scores on the Revised Conners improved from outside of normal limits prior to treatment to within one standard deviation of published normative data for 3 to 5 year olds after treatment (Goyette et al., 1978). Similarly, the TG's mean SESBI Intensity score and Problem score improved to within one standard deviation of normative data (Funderburk & Eyberg, in press) upon completion of treatment.
With respect to hyperactivity and distractibility in the classroom, the TG obtained significant pre- to post-treatment decreases on the Hyperactivity Index of the Revised Conners (p < .05) and demonstrated greater improvements on this measure than the NC group (p < .05) (see Table 5). However, the TG was not reported to have significantly greater decreases in hyperactive behaviors than the DC group. In summary, teachers reported that the TG improved more than both of the control groups in the area of oppositional and disruptive behavior, but they did not report improvements in hyperactivity and distractibility greater than those demonstrated by the DC children.
As was the case for the observational measures of school behavior, qualitative analyses of teacher report measures for the individual TG children were conducted (see Table 6) to evaluate whether a subgroup of children displayed either no improvement or behavioral decrements that could not be detected in the group comparisons.




75
Table 6. Pre- and Post-Treatment Scores of Individual Treatment Group Children on Teacher Report Measures of School Behavior
SESBI CONNERS
Intensity Problem Conduct Hyperactivity Subj. Score Score Problem Index
Pre/Post Pre/Post Pre/Post Pre/Post
#1 121 / 93 15 / 8 1.25 / 0.38+ 1.7 / 0.8+ #2 159 / 76+ 20 / 74 1.00 / 0.25+ 2.7 / 2.1 #3 159 / 118 27 / 154 1.88 / 1.38 2.0 / 1.7 #4 157 / 72+ 17 / 34 1.75 / 0.75+ 1.8 / 0.5+ #5 191 / 173 26 / 24 2.13 / 2.13 2.5 / 2.6 #6 139 / 139 14 / 13 1.88 / 1.13+ 1.8 / 1.4 #7 139 / 89+ 11 / 3 1.25 / 0.75+ 1.4 / 0.5+ #8 136 / 110 15 / 1+ 1.50 / 0.754 0.7 / 0.5 #9 180 / 144 30 / 16+ 2.00 / 1.25+ 2.0 / 1.9 #10 165 / 141 23 / 94 1.50 /1.38 2.6 / 2.2
Note. There were no decrements greater than 30%. 4Greater than 30% improvement relative to pre-treatment
Inspection of Table 6 indicates that six of the ten children showed significant improvements on at least two of the four teacher report measures of school behavior problems. No significant behavioral decrements were




76
reported by teachers, and only one child (#5) displayed no improvement on any of the teacher report measures. Interestingly, this child also was the only child whose mother did not report significant gains in home behavior problems (see ECBI scores in Table 1).
The finding that the teacher and parent of TG child #5 both reported only minimal improvement in his behavior raises the issue of whether there is a relationship between magnitude of change in the home and degree of change in the school. To evaluate this question, maternal ratings of behavior on the ECBI were compared to teacher ratings of behavior on the SESBI. Neither the pre-treatment nor post-treatment Pearson correlations between ECBI and SESBI Intensity scores were found to be significant (pre: r = .029, p = .896; post: r = .502, p = .137). However, when pre-treatment minus post-treatment difference scores were used, a significant Pearson correlation of .783 was obtained (p = .007). This correlation is indicative of a common shared variance of approximately 61% between parent and teacher reports of behavioral changes in their respective settings. Thus this result supports the presence of a significant relationship between magnitude of improvement in home behavior and magnitude of gain in classroom behavior.




77
Table 7. Comparisons of Treatment Group to Untreated Deviant Controls and Normal Controls on Teacher Report of Social Competence
TG DC NC
Pre Post Pre Post Pre Post
Measures M M M M M M F(2,24)
(SD) (SD) (SD) (SD) (SD) (SD)
Walker-McConnell
Total 101 129 120 133 176 183 3.99
(26) (33) (20) (24) (27) (13)
Teacher 35 45 41 46 64 65 2.87 Preferred (9) (11) (5) (7) (9) (5) Peer 40 53 49 56 69 74 1.97 Preferred (15) (13) (14) (13) (12) (7) School 23 31 28 32 43 44 2.78 Adjustment (8) (11) (7) (11) (6) (4)
Note. Higher scores represent more positive ratings.
All F values are nonsignificant (p > .05).
Between-group comparisons on the Walker-McConnell Test of Children's Social Skills are presented in Table
7. Although the TG had-significant pre- to post-treatment mean increases on all of the Walker-McConnell scales (all p values < .05), the TG




78
children did not display significantly greater improvements than the control groups on these scales. These results yield no evidence to suggest that PCIT leads to improved social skills at school. In addition to the group comparisons, results of the Walker-McConnell Total scale were examined qualitatively for each individual TG child (see Table 8). Four of the TG children showed greater than 30 percent improvement on this measure, and none of these children were reported by teachers to demonstrate poorer social behavior in the classroom following treatment.
In addition to the primary analyses presented above, a set of secondary analyses was conducted to examine a statistical issue. As already mentioned, data from the one NC child and the two DC children who were not available for post-treatment classroom observations were excluded from analyses. However, a decision was made to retain the data from all TG children and their one available classroom control in the results. It was felt that data from these subjects would contribute to the information gained from both the individual and group analyses. A potential problem with this decision is that for three of the TG children the effects of maturation and classroom factors are not completely controlled because they do not have both a NC and DC




79
Table 8. Pre- and Post-Treatment Scores of Individual Treatment Group Children on Teacher Report of Social Competence
Walker-McConnell Walker-McConnell Subj. Total Subj. Total
Pre / Post Pre / Post
#1 109 / 163+ #6 93 / 130+ #2 58/ 67 #7 128 / 145 #3 121 / 132 #8 141 / 166 #4 111 / 161+ #9 80 / 135+ #5 86 / 94 #10 78 / 98
Note. There were no decrements greater than 30%. +Greater than 30% improvement relative to pre-treatment
child for comparison. By including the three TG children who were missing one of their controls, an assumption was made that the drop-out children's data would have been approximately equivalent to that of like subjects (that is, the missing NC child would behave like the other NC children; the two missing DC children would be similar to others in the DC group). To ensure that this assumption was met, an additional set of ANOVAs was conducted that included only the seven TG children whose normal and deviant controls were both




80
available for the post-treatment data collection. Results of the secondary analyses were consistent with the original results. Specifically, the TG was found to display significantly greater improvements than both control groups on the following measures: appropriate classroom behavior, compliance, Revised Conners Conduct Problem factor, SESBI Intensity score, and SESBI Problem score (all p values < .05). No significant differences between the TG and either control group were found for on task behavior, the Revised Conners Hyperactivity Index, and all scales of the Walker-McConnell (all p values > .05). These results indicate that the findings of this study are not confounded by the inclusion of subjects from the drop-out children's classrooms.
In summary, the results of this study demonstrated that the TG improved significantly more than both the NC group and the DC group on all measures of oppositional and disruptive behavior (i.e., percent of time behaving appropriately in the classroom, percent compliance in the classroom, teacher report of conduct problems on the Revised Conners, and teacher report of the frequency and severity of conduct problems on the SESBI; all F's, p < .05). However, results in the areas of hyperactivity/distractibility and social competence were less favorable. Although the TG obtained significant




81
Table 9. Summary of School Generalization Results for Individual Treatment Group Children on Measures of Behavior Problems in the Classroom
Observational Measures Teacher Report Measures
Approp- Compli- On SESBI SESBI CONNERS CONNERS Subj. riate ance Task I.S.1 P.S.2 C.P.3 H.I.4
#1 + + + + + + #2 + + + + + + #3 + #4 + + + + + + + #5 + +
#6 + + #7 + + + + #8 + + + #9 + + + #10 + + +
+Greater than 30% improvement relative to pre-treatment
-Greater than 30% decrement relative to pre-treatment 'Intensity Score, 2Problem Score, 3Conduct Problem Factor, 4Hyperactivity Index
pre- to- post-treatment improvements in percentage of time on task (p < .05) and teacher report of hyperactivity on the Revised Conners (p < .05), TG children did not display significantly greater




82
improvements than the DC group on these measures. Similarly, the TG's improvements on the Walker-McConnell, a measure of social skills in the school setting, were not found to be significantly greater than those obtained by the two control groups. Thus, although this exploratory measure was expected to demonstrate improved social skills in the treatment children as a result of decreased levels of deviant behavior in the presence of peers, the results did not support this hypothesis.
Summary results of the pre- to post-treatment changes in classroom conduct problem behaviors for individual TG children are presented in Table 9. Inspection of the table indicates that all of the TG children improved on at least one of the seven measures of classroom behavior problems, and over half of the children improved on at least three of these measures. This suggests that the positive school generalization results in the area of oppositional/conduct problem behavior are due to relatively consistent improvements across individual TG children rather than to a positive response for only a small subgroup of the sample. Inspection of the summary table also indicates that a behavior decrement was evident for only one child and on only one measure, providing little evidence to support the presence of a behavioral contrast effect.




CHAPTER 4
DISCUSSION
Prior to the collection of data for this study, it was hypothesized that PCIT would result in home behavior improvements to within normal limits following treatment. It was important that this hypothesis be confirmed because a valid investigation of setting generality requires the documentation of clinically significant improvements in the primary setting. With respect to the treated group as a whole, clinical improvements were evidenced by the fact that parent report of home behavior problems and clinic observations of child compliance improved on average from outside of normal limits prior to treatment to within normal limits upon completion of therapy. Clinically significant improvements also were evident on the level of individual subjects. All ten of the treated children improved to within normal limits following treatment on either parent report of the frequency of conduct problems or clinic observations of compliance. Thus, the improvements noted on group mean scores are judged to represent uniform gains across the individual treatment children (as opposed to exceptional improvement in a subgroup of subjects which offsets the lack of responsiveness of a second subgroup).
83




84
Results of the school evaluation indicate that the successful treatment of home behavior problems using PCIT is associated with improvements in certain behaviors in the school setting. This is the first controlled study supporting the hypothesis that school generalization occurs following parent training for home behavior problems, and as such, the present findings contradict two previously reported conclusions. First, there was little evidence provided by this study to suggest the presence of a behavioral contrast effect (e.g., Forehand et al., 1979; Johnson et al., 1976). Behavioral improvements rather than decrements were found in group analyses, and only one of the individual subjects displayed any evidence of becoming more deviant in the classroom following treatment. Second, contrary to previous investigations, the present study provided evidence of clinically significant improvements (i.e., to within normal limits) in classroom noncompliance and disruptive behavior following PCIT. This finding calls into question prior conclusions that direct school intervention is necessary to obtain improvements in classroom behavior (e.g., Breiner & Forehand, 1981; Horn et al., 1987; Patterson, Cobb, & Ray, 1973). Because these findings were obtained using a methodology that




85
included control groups and two different methods of measurement (i.e., teacher report and observations conducted without the child's knowledge), the positive behavior changes in the classroom can be considered strong, objective evidence for the overall effectiveness of PCIT.
The two central hypotheses of this study were that the treated children would demonstrate greater classroom behavior improvements than either the normal or untreated deviant controls and that the school behavior of the treated children would improve to within normal limits upon completion of therapy. These hypotheses were confirmed in the area of conduct problem/oppositional behavior. The treated children improved to within normal limits (which represents a significantly greater improvement than that demonstrated by the two control groups) on observational and teacher report measures of compliance, disruptiveness, and oppositionality in the classroom. Specific behaviors assessed with these measures included disobeying teacher commands, sassing, teasing, hitting, talking out of turn, whining, yelling, and breaking school rules. Generalization in this area was expected because many of these behaviors also were considered problematic in the home and were addressed in treatment. In fact, child




86
noncompliance with instructions and rules, which is a basic component of conduct problem/oppositional school behavior, is the primary area of focus for the Parent Directed Interaction phase of PCIT.
In contrast to the above improvements in conduct problem/oppositional school behavior, generalization was not found in the areas of hyperactivity/inattention and peer relationships. With respect to the activity and attentional problems, there are several possible explanations for the lack of school generalization. One possibility is that treatment was effective for distractibility and overactivity in the home but that it did not generalize well because of the great disparity between the attentional demands of the school versus the home setting. An alternative possibility is that PCIT is less effective for the treatment of Attention-deficit Hyperactivity Disorder than Oppositional Defiant Disorder and Conduct Disorder. Because the sample size in this study did not permit an analysis of the differential responsiveness of children in these three diagnostic categories, this will be an important area for future research. If generalization in the hyperactivity area is found to occur for only a subgroup of children, then it would be beneficial to be able to identify the nonresponders so that they could be




87
provided with such additional treatments as medication and individualized classroom behavioral programs.
The lack of generalization in the area of peer
relationships was not completely unexpected given that second-order generalization would have to occur to obtain such findings. Not only would it be necessary for improvements in the home to generalize to the school, but those decreases in noncompliance and deviant behavior in the classroom would then have to lead to improvements in the children's ability to socialize effectively with peers. There also may have been a methodological reason for the lack of generalization in the social skills domain. The measure used to assess social competence in this study, the Walker McConnell Test of Children's Social Skills, was standardized on a sample of children in kindergarten through sixth grade (Walker et al., 1987). Although this test discriminated the three groups at pre-treatment, it may not have been particularly sensitive to treatment effects because the items were designed primarily for older children. There was one finding, however, suggesting that this measure may have been a valid indicator of social skills changes in this young sample. Although the treatment children did not demonstrate greater improvements in social skills than the control children, they did develop more




88
advanced social competencies over the course of treatment. It is possible that the peer relationships were improving but that adequate time had not passed for the long-standing negative interactions with classmates to be completely reversed. Because the other groups also improved, another possibility is that this measure assesses social skills that children are developing rapidly during their preschool years. Psychometric studies evaluating the stability of the Walker-McConnell in large samples of preschoolers would be helpful for future research in the area of social behavior in young children. Follow-up school generalization studies in this area also are needed to evaluate whether treated children eventually display improved social competence at school, particularly as they enter new classrooms containing children with whom they may not have established prior relationship expectancies. In the meantime, the lack of evidence for substantial improvement in peer relationships suggests that an additional social skills treatment component would be beneficial to the overall school adjustment of these children.
One potentially important finding from this study that has not been reported previously is that maternal report of the magnitude of improvement in home behavior problems was significantly related to teacher report of




89
the magnitude of improvement in school behavior problems (r = .783). This result was found despite the fact that teacher and parent reports were not highly correlated either at pre-treatment or at post-treatment, a finding that is typical of most parent-teacher informant comparisons (e.g., Achenbach, McConaughy, & Howell, 1987). The lack of agreement between parents and teachers about children's behavior at a single point in time is influenced by a number of factors that include environmental demands, the presence of other children, interpersonal and discipline styles, rater tolerance for disruptive behavior, and rater expectancies of child behavior based on knowledge of developmental norms. When the pre-treatment minus post-treatment difference scores were used as the measure of magnitude of improvement, rater and environmental variances were controlled in that the magnitude score was relative to the initial rating. The resulting correlation indicates that relative improvements reported by parents in the home environment were directly related to those noted by teachers in the school environment.
As this is the first controlled study demonstrating positive behavior changes in the classroom following parent training, it is important to examine factors that distinguish the present study from prior investigations.




90
One potential reason for the discrepant findings is that PCIT may have different therapeutic effects than traditional parent training programs. As mentioned previously, PCIT differs from the parent training programs used by Wahler (1969; 1975), Johnson et al. (1976), and Horn et al. (1987) in that the therapy is directed towards the implementation of general improvements in the parent-child relationship rather than the targeting of a small number of specific problem behaviors for modification. Also, a significant portion of PCIT is devoted to teaching parents a form of play therapy which they are encouraged to use with their children on a daily basis. In this play therapy, parents provide a great deal of specific praise for their children's positive behaviors, qualities, and abilities, while listening carefully to what their children have to say, avoiding criticism, and displaying affection.
From a behavioral perspective, it is possible that the play therapy component of PCIT enhances the likelihood of generalization by increasing the frequency of prosocial behaviors displayed by the child. In the daily play therapy sessions, parents consistently praise and attend to such prosocial behaviors as sharing, using polite manners, speaking quietly, controlling one's temper, and being helpful. As a result of these social




91
rewards, the frequency of these behaviors increases, thereby increasing the probability that the prosocial behaviors will be emitted at school. When the behaviors spontaneously occur at school, the child may receive naturally occurring positive consequences from teachers and possibly peers, resulting in a higher frequency of the behaviors at school as well as at home.
Alternatively, it is possible that the play therapy component of PCIT results in intrinsic changes within the children that are not situationally specific. For example, it could be speculated that daily play therapy leads to a decrease in the child's anger level and a more positive self-image, two factors that could result in improved behavior across situations. Unfortunately, empirical investigations of these speculations using self-esteem and affective measures for preschoolers currently are not available in the parent-child interaction literature. Future research of this type might shed light on the degree to which these "internal" child characteristics covary with the expression of positive behaviors in diverse situations and contribute to the explanation of cross-setting generality.
Another possible explanation for the discrepant findings amongst PCIT and the traditional parent training programs used in earlier studies of school generalization is that PCIT focuses on overlearning of




92
compliance to all parental instructions. During the second stage of treatment (PDI), parents engage in daily practice sessions in which they give many simple, specific commands and provide consistent consequences for compliance and noncompliance. Eventually, the parents are able to provide consistent consequences for every command given throughout the day. After many of these trials in which highly positive consequences consistently accompany compliance and an aversive time-out experience consistently is associated with noncompliance, the oppositional child's first response to parental commands appears to change from a determination to defy and disobey to an automatic, overlearned compliance response. As the compliance response increases, it is likely to occur in a greater number of situations and to obtain positive responses from others, as well. Over time, negative behaviors may no longer be necessary to receive sufficient attention.
One final factor that may relate to the finding of school generalization is that the behavioral consistency and structure that is put into place by the parents during the course of PCIT is similar to the rule-based structure in operation in most schools. This consistency between settings with regard to behavioral demands and expectations may help the child to adapt




93
more readily to the classroom routine and rules. Again, these are theoretical speculations that have not yet been empirically examined. Because the present study addressed only the question of whether school generalization occurs following PCIT, further research is necessary to identify and evaluate possible causes of this generalization.
In the above discussion, it is proposed that PCIT involves a different set of therapeutic techniques than did traditional parent training which might result in specific changes within the child that increase the likelihood of generalization across settings. However, the question still remains as to why school improvements were evident in the present study even though previous Hanf model investigations employing a similar therapeutic approach found no evidence of school generalization. One important difference between this study and the two Hanf model studies conducted by Forehand and colleagues (Breiner & Forehand, 1981; Forehand et al., 1979) is that the subjects in the current research specifically were selected because they were demonstrating both home and school behavior problems. Because the children in both of the previous Hanf studies displayed normal classroom behavior before treatment, behavioral changes may not have been found




Full Text

PAGE 1

PARENT-CHILD INTERACTION THERAPY WITH BEHAVIOR PROBLEM CHILDREN: GENERALIZATION OF TREATMENT EFFECTS TO THE SCHOOL SETTING By CHERYL ANN BODIFORD 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 1989

PAGE 2

ACKNOWLEDGMENTS I would like to thank each of my committee members, Dr. Sheila Eyberg, Dr. Randy Carter, Dr. Eileen Fennell, Dr. Jacque Goldman, and Dr. James Johnson, for providing conceptual guidance and critical reviews for this manuscript. I particularly would like to thank the chair of my committee, Dr. Eyberg, for the long evenings she put into this research, as well as for her knowledgeable contributions and support. Toni Hembree Eisenstadt, M.S., deserves special acknowledgment for her assistance and dedication to the study from its conception through completion. Other valued contributors to this work include Katharine Newcomb, Beverly Funderburk, Bill Eisenstadt, and Lori Rossi. I also would like to thank those individuals who provided much needed moral support throughout this research. I appreciate motivating phone calls from Dr. Shari Rediess that assisted me in meeting unreasonable deadlines. Dr. Dan McNeil also is very appreciated for the patience, support, consultation, and equipped office that he provided during the manuscript preparation phase of this project. Finally, I very much would like to ii

PAGE 3

thank my parents, Otis and Donna Bodiford, for providing continued support of my educational goals and making the attainment of this degree possible. iii

PAGE 4

TABLE OF CONTENTS ACKNOWLEDGMENTS . . . . . . . . ii ABSTRACT. . . . . . . . . . . vi CHAPTERS 1 INTRODUCTION. . . . . . . . 1 Review of Effectiveness of Treatment Components . . . . . . . . 4 Review of Treatment Efficacy Studies........ 10 Maintenance Literature (Generalization Over Time) . . . . . . . . 16 Generalization to Home, Untreated Behavior, and Siblings.................... 19 Overview of Hanf Model Treatment Outcome Literature. . . . . . . . 24 Review of Generalization to School.......... 26 Statement of Problem and Hypotheses......... 36 2 METHOD. . . . . . . . . . 40 Subjects. . . . . . . . . 4 0 Treatment Outcome Measures.................. 44 School Generalization Measures.............. 49 Procedures. . . . . . . . . 59 3 RESULTS. . . . . . . . . . 64 4 DISCUSSION. . . . . . . . . 83 APPENDICES A DSM-III-R CHECKLIST................... 99 B EYBERG CHILD BEHAVIOR INVENTORY....... 104 C CLASSROOM CODING SYSTEM............... 107 D REVISED CONNERS TEACHER RATING SCALE.. 113 E SUTTER-EYBERG STUDENT BEHAVIOR INVENTORY........................... 115 iv

PAGE 5

APPENDICES {continued) F THE WALKER-McCONNELL SCALE OF SOCIAL COMPETENCE AND SCHOOL ADJUSTMENT: A SOCIAL SKILLS RATING SCALE FOR TEACHERS. . . . . . . 118 G SAMPLE TREATMENT SESSION OUTLINE: SECOND PDI COACHING SESSION......... 124 REFERENCES. . . . . . . . . . 126 BIOGRAPHICAL SKETCH............................... 133 V

PAGE 6

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 PARENT-CHILD INTERACTION THERAPY WITH BEHAVIOR PROBLEM CHILDREN: GENERALIZATION OF TREATMENT EFFECTS TO THE SCHOOL SETTING By Cheryl Ann Bodiford May 1989 Chairman: Sheila Eyberg, Ph.D. Major Department: Clinical and Health Psychology Generalization of treatment effects from the home to school setting was evaluated in ten children between the ages of two and seven years who were referred for treatment of severe conduct problem behaviors that were occurring both at home and in the classroom. Referred children received 14 weeks of Parent-Child Interaction Therapy, a treatment program directed towards improving the parent-child relationship and modifying noncompliant and disruptive child behaviors. No direct classroom interventions were conducted. The referred children demonstrated clinically significant improvements to within normal limits on measures of home and clinic behavior problems upon completion of therapy. Post-treatment behavior changes vi

PAGE 7

in the school setting were evaluated through comparisons of the treatment children to ten untreated deviant classroom controls and ten normal classroom controls on multiple observational and teacher report measures. The treatment group was found to display significantly greater improvements than both control groups on all measures of conduct problem behavior in the classroom. These classroom behavioral gains were judged to be clinically significant in that the mean scores of treatment group children on these measures improved to within normal limits following therapy. Results in the areas of hyperactivity/distractibility and social behavior were less supportive of generalization. The positive school generalization results in this study contradict previous findings that children's behavior in the classroom either remains the same or worsens following parent training. Possible reasons for the discrepant findings are discussed, as well as the clinical and research implications of this investigation. vii

PAGE 8

CHAPTER 1 INTRODUCTION Parent-Child Interaction Therapy is based on a model of treatment devised by Constance Hanf, Ph.D. (1969) at the Oregon Health Sciences University. The Hanf model is based on operant conditioning and was designed to modify the dysfunctional interaction patterns between multiply handicapped children and their mothers. Hanf outlined a two-stage model of treatment to be implemented using structured laboratory situations. In the first stage mothers were taught to let the children lead the play and to use differential attention for positive and negative child behaviors. Mothers were instructed to give attention to all appropriate behavior and to ignore all inappropriate behavior. The second stage was devoted to teaching mothers to lead the play. Mothers were taught to give clear directions, to praise compliance, and to punish noncompliance with time-out. One factor that particularly set the Hanf parent-child interaction model apart from other parent training models was the emphasis on working with the mother and child together in the therapy sessions and providing immediate feedback to the 1

PAGE 9

2 mother to aid in the acquisition of the parenting skills. The Hanf model also differed from other parent training programs in that it emphasized the restructuring of family interaction patterns rather than the modification of specific targeted behaviors. Several of Hanf's students and colleagues have modified and extended the original model such that there currently are different versions of the Hanf parent-child interaction program (e.g., Barkley, 1987; Eyberg & Boggs, in press; Forehand & McMahon, 1981). Perhaps the most well-known modification of the Hanf program is the one proposed by Forehand and McMahon (1981) for the treatment of noncompliance in young children. In the book, Helping the Noncompliant Child (Forehand & McMahon, 1981), these authors describe their approach to treatment and present results from the programmatic research that they have conducted on the Hanf approach to therapy. Parent-Child Interaction Therapy (PCIT) is a specific treatment approach that also is based on the Hanf model (Eyberg, 1979). PCIT was designed by Sheila Eyberg, Ph.D., as a behavioral family therapy approach that integrates operant concepts with principles of behavior change used in traditional play therapy. While many of the Hanf techniques and skills for improving behavior problems in young children

PAGE 10

3 are included in PCIT, there also are several unique features to the program and a slightly different theoretical framework. Compared to other Hanf model approaches, PCIT places a greater emphasis on establishing a strong, positive relationship between the parent and child by providing relatively extensive training to parents in the use of specific play therapy skills (Eyberg & Matarazzo, 1980). PCIT also can be differentiated from other parent-child interaction programs by the emphasis on problem-solving skills training as a means of helping parents to learn effective parenting skills (Eyberg & Boggs, in press). The present study examines the issue of whether improvements in the home behavior of children following Parent-Child Interaction Therapy generalize to the children's behavior at school. Before the methodology and findings of the current research are discussed, a review of the literature is presented to provide a framework for analyzing the study. Because school generalization cannot be examined without first documenting that treatment has been effective in the primary setting, the literature review first addresses the issue of whether Hanf model treatment programs are effective for improving parenting skills and child behavior in the home setting. This literature includes

PAGE 11

4 studies of the effectiveness of individual components of the treatment package as well as studies of the overall efficacy of the treatment program as a whole. A review of the literature regarding treatment generalization is then presented. This literature includes generalization over time, to the home setting, to untreated behaviors, and to siblings. Finally, the school generalization literature is reviewed. Because the literature on school generalization with Hanf model programs is sparce, a review of school generalization in the general parent training literature also is provided. Review of Effectiveness of Treatment Components A number of studies have evaluated the effectiveness of specific components within each of the two stages of Hanf model programs. Particular emphasis has been placed on evaluating the effectiveness of various types of commands, time-out procedures, and praise. In a study of command training, Forehand and Scarboro (1975) found that increases in the number of parental commands were related to decreases in child compliance. This finding was viewed as supporting the initial use of didactic sessions to teach parents to reduce the number of commands given to their children (Forehand & Scarboro, 1975). Similarly, research on

PAGE 12

5 various time-out procedures has supported the time-out approach used in most Hanf model programs. Time-out has been found to be most effective when the child is removed from all potential sources of reinforcement (Scarboro & Forehand, 1975), when the length of time-out is between one minute and four minutes (Hobbs, Forehand, & Murray, 1978), and when release from time-out is under parental rather than child control (Bean & Roberts, 1981). In a study evaluating the relative contribution of time-out and commands, it was found that training parents to give specific, single commands led to an increase in child compliance when compared to a placebo control group, but that even greater improvements in child compliance were obtained when command training was combined with a time-out procedure (Roberts, McMahon, Forehand, & Humphreys, 1978). This finding supports the use of a combination of command and time-out training in the second stage of Hanf programs. With respect to the use of praise, Bernhardt and Forehand (1975) found that both labeled (specific) and unlabeled (nonspecific) praises led to increases in targeted child behaviors on a marble dropping task. However, labeled praise was found to result in a significantly greater increase in correct marble dropping than unlabeled praise (Bernhardt & Forehand, 1975). This finding supports the emphasis

PAGE 13

6 in Hanf programs on teaching parents to use specific, labeled praise to increase appropriate child behaviors. Walle, Hobbs, and Caldwell (1984) assessed the effectiveness of attention (in the form of praise following compliance) and time-out (in the f o r m of standing in a corner following noncompliance) in various sequences for reducing noncompliance. Twenty-eight mother-child dyads were assigned to groups that received either no treatment or various sequences of attention and time-out training. Contrary to studies indicating that positive reinforcement techniques are more appealing to parents than punishment techniques (e.g., Calvert & McMahon, 1987; Kazdin, 1980, 1981), parents in this study rated praise and time-out as equally acceptable treatment components. The sequence in which the components were presented did not affect parental ratings of overall treatment acceptability. Attention alone was not found to produce significant reductions in noncompliance, and attention combined with time-out was not found to be superior to time-out alone. Analyses of sequencing effects, however, indicated that attention is an important variable. Receivin g attention b efore time-out was found to enhance the efficacy of time-out for reducing noncompliance (i.e. the group that received attention before time-out was found to have greater

PAGE 14

7 compliance in the time-out phase of the intervention than the group that r e c eived attentio n training after time-out). Interestingly, receiving attention after time-out also was found to have specific therapeutic benefits. Improvements in compliance were found to maintain better over time when the attention phase followed the time-out phase. As specific therapeutic benefits were obtained both when the attention stage preceded the time-out stage and when time-out was presented first, neither stage sequence was shown to be superior for the treatment of noncompliance. The authors concluded that teaching parents to use a combination of time-out for noncompliance and praise f o r compliance, regardless of stage sequence, is an effective strategy for producing immediate behavior change (Walle et al., 1984). Another component of the Hanf program that has been evaluated is the relative effectiveness of various instructional techniques for teaching parenting skills. Flanagan, Adams, and Forehand (1979) compared the following four methods of teaching parents time-out: writ t e n presentation, lecture presentation, videotape d modeling, and role-playing. After time-out training, parents in the four groups were not found to differ on a written test of knowledge of time-out principles.

PAGE 15

8 However, role-playing was found to be superior to the other presentations when outcome was measured by parental responses to an audiotaped analogue task. When outcome was assessed through parental performance of time-out with their own children in the home, the videotaped modeling group was found to implement the time-out more effectively than the other groups (Flanagan et al., 1979). Studies evaluating the relative effectiveness of group versus individual instruction and videotaped training versus live modeling have reported contradictory results. Webster-Stratton, Hollinsworth, and Kolpacoff (in press) compared the long-term effectiveness and clinical significance of the following three cost-effective presentations of a Hanf model program: a) group videotaped modeling with therapist-led discussion, b) an individually self-administered videotaped modeling presentation, and c) a group discussion presentation with no videotaped modeling. All presentations resulted in significant behavioral improvements as measured by parent report of behavior problems and observations in the home, and these improvements were found to be maintained at one year follow-up. The only significant finding differentiating the three presentations was that parents in the group

PAGE 16

9 videotaped modeling presentation with therapist-led discussion reported greater satisfaction with treatment (Webster-Stratton et al., in press). In another study of the relative effectiveness of individual and group presentations, Webster-Stratton (1984) randomly assigned 35 families with conduct disordered children to the following groups: a) a waiting-list control group, b) a 9 week individual parent-child interaction program, and c) a 9 week standardized group videotape modeling program. Improvements in parenting skills and parental attitudes toward the children, as well as reductions in deviant and noncompliant behavior, were noticed equally in both groups. As no between group differences were found on the treatment outcome measures, it was concluded that individual treatment with live modeling and group treatment with videotape modeling are equally effective for treating the behavior problems of young children (Webster-Stratton, 1984). Contradictory findings were evident in a study comparing individual Parent-Child Interaction Therapy with group didactic training using 29 families with children who had speech and language disorders (Eyberg & Matarazzo, 1980). In the individual training group, parents demonstrated improved parenting skills, reported significantly fewer home problems, and a more positive

PAGE 17

10 attitude toward their children. Additionally, child behaviors improved significantly with fewer children exhibiting inappropriate behavior and a mean compliance increase from 79% at pre-treatment to 91% at post-treatment. In the group didactic training, mothers reported significantly fewer home problems and a more positive attitude toward their children, but no pre-to post-treatment behavioral differences were noted in parenting skills or child behavior. Also, parents expressed more satisfaction with the individual presentation than the group didactic training. These authors concluded that parenting skills are improved most readily through direct observation and immediate feedback (Eyberg & Matarazzo, 1980). When studies of the relative effectiveness of different instructional methods are considered together, there are consistent data to support that the Hanf model is effective in both an individual and group training format. However, the issue of whether individual direct feedback training is more effective than group videotape/didactic training remains unclear. Review of Treatment Efficacy Studies In an initial study of the overall effectiveness of the Hanf program, Hanf and Kling (1974) treated 40 pairs

PAGE 18

11 of mothers and their severely physically handicapped, noncompliant children. Comparisons of pre-and post-treatment data demonstrated that maternal verbalizations changed in the direction expected based on the skills taught in the program. Mothers significantly increased their use of verbal rewards and decreased their use of commands and questions. The program's focus on noncompliance also was found to be effective in that the children displayed significant increases in overall rate of compliance following treatment. Improvements in parenting skills and child compliance were found to maintain over a three month time interval (Hanf & Kling, 1974). Further evidence for the effectiveness of the Hanf program for treatment of behavior problems in handicapped children was provided in a case study of a deaf child. Consistent with the group findings, the compliance rate of this child was found to increase significantly following the parent-child interaction program (Forehand, Cheney, & Yoder, 1974). Forehand and King (1974) empirically demonstrated the effectiveness of a modified version of Hanf's program for the treatment of noncompliant preschoolers without physical handicaps. After approximately six sessions, mothers of eight noncompliant preschool

PAGE 19

12 children were found to increase significantly their use of rewards and decrease their use of commands and questions in the first stage of treatment (which Forehand and colleagues refer to as the Child's Game). Maternal verbal rewards also increased significantly in the second stage of treatment (i.e. Parent's Game), as did child compliance (Forehand & King, 1974). In a subsequent study Forehand and King (1977) again used the modified Hanf program for the treatment of noncompliant preschoolers. However, this study used parent perception measures as well as clinic observations of mother-child interactions to evaluate treatment outcome. As in prior studies, mothers significantly increased their use of rewards and decreased their use of commands and questions while children's compliance to parental commands increased. Results of the parent perception measures demonstrated that mothers perceived their children as significantly better adjusted following treatment. An additional comparison of the 11 treatment children to 11 nonclinic normal controls showed that the treated children were less compliant than the normals before treatment and more compliant following treatment. Although treatment mothers had more negative perceptions of their children than normal control mothers prior to treatment, scores on parent

PAGE 20

13 perception measures were not found to differ between the two groups after the Hanf program. All improvements in parent and child behaviors were found to maintain over a three month time interval (Forehand & King, 1977). In addition to behavioral observations and measures of parental perceptions of their children, researchers have examined consumer satisfaction and treatment acceptability as indicators of treatment efficacy. Several studies have used either the Parent's Consumer Satisfaction Questionnaire (Forehand & McMahon, 1981) or the Therapy Attitude Inventory (Eyberg, 1974) to evaluate parental satisfaction. Results indicate that Hanf parent-child interaction programs are typically rated as moderately to highly satifactory by parents (Eyberg & Matarazzo, 1980; Eyberg & Robinson, 1982; McMahon, Forehand, & Griest, 1981). While one study found that parents perceived praise and time-out as equally acceptable methods of modifying child behavior (Walle et al., 1984), most studies indicate that parents rate positive techniques for increasing deficit behavior (e.g., rewards, commands, attends) as more acceptable than strategies for reducing behavioral excesses (e.g., time-out, ignoring) (Calvert & McMahon, 1987; Kazdin, 1980, 1981). Results of studies evaluating the acceptability of various aspects of Hanf model programs

PAGE 21

14 must be considered with caution, however, because subjects typically are not clinical subjects. Parents of conduct disordered children might view the acceptability of the Hanf treatment strategies differently than parents whose children are not in need of treatment. Some researchers have suggested that drop-out rates can be viewed as an outcome measure or as an indicator of consumer satisfaction. In a review of 45 parent training studies in eight journals from 1972 to 1982, it was found that about half of the studies reported drop-out data (Forehand, Middlebrook, Rogers, & Steffe, 1983). Using the drop-out statistics from these studies, it was estimated that the drop-out rate in parent training is approximately 28% (Forehand et al., 1983). Although it is unknown how this rate compares to drop-out rates in other types of child psychotherapy because of the lack of research in the area, the 28% drop-out figure compares favorably to drop-out rates reported in the adult literature (Baekeland & Laundwall, 1975). Preliminary data reported by McMahon, Forehand, Griest, and Wells (1981) suggest that parents who are depressed or socioeconomically disadvantaged account for a substantial portion of the parent training drop-out

PAGE 22

15 population. A meta-analysis of the overall drop-out rate specifically for Hanf parent-child interaction programs has not yet been conducted. In addition to evaluating whether statistically significant improvements in parenting skills and child behavior occur following treatment, studies have examined the issue of whether treatment gains are socially and clinically valid. Forehand, Wells, and Griest (1980) examined the social validity of a Hanf treatment program in the following ways: a) comparing referred families to nonclinic normal controls, b) having parents rate their children's progress ih treatment, and c) having parents provide consumer satisfaction data. Results demonstrated that the clinic children were less compliant and more deviant than the normal controls prior to treatment but not after treatment or at the two month follow-up evaluation. Parental ratings indicated that parents perceived improvements in their children's behavior and were satisfied with the treatment program. Because the treatment gains represented not only statistical changes but normative improvements to within the normal range, the authors were able to conclude that the Hanf program resulted in a clinically significant (i~e. socially valid) outcome (Forehand et al., 1980).

PAGE 23

16 Maintenance Literature (Generalization Over Time) In a review of the generalization research in the parent training literature, Forehand and Atkeson (1977) discuss maintenance of treatment effects as a form of generalization termed "temporal generalization.'' In one study addressing the temporal generalization issue, home observations of ten parent-child pairs were conducted immediately after a Hanf parent-child interaction program and again at 6 and 12 month follow-up intervals (Forehand, Sturgis, McMahon, Aguar, Green, Wells, & Breiner, 1979). Post-treatment changes in parental attitudes toward the children, parental use of commands, attends, and questions, and child behavior were found to be maintained both at 6 and 12 months following termination of therapy (Forehand et al., 1979). However, maternal rewards (i.e. labeled and unlabeled praise) at the follow-up evaluations did not differ significantly from the pre-treatment level. This finding suggests that maternal use of praise, a central parenting skill of Hanf programs, did not generalize over time as expected. Positive temporal generality results were obtained i~ several more recent maintenance studies. For example, Baum and Forehand (1981) conducted follow-up

PAGE 24

17 assessments on 36 families at intervals ranging from 1 1/2 to 4 1/2 years following treatment. Although parental use of attends, rewards, and questions decreased from the post-treatment to follow-up evaluations, the frequency with which parents demonstrated these skills at follow-up remained significantly improved from the pre-treatment levels. Improvements in child compliance, maternal use of indirect commands, and maternal report of child behavior were found to be maintained at follow-up. Consumer satisfaction measures indicated that the parents continued to have a favorable view of treatment up to 4 1/2 years after termination of therapy. Additionally, measures of child deviant behavior taken both in the clinic and the home demonstrated that the behavior of the children further improved from the post-treatment to follow-up evaluation (Baum & Forehand, 1981). Similar results were obtained in a study by Webster-Stratton (1984) in which one year follow-up evaluations were conducted on families who had received either individual Parent-Child Interaction Therapy or a videotape modeling program. Both groups of treated mothers were found to be significantly more positive, less critical, and less negative in interactions with their children. As in the Baum and Forehand (1981)

PAGE 25

18 study, treated children displayed significant reductions in noncompliant and deviant behaviors from the post-treatment to follow-up evaluation. Forehand and Long (1988) conducted a long term follow-up study addressing the adjustment and functioning of 11 to 14 year old children and their families who had participated in a Hanf model treatment program for child noncompliance 4 1/2 to 10 1/2 years earlier. Of the 43 families who previously had been treated and were eligible for the study, 21 families agreed to participate. These families were compared to a control group of 21 families who had a child in early adolescence who had never received psychological treatment. The two subject groups were matched on age, gender of the adolescent, and socioeconomic status. Multiple measures of child and parent functioning were used that included observational measures, self-report, parent report, teacher report, and academic grades from school files. With respect to externalizing behaviors, Forehand and Long (1988) found no differences between the treated children and controls on parent report of conduct, attent~onal, and delinquency problems in the home. However, parents who had been in treatment reported a greater amount of conflict with their adolescents than

PAGE 26

19 was reported by control parents. Teachers reported significantly more attentional problems and lower grades in the children who had been treated than the controls, but no differences were noted in the areas of delinquency and aggression. Observational, self-report, and parent report measures showed no differences between the groups on depression, anxiety, and social behaviors. The only internalizing or social behavior difference noted by teachers was that children who had been in treatment were viewed as being more anxious than the control children. No differences between the groups were found in the areas of parenting skills and parental adjustment. The authors concluded that these findings are generally positive and are indicative of "moderate" long-term effectiveness (Forehand & Long, 1988). Taken together, these maintenance studies support the presence of temporal generalization following the successful treatment of behavior problems in young children using a Hanf model treatment approach. Generalization to Home, Untreated Behavior, and Siblings Most researchers agree that an adequate evaluation of parent training outcome requires an assessment of the functioning of the parent and child outside of the clinic setting. The two most common methods of

PAGE 27

20 assessing home behavior are the use of parent questionnaires and direct observations i n the home setting. An example of the use of parent questionnaires is a study by Eyberg and Robinson (1982) evaluating the effectiveness of Parent-Child Interaction Therapy for reducing child behavior problems at home. Parents reported a significant decline in the number and intensity of home behavior problems on the Eyberg Child Behavior Inventory (Robinson, Eyberg, & Ross, 1980) upon completion of treatment, indicating that behavioral improvements generalize from the clinic to home setting. With respect to direct observations, Peed, R o b erts, and Forehand (1977) evaluated setting generality by conducting behavioral observations in both the clinic and home setting. Six mother-child pairs who received a Hanf model treatment program were compared to six mother-child pairs who were on a waiting-list to receive treatment. Clinic behavioral observations demonstrated improvements in maternal parenting skills and child behavior. Similarly, home behavioral observations demonstrated significant increases in child compliance and maternal skills following treatment. In contrast, no significant changes were found in the waiting list children or parents after completion of the waiting period. The authors concluded that the positive

PAGE 28

21 treatment results were not attributable to the passage of time, nonspecific environmental events, or the effects of observation, and that generalization occurred from the clinic to home setting (Peed et al., 1977). Several more recent studies using behavioral observations in the home have provided additional evidence that improvements noted in the clinic generalize to the home setting (e.g., Baum & Forehand, 1981; Forehand, Sturgis, et al., 1979). Treatment generalization also has been examined by assessing whether nontargeted deviant behavior improves following treatment of noncompliance using a Hanf parent-child interaction program. Wells, Forehand, and Griest (1980) compared 12 mother-child dyads who were referred for treatment of noncompliance to 12 mother-child normal controls on measures of noncompliance and deviant behavior. With respect to compliance (which was a behavior targeted for change in treatment), the clinic children were found to become significantly more compliant afte r treatment whereas no compliance differences were found in the nonclinic children after a comparable time period. Additionally, although th~ clinic and nonclinic groups differed significantly on rate of compliance before treatment, no between-group compliance differences were noted upon

PAGE 29

22 completion of therapy. Once it was determined that treatment had b een effectiv e for the targeted behavior, a further analysis was conducted to examine generalization of improvements in child noncompliance to other deviant behaviors. The treated children were found to have significant decreases in such deviant behaviors as aggressiveness, yelling, and whining whereas the nonclinic group demonstrated no change in deviant behavior. The authors concluded that the successful treatment of noncompliance is sufficient in many cases to reduce other untreated deviant behaviors (Wells, Forehand, and Griest, 1980). Several studies have examined the hypothesis that generalization of treatment effects from the target child to untreated siblings occurs following improvements in parental child management skills. Two case studies have been reported in which successful parent training was found to result not only in improvements in the behavior of the child targeted as having behavior problems but in the untargeted siblings as well {Laviqueur, Peterson, Sheese, & Peterson, 1973; Resnick, Forehand, & Mcwhorter, 1976) In a more controlled study of 8 clinic referred children, their parents, and untreated siblings, positive behavior changes in parental interactions with nontargeted

PAGE 30

23 siblings were noted (Humphreys, Forehand, McMahon, & Roberts, 1978). Observations of the parent with the untreated sibling revealed that the parents used significantly more rewards and contingent attention and fewer indirect commands following a Hanf treatment program. These changes in parenting skills were found to be accompanied by compliance rate increases from 34% to 44% in the untreated sibling, suggesting the presence of generalization across children (Humphreys et al., 197 8) In a second controlled group study of sibling generalization, Eyberg and Robinson (1982) evaluated the impact of treatment on nontargeted siblings in seven families. Prior to treatment, the children targeted for treatment of behavior problems were found to be 39% compliant to total parental commands and their nontargeted siblings demonstrated compliance rates of 53%. Following treatment, the rates of compliance in both groups of children improved significantly with post-treatment compliance rates of 89% and 73% respectively. Thus this study supports previous findings that improvements in parenting skills and the target child's behavior generalize to the behavior of siblings who are not directly participating in the parent-child interaction program.

PAGE 31

24 Overview of Hanf Model Treatmen t Outcome Literature Individual studies in the Hanf parent-child interaction literature can be criticized for methodological inadequacies that bring into question specific results. Examples of methodological problems in this literature include the following: a) failure to demonstrate that children had clinically significant behavior problems prior to treatment according to DSM-III criteria, b) use of improvements in child noncompliance and deviant behavior as primary outcome criteria with little attention given to other aspects of child functioning (e.g., activity level, attention span, affectionate behavior, self-esteem, attachment to parents, attitude, social skills, mood), and c) failure to use standardized measures to evaluate the social validity of findings (i.e. outside of normal limits at pre-treatment and within normal limits at post-treatment). While these criticisms are valid for individual studies, the parent-child interaction literature as a whole is considered to be relatively well-researched and composed primarily of methodoiogically sound, empirical investigations. The major results in the field have been consistently obtained across research groups and replicated in

PAGE 32

25 multiple subsequent studies, supporting the validity of the positive treatment outcome findings reviewed above. In summary, researchers have consistently documented the effectiveness of the Hanf approach for treating behavior problems in young children. Studies of the individual components of the treatment support the use of single, direct commands (e.g., Forehand & Scarboro, 1975), time-out with release contingencies under parental control (e.g., Bean & Roberts, 1981), and specific, labeled praise (e.g., Bernhardt & Forehand, 1975). Many studies have demonstrated that the Hanf model is effective for reducing noncompliance and deviant behavior in the clinic setting (e.g., Eyberg & Robinson, 1982), and generalization of these positive behavior changes from the clinic to the home has been consistently documented (e.g., Peed et al., 1977). Treatment effects also have been shown to generalize over time (e.g., Webster-Stratton, 1984), with follow-up studies demonstrating maintenance of up to 10 and 1/2 years following treatment (Forehand & Long, 1988). Furthermore, positive gains resulting from tre~tment have been shown to generalize to nontargeted deviant behavior (e.g., Wells et al., 1980) and to the behavior of untreated siblings (e.g., Eyberg & Robinson, 1982).

PAGE 33

26 Given the positive outcome results and generalization findings, it would be expected that treatment gains also would result in positive behavior changes within the school setting. As the following literature review demonstrates, however, the research on generalization of treatment effects to the school setting has produced contradictory and confusing findings. Because few studies directly have addressed school generalization with Hanf approaches to treatment, school generalization in the general parent training literature will be reviewed along with the Hanf studies in the following section. Review of Generalization to School In the first published parent training study addressing whether improvements in home behavior generalize to the school setting, Wahler (1969) presented case studies of two boys (ages 5 and 8) referred by the school for treatment of behavior problems both at home and in the classroom. The stated hypothesis for this study was that generalization would not be expected because of the situational specificity of behavior. According to Wahler, "If a child's behavior is a principal function of its short-term environmental consequences and antecedents, one could argue that the behavior is situation-specific. That is,

PAGE 34

27 the child's behav i c r in various settings should conform t o the contingencies present, regardless of between-setting contingency differences (Wahler, 1969, p. 239) ." The treatment for the two subjects in this study involved training the parents in the home environment to use differential attention, verbal and physical rewards, and time-out to modify specific behaviors. Each of the families received only two to three one-hour sessions. Behavioral observations in the home demonstrated higher levels of cooperative behavior during treatment sessions than baseline sessions. Yet, behavioral observations in the schools of these boys revealed no improvements in the classroom until a direc t behavioral intervention was conducted with the teachers. The study thus provided no evidence of setting generalization following this brief parent training intervention. Wahler (1969) concluded that the home and school settings were functionally independent such that the behavior of the child only improved in the setting in which the contingencies were operating. Wahler (1975) conducted a second study t o further investigate the question of whether the home and school settings are functionally independent. Subjects were two boys (ages 10 and 11) who were both referred by teachers for predelinquent behavior problems in the

PAGE 35

28 classroom. One of the boys also displayed acting-out behav i o r problems in the home. A parent training program was used in which parents were taught a combination of reinforcement and time-out procedures. Baseline behavioral observations showed that the behavioral clusters evident at home were different from those being displayed in the school setting. As in the previous school generalization study (Wahler, 1969), the successful management of one of the subject's behavior in the home resulted in no significant changes in school behavior. For the second subject, improvements at home were found to be associated with increased oppositional behavior in the classroom, a result which Wahler referred to as covariation (later termed a "behavioral contrast effect"). Because this was a case study and one of the subjects did not demonstrate the covariation response, conclusions that home to school generalization either does not occur or that children become more deviant at school following successful parent management training were judged to be premature. Instead, Wahler (1975) emphasized the need for additional research on setting generality with parent training programs. Jo~nson, Bolstad, and Lobitz (1976) also examined school generalization following a parent training program. Treatment subjects were eight children (mean

PAGE 36

29 age= 8.5 years; range= 5.3 to 12.8) who were referred for treatment of home behavior problems. Control subjects were eight untreated children (mean age= 7.5 years; range= 7.2 to 8.3) who were exhibiting deviant behavior at school but were not classmates of the referred children. Initial school observations and teacher report measures showed that the school behavior of the nonreferred control subjects was significantly more deviant than the pre-treatment school behavior of children who were referred for treatment. Thus it is unclear whether the children referred for treatment of home behavior problems also were demonstrating significant behavior problems at school. The referred children received a treatment program that involved teaching the parents to modify specific, targeted behaviors using behavioral principles. After three months of treatment, parents displayed more positive attitudes toward their children and improvement was evident in targeted behaviors. However, no overall changes in deviant behavior in the home were observed. Post-treatment classroom evaluations were conducted 3.5 months after the pre-treatment evaluation for referred children, whereas the deviant control children were reevaluated after a 1.8 month time interval. The post-treatment classroom assessment showed a

PAGE 37

30 nonsignificant increase in observed deviant behavior and no change in teacher r eport of behavior problems for the referred subjects. The untreated control group demonstrated more classroom deviant behavior at pre-treatment and less at post-treatment than the referred children. These pre-and p ost-treatment between-group differences did not attain significance. However, the slight improvement in the control group's behavior concurrent with the slight deterioration in the behavior of the treated group represented a significant Group by Time interaction (Johnson et al., 1976). Although Johnson et al. (1976) concluded that their data supported a "behavioral contrast effect," they cautioned against drawing definitive conclusions from their study due to the small sample size and small magnitude of improvement in home behavior. In addition to these problems, the method of selecting and assessing the treatment and control children makes between group comparisons of school behavior difficult to interpret. Because the treatment subjects were not referred for school problems, it is not clear whether their classroom behavior was sufficiently deviant before treatment to be affected by improvements in home behavior. Also, because the control children were not selected from the referred children's classrooms, between-classroom

PAGE 38

31 differences in teaching and disciplining styles may have contributed to the behavioral changes. Another consideratio~ in interpreting these findings is that the control group received their "post-treatment'' assessment after a shorter time-interval than the treatment group. It is unknown how the control group's behavior would have compared to that of the treatment group after a comparable amount of time. In a well-designed study of the overall effectiveness of behavioral parent training and cognitive-behavioral self-control therapy for ADD-H children, school generalization was examined as one indicator of treatment outcom~. Horn, Ialongo, Popovich, and Peradotto (1987) randomly assigned 24 families with ADD-H children between the ages of seven and eleven years to either group behavioral parent training only, group self-control training only, or a combination of group parent training and self-control training. The treatment groups each met for eight weekly 90-minute sessions. Pre-treatment assessment of classroom performance included achievement testing, teacher report of activity/attentional problems (Conners' Te~cher Questionnaire), and a 30-minute behavioral observation of disruptiveness, demanding the teacher's attention, time off-task, and gross motor

PAGE 39

32 movemeuts. All three treatment groups evidenced significant behavior improvments in the home following treatment and at 1-month follow-up, but no treatment was found to be superior to the others. With respect to behavior in the classroom, no evidence f o r school generalization or a behavioral contrast effect was found for any of the three treatment groups. The authors concluded that because of the situationally specific nature of childhood disorders, "a successful treatment program for a child with behavioral problems at home and in school must develop coordinated treatment components for home and school settings" (Horn et al., 1987, p. 65) Sayger and Horne (1987} obtained more favorable school generalization findings in a study of the long-range effectiveness of a social learning family therapy model. The sample consisted of 20 aggressive boys in the second through sixth grade whose parents chose to participate in treatment following referral by school personnel. All families participated in a 10-session treatment program which included training in the following five areas: setting-up for success, self-control, discipline, reinforcement, and communication. Children were assessed before and after treatment and at a nine month follow-up. A number of

PAGE 40

33 improvements were found in the home behavior of the children, as well as in family relationships and problem-solving skills. School behavior was assessed using the Daily Behavior Checklist (Prinz, O'Connor, & Wilson, 1981) which allows teachers to rate the occurrence of classroom behavior problems over three consecutive days. The children's classroom behavior was found to improve significantly from pre-treatment to post-treatment (22.SO to 13.25 on the DBC) and from post-treatment to follow-up (13.25 to 7.75) (Sayger & Horn, 1987). These results are difficult to interpret, however, because classroom behavior was assessed using only a single teacher report measure, and a classroom control group was not included. For the most part, the school generalization studies reviewed thus far have provided little evidence to suggest that treatment effects resulting from Parent Child Interaction Therapy might generalize to classroom behavior. However, none of these studies directly examined the effectiveness of the Hanf model. Instead, they represent such diverse treatment approaches as behavioral parent training targeting single behaviors, social learning family therapy, and group parent management training. It is unknown whether the results of studies using these approaches are applicable to PCIT

PAGE 41

34 because PCIT is based on a different treatment model and involves a different set of procedures. More favorable classroom findings actually could be expected from a Hanf approach because behavior problems are managed through the implementation of general improvements in the parent-child relationship, rather than targeting a small number of specific behaviors. Positive school behavior changes also would be expected because the Hanf model is used primarily with very young children whose relatively short learning histories may make them particularly amenable to the types of behavioral and personality changes that are necessary for generalization. In fact, the treatment outcome literature on the Hanf model has provided strong evidence for generalization across time (e.g., Forehand et al., 1979), and behaviors (e.g., Wells et al., 1980), as well as from the clinic to home setting (e.g., Eyberg & Robinson, 1982; Webster-Stratton, 1984). Additionally, multiple studies have demonstrated that Hanf model programs lead to decreases in child noncompliance and disruptiveness (e.g., Eyberg & Ross, 1978), two areas of behavioral improvement that would be most beneficial to the children's adjustment to the behavioral demands of the classroom setting.

PAGE 42

35 Despite the above evidence suggesting the likelihood of school generalization, Hanf model studies to date have provided no evidence of improvement in classroom behavior following this clinic-based treatment approach. In the first of two controlled group studies using the Hanf model, the classroom behavior of eight children referred for treatment of noncompliance was compared to that of eight normal control children (Forehand et al., 1979). The treatment children were not specifically referred for school behavior problems, and the treatment and control groups were not found to differ on level of inappropriate school behavior before treatment. Following treatment, the referred children's rate of compliance in the home increased from 88% to 93%. In the classroom, nonsignificant increases in inappropriate behavior occurred for both groups of children over the eight week treatment period. Although there were no significant differences between the groups, five of the treated children, in contrast to only four of the control children, were found to demonstrate more inappropriate behavior after treatment. It was suggested that this finding could be indicative of a behavioral contrast effect (Forehand et al., 1979). In the second school generalization study, Breiner and Forehand (1981) evaluated noncompliance and inappropriate behavior in the classroom using a larger

PAGE 43

36 sample of subjects. As in the previous study (Forehand et al., 1979), the treatment children were not specifically referred for school problems, and pre-treatment school observations revealed no significant differences between the 16 treatment children and their normal classroom controls (Breiner & Forehand, 1981). After approximately eight weeks of treatment, 11 of the 16 treated children became less deviant and more compliant in the home. In the classroom, both the treatment and control children became more compliant and less oppositional, and there still were no significant differences between the groups. It was concluded that, "if school problems exist, they will not be reduced by treatment directed toward home problems. School problems will have to be directly programmed into an overall treatment package in order to successfully reduce them" (Breiner & Forehand, 1981, p. 41). Statement of Problem and Hypotheses Because Hanf model programs differ from other parent training programs in that they have a different set of procedures, the Hanf model generalization literature is considered most applicable to the question of whether home behavioral improvements generalize to

PAGE 44

37 the school setting following Parent-Child Interaction Therapy. The two Hanf model generalization studies both provided no evidence of school generalization and concluded that direct classroom interventions are necessary for school behavior change. However, two methodological issues of concern in the previous Hanf model studies indicate that this negative conclusion regarding school generalization may be premature. First, to evaluate cross setting generalization, it is necessary to document that treatment resulted in clinically significant improvements in the primary setting. In the previous Hanf model investigations of school generalization, child compliance in the home increased by only 5% (Forehand et al., 88% to 93%; Breiner & Forehand, 30% to 35%) and home deviant behavior decreased only from 9% to 6% after treatment (Breiner & Forehand, 1981). Because of the apparently limited magnitude of change in home behavior, it may be that the children's overall adjustment was not affected to such a degree that improvements could be noted across settings. Second, to evaluate school generalization, it is necessary to document that the treatment children have conductproblems in the classroom to which the positive changes in home behavior can generalize. Children in both of these studies displayed normal

PAGE 45

38 classroom behavior before treatment, making it difficult to demonstrate significant improvements in the school setting. Therefore, the issue of school generalization with the Hanf model, and more specifically Parent-Child Interaction Therapy, is considered unanswered and in need of further investigation. The purpose of the present study was to evaluate school generalization in referred children who demonstrated severe conduct problem behaviors in both the home and school settings before treatment and who demonstrated clinically significant improvements in home behavior after treatment. Comparisons of the treatment children to both normal and deviant classroom controls were conducted on multiple measures to evaluate the clinical significance of school changes. It was hypothesized that a) PCIT would result in home behavior improvements to within normal limits following treatment, b) the treated children would demonstrate greater classroom behavior improvements in pre-treatment problem behaviors than either the normal or untreated deviant classroom controls, and cl the treated children would demonstrate clinically significant improvements in their school behavior problems upon completion of therapy (i.e., outside of normal limits before treatment and within normal limits after treatment). Social

PAGE 46

39 behavior was an exploratory variable in this study that also was hypothesized t o improve following treatment, but as a result of second order, rather than first order, generalization. That is, the generalization of behavioral improvements to the school setting was hypothesized to lead to improved social skills because of the increased opportunity for the treated children to display prosocial, rather than deviant, behaviors in interactions with peers.

PAGE 47

CHAPTER 2 METHOD Subjects Subjects were 30 children between the ages of 2 years, 0 months and 7 years, 0 months. Twenty-four of the children attended preschool, three were in kindergarten, and three were in first grade. There were three subject groups: Treatment Group (TG) (n=lO), Normal Classroom Controls (NC) (n=lO), and Untreated Deviant Classroom Controls (DC) 127; Problem Score> 11) (Eyberg & Ross, 1978): b) demonstrated a compliance ratio (averaged across three semi-structured situations) less than that -typically shown by nonreferred children (i.e., < 62%) (Robinson & Eyberg, 1981) on at least one of the two baseline clinic observations using the Dyadic 40

PAGE 48

41 Parent-Child Interaction Coding System (DPICS) (Robinson & Eyberg, 1981); c) received a DSM-III-R diagnosis (via structured interview) of either Oppositional Defiant Disorder, Attention-deficit Hyperactivity Disorder, or Conduct Disorder (American Psychiatric Association, 1987); and d) received teacher ratings on the Revised Conners Teacher Rating Scale at least one standard deviation above the mean for nonreferred children on either the Conduct Problem factor (i.e., for 3 to 5 year olds, > 1.23; for 6 to 8 year olds, > .71) or Hyperactivity Index (i.e., for 3 to 5 year olds, > 1.60; for 6 to 8 year olds, > 1.05) (Goyette, Conners, & Ulrich, 1978). Families were excluded from participation if either the child or the primary caretaker had a history of moderate to profound mental retardation
PAGE 49

42 child's particular problems and whether the child demonstrated each behavior rarely, occasionally, pretty often, or very often (Campbell, Ewing, Breaux, & Szumowski, 1986). A child was considered to display a particular symptom only if the parent indicated that it occurred pretty often or very often (Campbell et al., 1986). Diagnoses were made according to whether the child demonstrated the minimum number of symptoms specified in the DSM-III-R for each disorder. Interrater agreement was evaluated by comparing the interview checklist data collected by the primary interviewer with that generated by an independent observer (the co-therapist). Interrater agreement was calculated by dividing the number of agreements by the number of agreements plus disagreements. The percentages of agreement were found to be 100% for the duration of the behavior problems and 99% for the frequency of specific diagnostic symptoms. Structured parental interviews addressing the DSM-III-R Disruptive Behavior Disorders (APA, 1987) resulted in the following diagnostic breakdown for the TG children: a) five children met the criteria for both Oppositional Defiant Disorder (ODD) and Attention-deficit Hyperactivity Disorder (ADHD), b) three children met the criteria for ODD, ADHD, and

PAGE 50

43 Conduct Disorder, c) one child met only the criteria for ODD, and d) one child met only the criteria for ADHD. The mean age of TG children was 4.5 years (range= 31 to 79 months; SD= 14.8 months). All of the TG children were male, and nine were White (one child was Oriental). The mean income for TG families was $21,360 (SD= $25,370), with a median income of $12,000. Welfare funds were the sole source of income for four of the TG families, and four of the TG children were being raised by single mothers. Of the six two parent families, four of the fathers participated in treatment. A Normal Classroom Control (NC) subject and an Untreated Deviant Classroom Control (DC) subject were selected from each of the classrooms of the TG children. To reduce the possibility of expectancy effects, the two classroom controls were selected by an individual who was familiar with the students in the classroom (e.g., classroom aide, counselor, daycare director) but was not the main classroom teacher who completed the teacher report measures. This individual was asked to select a child in the TG child's classroom who demonstrated average behavior as the NC subject and a child who demonstrated behavior problems in that classroom as the DC subject. Teachers who completed questionnaires were reimbursed $20 for their time.

PAGE 51

44 The mean age of NC subjects was 4.9 years (range= 30 to 73 months; SD= 13.9 months). Six of the NC subjects were male, and eight were White (two were Black). The mean age of DC subjects was 4.4 years (range= 31 to 86 months; SD= 13.1 months). Seven of the DC children were male, and six were White (three were Black, one was Spanish). Treatment Outcome Measures Dyadic Parent-Child Interaction Coding System (DPICS) The DPICS is a measure of parent-child social interaction that specifically was designed to assess treatment progress and to be used as a pre-and post-treatment observational measure (Eyberg & Robinson, 1983a; Robinson & Eyberg, 1981). Parent-child interactions are coded during three standard five-minute situations: Child-Directed Interaction (CDI), Parent-Directed Interaction (PDI), and Clean-Up. The following instructions were given to parents for CDI: "In this situation, tell (child's name) that he/she may play whatever he/she chooses. Let him/her choose any activity he/she wishes. You just follow his/her lead and play along with him/her." For PDI, the parents were given the following instructions: "That was fine. Now we'll switch to another situation. Tell (child's name) that it is your turn to choose the game. You may choose

PAGE 52

45 any activity. Keep him/her playing with you according to your rules." For the Clean-Up situation, the parents were given the following instructions (slightly modified from those provided in the manual): "That was fine. Now I'd like you to tell (child's name) that it is time to put the toys away. Get him/her to put all the toys away without your help. Make sure that you have him/her put them away all by himself/herself. You should get (child's name) to put _each toy in its container and then put all the containers into the big toy box (Robinson & Eyberg, 1981) ." The DPICS was standardized for the CDI and PDI clinic situations with a sample of 22 normal families and 20 families referred for treatment of a conduct problem child. All children in the standardization samples were between the ages of 2.0 and 7.0 years (Robinson & Eyberg, 1981). Norms are reported in the DPICS manual for both parent and child behaviors (Eyberg & Robinson, 1983a). The mean compliance ratio for children in the normal control families during ten minutes of PDI was 62 percent. As norms have not yet been reported for the mean compliance ratio in Clean-up, 62 percent was used in this study as the best available estimate of normative compliance across the three DPICS situations. The mean inter-rater reliability for the

PAGE 53

46 child behaviors was found to be .92 {range= .76 to 1. 0) Several studies provide supportive evidence for the validity of DPICS. The coding system has been found to distinguish the behavior of conduct problem children from their siblings and normal children {Robinson & Eyberg, 1981). In a study by Robinson, Eyberg, and Ross (1980), the coding system was found to predict 61% of the variance in parent report of home behavior problems on the ECBI. The validity of DPICS is further supported by data demonstrating its sensitivity to treatment effects. Child compliance, as measured by DPICS, has been shown to increase from 39% at pre-treatment to 89% at post-treatment {Eyberg & Robinson, 1982). All DPICS coders for the present study were blind to the stage of intervention and coded all observations from videotapes. Coders received approximately 30 hours of training and met a 75% level of agreement with the criterion coding of a prerecorded videotape. DPICS reliability checks (using all three of the structured situations) were conducted on five of the ten TG subjects for a total of 11 of the 40 observations {i.e., 27. 5%) F i ve of the reliability checks were conducted on pre-treatment data, while the remaining six of the checks were conducted on post-treatment data.

PAGE 54

47 Interceder reliability coefficients were as follows: total~= .982, compliance~= .929, and deviant behavior~= .979. Although a number of parent behaviors are included in DPICS, only the child compliance (total# of complies across three DPICS situations divided by total# of commands) and deviant behavior (total# of deviant behaviors across three DPICS situations) were analyzed in this study. The DPICS observations were conducted twice before treatment and twice after treatment (with a one week interval both pre-and post-treatment). Pearson stability correlation coefficients were as follows: a) pre-treatment compliance~= .537 ( = .21), b) pre-treatment deviance~= .978 ( < .01), c) post-treatment compliance~= .438 ( = .24), d) post-treatment deviance~= .572 ( = .11), e) total (pre-and post-treatment combined) compliance~= .664 ( < .01), f) total deviance~= .888 ( < .01). The compliance and deviant behavior scores used in the statistical analyses were mean scores that had been averaged across the two DPICS observations. Eyberg Child Behavior Inventory (ECBI) The ECBI is a 36-item parent report inventory measuring a variety of behavior problems of children between the ages of 2 and 16 years (Eyberg & Robinson,

PAGE 55

48 1983b; Eyberg & Ross, 1978; Robinson, Eyberg, & Ross, 1980) (see Appendix B). The ECBI includes such items as "refuses to go to bed on time," "teases or provokes other children," "acts defiant when told to do something," and "is overactive or restless." This inventory allows the parent to indicate both the current frequency of occurence of particular behaviors and whether they are considered to be problems. Frequency ratings are scored on a scale of 1 ("never" occurs) to 7 ("always" occurs) and are summed to yield an overall problem behavior Intensity Score. For each item, the parent also indicates whether the behavior is currently a problem by circling either "yes" or "no," and an overall Problem Score is computed by summing the number of "yes" responses. The psychometric properties of the ECBI originally were established on samples of 42 normal preschool children and 43 behavior problem preschoolers (age range for both samples = 2. 0 to 7. 0 years) (Eyberg & Ross, 1978). The mean Intensity Score for the normal preschoolers was 102.6 (SD= 25.5), while the mean Intensity Score for behavior problem preschoolers was 158.3 (SD= 31.7). The mean Problem Score for the normal subjects was 4.62 (SD= 5.7), while the mean for behavior problem subjects was 18.6 (SD= 7.2). Several

PAGE 56

49 studies have demonstrated that the ECBI has satisfactory test-retest reliability {~ = .86 at an interval of 3 weeks) and internal consistency {~ = .98 for both scales) {Eyberg & Ross, 1978; Robinson, Eyberg, & Ross, 1980). Findings of significant inter-item correlations, high internal consistency coefficients, and homogeneity of each of the ECBI scales provide good evidence for internal validity (Robinson, Eyberg, & Ross, 1980). External validity is supported by findings that the ECBI correlates well with independent observations of children's behavior and discriminates between normals and children referred for treatment of conduct problems (Eyberg & Ross, 1978; Robinson, Eyberg, & Ross, 1980). School Generalization Measures Classroom Coding System This system was developed from coding systems used by Forehand and colleagues (Breiner & Forehand, 1981; Forehand et al., 1979) and those used by Walker, Shinn, O'Neill, and Ramsey (1987). Three behavior categories were coded for each child: a) Appropriate Behavior vs. Oppositional Behavior, b) Comply vs. Noncomply vs. Unsure/ No Command Given, and c) On Task vs. Off Task vs. Not Applicable. The Appropriate Behavior category and the Comply category were used to code classroom behavior in two previous studies of school

PAGE 57

50 generalization (Breiner & Forehand, 1981; Forehand et al., 1979). The On Task category is based on the definition of academically engaged time proposed by Walker et al. (1987). (See Appendix C for definitions of each of the coding categories.) The classroom coding system yields three scores for each child: percent appropriate behavior (# of intervals child was appropriate divided by total# of intervals), percent compliance (# of commands obeyed divided by total# of commands), and percent on task (# of intervals on task divided by the total# of intervals in which the child was expected to be on task). Data were coded using an interval sampling procedure similar to that used by Forehand et al. (1979) and Breiner and Forehand (1981). Coders used a tape recorder and earphones to listen to a prerecorded audiotape that cued them to the beginning and end of each interval. The TG child and the two control children were observed sequentially. Each target child was observed during a 10-second period followed by a 10-second marking period. Two 45 interval classroom observations were conducted prior to treatment and two 45 interval sessions were conducted at the end of treatment. The 45 interval sessions typically were conducted two days apart and during the morning hours. Pearson stability

PAGE 58

51 coefficients (first 45 interval observation compared to second 45 interval observation) for the three behavioral categories were as follows: Compliance~= .392 ( < .01), Appropriate~= .734 ( < .01), and On Task~= .745 ( < .01). Breiner and Forehand (1981) obtained high interrater reliability scores for both the Appropriate vs. Oppositional category (84%) and the Comply vs. Noncomply behavioral category (88%). Forehand et al. (1979) did not report reliability for individual categories but obtained an overall interrater reliability of 93%. High interrater reliability has also been obtained using the On Task vs. Off Task category. Walker et al. ( 1987) obtained in terr a ter reliability coefficients ranging from 91% to 100% with an average agreement of 97% for the On Task category. This. study also provided evidence for the discriminant validity of the On Task category. Walker et al. (1987) found that the sixteen fifth grade antisocial subjects obtained a mean On Task time of 68% (SD= 16.8%) whereas the non-antisocial control children achieved a mean On Task time of 85% (SD= 12.7%). Coders received approximately 25 hours of training and met an 80% interrater agreement criterion with the principal investigator before being permitted to code in

PAGE 59

52 the classroom. School coders were blind to group assignment of individual children and stage of intervention. Reliability checks were conducted for 25 of the 40 observations (62.5%) by using a dual-jack earplug that cued both coders to the beginning and end of the 10-second intervals. Reliability checks were conducted on 81% or pre-treatment observations and 40% of post-treatment observations. Intermittent training sessions were conducted throughout the data collection to minimize the possibility of observer drift. Interrater reliability percentages (# of Agreements divided by# of Agreements plus Disagreements were as follows: Total= 92.3% (range 85 to 96%), Comply= 93.2% (range 83 to 99%), Appropriate= 92.6% (range 87 to 97%), and On Task= 90.9% (range 80 to 96). Pre-treatment total agreement was 90.8% (SD= 4.2), and post-treatment total agreement was 93.9% (SD= 2.2). Revised Conners Teacher Rating Scale The Revised Conners is a 28 item teacher report inventory that is widely used for the assessment of hyperactivity in children ages 3 through 17 (Goyette, Conners, & Ulrich, 1978) (see Appendix D). This inventory is based on an earlier 39 item scale (Conners, 1969). The Revised Conners includes such items as "temper outbursts and unpredictable behavior," "mood

PAGE 60

53 changes quickly and drastically," "excitable, impulsive," ''excessive demands for teacher attention," and "uncooperative with classmates." For each item the teacher marks whether the behavior is present "not at all," ''just a little," "pretty much," or "very much." These ratings are given weights of 0, 1, 2, and 3 respectively, with the higher score representing greater symptomatology. A total score is obtained by summing the weights of the scored items and dividing the sum by 28. Scores also are obtained for the Hyperactivity (i.e. Hyperkinesis) Index as well as the following three factors: Conduct Problems, Hyperactivity, and Inattentive-Passive. Only the Conduct Problems and Hyperactivity Index factor scores were analyzed in this study. These scores are obtained by summing the points for each of the items assigned to the factor and dividing by the number of factor items. The Revised Conners was standardized on a sample of 570 children ranging in age from 3 to 17 (Goyette, Conners, & Ulrich, 1978). Three to five year old children (n = 24) in the normative sample obtained the following mean factor scores: Conduct Problem= .49 (SD = .74); Hyperactivity= .74 (SD= .74); Inattentive-Passive= .83 (SD= .87); Hyperactivity Index= .78 (SD= .82). Six to eight year old children

PAGE 61

54 (g = 102) obtained the following mean factor scores: Conduct Problem= .30 (SD= .41); Hyperactivity= .46 (SD= .57); Inattentive-Passive= .64 (SD= .71); Hyperactivity Index= .49 (SD= .56). The Revised Conners has been shown to have adequate reliability. Test-retest reliability scores for a one week interval have ranged from .88 to .98 for the various factors (Edelbrock, Greenbaum, & Conover, 1985; Edelbrock & Reed, 1984). The 39-item version was found to have adequate test-retest reliability for all three factors over a one month interval (ranging from~= .72 to .91) (Conners, 1969). Inter-rater reliability gathered from parents and teachers on the Hyperkinesis Index (a 10 item subscale that is identical for parent and teacher forms) of the Revised Conners was found to be~= .49 (Goyette et al., 1978). Acceptable correlations between parent and teacher factor scores also have been obtained providing evidence of convergent validity (Conduct Problem= .33; Inattentive-Passive/ Learning Problem= .45; Hyperactivity/ Impulsive-Hyperactive= .36) (Goyette et al., 1978). Numerous studies have shown that the original form of this rating ~cale (Conners, 1969) discriminates between normal and hyperactive children (e.g., Conners, 1970), and both the original and the Revised Conners have been

PAGE 62

55 shown to be sensitive to drug treatment effects (e.g., Barkley, 1977; Barkley, Fischer, Newby, & Breen, 1988). Sutter-Eyberg Student Behavior Inventory (SESBI) The SESBI is a 36-item teacher report inventory measuring a variety of conduct problem behaviors of young children (Sutter & Eyberg, 1984) (see Appendix E). The SESBI includes such items as "argues with teachers about rules or instructions," "sasses teacher," "teases or provokes other students," "has difficulty staying on task," and "demands teacher attention." The teacher indicates both the frequency of occurrence of particular behaviors and whether they currently are considered to be problems. As is the case for the ECBI, frequency ratings are scored on a scale of 1 ("never" occurs) to 7 ("always" occurs) and are summed to yield an overall Intensity Score. The teacher indicates whether each behavior is a problem by circling either "yes" or "no," and the overall Problem Score is the sum of the number of "yes" responses. The SESBI was standardized on 55 three to five year old children (Funderburk & Eyberg; in press). The mean Intensity Score was 100.9 (range: 36 to 228) with a standard deviation of 47.6. The Problem Scores ranged from Oto 33 with a mean of 6.0 (SD= 8.8). The Intensity Score and the Problem Score were significantly

PAGE 63

56 correlated (r = .65, < .0001), indicating that the scales measure similar but not identical dimensions. The teacher completed SESBI ratings and the parent completed ECBI ratings were not found to be significantly correlated (Intensity Scores r = -.03). However, the SESBI has been shown to correlate significantly with two other teacher measures o f behavior problems. The correlation between the SESBI and the Preschool Behavior Questionnaire (Behar & Stringfield, 1974) was .73 (Funderburk & Eyberg, in press). Correlations between the SESBI and the Revised Conners Teacher Rating Scale were as follows: Conduct Facto r with Intensity Score= .90 (Problem Score= .88), Hyperactivity Factor with Intensity Score= .89 (P.S. = .88), and Inattentive/Passive Factor with Intensity Score= .75 (P .S. = .67) (Sosna, Ladish, Warner, & Burns, 1989). Walker-McConnell Test of Children's Social Skills The Walker-McConnell consists of 43 positively stated items describing social-behavioral competencies of children in their interactions with pe~rs and teachers (Walker & McConnell, 1987) (see Appendix F). This scale includes such items as "makes friends easily with other children," "expresses anger appropriately,

PAGE 64

57 "listens carefully to teacher instructions and directions for assignments," "listens while others are speaking," "controls temper," "can accept not getting his or her own way," and "gains peer attention in an appropriate manner" (Walker & McConnell, 1987). Teachers rate the occurrence of each behavior on a Likert-type scale ranging from 1 ("never") to 5 ("frequently"). The scale produces a total score and the following three subscale scores: 1) Teacher-preferred social behavior, which is defined as "peer-related behavioral competencies which teachers value as appropriate to an academic setting'' (e.g., sharing, assisting others, taking turns), 2) Peer-preferred social behavior, which is defined as "peer-related behavioral competencies that facilitate the development of friendships and social acceptance" (e.g., complimenting others, having extended conversations, playing games skillfully), and 3) School adjustment behavior, which is defined as "behavioral competencies that determine a positive teacher-pupil adjustment within instructional contexts" (e.g., making assistance needs known in an appropriate manner, following classroom rules, compliance). The Walker-McConnell was standardized on a sample of 762 children in grades kindergarten through sixth.

PAGE 65

58 This scale was chosen as the measure of social behavior for this study despite that fact that preschoolers were not included in the norm group for three reasons: a) the lack of available social behavior measures designed for young children, b) the individual items on the Walker-McConnell appear appropriate for preschoolers, and c) the measure has been shown to have good psychometric properties for children as young as five years of age. The norm sample for the Walker-McConnell obtained the following scores: mean Total Adjustment score= 159.78 (SD= 33.16), mean Teacher-preferred social behavior= 58.66 (SD= 13.31), mean Peer-preferred social behavior= 66.98 (SD= 12.92), and mean School Adjustment behavior= 38.12 (SD= 10.27). Test-retest reliability over intervals of two weeks to one month have been found to be at or above .80 (Walker et al., 1987). Estimates of internal consistency yield alpha coefficients exceeding .90 for the total score as well as the three subscale scores. The Walker et al. (1987) study supported the discriminant validity of the scale. Antisocial subjects were rated by teachers as substantially less competent/adjusted than non-antisocial subjects on all scales except for the peer-preferred subscale. Mean scores for the 16

PAGE 66

59 antisocial subjects were as follows: Total Adjustment score= 134.94 (SD= 33.74), Teacher-preferred social behavior= 47.25 (SD= 11.88), Peer-preferred social behavior= 57.25 (SD= 14.55}, and School Adjustment behavior= 30.44 (SD= 9.38) (Walker et al., 1987). Procedures Preparation for Classroom Evaluation Parents of children referred for treatment of both home and school behavior problems initially completed the ECBI and the DSM-III-R Structured Interview to determine eligibility for inclusion in the TG. Then, the parent-child dyads were observed interacting in the three standard DPICS situations. These clinic behavioral observations were repeated one week later to increase the likelihood of obtaining a representative sample of behavior. DPICS data from the three structured situations were combined, and the resulting data for the two observation sessions were averaged. Families who met the inclusion criteria for clinically significant behavior problems in the home were then given the Revised Conners to be completed by the teacher. If the scores on the Revised Conners met the inclusion criteria for clinically significant behavior problems in the classroom, the family was invited to participate in the study. All families asked to participate agreed to be included in the research.

PAGE 67

60 Pre-Treatment Classroom Evaluation The Treatment Group child and the two control children were observed in the classroom for approxiffiately 45 minutes on two separate occasions. To minimize the children's reactivity to observers in the classroom, the children were told by their teacher that the data collectors were "people interested in finding out what we do in a preschool class." Because previous studies on school generalization have found that teachers sometimes give too few commands to code noncompliance reliably (e.g., Forehand et al., 1979), teachers were asked to give as many group commands as possible to the classroom as a whole during the observation period. The classroom observations were conducted as described in the section on the Classroom Coding System with the observers seated in an unobtrusive location in the room. Teachers completed the SESBI, Revised Conners, and Walker-McConnell for the Treatment Group ~hild, the Deviant Control Child, and the Normal Control Child. Each teacher was reimbursed $20 for the time that was required to complete the questionnaires. Description ~f Treatment Parent-Child Interaction Therapy (Boggs & Eyberg, in press; Eyberg & Robinson, 1982) is conducted in two

PAGE 68

61 treatment phases: Child-Directed Interaction (CDI) and Parent-Directed Interaction (PDI) (Eyberg & Robinson, 1982). Both phases involve didactic and coaching sessions. During didactic sessions, therapists first model and then have the parents role play skills. In subsequent sessions, parents are actively coached with their child using positive feedback through a bug-in-the-ear device. In CDI, the parents are taught to allow their child to lead the play activity. Parents are instructed to describe and praise the child's appropriate behavior, reflect appropriate child speech, and ignore inappropriate behavior. They learn not to criticize the child and not to use commands and leading questions which make it difficult for the child to lead the play. The major goal of CDI is to "create or strengthen a positive and mutually rewarding relationship" between the parent and child (Eyberg & Robinson, 1982). In PDI, the parents are taught how to direct their child's activity. They are instructed in the use of clear, positively stated direct commands and consistent consequences for behavior (i.e., praise for compliance, time-out in a chair for noncompliance). Parents learn to establish and enforce "house-rules" and to manage their child's behavior both at home and in

PAGE 69

62 public places. The major goal of PDI is to decrease problematic behavior while increasing lowrate prosocial behaviors. For a more complete description of Parent-Child Interaction Therapy, see Eyberg and Boggs (in press) and Eyberg (1988). All TG families received both the CDI and the PDI treatment phases. Because these subjects also were participating in a study comparing the effects of the two treatment phases, six of the subjects received CDI before PDI and the remainder received PDI before CDI. Thus, by the end of therapy all families had been taught the same child management skills, b~t there were two different phase order presentations. All TG families received 14 weekly one-hour sessions of Parent-Child Interaction Therapy (7 in CDI and 7 in PDI). No direct intervention was conducted in the classrooms. Session outlines were followed to avoid divergence from the treatment protocol (see Appendix G for a sample outline). Clinical psychology graduate students trained in the therapeutic approach served as the therapists for the study. Post-Treatment Clinic and Classroom Evaluations Upon completion of treatment (approximately four months after the initial intake evaluation) parents of TG children again completed the ECBI and were observed

PAGE 70

63 interacting with their children in the three standard DPICS situations. As in the pre-treatment clinic evaluation, DPICS observations were repeated one week later and the scores were averaged. A second classroom evaluation was conducted approximately one week after treatment. The TG children and the two controls were observed again on two separate days approximately two days apart. Teachers again completed the SESBI, Revised Conners, and Walker-McConnell for each of the three subjects and were reimbursed $20 for their time. Data Coding and Reliability Checks The primary coders were graduate students blind to group assignment and stage of intervention. To keep coders blind to stage of intervention, different coders were used for a particular TG subject's pre-and post-treatment observations. For each classroom observation session, there was one primary coder whose data were used in the study. For 25 of the 40 observations, the primary coder was accompanied by a reliability coder. The reliability coder was an advanced graduate student skilled in the use of the coding system. This reliability coder was not always blind to group assignment and stage of intervention. Thus, in all cases it was the data of the primary coder, not the reliability coder, that were analyzed for the study.

PAGE 71

CHAPTER 3 RESULTS To be included in the analyses, TG children were required to improve to within normal limits following treatment on either the DPICS compliance ratio or the ECBI Intensity score {see Method for information regarding normative data). All ten of the TG children met this criterion, indicating that each TG child demonstrated clinically significant improvements in behavior with their parents {see Table 1). Statistical significance was tested using paired-comparison t-tests evaluating pre-and post-treatment TG mean scores on DPICS compliance and deviance and ECBI Intensity and Problem scores {see Table 2). Significant pre-to post-treatment improvements were found on all variables except DPICS deviant behavior. Although mean deviant behavior change was in the positive direction, this change failed to reach statistical significance using a two-tailed test ( < .10). The TG mean scores for measures of home and clinic behavior {see Table 2) improved to within normal limits following treatment (Eyberg & Ross, 1978; Robinson & Eyberg, 1981). As mentioned previously in this paper, in addition to this generalization study TG subjects participated in a study evaluating the relative effectiveness of the 64

PAGE 72

65 Table 1. Pre-and Post-Treatment Scores of Individual Treatment Group Children on Measures of Home and Clinic Behavior Probleffis Subj. #1 #2 #3 #4 #5 #6 #7 #8 #9 #10 ECBI Intensity Pre I Post 175 I 118* 189 I 85* 220 I 133 221 I 105* 185 I 159 154 I 117* 171 I 56* 194 I 106* 170 I 76* 128 I 104* *Within normal limits ECBI Problem Pre I Post 24 I 8* 22 I 2* 30 I 1* 24 I 6* 33 I 26 25 I 8* 22 I 0* 28 I 3* 12 I 0* 13 I 7* DPICS Co mpliance1 Pre I Post 44 I 85* 19 I 47 51 I 73* 58 I 100* 52 I 82* 18 I 50 52 I 75* 46 I 68* 9 I 59 58 I 65* 1Total # of complies in the three DPICS situations divided by the total# of commands (averaged across two observations) two phases of PCIT and the impact of treatment phase order on overall outcome. As such, six of the TG subjects received CDI before PDI and the remainder received PDI before CDI. An analysis of possible differences between these two groups was conducted to

PAGE 73

66 Table 2. Pre-and Post-Treatment Comparisons of Treatment Group Means on Measures of Home and Clinic Behavior Problems Pre-Treatment Post-Treatment Measures M (SD) M ( SD) _t(df = 9) DPICS Comply1 40.7 (18.2) 70.4 (16.3) 7. 59* Deviance2 23.8 (27.6) 9.8 (12.6) 1. 42 ECBI Intensity 180.7 (28.2) 105.9 (29.2) 6. 60* Problem 23.3 ( 6. 7 l 6.1 (7. 7 l 7. 37* < 01 1Total # of complies in the three DPICS situations divided by total# of commands (averaged across two observations) 2Total # of deviant behaviors across three DPICS situations (averaged across two observations) determine whether treatment phase order influenced the degree of improvement in home and clinic behavior problems. No significant differences were found between the CDI-PDI group and the PDI-CDI group on pre-to post-treatment difference scores for the following measures of home and clinic behavior: ECBI Intensity score, ECBI Problem score, DPICS Compliance, DPICS Deviant Behavior (all 's > .05). Because no between

PAGE 74

67 group differences were found on magnitude of behavior change using measures of hom e and clinic behavior problems, there was no e vidence to suggest that treatment phase order would affect degree of school generalization. Thus, the CDI first group (rr = 6) was combined with the PDI first group (rr = 4) for the analyses of the school generalization measures. By the time that the TG children completed the PCIT program (approximately four months after the pre-treatment schoo l evaluation), one NC child and two DC children were no longer members of the classrooms in which they originally had been observed. Because inclusion of the incomplete data of these children could skew the findings, all data collected on subjects who were unavailable for the post-treatment school evaluation were excluded from analyses. To determine whether the improvements in home behavior generalized to the school setting, a multivariate analysis of variance was conducted that included the following measures of classroom behavior in the statistical package: percent appropriate behavior, percent compliance, percent of time on task, SESBI Intensity score, SESBI Problem score, Revised Conners Conduct Problem factor, Revised Conners Hyperactivity Index, and the Walker-McConnell scales (i.e., Total

PAGE 75

68 score, Teacher-preferred behavior, Peer-preferred behavior, and School Adjustment). This analysis yielded a significant result ( < .01), supporting the use of separate analyses of variance for individual measures. Thus, a 3 X 2 analysis of variance with one nonrepeated factor (group) and one repeated factor (time) was conducted for each of the dependent school variables listed above. For those scales displaying significant group by time interaction effects, pairwise group comparisons of difference scores (pre-treatment minus post-treatment) were done using the Duncan's Multiple Range Test. Results of analyses comparing behavioral changes of TG children with the DC and NC groups on observational measures of classroom behavior are presented in Table 3. TG subjects were found to demonstrate significantly greater improvements in a) percent of time behaving appropriately in the classroom ( < .01) and b) compliance to teacher commands ( < .05) than either the DC or NC groups. The average percentage of time that the TG was found to display appropriate behavior during school observations after treatment was within one-half of one standard deviation of that obtained by the NC group, suggesting that the behavior improved to within normal limits. Similarly, although the TG obtained a

PAGE 76

69 Table 3. Comparisons of Treatment Group to Untreated Deviant Controls and Normal Controls on Observational Measures of Classroom Behavior TG8 Pre Post Measures M M (SD) (SD) % Appropriate 65 87 (14) (14) % Compliance 54 87 (23) (16) % On Task 69 82 (14) (16) 8!l = 10. b!l = 8. c!l = 9. < .05, ** < .01 Pre Post M M (SD) (SD) 78 (12) 64 89 ( 5) 75 (18) (15) 80 80 (16) (10) Pre Post M M (SD) (SD) 89 ( 8) 73 90 ( 9) 80 (14) (13) 86 84 (6) (16) _E.(2,24) 8.09**1 4.24*1 1. 75 1Results of pairwise comparisons: TG difference score greater than both NC and DC (all 2.'s < .05). No difference between NC and DC ( > .05). significantly lower mean compliance ratio (54%) than the NC group (73%) or the DC group (64%) prior to treatment (both values< .05), the post-treatment compliance ratio (87%) was not significantly different from those obtained by the NC (80%) and the DC (75%) groups. These findings suggest that in addition to demonstrating

PAGE 77

70 statistically greater improvements in appropriate classroom behavior and compliance than the control groups, the TG displayed clinically significant gains to within normal limits on these two measures of school behavior. In contrast, significant group by time interaction effects were not found for the observational measure of on task behavior. Although the TG obtained significant pre-to post-treatment increases in percentage of time on task (~ < .05), the TG was not found to display significantly greater improvements on this measure than the two control groups. Qualitative analyses of the individual TG children were conducted to evaluate whether a subgroup of children displayed either no improvement or behavioral decrements that could not be detected in the group comparisons. As Eyberg and Johnson (1974) found a 30 percent change in pre-to post-treatment scores to be an effective cut-off value for determining the clinical significance of behavioral change in individual families, the 30 percent criterion was used in the present study for interpreting the classroom behavioral changes of individual TG children. The percent change values were obtained by dividing the pre-to post-treatment difference score by the pre-treatment score. A 30 percent improvement was interpreted as

PAGE 78

71 behavioral gain, whereas a 30 percent decrement was interpreted as behavioral deterioration. Pre-to post-treatment changes in the scores of individual TG children on the classroom observational measures are presented in Table 4. While four children improved in the area of appropriate classroom behavior and four children improved in the area of on-task time, the greatest number of behavioral improvements (~ = 8) was evidenced in the area of compliance with teacher commands. Eight of the ten children demonstrated behavioral improvements on at least one of the observational measures (i.e., appropriate behavior, compliance, time on-task}, and one child displayed a behavioral decrement on one of the measures (i.e., time on-task}. In general, these data support the presence of behavioral improvements in the majority of TG children at post-treatment, while providing only minimal evidence to suggest that individual TG children became more deviant in the classroom after therapy. Results of analyses comparing the TG to the two control groups on teacher report measures of school behavior problems are presented in Table 5. Teachers rated TG children as demonstrating significantly greater improvements than both the DC and NC children on three measures of oppositional and disruptive behavior (i.e.,

PAGE 79

72 Table 4. Pre-and Post-Treatment Scores of Individual Treatment Group Children on Observational Measures of School Behavior Subj. % Appropriate % Compliance % On Task Pre/ Post Pre/ Post Pre/ Post #1 40 I 91+ 45 I 100+ 41 I 93+ #2 46 I 89+ 31 I 85+ 61 I 86+ #3 74 I 93 100 I 83 82 I 70 #4 70 I 96+ 53 I 91+ 62 I 95+ #5 50 I 72+ 55 I 82+ 58 I 71 #6 80 I 97 62 I 86+ 85 I 80 #7 76 I 95 83 I 100 78 I 86 #8 76 I 96 37 I 100+ 78 I 96 #9 60 I 55 23 I 47+ 75 I 46#10 73 I 86 54 I 100+ 73 I 96+ Greater than 30% improvement relative to pre-treatment -Greater than 30% decrement relative to pre-treatment Revised Conners Conduct Problem factor, SESBI Intensity score, SESBI Problem score) (all~ values < .05). The TG's mean scores on the Conduct Problem factor of the Revised Conners and the Intensity and Problem scores of the SESBI did not fall within one standard deviation of the NC group at post-treatment. However, the TG's mean

PAGE 80

73 Table 5. Comparisons of Treatment Group to Untreated Deviant Controls and Normal Controls on Teacher Report Measures of Classroom Behavior TG DC NC Pre Post Pre Post Pre Post Measures M M M M M M f(2,24) (SD) (SD) (SD) (SD) (SD) (SD) Revised Conners Conduct Probs. 1. 6 1. 0 1. 3 1. 2 .35 .33 13.90**1 ( 4) (. 6) (. 7) ( 6) ( 2) (. 2) Hyper. Index 1. 9 1.4 1. 5 1. 3 .28 .36 4. 51 2 (. 6) (. 8) ( 5) ( 7) ( 3) ( 4) SESBI Intensity 155 116 126 115 57 67 9.88**1 (21} (34) (36) (35) ( 15) (21) Problem 20 10 13 10 .67 1.1 14.00**1 ( 6) ( 7) ( 8) ( 9) ( 1) (2) *p < .05, **p < .01 1Results of pairwise comparisons: TG difference score greater than both NC and DC (all Q's< .05). No difference between NC and DC (Q > .05). 2Results of pairwise comparisons: TG difference score greater than NC (Q < .05). DC difference score greater than NC (Q < .05). No difference between TG and DC (Q > .05).

PAGE 81

74 scores on the Revised Conners improved from outside of normal limits prior to treatment to within one standard deviation of published normative data for 3 to 5 year olds after treatment (Goyette et al., 1978). Similarly, the TG's mean SESBI Intensity score and Problem score improved to within one standard deviation of normative data (Funderburk & Eyberg, in press) upon completion of treatment. With respect to hyperactivity and distractibility in the classroom, the TG obtained significant pre-to post-treatment decreases on the Hyperactivity Index of the Revised Conners (~ < .05} and demonstrated greater improvements on this measure than the NC group (~ < .05} (see Table 5). However, the TG was not reported to have significantly greater decreases in hyperactive behaviors than the DC group. In summary, teachers reported that the TG improved more than both of the control groups in the area of oppositional and disruptive behavior, but they did not report improvements in hyperactivity and distractibility greater than those demonstrated by the DC children. As was the case for the observational measures of school behavior, qualitative analyses of teacher report measures for the individual TG children were conducted (see Table 6) to evaluate whether a subgroup of children displayed either no improvement or behavioral decrements that could not be detected in the group comparisons.

PAGE 82

75 Table 6. Pre-and Post-Treatment Scores of Individual Treatment Group Children on Teacher Report Measures of School Behavior SESBI CONNERS Subj. #1 #2 #3 #4 #5 #6 #7 #8 #9 #10 Note. Intensity Score Pre/Post 121 I 93 159 I 76 159 I 118 157 I 72+ 191 I 173 139 I 139 139 I 89+ 136 I 110 180 I 144 165 I 141 There were Greater than 30% Problem Score Pre/Post 15 I 8+ 20 I 7+ 27 I 15+ 17 I 3+ 26 I 24 14 I 13 11 I 3+ 15 I i+ 30 I 16+ 23 I 9 Conduct Problem Pre/Post 1. 25 I 0. 38+ 1. 00 I 0. 25+ 1. 88 I 1. 38 1. 75 I 0. 75+ 2.13 I 2.13 1. 88 I 1.13+ 1. 25 I 0. 75+ 1. 50 I 0. 75 2.00 I 1. 25 1. 50 I 1. 38 Hyperactivity Index Pre/Post 1. 7 I 0. 8+ 2.7 I 2.1 2.0 I 1.7 1. 8 I 0. 5+ 2.5 I 2.6 1. 8 I 1. 4 1. 4 I 0. 5+ 0.7 I 0.5 2.0 I 1. 9 2.6 I 2.2 no decrements greater than 30%. improvement relative to pre-treatment Inspection of Table 6 indicates that six of the ten children showed significant improvements on at least two of the four teacher report measures of school behavior problems. No significant behavioral decrements were

PAGE 83

76 reported by teachers, and only one child (#5) displayed no i mprovement on any of the teacher report measures. Interestingly, this child also was the only child whose mother did not report significant gains in home behavior problems (see ECBI scores in Table 1). The finding that the teacher and parent of TG child #5 both reported only minimal improvement in his behavior raises the issue of whether there is a relationship between magnitude of change in the home and degree of change in the school. To evaluate this question, maternal ratings of behavior on the ECBI were compared to teacher ratings of behavior on the SESBI. Neither the pre-treatment nor post-treatment Pearson correlations between ECBI and SESBI Intensity scores were found to be significant (pre:~= .029, = .896; post:~= .502, = .137). However, when pre-treatment minus post-treatment difference scores were used, a significant Pearson correlation of .783 was obtained ( = .007). This correlation is indicative of a common shared variance of approximately 61% between parent and teacher reports of behavioral changes in their respective settings. Thus this result supports the presence of a significant relationship between magnitude of improvement in home behavior and magnitude of gain in classroom behavior.

PAGE 84

77 Table 7. Comparisons of Treatment Group to Untreated Deviant Controls and Normal Controls on Teacher Report of Social Competence TG DC NC Pre Post Pre Post Pre Post Measures M M M M M M [(2,24) (SD) (SD) (SD) (SD) (SD) (SD) Walker-McConnell Total 101 129 120 133 176 183 3.99 (26) (33) (20) (24) (27) (13) Teacher 35 45 41 46 64 65 2.87 Preferred ( 9) (11) ( 5) ( 7) ( 9) ( 5) Peer 40 53 49 56 69 74 1. 97 Preferred (15) (13) (14) (13) (12) ( 7) School 23 31 28 32 43 44 2.78 Adjustment ( 8) (11) ( 7) (11) ( 6) ( 4) Note. Higher scores represent more positive ratings. All E values are nonsignificant ( > .05). Between-group comparisons on the Walker-McConnell Test of Children's Social Skills are presented in Table 7. Although the TG had significant pre-to post-treatment mean increases on all of the Walker-McConnell scales (all values < .05), the TG

PAGE 85

78 children did not display significantly greater improvements than the control groups on these scales. These results yield no evidence to suggest that PCIT leads to improved social skills at school. In addition to the group comparisons, results of the Walker-McConnell Total scale were examined qualitatively for each individual TG child (see Table 8). Four of the TG children showed greater than 30 percent improvement on this measure, and none of these children were reported by teachers to demonstrate poorer social behavior in the classroom following treatment. In addition to the primary analyses presented above, a set of secondary analyses was conducted to examine a statistical issue. As already mentioned, data from the one NC child and the two DC children who were not available for post-treatment classroom observations were excluded from analyses. However, a decision was made to retain the data from all TG children and their one available classroom control in the results. It was felt that data from these subjects would contribute to the information gained from both the individual and group analyses. A potential problem with this decision is that for three of the TG children the effects of maturation and classroom factors are not completely controlled because they do not have both a NC and DC

PAGE 86

79 Table 8. Pre-and Post-Treatment Scores of Individual Treatment Group Children on Teacher Report of Social Competence Subj. #1 #2 #3 #4 #5 Walker-McConnell Total Pre I Post 109 I 163+ 58 / 67 121 / 132 111 I 161+ 86 / 94 Subj. #6 #7 #8 #9 #10 Walker-McConnell Total Pre I Post 93 I 130+ 128 / 145 141 / 166 80 / 135+ 78 / 98 Note. There were no decrements greater than 30%. Greater than 30% improvement relative to pre-treatment child for comparison. By including the three TG children who were missing one of their controls, an assumption was made that the drop-out children's data would have been approximately equivalent to that of like subjects (that is, the missing NC child would behave like the other NC children; the two missing DC children would be similar to others in the DC group). To ensure that this assumption was met, an additional set of ANOVAs was conducted that included only the seven TG children whose normal and deviant controls were both

PAGE 87

80 available for the post-treatment data collection. Results of the secondary analyses were consistent with the original results. Specifically, the TG was found to display significantly greater improvements than both control groups on the following measures: appropriate classroom behavior, compliance, Revised Conners Conduct Problem factor, SESBI Intensity score, and SESBI Problem score (all values< .05}. No significant differences between the TG and either control group were found for on task behavior, the Revised Conners Hyperactivity Index, and all scales of the Walker-McConnell (all values> .05}. These results indicate that the findings of this study are not confounded by the inclusion of subjects from the drop-out children's classrooms. In summary, the results of this study demonstrated that the TG improved significantly more than both the NC group and the DC group on all measures of oppositional and disruptive behavior (i.e., percent of time behaving appropriately in the classroom, percent compliance in the classroom, teacher report of conduct problems on the Revised Conners, and teacher report of the frequency and severity of conduct problems on the SESBI; all E's,< .05}. However, results in the areas of hyperactivity/distractibility and social competence were less favorable. Although the TG obtained significant

PAGE 88

81 Table 9. Summary of School Generalization Results for Individual Treatment Group Children on Measures of Behavior Problems in the Classroom Observational Measures Teacher Report Measures Approp-Compli-On SESBI SESBI CONNERS CONNERS Subj. riate ance Task I.S.1 P.S.2 C.P.3 H. I. 4 #1 + + + + + + #2 + + + + + + #3 + #4 + + + + + + + #5 + + #6 + + #7 + + + + #8 + + + #9 + + + #10 + + + Greater than 30% improvement relative to pre-treatment -Greater than 30% decrement relative to pre-treatment 1Intensity Score, 2Problem Score, 3Coriduct Problem Factor, 4Hyperactivity Index pre-to. post-treatment improvements in percentage of time on task ( < .05) and teacher report of hyperactivity on the Revised Conners ( < .05), TG children did not display significantly greater

PAGE 89

82 improvements than the DC group on these measures. Similarly, the TG's improvements on the Walker-McConnell, a measure of social skills in the school setting, were not found to be significantly greater than those obtained by the two control groups. Thus, although this exploratory measure was expected to demonstrate improved social skills in the treatment children as a result of decreased levels of deviant behavior in the presence of peers, the results did not support this hypothesis. Summary results of the pre-to post-treatment changes in classroom conduct problem behaviors for individual TG children are presented in Table 9. Inspection of the table indicates that all of the TG children improved on at least one of the seven measures of classroom behavior problems, and over half of the children improved on at least three of these measures. This suggests that the positive school generalization results in the area of oppositional/conduct problem behavior are due to relatively consistent improvements across individual TG children rather than to a positive response for only a small subgroup of the sample. Inspection of the summary table also indicates that a behavior decrement was evident for only one child and on only one measure, providing little evidence to support the presence of a behavioral contrast effect.

PAGE 90

CHAPTER 4 DISCUSSION Prior to the collection of data for this study, it was hypothesized that PCIT would result in home behavior improvements to within normal limits following treatment. It was important that this hypothesis be confirmed because a valid investigation of setting generality requires the documentation of clinically significant improvements in the primary setting. With respect to the treated group as a whole, clinical improvements were evidenced by the fact that parent report of home behavior problems and clinic observations of child compliance improved on average from outside of normal limits prior to treatment to within normal limits upon completion of therapy. Clinically significant improvements also were evident on the level of individual subjects. All ten of the treated children improved to within normal limits following treatment on either parent report of the frequency of conduct problems or clinic observations of compliance. Thus, the improvements noted on group mean scores are judged to represent uniform gains across the individual treatment children (as opposed to exceptional improvement in a subgroup of subjects which offsets the lack of responsiveness of a second subgroup). 83

PAGE 91

84 Results of the school evaluation indicate that the successful treatment of home behavior problems using PCIT is associated with improvements in certain behaviors in the school setting. This is the first controlled study supporting the hypothesis that school generalization occurs following parent training for home behavior problems, and as such, the present findings contradict two previously reported conclusions. First, there was little evidence provided by this study to suggest the presence of a behavioral contrast effect (e.g., Forehand et al., 1979; Johnson et al., 1976). Behavioral improvements rather than decrements were found in group analyses, and only one of the individual subjects displayed any evidence of becoming more deviant in the classroom following treatment. Second, contrary to previous investigations, the present study provided evidence of clinically significant improvements (i.e., to within normal limits) in classroom noncompliance and disruptive behavior following PCIT. This finding calls into question prior conclusions that direct school intervention is necessary to obtain improvements in classroom behavior (e.g., Breiner & Forehand, 1981; Horn et al., 1987; Patterson, Cobb, & Ray, 1973). Because these findings were obtained using a methodology that

PAGE 92

85 included control groups and two different methods of measurement (i.e., teacher report and observations conducted without the child's knowledge), the positive behavior changes in the classroom can be considered strong, objective evidence for the overall effectiveness of PCIT The two central hypotheses of this study were that the treated children would demonstrate greater classroom behavior improvements than either the normal or untreated deviant controls and that the school behavior of the treated children would improve to within normal limits upon completion of therapy. These hypotheses were confirmed in the area of conduct problem/oppositional behavior. The treated children improved to within normal limits (which represents a significantly greater improvement than that demonstrated by the two control groups) on observational and teacher report measures of compliance, disruptiveness, and oppositionality in the classroom. Specific behaviors assessed with these measures included disobeying teacher commands, sassing, teasing, hitting, talking out of turn, whining, yelling, and breaking school rules. Generalization in this area was expected because many of these behaviors also were considered problematic in the home and were addressed in treatment. In fact, child

PAGE 93

86 noncompliance with instructions and rules, which is a basic component of conduct problem/oppositional school behavior, is the primary area of focus for the Parent Directed Interaction phase of PCIT. In contrast to the above improvements in conduct problem/oppositional school behavior, generalization was not found in the areas of hyperactivity/inattention and peer relationships. With respect to the activity and attentional problems, there are several possible explanations for the lack of school generalization. One possibility is that treatment was effective for distractibility and overactivity in the home but that it did not generalize well because of the great disparity between the attentional demands of the school versus the home setting. An alternative possibility is that PCIT is less effective for the treatment of Attention-deficit Hyperactivity Disorder than Oppositional Defiant Disorder and Conduct Disorder. Because the sample size in this study did not permit an analysis of the differential responsiveness of children in these three diagnostic categories, this will be an important area for future research. If generalization in the hyperactivity area is found to occur for only a subgroup of children, then it would be beneficial to be able to identify the nonresponders so that they could be

PAGE 94

87 provided with such additional treatments as medication and individualized classroom behavioral programs. The lack of generalization in the area of peer relationships was not completely unexpected given that second-order generalization would have to occur to obtain such findings. Not only would it be necessary for improvements in the home to generalize to the school, but those decreases in noncompliance and deviant behavior in the classroom would then have to lead to improvements in the children's ability to socialize effectively with peers. There also may have been a methodological reason for the lack of generalization in the social skills domain. The measure used to assess social competence in this study, the Walker McConnell Test of Children's Social Skills, was standardized on a sample of children in kindergarten through sixth grade (Walker et al., 1987). Although this test discriminated the three groups at pre-treatment, it may not have been particularly sensitive to treatment effects because the items were designed primarily for older children. There was one finding, however, suggesting that this measure may have been a valid indicator of social skills changes in this young sample. Although the treatment children did not demonstrate greater improvements in social skills than the control children, they did develop more

PAGE 95

88 advanced social competencies over the course of treatment. It is possible that the peer relationships were improving but that adequate time had not passed for the long-standing negative interactions with classmates to be completely reversed. Because the other groups also improved, another possibility is that this measure assesses social skills that children are developing rapidly during their preschool years. Psychometric studies evaluating the stability of the Walker-McConnell in large samples of preschoolers would be helpful for future research in the area of social behavior in young children. Follow-up school generalization studies in this area also are needed to evaluate whether treated children eventually display improved social competence at school, particularly as they enter new classrooms containing children with whom they may not have established prior relationship expectancies. In the meantime, the lack of evidence for substantial improvement in peer relationships suggests that an additional social skills treatment component would be beneficial to the overall school adjustment of these children. One potentially important finding from this study that has not been reported previously is that maternal report of the magnitude of improvement in home behavior problems was significantly related to teacher report of

PAGE 96

89 the magnitude of improvement in school behavior problems (~ = .783). This result was found despite the fact that teacher and parent reports were not highly correlated either at pre-treatment or at post-treatment, a finding that is typical of most parent-teacher informant comparisons {e.g., Achenbach, Mcconaughy, & Howell, 1987). The lack of agreement between parents and teachers about children's behavior at a single point in time is influenced by a number of factors that include environmental demands, the presence of other children, interpersonal and discipline styles, rater tolerance for disruptive behavior, and rater expectancies of child behavior based on knowledge of developmental norms. When the pre-treatment minus post-treatment difference scores were used as the measure of magnitude of improvement, rater and environmental variances were controlled in that the magnitude score was relative to the initial rating. The resulting correlation indicates that relative improvements reported by parents in the home environment were directly related to those noted by teachers in the school environment. As this is the first controlled study demonstrating positive behavior changes in the classroom following parent training, it is important to examine factors that distinguish the present study from prior investigations.

PAGE 97

90 One potential reason for the discrepant findings is that PCIT may have different therapeutic effects than traditional parent training programs. As mentioned previously, PCIT differs from the parent training programs used by Wahler (1969; 1975), Johnson et al. (1976), and Horn et al. (1987) in that the therapy is directed towards the implementation of general improvements in the parent-child relationship rather than the targeting of a small number of specific problem behaviors for modification. Also, a significant portion of PCIT is devoted to teaching parents a form of play therapy which they are encouraged to use with their children on a daily basis. In this play therapy, parents provide a great deal of specific praise for their children's positive behaviors, qualities, and abilities, while listening carefully to what their children have to say, avoiding criticism, and displaying affection. From a behavioral perspective, it is possible that the play therapy component of PCIT enhances the likelihood of generalization by increasing the frequency of prosocial behaviors displayed by the child. In the daily play therapy sessions, parents consistently praise and attend to such prosocial behaviors as sharing, using polite manners, speaking quietly, controlling one's temper, and being helpful. As a result of these social

PAGE 98

91 rewards, the frequency of these behaviors increases, thereby increasing the probability that the prosocial behaviors will be emitted at school. When the behaviors spontaneously occur at school, the child may receive naturally occurring positive consequences from teachers and possibly peers, resulting in a higher frequency of the behaviors at school as well as at home. Alternatively, it is possible that the play therapy component of PCIT results in intrinsic changes within the children that are not situationally specific. For example, it could be speculated that daily play therapy leads to a decrease in the child's anger level and a more positive self-image, two factors that could result in improved behavior across situations. Unfortunately, empirical investigations of these speculations using self-esteem and affective measures for preschoolers currently are not available in the parent-child interaction literature. Future research of this type might shed light on the degree to which these "internal" child characteristics covary with the expression of positive behaviors in diverse situations and contribute to the explanation of cross-setting generality. Another possible explanation for the discrepant findings amongst PCIT and the traditional parent training programs used in earlier studies of school generalization is that PCIT focuses on overlearning of

PAGE 99

92 compliance to all parental instructions. During the second stage of treatment (PDI), parents engage in daily practice sessions in which they give many simple, specific commands and provide consistent consequences for compliance and noncompliance. Eventually, the parents are able to provide consistent consequences for every command given throughout the day. After many of these trials in which highly positive consequences consistently accompany compliance and an aversive time-out experience consistently is associated with noncompliance, the oppositional child's first response to parental commands appears to change from a determination to defy and disobey to an automatic, overlearned compliance response. As the compliance response increases, it is likely to occur in a greater number of situations and to obtain positive responses from others, as well. Over time, negative behaviors may no longer be necessary to receive sufficient attention. One final factor that may relate to the finding of school generalization is that the behavioral consistency and structure that is put into place by the parents during the course of PCIT is similar to the rule-based structure in operation in most schools. This consistency between settings with regard to behavioral demands and expectations may help the child to adapt

PAGE 100

93 more readily to the classroom routine and rules. Again, these are theoretical speculations that have not yet been empirically examined. Because the present study addressed only the question of whether school generalization occurs following PCIT, further research is necessary to identify and evaluate possible causes of this generalization. In the above discussion, it is proposed that PCIT involves a different set of therapeutic techniques than did traditional parent training which might result in specific changes within the child that increase the likelihood of generalization across settings. However, the question still remains as to why school improvements were evident in the present study even though previous Hanf model investigations employing a similar therapeutic approach found no evidence of school generalization. One important difference between this study and the two Hanf model studies conducted by Forehand and colleagues (Breiner & Forehand, 1981; Forehand et al., 1979) is that the subjects in the current research specifically were selected because they were demonstrating both home and school behavior problems. Because the children in both of the previous Hanf studies displayed normal classroom behavior before treatment, behavioral changes may not have been found

PAGE 101

94 because the behavior was not sufficiently deviant for significant improvements to be possible. Also, unlike the investigations conducted by Breiner and Forehand (1981) and Forehand et al. (1979), PCIT was only considered successful for home behavior problems if the children's behavior came to within normal limits after therapy. To achieve this degree of effectiveness, children received 14 weekly one-hour treatment sessions, a greater number of sessions than that of the other Hanf generalization studies (Breiner & Forehand= 8; Forehand et al.= 9.5). Upon completion of PCIT, treated subjects demonstrated a mean compliance increase during parent-child interactions of 29.7%, while the mean increase for both of the previous Hanf model studies was only 5% (Breiner & Forehand, 1981; Forehand et al., 1976). This discrepancy suggests that subjects in the present study may have obtained a greater magnitude of improvement in home behavior problems, thereby providing a better opportunity for generalization to occur. Several methodological problems arose during the course of this investigation that should be CQilsidered in interpreting the findings and planning future research in the school generalization area. One concern is that the teachers were aware of which child in their class was in treatment, and thus their reports of

PAGE 102

95 behavioral changes of treated children may have been biased by expectancies that the children should improve as a result of receiving therapy. This occurred because teacher report on the Revised Conners was required to evaluate the treatment child's appropriateness for the study and because teachers were given an explanation of the purpose of the study before permission was requested to conduct observations in the classroom. This knowledge may have led teachers to attend selectively to behavioral improvements in the treated children. Although an attempt was made to minimize expectancy effects regarding control children by having an individual other than the primary teacher select these subjects, it is possible that the teachers' impressions of the controls also were biased by preconceived expectations about the severity of behavior problems and the likelihood of behavior change. In considering the extent to which expectancy effects may have confounded the results, it is helpful to look at the degree to which behavioral observations confirmed the teacher reports. Given that the results of the classroom behavioral measures supported teacher reports that treated children demonstrated clinically significant improvements in the area of oppositional behavior but not in the area of hyperactivity, it is

PAGE 103

96 doubtful that expectancy effects rendered the teacher report measures used in this study invalid. However, to ensure that the results were not negatively affected by teacher bias, it will be important for future investigations to obtain reports from teachers who are unaware of the treatment status of subjects. A second potential problem with this study is that the untreated deviant control subjects did not have behavioral problems in the classroom that were as severe as those demonstrated by the treatment subjects. Because the treatment children were required to demonstrate clinically significant behavioral problems at school to be included in the study, there were few cases in which there was a potential control child in the classroom who was deviant enough to be a close behavioral match for the treatment child. Thus, there was greater room for improvement in the behavior of the Treatment Group than the Deviant Controls. This raises the question of whether the greater magnitude of improvement in the treatment children as compared to the deviant control children was due to generalization of treatment effects to the school setting or regression towards the mean. One result that could be used as evidence for generalization is that the treatment children's mean scores improved to within normal limits

PAGE 104

97 on measures of disruptive and oppositional behavior (nonstandardized observational measures improved to within one standard deviation of the normal controls; standardized teacher report measures improved to within the normative range). These changes represent clinically significant improvements that would be unlikely to result solely from regression towards the mean. Nevertheless, a thorough investigation of this issue is needed in which replication of the results is attempted using a behaviorally equivalent deviant control group. Despite the problems mentioned above with the selection of the control children, the inclusion of normal and deviant classroom control groups was found to be helpful for the interpretation of the school generalization results. Comparisons with normal control children allowed for an analysis of the clinical significance of the findings, and comparisons with deviant control children provided some information about the degree to which behavioral changes could be attributed to treatment effects versus extraneous factors (e.g., maturation, the effectiveness of teacher-initiated behavioral programs, adjustment to school routine).

PAGE 105

98 Follow-up studies are needed to determine whether the i mprovements in oppositional classroom behavior found in this study maintain over time. Although it is possible that children's overlearned behaviors will maintain as a result of intermittent reinforcement in the school setting, there also is a possibility that classroom improvements diminish as parents practice the learned skills less consistently and intensively in the home setting. With the inclusion of affective, personality, and self-esteem measures in this type of follow-up study, an evaluation could be made of possible internal and enduring child changes or trait factors that might be contributing to school generalization. Additionally, research is needed to evaluate whether the improvements demonstrated without direct school intervention are comparable to those that are obtained with individualized classroom behavioral intervention programs. If future studies show that more efficient results are obtained through direct school interventions than through generalization, then consideration should be given to expanding PCIT such that classroom programming would be an integral component of treatment for children with both home and school behavior problems.

PAGE 106

Child's Na me: Date: Rater's Na me: APPENDIX A DSM-III-R CHECKLIST l=rarely 2=occasionally 3=pretty often 4=very often Oppositional Defiant Disorder (313.81) A. Have the problems been present for at least six months? B. Co mpared with other children the same age, are at 1 2 1 2 1 2 1 2 1 2 least five of these problems? "4" must be circled) 3 4 1) often loses temper 3 4 2) often argues with adults 3 4 3) often actively defies or requests or rules, e.g. chores at home 3 4 4) often deliberately does (either "3" or refuses adult refuses to do things that annoy other people, e.g., grabs other children's hats 3 4 5) often blames others for his or her own mistakes 99

PAGE 107

1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 100 6 ) is often t ouchy o r easily annoyed by others 7) is ofte n angry and resentful 8 ) is often spiteful or vindictive 9) often swears or uses obscene language Check the appropriate level of severity (make this rating at the conclusion of the intake). Mild: Few, if any, symptoms in excess of those required to make the diagnosis and only minimal or no impairment in school and social functioning. Moderate: Symptoms or functional impairment intermediate between mild and severe. Severe : Many symptoms in excess of those required to make the diagnosis and significant and pervasive impairment in functioning at home and in school and with other adults and peers. Attention-deficit Hyperactivity Disorder (314.01) A. Have the symptoms been present for at least six months? B. Compared to other children the sam e age, are at least eight of these problems? must be circled) (either "3" or "4" 1 2 3 4 1 2 3 4 1 2 3 4 1) often fidgets with hands or feet or squirms in seat 2) has difficulty remaining seated when it is required 3) is easily distracted by extraneous stimuli

PAGE 108

1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 101 4) has difficulty awaiting turn in games or group situations 5) often blurts out answers to questions before they have been completed 6) has difficulty following through on instructions from others (not due to oppositional behavior or failure of comprehension), e.g., fails to finish chores 7) has difficulty sustaining attention in tasks or play 8) often shifts from one uncompleted activity to another 9) has difficulty playing q uietly 10) often talks excessively 11) often interrupts or intrudes on others, e.g. butts into other children's games 12) often does not seem to listen to what is being said to him or her 13) often loses things necessary for tasks or activities at school or home, eg. toys, pencils, books, homework 14) often engages in physically dangerous activities without considering possible consequences (not for the purpose of thrill seeking), e.g. runs into street without looking

PAGE 109

102 Check the appropriate level of severity (make this rating at the conclusion of the intake). Mild: Few, if any, symptoms in excess of those required to make the diagnosis and only minimal or no impairment in school and social functioning. Moderate: Symptoms or functional impairment intermediate between mild and severe. Severe: Many symptoms in excess of those required to make the diagnosis and significant and pervasive impairment in functioning at home and in school and with other adults and peers. Conduct Disorder (group type: 312.20; solitary aggressive type: 312.00; undifferentiated type: 312.90) A. Have the symptoms been present for at least six months? ____ B. Are at least three of the following present? ____ (either "3" or "4" must be circled) 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1) has stolen without confrontation of a victim on more than one occasion 2) has run away from home overnight at least twice 3) often lies 4) has deliberately engaged in firesetting 5) is often truant from school

PAGE 110

103 1 2 3 4 6) has broken into someone's house, building or car 1 2 3 4 7) has deliberately destroyed other's property 1 2 3 4 8) has been physically cruel to animals 1 2 3 4 9) has used a weapon in more than one fight 1 2 3 4 10) often initiates physical fights 1 2 3 4 11) has stolen with confrontation of a victim 1 2 3 4 12) has been physically cruel to other people Check the appropriate level of severity (make this rating at the conclusion of the intake). Mild: Few, if any, conduct problems in excess of those required to make the diagnosis, and conduct problems cause only minor harm to others. Moderate: Between mild and severe. Severe: Many conduct problems in excess of those required to make the diagnosis, or conduct problems cause considerable harm to others (e.g. serious physical injury to victims, extensive vandalism or theft, prolonged absence from home).

PAGE 111

APPENDIX B EYBERG CHILD BEHAVIOR INVENTORY Rater's Name: Relationship to Child: Date of Rating: Child's Name: Child's Age: Birthdate: Directions: Below are a series of phrases that describe children's behavior. Please (1) circle the number describing how often the behavior currently occurs with your child (1 = Never; 2 to 3 = Seldom; 4 = Sometimes; 5 to 6 = Often; 7 =Always), and (2) circle either "Y" (yes) or "N" (no) to indicate whether the behavior is currently a problem. 1. Dawdles in getting dressed 2. Dawdles or lingers at mealtime 3. Has poor table manners 4. Refuses to eat food presented 5. Refuses tb do chores when asked 104 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 y N y N y N y N y N

PAGE 112

105 6. Slow in getting ready for 1 2 3 4 5 6 7 y N bed 7 Refuses to go to bed on time 1 2 3 4 5 6 7 y N 8 Does not obey house rules on 1 2 3 4 5 6 7 y N his own 9. Refuses to obey until 1 2 3 4 5 6 7 y N threatened with punishment 10. Acts defiant when told to 1 2 3 4 5 6 7 y N do something 11. Argues with parents about 1 2 3 4 5 6 7 y N rules 12. Gets angry when doesn't get 1 2 3 4 5 6 7 y N his own way 13. Has temper tantrums 1 2 3 4 5 6 7 y N 14. Sasses adults 1 2 3 4 5 6 7 y N 15. Whines 1 2 3 4 5 6 7 y N 16. Cries easily 1 2 3 4 5 6 7 y N 17. Yells or screams 1 2 3 4 5 6 7 y N 18. Hits parents 1 2 3 4 5 6 7 y N 19. Destroys toys and other 1 2 3 4 5 6 7 y N objects 20. Is careless with toys or 1 2 3 4 5 6 7 y N other objects 21. Steals 1 2 3 4 5 6 7 y N 22. Lies 1 2 3 4 5 6 7 y N

PAGE 113

106 23. Teases or provokes other 1 2 3 4 5 6 7 y N children 24. Verbally fights with friends 1 2 3 4 5 6 7 y N his own age 25. Verbally fights with sisters 1 2 3 4 5 6 7 y N and brothers 26. Physically fights with 1 2 3 4 5 6 7 y N friends his own age 27. Physically fights with 1 2 3 4 5 6 7 y N sisters and brothers 28. Constantly seeks attention 1 2 3 4 5 6 7 y N 29. Interrupts 1 2 3 4 5 6 7 y N 30. Is easily distracted 1 2 3 4 5 6 7 y N 31. Has short attention span 1 2 3 4 5 6 7 y N 32. Fails to finish tasks or 1 2 3 4 5 6 7 y N projects 33. Has difficulty entertaining 1 2 3 4 5 6 7 y N himself alone 34. Has difficulty concentrating 1 2 3 4 5 6 7 y N on one thing 35. Is overactive or restless 1 2 3 4 5 6 7 y N 36. Wets the bed 1 2 3 4 5 6 7 y N

PAGE 114

APPENDIX C CLASSROOM CODING SYSTEM There are three behavior categories that must be coded for each child: 1) Appropriate Behavior vs. Oppositional Behavior, 2) Comply vs. Noncomply vs. Unsure/No Command Given, and 3) On Task vs. Off Task vs. Not Applicable. This is a "forced choice" system in that the coder must enter only one mark in each of the three categories. Coding is done using an interval sampling procedure. The coder observes the target child for a 10-second time interval then makes one mark in each of the three categories. The coder is given ten seconds to mark the categories before the next 10-second oberservation interval begins. The coder will listen to an audiotape (using an ear jack) which will deliver prompts to "start" and "stop" for each observation interval. Definitions of Behavioral Categories Appropriate Behavior: The absence of all Oppositional Behaviors. Behavior must be appropriate for the entire 10-second interval. If unsure as to whether ~ehavior was appropriate or oppositional, code Appropriate Behavior. 107

PAGE 115

108 Oppositional Behavior: All of the following behaviors are coded as Oppositional Behaviors because they are annoying or disruptive to the target child, the teacher, or other children: a) Whining -Words uttered by the child in a slurring, nasal, high-pitched, falsetto voice. b) Crying -Inarticulate utterances of distress (audible weeping) which may or may not be accompanied by tears. c) Yelling -Loud screeching, screaming, shouting, or crying. The sound must be loud enough so that it is clearly above the intensity of normal indoor conversation. Not coded during outdoor recess observations. d) Tantruming Any combination of whining, yelling, crying, hitting, and/or kicking. e) Destructiveness -Behaviors in which the child damages or destroys an object or attempts or threatens to damage an object or injure a person. Do not code if it is appropriate within the context of the play situation, e.g. ramming cars in a car crash. Examples of aggression toward persons include fighting, kicking, slapping, hitting, or grabbing an object roughly away from another person, or threatening to do any of the preceding. f) Negativism -A verbal or nonverbal negative behavior. May be scored when the child makes a statement in which

PAGE 116

109 the verbal message may be neutral but which is delivered in a tone of voice that conveys an attitude of "don't bug me," or "don't bother me." Negativism may be expressed in a derogatory, uncomplimentary, or angry manner. Also included are defeatist statements such as "I give up," contradictions of what another person says (e.g., teacher says: "Johnny did a nice job," child says: "He did not."), and teasing or mocking behaviors or verbalizations. g) Pathological Self-Stimulation -Repetitive behavior that may be harmful and interfere with a child's ability to attend or complete a task. Examples of pathological self-stimulation include head-banging, thumb-sucking, and masturbation. h) Demanding Attention -Includes repetitive verbal and nonverbal requests for attention from the teacher or other students (e.g., "Call on me! Call on me! Call on me!). Other behaviors that are coded in this category include making faces, making disruptive noises, repetitively tugging on teacher's sleeve, tapping neighbor on the shoulder, waving arms in air, passing notes to another child, and clowning. i) High-Rate Behavior Any very physically active, repetitive behavior that has been carried on sufficiently long that it has become disruptive to

PAGE 117

110 either the target child or others. Examples include kicking a child's chair repeatedly, drumming on the table loudly, and spinning a pencil on the desk. j) Talking Out of Order Any verbalization made in a situation in which the children are clearly expected to be silent unless asked to speak. Talking Out of Order includes whispering to a neighbor, answering a question not directed toward the target child, talking, singing, or humming to oneself, and calling out to another child. k) Being Out of Area -Coded when the target child without permission leaves the area that he is clearly expected to stay in. Examples include standing up when rest of class is seated, leaving desk, approaching the teacher without permission, playing with an attractive toy that is not in the work area the child is supposed to be in. When coding, be certain that the out of area behavior is inappropriate for the context or classroom norms (e.g., in some classrooms the teacher may not be disturbed if the child spontaneously walks to the teacher's desk if he obviously needs help on a math problem). 1) Cheating -Child borrows another child's work when such behavior is clearly not allowed. Examples include looking at another child's paper during a spelling quiz and copying another child's work.

PAGE 118

111 Comply: The target child obeys, begins to obey, or attempts to obey (within 5-seconds) of a direct or indirect teacher command. The command can be one directed toward the target child individually or to a group of children that includes the target child. To be coded, the command must be given during the 10-second observation interval. If the command is given near the end of the 10-second observation interval, continue to watch for 5-seconds to determine whether the child complies. Noncomply: Target child makes no movement toward obeying a direct or indirect teacher command during a 5-second period following the command. Unsure/No Command: This category is coded when the observer is unsure about whether the child obeyed the command, or no command was issued during the 10-second observation interval. Examples of commands that the coder is likely to be unsure about whether the child obeyed include: "All children who got stars last week come to the front of the room," "Say the word h-a-p-p-y" (target child's back is turned to observer), "If you haven't fininshed your worksheet, do it now" (observer is sitting at a distance and can't tell whether target child's paper is finished).

PAGE 119

112 On Task: The child is considered to be On Task if he is (a) attending to the material and the task, (b) making appropriate motor responses (e.g., writing, computing, pasting), and (c) asking for assistance (where appropriate) in an acceptable manner. Interacting with the teacher or classmates about academic matters or listening to teacher instructions and directions are considered to be On Task behaviors. To be coded as On Task, the child must remain on task for the full 10-second observation interval. Off Task: Coded if at any point during the 10-second interval the child is engaging in behavior that does not meet the definition for On Task behavior. Examples of Off Task include failure to attend to or work on the assigned task, breaking classroom rules (out of seat, talking out, disturbing others, etc.), laying head on desk passively when there is a task to complete, and daydreaming. If the child is in time out during the observation interval, he or she is automatically coded as Off Task. Not Applicable: Coded when there is no readily identifiable task that the child is expected to perform. Examples of Not Applicable activities include free play and unstructured recess time.

PAGE 120

APPENDIX D REVISED CONNERS TEACHER RATING SCALE Nam e of Child: Date of Evaluation: Grade: Please answe r all questions. Beside each item, indicate the degree of the problem by a check mark. 1 = Not at all 2 =Justa little 3 = Pretty much 4 = Very much 1. Restless in the "squirmy" sense. 2. Makes inappropriate noises when he shouldn't. 3. Demands must be met immediately. 4. Acts "smart" ( impudent or sassy) 5. Temper outbursts and unpredictable behavior. 6. Overly sensitive to criticism. 7. Distractibility or attention span a problem. 8. Disturbs other children. 9. Daydreams. 113 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4

PAGE 121

114 10. Pouts and sulks. 11. Mood changes quickly and drastically. 12. Quarrelsome. 13. Submissive attitude toward authority. 14. Restless, always "up and on the go." 15. Excitable, impulsive. 16. Excessive demands for teacher' s attention. 17. Appears to be unaccepted by group. 18. Appears to be easily led by other children. 19. No sense of fair play. 20. Appears to lack leadership. 21. Fails to finish things that he started. 22. Childish and immature. 23. Denies mistakes or blames others. 24. Does not get along well with other children. 25. Uncooperative with classmates. 26. Easily frustrated in efforts. 27. Uncooperative with teacher. 28. Difficulty in learning. 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 1 2 1 2 1 2 3 4 3 4 3 4 3 4

PAGE 122

APPENDIX E SUTTER-EYBERG STUDENT BEHAVIOR INVENTORY Rater's Na me: Relationship to Child: Date of Rating: Child's Name: Child's Age: Birthdate: Directions: Below are a series of phrases that describe children's behavior. Please (1) circle the number describing how often the behavior currently occurs with this student (1 = Never; 2 to 3 = Seldom ; 4 = Sometimes; 5 to 6 = Often; 7 =Always), and (2) circle either "Y" (yes) or "N" (no) to indicate whether the behavior is currently a problem. 1 Dawdles in obeying rules or instructions 2. Argues with teachers about rules or instructions 3. Has difficulty accepting criticism or correction 4. Does not obey schooi rules on his/her own 115 1 2 3 4 5 6 7 y N 1 2 3 4 5 6 7 y N 1 2 3 4 5 6 7 y N 1 2 3 4 5 6 7 y N

PAGE 123

116 5. Refuses to obey until threatened with punishment 6 Gets angry when doesn't get his/her own way 7. Acts defiant when told to do something 8 Has temper tantrums 9. Sasses teacher(s) 10. Whines 11. Cries 12. Pouts 13. Yells or screams 14. Hits teacher(s) 15. Is careless with books and other objects 16. Destroys books and other objects 17. Steals 18. Lies 19. Makes noises in class 20. Teases or provokes other students 21. Acts bossy with other students 1 2 3 4 5 6 7 y N 1 2 3 4 5 6 7 y N 1 2 3 4 5 6 7 y N 1 2 3 4 5 6 7 y N 1 2 3 4 5 6 7 y N 1 2 3 4 5 6 7 y N 1 2 3 4 5 6 7 y N 1 2 3 4 5 6 7 y N 1 2 3 4 5 6 7 y N 1 2 3 4 5 6 7 y N 1 2 3 4 5 6 7 y N 1 2 3 4 5 6 7 y N 1 2 3 4 5 6 7 y N 1 2 3 4 5 6 7 y N 1 2 3 4 5 6 7 y N 1 2 3 4 5 6 7 y N 1 2 3 4 5 6 7 y N

PAGE 124

117 22. Verbally fights with other 1 2 3 4 5 6 7 y N students 23. Physically fights with 1 2 3 4 5 6 7 y N other students 24. Demands teacher attention 1 2 3 4 5 6 7 y N 25. Interrupts teachers 1 2 3 4 5 6 7 y N 26. Interrupts other students 1 2 3 4 5 6 7 y N 27. Has difficulty entering 1 2 3 4 5 6 7 y N groups 28. Has difficulty sharing 1 2 3 4 5 6 7 y N materials 29. Is uncooperative in group 1 2 3 4 5 6 7 y N activities 30. Blames others for problem 1 2 3 4 5 6 7 y N behaviors 31. Is easily distracted 1 2 3 4 5 6 7 y N 32. Has difficulty staying on 1 2 3 4 5 6 7 y N task 33. Acts frustrated with 1 2 3 4 5 6 7 y N difficult tasks 34. Fails to finish tasks or 1 2 3 4 5 6 7 y N projects 35. Impulsive, acts before 1 2 3 4 5 6 7 y N thinking 36. Is overactive or restless 1 2 3 4 5 6 7 y N

PAGE 125

APPENDIX F THE WALKER-McCONNELL SCALE OF SOCIAL COMPETENCE AND SCHOOL ADJUSTMENT: A SOCIAL SKILLS RATING SCALE FOR TEACHERS I. Student Demographic Information Date Administered: Classroom Type: Regular Student Name: Teacher: Sex: M F Age: Years II. Rating Instructions Resource Self-contained Months Please read each item below carefully and rate the child's behavioral status in relation to it. If you have not observed the child displaying a particular skill or behavioral competency defined by an item, check 1, indicating Never. If the child exhibits the skill at a high rate of occurrence, check 5, for Frequently. If the child's frequency is in between these two extremes, please check 2, 3, or 4, indicating your best estimate of its rate of occurrence. Please answer each item. DO NOT MARK BETWEEN THE NUMBERS ON THE RATING SCALE. Check one of the 118

PAGE 126

119 numbers from 1-5 to indicate your frequency estimate. Numbers in parentheses to the left of the item number represent the subscale for that particular item. III. Items and Rating Formats ( 2) 1. Other children seek child out to involve her/him in activities. (2) 2. Changes activities with peers to permit continued interaction. (3) 3. Uses free time appropriately. (2) 4. Shares laughter with peers. (1) 5. Shows sympathy for others. (2) 6. Makes friends easily with other children. (3) 7. Has good work habits, e.g., is organized, makes efficient use of class time, etc. ( 2) ( 1) 8. Asks questions that request information about someone or something. 9. Compromises with peers when situation calls for it. 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 1. .2 .. 3 .. 4 .. 5 1. .2 .. 3 .. 4 5 1. .2 .. 3 .. 4 .. 5 1. 2 .. 3 .. 4 .. 5

PAGE 127

120 (1) 10. Responds to teasing or name calling by ignoring, changing the subject, or some other constructive means. (2) 11. Spends recess and free time interacting with peers. ( 1} ( 2} 12. Accepts constructive criticism from peers without becoming angry. 13. Plays or talks with peers for extended periods of time. (2) 14. Voluntarily provides assistance ( 2) ( 1) to peers who require it. 15. Assumes leadership role in peer activities. 16. Is sensitive to the needs of others. (2) 17. Initiates conversation(s} with ( 1} peers in informal situations. 18. Expresses anger appropriately, e.g., reacts to situation without becoming violent or destructive. (3) 19. Listens carefully to teacher instructions and directions for assignments. 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 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

PAGE 128

121 (3) 20. Answers or attempts to answer a question when called on by the teacher. ( 3) ( 1) ( 3) ( 1) 21. Displays independent study skills, e.g., can work adequately with minimum teacher support. 22. Appropriately copes with aggression from others, e.g., tries to avoid a fight, walks away, seeks assistance, defends self. 23. Responds to conventional behavior management techniques, e.g., praise, reprimands, time-out. 24. Cooperates with peers in group activities or situations. (2) 25. Interacts with a number of different peers. (1) 26. Uses physical contact with peers appropriately. (3) 27. Responds to requests promptly. (1) 28. Listens while others are speaking, e.g., as in circle or sharing time. 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 1 .. 2 .. 3 .. 4 .. 5 1. .2 .. 3 .. 4 .. 5 1. .2 3 .. 4 .. 5 1. .2 3 .. 4 .. 5

PAGE 129

122 (1) 29. Controls temper. (2) 30. Compliments others regarding personal attributes, e.g., appearance, special skills, etc 1. .2 .. 3 .. 4 .. 5 1. .2 .. 3 .. 4 .. 5 (1) 31. Can accept not getting her/his 1 .. 2 .. 3 .. 4 .. 5 own way. ( 2) 32. Is socially perceptive, e.g. reads social situations accurately. (3) 33. Attends to assigned tasks. (2) 34. Plays games and activities at recess skillfully. (2) 35. Keeps conversation with peers going. (1) 36. Finds another way to play when requests to join others are refused. (1) 37. Is considerate of the feelings of others. (2) 38. Maintains eye contact when speaking or being spoken to. (1) 39. Gains peers' attention in an appropriate manner. (1) 40. Accepts suggestions and assistance from peers. 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 1. .2 .. 3 .. 4 .. 5 1. .2 .. 3 4 5 1. .2 .. 3 4 .. 5 1. .2 .. 3 .. 4 .. 5

PAGE 130

1 2 3 ( 2) 41. Invites peers to play or share 1. .2 .. 3 .. 4 .. 5 activ i t i es. ( 3) 42. Does seatwork assignments as 1. .2 3 .. 4 .. 5 directed. ( 3 ) 43. Produces work of acceptable 1. 2 .. 3 .. 4 5 quality given her/his skills level

PAGE 131

APPENDIX G SAMPLE TREATMENT SESSION OUTLINE: SECOND PDI COACHING SESSION 1) The first 15 minutes or so will be spent in interview and review of PDI homework from last week. 2) The next 35 minutes will be used for coaching PDI skills. Record the exact amount of coaching time 3) In addition to coaching the use of play commands (e.g., "Put the red block on top of the blue block"), parents will be coached to use an many "real life" commands as possible (e.g., "Sit in this chair," "Put that toy back in its box"), including clean-up commands at the end of the coaching period. 4) If PDI is being taught as phase #1, no CDI skills are coached during this session. In other words, labeled praise is coached only following compliance by the child (not for other things he does during the playtime), and imitating, describing, reflecting, and eliminating questions are not coached. 5) No formal coding will be done during this session. The co-therapist should informally record throughout the 124

PAGE 132

125 session: command given, obey or disobey, warnings, time on chair, refusals, and spanks. 6) The remaining 10 minutes will be spent providing feedback and explaining the new homework assignment. Parents will be asked to practice PD! at home in a 10-minute daily clean-up session and to use their PD! skills for 3-5 carefully selected direct commands each day. 7) Parents will be given a homework sheet on which to record their daily practice of PD! skills.

PAGE 133

REFERENCES Achenbach, T. M., Mcconaughy, S. H., & Howell, C. T. (1987). Child/adolescent behavioral and emotional problems: Implications of cross-informant correlations for situational specificity. Psychological Bulletin, 101,. 213-232. American Psychiatric Association (1987). Diagnostic and statistical manual of mental disorders (Third Edition Revised). Washington, DC: Author. Baekeland, R., & Laundwall, L. (1975). Dropping out of treatment: A critical review. Psychological Bulletin, 82, 738-783. Barkley, R. A. (1977). The effects of methylphenidate on various measures of activity level and attention in hyperkinetic children. Journal of Abnormal Child Psychology,~, 351-369. Barkley, R. A. (1987). Defiant children: A clinician's manual for parent training. New York: Guilford. Barkley, R. A., Fischer, M., Newby, R. F., & Breen, M. J. (1988). Development of a multimethod clinical protocol for assessing stimulant drug response in children with attention deficit disorder. Journal of Clinical Child Psychology, 17, 14-24. Baum, C., & Forehand, R. (1981). Long term follow-up assessment of parent training by use of multiple outcome measures. Behavior Therapy, 12, 643-652. Bean, A. W., & Roberts, M. W. (1981). The effect of time-out release contingencies on changes in child noncompliance. Journal of Abnormal Child Psychology, i, 95-105. Behar, L. & Stringfield, S. ( 197 4) A behavior rating scale for the preschool child. Developmental Psychology, 10, 601-610. Bernhardt, A. J., & Forehand, R. (1975). The effects of labeled and unlabeled praise upon lower and middle class children. Journal of Experimental Child Psychology, 19, 536-543. 126

PAGE 134

127 Breiner, J., & Forehand, R. (1981). An assessment of the effects of parent training on clinic-referred children's school behavior. Behavioral Assessment, J, 31-42. Calvert, S. C., & McMahon, R. J. (1987). The treatment acceptability of a behavioral parent training program and its components. Behavior Therapy,~. 165-179. Campbell, S. B., Ewing, L. J., Breaux, A. M., & Szumowski, E. K. (1986). Parent-referred problem three-year olds: Follow-up at school entry. Journal of Child Psychology and Psychiatry, 27, 473-488. Conners, C. K. (1969). A teacher rating scale for use in drug studies with children. American Journal of Psychiatry, 126, 884-888. Conners, C. K. (1970). Symptom patterns and hyperkinetic, neurotic, and normal children. Child Development, 41, 667-682. Edelbrock, C., Greenbaum, R., & Conover, N. C. (1985). Reliability and concurrent relations between the teacher version of the Child Behavior Profile and the Conners Revised Teacher Rating Scale. Journal of Abnormal Child Psychology, 13, 295-303. Edelbrock, C., & Reed, M. L. (1984). Reliability and concurrent validity of the Teacher Version of the Child Behavior Profile. Unpublished manuscript, University of Pittsburgh. Eyberg, S. M (1974). Therapy Attitude Inventory. (Available from S. M. Eyberg, Dept. of Clinical & Health Psychology, University of Florida.) Eyberg, S. M. (1979, April). A parent-child interaction model for the treatment of psychological disorders in young children. Paper presented at the meeting of the Western Psychological Association, San Diego, CA. Eyberg, S. M. (1988). Parent-Child Interaction Therapy: Integration of traditional and behavioral concerns. Child and Family Behavior Therapy, 10, 33-46.

PAGE 135

128 Eyberg, S. M., & Boggs, S. R. (In press). Parent trairiing for oppositional preschoolers. In C. E. Schaefer & J.M. Briesmeister (Eds.), Handbook of parent training: Parents as cotherapists for children's behavior problems. New York: John Wiley & Sons. Eyberg, S. M., & Johnson, S. M. (1974). Multiple assessment of behavior modification with families: Effects of contingency contracting and order of treated problems. Journal of Consulting and Clinical Psychology, !l, 594-606. Eyberg, S. M., & Matarazzo, R. G. (1980). Training parents as therapists: A comparison between individual parent-child interaction training and parent group didactic training. Journal of Clinical Psychology, 36, 492-499. Eyberg, S. M., & Robinson, E. A. (1982). Parent-child interaction training: Effects on family functioning. Journal of Clinical Child Psychology, 11, 130-137. Eyberg, S. M., & Ross, A. W. (1978). Assessment of child behavior problems: The validation of a new inventory. Journal of Clinical Child Psychology, I, 113-116. Flanagan, S., Adams, H. E., & Forehand, R. (1979). A comparison of four instructional techniques for teaching parents the use of time-out. Behavior Therapy, 10, 94-102. Forehand, R., & Atkeson, B. M. (1977). Generality of treatment effects with parents as therapists: A review of assessment and implementation procedures. Behavior Therapy, ~' 575-593. Forehand, R., Cheney, T., & Yoder, P. (1974). Parent behavior training: Effects on the non-compliance of a deaf child. Journal of Behavior Therapy and Experimental Psychiatry,~' 281-283. Forehand, R., & King, H. E. (1974). Pre-school children's non-compliance: Effects of short-term therapy. Journal of Community Psychology, ~' 42-44. Forehand, R., & Long, N. (1988). Outpatient treatment of the acting out child: Procedures, long term follow-up data, and clinical problems. Advances in Behaviour Research and Therapy, 10, 129-177.

PAGE 136

129 Forehand, R., & McMahon, R. J. (1981). Helping the noncompliant child. New York: Guilford. Forehand, R., Middlebrook, J., Rogers, T., & Steffe, M. (1983). Dropping out of parent training. Behaviour Research and Therapy, 21, 663-668. Forehand, R., & Scarboro, M. E. (1975). An analysis of children's oppositional behavior. Journal of Abnormal Child Psychology, l, 27-31. Forehand, R., Sturgis, E., McMahon, R., Aguar, D., Green, K., Wells, K., & Breiner, J. (1979). Parent behavioral training to modify child noncompliance: Treatment generalization across time and from home to school. Behavior Modification, l, 3-25. Forehand, R., Wells, K. C., & Griest, D. L. (1980). An examination of the social validity of a parent training program. Behavior Therapy, 1980, 11, 488-502. Funderburk, B., & Eyberg, S. (In press). Psychometric characteristics of the Sutter-Eyberg Student Behavior Inventory: A school behavior rating scale for use with preschool children. Behavioral Assessment. Goyette, C. H., Conners, C. K., & Ulrich, R. F. (1978). Normative data on Revised Conners Parent and Teacher Rating Scales. Journal of Abnormal Child Psychology, ~, 221-236. Hanf, C. A. (1969). A two-stage program for modifying maternal controlling during mother-child (M-C) interaction. Paper presented at the meeting of the Western Psychological Association, Vancouver, B.C., 1969. Hanf, C., & Kling, J. (1974}. Facilitating parent-child interaction: A two-stage training model. Unpublished manuscript, University of Oregon Medical School. Hobbs, S. A., Forehand, R., & Murray, R. G. (1978). Effects of various durations of time-out on the non-compliant behavior of children. Behavior Therapy, 1, 652-656. Horn, W. F., Ialongo, N., Popovich, S., & Peradotto, D. (1987}. Behavioral parent training and cognitive-behavioral self-control therapy with ADD-H children: Comparative and combined effects. Journal of Clinical Child Psychology, 16, 57-68.

PAGE 137

130 Humphreys, L., Forehand, R., McMahon, R., & Roberts, M. (1978). Parent behavioral training to modify child noncompliance: Effects on untreated siblings. Journal of Behavior Therapy and Experimental Psychiatry,~. 235-238. Johnson, S. M., Bolstad, 0. D., & lobitz, G. K. (1976). Generalization and contrast phenomena in behavior modification with children. In E. J. Mash, L.A. Hamerlynck, & L. C. Handy (Eds.), Behavior modification and families (pp. 160-188). New York: Brunner/Mazel. Kazdin, A. E (1980). Acceptability of timeout and reinforcement procedures for disruptive child behavior. Behavior Therapy, 11, 329-344. Kazdin, A. E. (1981). Acceptability of child treatment techniques: The influence of treatment efficacy and adverse side-effects. Behavior Therapy, 12, 493-506. Laviqueur, H., Peterson, R. F., Sheese, J. G., & Peterson, L. W. (1973). Behavioral treatment in the home: Effects on an untreated sibling and long-term followup. Behavior Therapy, !, 431-441. McMahon R J., Forehand, R., & Griest, D. L. (1981). Effects of knowledge of social learning principles on enhancing treatment outcome and generalization in a parent training program. Journal of Consulting and Clinical Psychology, 49, 526-532. McMahon, R. J., Forehand, R., Griest, D. L., & Wells, K. C. (1981). Who drops out of treatment during parent behavioral training? Behavioral Counseling Quarterly, 1, 79-85. Patterson, G R., Cobb, J. A., & Ray, R. S. (1973). A social engineering technology for retraining the families of aggressive boys. In H. E. Adams & I. P. Unikel (Eds.), Issues and trends in behavior therapy (pp. 139-210). Springfiled IL: Charles C. Thomas Peed, S., Roberts, M., & Forehand, R. (1977). Evaluation of the effectiveness of a standardized parent training program in altering the interaction of mothers and their noncompliant children. Behavior Modification, 1, 323-351.

PAGE 138

131 Prinz, R. J., O'Connor, P. A., & Wilson, C. C. (1981). Hyperactive and aggressive behaviors in childhood: Intertwined dimensions. Journal of Abnormal Child Psychology,~, 287-295. Resnick, P.A., Forehand, R., & Mcwhorter, A. Q. (1976). The effect of parental treatment with one child and an untreated sibling. Behavior Therapy, I, 544-548. Roberts, M. W., McMahon, R. J., Forehand, R., & Humphreys, L. (1978). The effect of parental instruction-giving on child compliance. Behavior Therapy, ~. 793-798. Robinson, E. A., & Eyberg, S. M. (1978). Systematic behavioral observation of normal and behavior problem children. Paper presented at the annual meeting of the American Psychological Association, Toronto, Ontario. Robinson, E. A., & Eyberg, S. M. (1981). The Dyadic Parent-Child Interaction Coding System: Standardization and validation. Journal of Consulting and Clinical Psychology, 49, 245-250. Robinson, E. A., Eyberg, S. M., & Ross, A. W. (1980). The standardization of an inventory of child conduct problem behaviors. Journal of Clinical Child Psychology, ~, 22-28. Sayger, T. V., & Horne, A. M. (1987). The maintenance of treatment effects for families with aggressive boys participating in social learning family therapy. Paper presented at the annual meeting of the American Psychological Association, New York. Scarboro, M. E., & Forehand, R. (1975). Effects of response-contingent isolation and ignoring on compliance and oppositional behavior of children. Journal of Experimental Child Psychology, 19, 252-264. Sosna, T. D., Ladish, C., Warner, D., & Burns, G. L. (1989, August). Psychometric properties of the SESBI in a -preschool sample. Paper to be presented at the meeting of the American Psychological Association, New Orleans, LA.

PAGE 139

132 Sutter, J., & Eyberg, S. (1984). Sutter-Eyberg Student Behavior Inventory. (Available from Sheila Eyberg, Dept. of Clinical and Health Psychology, Box J-165, JHMHC, Unive~sity of Florida.) Wahler, R. G. (1969). Setting generality: Some specific and general effects of child behavior therapy. Journal of Applied Behavior Analysis, ~, 239-246. Wahler, R. G. (1975). Some structural aspects of deviant child behavior. Journal of Applied Behavior Analysis, ~. 27-42. Walker, H. M., & McConnell, S. (1987). The Walker-McConnell Scale of Social Competence and School Adjustment. Austin, TX: Pro-Ed. Walker, H. M., Shinn, M. R., O'Neill, R. E., & Ramsey, E. (1987). A longitudinal assessment of the development of antisocial behavior in boys: Rationale, methodology, and first year results. Remedial and Special Education,~' 7-16. Walle, D., Hobbs, S., & Caldwell, H. S. (1984). Sequencing of parent training procedures: Effects on child noncompliance and treatment acceptability. Behavior Modification, ~, 540-552. Webster-Stratton, C. {1984). Randomized trial of two parent-training programs for families with conduct-disordered children. Journal of Consulting and Clinical Psychology, 52, 666-678. Webster-Stratton, C., Hollinsworth, T., & Kolpacoff, M. (In press). The long-term effectiveness and clinical significance of three cost-effective training programs for families with conduct problem children. Journal of Consulting and Clinical Psychology. Wells, K. C., Forehand, R., & Griest, D. L. (1980). Generality of treatment effects from treated to untreated behaviors resulting from a parent training program. Journal of Clinical Child Psychology, ~, 217-219.

PAGE 140

BIOGRAPHICAL SKETCH Cheryl Ann Bodiford was born on June 17, 1961, in Miami, Florida, to Otis and Donna Bodiford. She has two older siblings, a sister, Becky, and a brother, Jack, as well as two living grandparents, Minnie Curry and Eldonese Lingerfelt. Cheryl was graduated as valedictorian from Louisiana State University in May 1983, earning a B.S in psychology. As an undergraduate, Cheryl was initiated into Phi Beta Kappa. Her graduate studies in clinical psychology began at the University of Florida in August 1983. Cheryl was granted a M.S. in clinical psychology in May 1986 after completing a master's thesis in the area of childhood depression and learned helplessness. In 1986, the University of Florida's Department of Clinical and Health Psychology awarded Cheryl the Molly Harrower award for excellence in psychodiagnostic assessment. Cheryl completed a clinical internship in 1988 and 1989 at the University of Oklahoma Health Sciences Center, with a specialization in clinical child Psychology. Cheryl's major professional interest is child psychotherapy, particularly Parent-Child Interaction Therapy. Personal interests include swimming and swim instruction, playing tennis, fishing, boating, travel, poker, and film. 133

PAGE 141

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholariy presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Sheila M. Ey~~. C ir Professor of Clinical and Health 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. Associate Professor of Statistics 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. Eileen B. Fennell Professor of Clinical and Health 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. J~Goldman Professor of Clinical and Health Psychology

PAGE 142

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. fessor of 'nical and alth Psychology This dissertation was submitted to the Graduate Faculty of the College of Health Related Professions and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. May 1989 Dean, College of Health Related Professions Dean, Graduate School