THESI‘S Date 0-7639 \l\\ \\\\\\l\\\\\\ :\3\\\\\1\\293 10064 24 0 j' LIBRARY \\\\\\\W\\\\\\\\\\\\\l\\\\W UM _ This is to certify that the thesis entitled AN EVALUATION OF TWO TREATMENT PROGRAMS FOR FAMILIES WITH ACTING-OUT ADOLESCENTS presented by Linda Maria Garcia-Shelton has been accepted towards fulfillment of the requirements for Ph.D. degreein Counseling Psychology /// (xx/V f r / " Major£ rofessor/y’M October 22, 1979 OVERDUE FINES ARE 25; PER DA: ‘ 9 HR IT EM Return to book drop to Ic¢0¥t this checkout from your record. JAN 13320051; AN EVALUATION OF TWO TREATMENT PROGRAMS FOR FAMILIES WITH ACTING-OUT ADOLESCENTS By Linda Maria Garcia-Shelton A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Counseling, Personnel Services, and Educational Psychology 1979 ABSTRACT AN EVALUATION OF TWO TREATMENT PROGRAMS FOR FAMILIES WITH ACTING-OUT ADOLESCENTS By Linda Maria Garcia-Shelton Effects of a short-term family treatment program for families with an acting-out adolescent were compared to the effects of an adolescent group treatment program and a no-treatment control group. Thirty families, randomly assigned to three experimental groups using referral source as a blocking variable, were pretested on four measures, treated, and then posttested on the same measures. The measures used to evaluate change as a result of treatment were: (l) the Wynne and Singer Family Interaction Task, (2) the Family Behavior Change Scale, (3) the Tennessee Self-Concept Scale, and (4) the Drug and Delinquent Behavior Questionnaire. Posttest data were analyzed using a Multivariate Analysis of Covariance applied to two orthogonal contrasts, using pretest data as covariates. All analyses failed to demonstrate any difference between the control group score and the average score of the two treatment groups. They also failed to demonstrate any difference between the two treat- ment groups. Furthermore, the nonsignificant differences found between groups were uniformly small, indicating that the problem was not lack 0f power. Linda Maria Garcia-Shelton In conclusion, both treatment programs evaluated in this study failed to change family interaction patterns, family attitudes, or adolescent acting-out behavior and self-concept. TABLE OF CONTENTS LIST OF TABLES ......................... LIST OF FIGURES ......................... Chapter l. INTRODUCTION ...................... Specific Questions .................. Definitions of Terms ................. Delimitations ..................... Basic Hypotheses ................... Need for Adolescent Counseling ............ Difficulties in Adolescent Families ......... Common Treatment Approaches .............. Limitations of Traditional Social Institutions New Developments in Alternative Community Agencies . Theory ........................ Family Tasks .................... Family Strengths .................. Family Change .................... Differences Between Normal and Dysfunctional Adolescent Families ................ Summary ........................ Theory Applied to Treatment .............. Overview of Remaining Chapters ............ 2. REVIEW OF THE TREATMENT LITERATURE ........... Change as a Result of Family Treatment ........ Parsons and Alexander Study ............. Katz, Krasinski, Philip, and Nieser Study ...... Beal and Duckro Study ................ Stanley Study .................... Parent Effectiveness Training ............ Discussion of Family Treatment Research ....... Change as a Result of Adolescent Group Treatment . . . Gilliland Study ................... Warner and Hansen Study ............... Sarason and Ganzer Study .............. ii —I SDNO‘U‘IU'l-bWNN Chapter Page Block Study .................... 54 Discussion of Adolescent Group Treatment Research . 56 Treatment by Paraprofessionals in a Community- Sponsored Crisis Center .............. 58 Berthea Study ................... 58 Implications of Previous Research .......... 61 Suggested Treatment Program ............ 61 Research Design .................. 63 3. REVIEW OF MEASUREMENT METHODS Family Interaction Measures ............. 66 Family Tasks ................... 66 Unit of Measurement ................ 68 Family Interaction Scoring Codes ......... 70 Self-Concept Measures ................ 85 Tennessee Self-Concept Scale ........... 86 California Test of Personality .......... 89 Security-Insecurity Inventory ........... 91 Test of Personality Adjustment .......... 92 Self-Perception Inventory ............. 92 Teen Drug and Delinquent Behavior Questionnaires . . 94 4. STUDY DESIGN ..................... 99 Population Description and Recruitment Procedures . . 99 Sample Description and Group Assignment Procedure . . 100 Measures ...................... 104 Family Interaction Task .............. 104 Problem Behavior Scale .............. 109 Tennessee Self-Concept Scale ........... 112 Drug and Delinquent Behavior Questionnaire . . . . 112 Description of the Agency Offering Treatment . . . . 114 Description of Treatment Programs .......... 114 Treatment I: The Family Program .......... 114 Treatment 11: The Adolescent Program ....... 119 Treatment 111: The Control Group ......... 119 Completion Requirements for Treatment Programs . . 120 Intake Procedures .................. 120 Posttest ...................... 122 Research Design ................... 122 Hypotheses ..................... 122 Null Hypothesis l ................. 122 Null Hypothesis 2 ................. 124 Data Analysis .................... 125 Summary ....................... 126 iii Chapter Page 5. RESULTS ......................... 128 Analysis of the Experimental Hypotheses ........ 129 Hypothesis 1 .................... 129 Hypothesis 2 .................... 139 Hypothesis 3 .................... 142 Hypothesis 4 .................... 145 Summary of Research Results .............. 145 6. SUMMARY AND CONCLUSIONS ................. 148 Brief Literature Summary ............... 148 Adolescent Family Problems ............. 148 Family Interaction Studies ............. 149 Family Treatment Studies .............. 150 Current Research Study and Results ......... 152 Discussion ...................... 154 Selection as an Independent Variable ........ 155 Consistency of Analysis and Design ......... 157 Cluster Patterns of Dependent Variables ....... 158 No Difference--Why? ................. 161 Implications for Future Research ........... 163 Observations on Clinical Effectiveness Research . . . . 165 APPENDICES ........................... 166 A. PROBLEM BEHAVIOR RATING FORM .............. 167 B. DRUG AND DELINQUENT BEHAVIOR QUESTIONNAIRE ....... 169 BIBLIOGRAPHY .......................... 180 iv Table 2.1 4.1 4.2 5.1 5.2 5.3 5.4 5.5 5.6 5.7 LIST OF TABLES Group Posttest Means and §_Ratios for Interaction and Content Measures .................. Comparison of Families Offered Each Treatment Program, Beginning Each Treatment Program, and DrOpping Out of Each Treatment Program ............... Internal Consistency Reliability Coefficients for the Problem Behavior Rating ................ Posttest Mean and Standard Deviation of Each Group on Each Variable .................... Statistics for Regression Analysis Using Eight Covariates Simultaneously ............... Pretest Mean and Standard Deviation of Each Group on Each Variable ..................... Correlations Between A11 Dependent Variables ....... Summary of Results of the Clustered Dependent Variables, Hypothesis 1 ...................... Summary of Results of the Clustered Dependent Variables, Hypothesis 2 ...................... Mean and Standard Deviation of Each Group on the Scales of the Drug and Delinquent Behavior Questionnaire Page 28 103 111 130 133 135 136 140 143 144 LIST OF FIGURES Figure Page 4.1 Example of Transcription Sheet for Family Interaction Task ................... 106 4.2 Summary of Data-Collection Procedure .......... 123 4.3 Data-Analysis Design .................. 127 5.1 Summary Outline of the Multivariate Analysis of Covariance ...................... 131 5.2 Dependent Variable Clusters and Their Correlations . . . 137 vi CHAPTER 1 INTRODUCTION The purpose of this study is to determine the relative effec- tiveness of two counseling programs. Each program was offered by a community service agency. The agency is staffed by paraprofessionals and the programs serve families with acting-out adolescents. Both counseling programs are intended (1) to reduce strain in the partici- pating families by improving communication and problem-solving skills, and (2) to improve the self-concept of the acting-out adolescent. One program offers services to the whole family; the other serves only the acting-out adolescent. Chapter 1 includes background material related to both the traditional treatment of adolescent families and also the aims of this study. This background material includes: 1. Questions to be applied to both treatments. Definition of terms. Limitations of the study. Basic hypothesis. Problems of adolescent families. Traditional treatment approaches. New treatment approaches. CDNO‘U'I-DMN Theory: clinical perspective. 1 9. Normal versus dysfunctional families. 10. Theory applied to treatment. Chapter I ends by outlining the remaining chapters of the paper. Specific Questions This study hopes to answer several questions about the merits of each program. First, does inclusion of additional family members in a counseling program improve outcome? Second, will parents and ado- lescents consent to family counseling as readily as they consent to counseling for the adolescent alone? Third, will families remain in counseling (not drop out) at the same rate at which individual adoles- cents remain in counseling? These questions are critical to the program's effectiveness because a program which no one accepts or completes is useless. Definitions of Terms The following words are defined as used in this study: family_is used to mean those people, residing together, who have a long-term commitment to each other. Generally, the term family refers to parents and their children; however, the population studied in this research project evidenced many variations of the traditional nuclear family. These variations included the single-parent family, the multigenerational family, and the foster family. An adolescent is a child between the ages of 11 and 19. An adolescent family is a family including at least one adolescent. An acting-out adolescent is a child whose behavior is in conflict with rules set by either the family or society (through the institutions of schools or courts). Acting-out behaviors fall into three cate- gories: intrafamily transgressions, status offenses, and nonstatus offenses. Intrafamily transgressions are violations of parental rules such as those relating to curfew, nonpermissible activities, and non- permissible locations. Status offenses are those behaviors that are legally acceptable for adults but not for minors. Examples of status offenses are running away from home, school truancy, and sexual promis- cuity. Nonstatus offenses are those behaviors that are always illegal, such as theft, breaking and entering, or assault and battery. Family interaction refers to the many ways family members behave toward one another. However, when assessing family interaction in this study, only verbal interaction will be examined. A triadic family consists of two parents and a child. This triad may be the entire family or may be a subset of the entire family. An extended family is a multiple family unit consisting of a number of nuclear families related through the sibship of the adult members and often including a grandparental generation. The extended family therefore includes the roles of grandparent, aunt, uncle, cousin, niece and nephew, in addition to the roles of parent, spouse and child which are present in the nuclear family. Delimitations The treatment programs evaluated in this study were both carried out by a paid staff of paraprofessional counselors supported by funds from four sources: (1) a $52,000 grant from the Attorney General of "Michigan Trust Fund, (2) the Ingham County Probate Court, (3) the City of East Lansing, and (4) the Ingham County Board of Commissioners. Counselor training, education and supervision data are discussed in Chapter 4. Due to this very adequate funding situation, results from this study cannot be necessarily applied to low-budget free clinics staffed by volunteers. A paid staff has a degree of commitment to the job that a volunteer staff may not possess, and an agency with stable financial support has a measure of community trust and recognition that is perhaps not present in a volunteer agency. In Michigan, many medium-to-large communities have well-funded crisis centers. Results from this study may be applied to similar populations treated at those centers. The programs evaluated in this study were highly thought of by local professional community organizations and enjoyed their active support. Results from this study cannot be generalized to programs which are merely tolerated or marginally accepted by the larger pro- fessional community. The subjects of this study were all volunteers from one city in central Michigan. Hence results may not be generalizable to the population as a whole, but only to those families who identify them- selves as needing and wanting help with their problems. Basic Hypotheses It is hypothesized that families receiving counseling as a family unit will increase the quality of their communication, improve the self-concept of the acting-out adolescent and decrease the amount of adolescent acting-out more than families in which only the adolescent receives counseling. It is further expected that families in both family and adolescent-only counseling programs will improve more on these variables than families receiving no counseling at all. Need for Adolescent Counseling The need for the programs evaluated in this study is based pri- marily on (1) the difficulties facing adolescent families today and on (2) the failure of other existing programs to deal with those problems. Difficulties in Adolescent Families There are three problem areas to which many families are forced to respond: (1) increasing adolescent drug use (Light, 1975), (2) delinquency (both amount and gravity of offense) (National Commis- sion on Marijuana and Drug Abuse, 1972), and (3) a general increase in the number of families broken apart through divorce (Epstein, 1974). The social thinkers whose ideas are discussed on the following pages have stated that these problems are responses to stress, both in the family and in the larger society. Constant change in our society has resulted in many families being caught between conflicting demands, unstable supports and ever-increasing expectations. Winter (1961) agrees that modern life has created a great strain in intimate family relationships. As the nuclear family moves away from the extended family, the number of possible intimate relation- ships within the family diminishes. Winter states that we continue to expect the family to meet all of our basic needs, even as family resources to meet those needs diminish. This situation produces conditions favorable to conflict. Stoller (1970) also commented on this phenomenon, and on the resulting family burdens: The increasing isolation and independence of the conjugal family has resulted in an impoverished experience, thus placing an almost impossible burden on the family members by forcing them to carry out without aid tasks not easily accomplished in an insular fashion (p. 146). Buckland (1972) states that families are very much in touch with the stresses of their modern-day life, but are not aware of the resources they have to deal with those stresses. She said: Families are more attuned to stress, than they are familiar with resources; . . . they feel at the mercy of bureaucratic institutions without knowing how to utilize them to increase their options. Many affluent middle-class families function at a psychological survival level, while other families still struggle at a physical survival level in a society which has not yet established priorities in terms of human well-being. Families tend to feel bewildered by change, resistant and reactive, not having been taught any skills for the manage- ment of change (p. 151). This undercurrent of stress will manifest itself differently in different families. Common Treatment Approaches Most approaches to helping the adolescent family deal with stress focus on the family member most obviously evidencing that stress-- very often the acting-out adolescent. The adolescent in seen indi- vidually for counseling, or in some cases, in a group with his/her agemates. Increasingly, though, the family unit has been viewed as the appropriate unit of care (Lewis et a1., 1976). By focusing on the family, the change agent attempts to influence the way in which the family deals with the issues of intimacy, values, and role definition, which may be closer to the causes of family stress than is the behavior of the acting-out adolescent. Few concerted efforts have yet been directed at influencing external causes of family stress. Limitations of Traditional .Social Institutions Schools, courts, the mental health system, organized religion, and Parent Effectiveness Training (PET) have attempted to deal with the problems of families with acting-out adolescents. All are very limited in their attempt to address the entire problem. Some are limited by mandate, others by skill or time limitations of their workers, and still others by families themselves, who may not consider a particular institution to be a legitimate source of help. Schools. Schools, the adolescent family's most important social institution, focus almost exclusively on individual students rather than on their families. At best, attention to and help for problems is limited to child-centered rather than family-system-centered issues. Secondary schools are prepared to educate interested and able students. Pupils not exhibiting those qualities present problems. Schools typically choose to solve these problems with Options that are in the schools', not the pupils', best interest. For instance, chronic truancy, a common behavior of the acting-out adolescent, often results in temporary or permanent suspension from school. This is a solution for the school, perhaps, but not for the family. QQEEEE: The probate or juvenile court is another institution that frequently comes in contact with families or acting-out adoles- cents. The court also finds itself unequal to the demands before it. According to a Lansing Probate Court official, court staff workers are notoriously overloaded, and often have little or no advanced professional training.1 Because of this, and the complexity of prob- lems in the families of their clients, the courts most often choose to work with the adolescent individually, or in groups. Family treat- ment is still viewed as too complicated and time-consuming for courts to attempt. Traditional mental health system. Public and private mental health agencies and practitioners are involved in the treatment of acting-out adolescents. They occasionally work with the family as well as the individual adolescent. Their efforts are largely limited to clients who are able and willing to come in for treatment. This is a major limitation when the target population includes large numbers of disorganized families, and families with multiple problems. In addition, treatment at a mental health agency or through a private practitioner may carry with it a stigma of failure or of mental weak- ness, which many families won't accept. The cost of private treatment can also be prohibitive. Costs range from $35-$60 an hour, and many families do not have insurance to cover psychotherapy. Organized religion. The church or synagogue has traditionally been a source of support and guidance to families facing difficult 1Summarized from conversations during 1975-76 with David Brown, Director of Training, Ingham County Probate Court. problems. Their support, however, is most useful for those families who seek help before major family problems emerge. The family with a chronically truant, runaway adolescent needs an intensive remedial program. The growth-enhancing, educationally oriented programs offered through organized churches and synagogues are not enough. Another limitation of church-related programs is that many families are not affiliated with an organized religion, and would not participate in programs offered through those institutions. Also, some families prefer not to reveal the extent of their family diffi- culties to their local minister, priest, or rabbi. Parent Effectiveness Training. Within the last decade, Gordon (1975) has developed an educational program for addressing many common communication and problem-solving difficulties. Their model includes work with parents only and requires intellectual skill and interest. An additional limitation is the high cost of the course ($50 for a series of eight classes for one parent, $75 if both parents take the class).2 In spite of these limitations, this program has been effec- tive in helping highly motivated parents increase their communication and problem-solving skills (Effectiveness Training Associates, 1975). New Develgpments in Alternative Community_Agencies It is clear from recent research that a portion of the families with acting-out adolescents is not being served by traditional social 2From conversations with Eleanor Morrison, a local P.E.T. instructor. 1O agencies (Garfield & Bergin, 1978). As family needs become increas- ingly complex, perhaps new institutions need to be developed to address those needs. Crisis centers. One response during the last decade has been a dramatic increase in the number of crisis centers. Some of these centers meet the needs of troubled adolescents who are not being served by established social agencies (Garfield & Bergin, 1978). Other centers are outreach programs of existing social agencies. Rabinowitz (1973) described the need for such programs: There is a growing community demand for the mental health establishment to respond to the adolescents' needs by set- ting up community-based outreach programs which, although clinically oriented, do not follow the usual clinical model for the delivery of service (p. 90). Most of the crisis-centered agencies were originally staffed primarily by volunteers and began by offering crisis service for acute problems. As they became more experienced in the areas of drug abuse, drug overdose, attempted suicide, domestic violence and ado- lescent runaways, many of these agencies expanded their programs to provide short— and/or long-term counseling for these and other prob- lems. There are four reasons for the expansion of services through this relatively new social institution. 1. Communities have increasingly shared responsibility for the increase in family problems. The paraprofessional crisis center is likely to be a small, loosely structured group of local residents who are concerned with helping the immediate community. These residents share the problems, and try to find solutions. 11 2. People have become more actively involved in their own care. Clients often demand an equal share in the planning and execution of their treatment. New crisis-centered agencies often allow the involved pe0ple, both clients and workers, to develop a treatment program unique to each new situation, unhampered by custom and bureaucracy. 3. These agencies cost less to run than many other institutions that address the same needs. Most counseling in these agencies is done by trained and supervised paraprofessionals, both paid and volunteer, and the savings in wages alone are great. Additionally, these agencies most often function in houses or store-front offices with low overhead costs and minimal supplies. 4. Paraprofessional crisis centers have a special attraction for adolescents. The agency itself is often viewed by the surrounding community as having a kind of fringe legitimacy, just as adolescents have a kind of fringe legitimacy in the adult world. The workers are often young adults searching for a way to make their adult experiences relevant to the hopes and ideals of their own adolescence. These two factors combine in helping a troubled adolescent identify with the worker and the institution in a way that allows each to help the other. The acknowledged mutuality of the relationship is unique to this insti- tution and is one of its greatest strengths. Funding. Federal and state governments have added funds to those the communities provide through programs such as the Comprehensive Employment and Training Act (CETA) and grants for specific projects. The additional funds have allowed many crisis centers to expand their services. 12 Effectiveness. Research on the effectiveness of counseling pro- grams offered by trained and supervised paraprofessionals in crisis centers is severely limited. The evaluation skills necessary for such an endeavor are not generally available to these low-budget agencies and the methodological discipline required is somewhat in conflict with their independent and individualistic style. Therefore, a well- designed evaluation of treatment programs offered by one such agency would clearly be useful. This study evaluates two programs of such an agency. Theory The treatment programs evaluated in this study are evaluated from a societal and interactional theoretical viewpoint. Families are viewed as having basic tasks, assigned to them by society, which they perform using strengths inherent to the family system. When the family system itself does not work well enough, outside intervention can help alter the dysfunctional system, thereby returning it to health. The next section contains discussions of societally assigned family tasks, family strengths, methods of family change and the inter- action differences between normal and dysfunctional adolescent families. Family Tasks The family is the basic unit of our society and according to Geismar and LaSorte (1964) it is responsible for "reproduction, shelter and physical care for family members, emotional care and the socialization of the young" (p. 38). Ackerman (1966) elaborated on these tasks. He stated that the family is responsible for "union and 13 individuation, the care of the young, the cultivation of a bond of affection and identity, reciprocal need satisfaction, and training for the tasks of social participation" (p. 62). Family responsibilities lie simultaneously in the current physi- cal and emotional support a family offers its members and the promise it gives the future through the raising of children. Rapid social change threatens the family's ability to function at the same time it requires the family's continued existence. Minuchin (1974) addressed this paradox: The occidental world is in a state of transition, and the family, which must always accommodate to society, is chang- ing with it. But because of transitional difficulties, the family's major psychosocial task-~to support its members-- has become more important than ever. Only the family, society's smallest unit, can change and yet maintain enough continuity to rear children who will not be "strangers in a strange land," who will be rooted firmly enough to grow and to adapt (p. 47). In summary, society needs the family to provide an environment that is both stable and flexible, in order to meet the physical and emotional needs of its members and to raise children. The family has the potential to change quickly enough to meet the shifting demands of our society and slowly enough to allow its members to feel secure in their growth. This concept was enacted by Tevye in Fiddler on the Egg: (Jewison, 1971). He struggled with just how much his “traditions" must bend in order to remain relevant to his family in their changing world. He also recognized that there is a point at which the ”tradi- tions" will no longer bend, but will break; that to pass that point meant to float anchorless with no signposts directing the way to safety. 14 Family Strengths The dynamic, changing nature of the family is a strength to be used in a family treatment program. It is easier to influence a moving system than a static one. The family has many internal resources. These resources stem from the individual members, who bring different skills, demands, and insights to the group as a whole. This behavioral and emotional diversity, combined with the caring within the group, contributes to the solution of systemic problems. The mutual caring that exists in all families can lead to a commitment to overcome the problems that beset the system. This commitment allows and encourages the family to examine sometimes painful or unacceptable feelings and behaviors in order to change or understand them. Family Change Ackerman (1966) outlined three ways that families can change: (1) Re-education of the family through guidance, (2) Re- organization through a change in the patterns of family communication, and (3) Resolution of pathogenic conflict and induction of change and growth by means of a dynamic approach in depth to the affective currents of family life (p. 92). These avenues of change are not mutually exclusive. The third avenue for change describes the process of in-depth family psychotherapy. This approach is thorough but time consuming; it requires the skill of a well-trained psychotherapist. The first avenue for change, that of guidance, offers new ideas and expanded limits that may increase a family's problem solving. However, guidance has a connotation of receiver passivity, and may not in 15 itself produce change. The second avenue for change, reorganization through a change in the patterns of family communication, requires the active participation of family members and, by definition, involves an alteration in family behavior. The guidance and communication methods for inducing family change, avenues 1 and 2, work well in group settings. They can be facilitated by trained leaders. Gordon's Parent Effectiveness Train- ing program (1975) uses these two methods to work with a large group of parents simultaneously. Similarly, the empathy training model outlined in A Survival Manual for the Drug Center Volunteer (Hinds et a1., 1972) blends guidance and communication methods, and is advo- cated in a group setting. These education/communication programs, when applied to the family by paraprofessional leaders, have the potential to influence many families. Differences Between Normal and Dysfunctional Adolescent Families Ferreira and Winter (1965) studied interaction in triadic families with the index child diagnosed as either (1) schizophrenic, (2) delin- quent, (3) maladjusted, or (4) normal. Using the Unrevealed Differ- ence Technique, which the researchers developed themselves, they assessed family interaction on 13 different variables. Ferreira and Winter discovered that normal families had higher Spontaneous Agree- ment scores, lower Decision Time scores, and higher Choice Fulfillment scores than delinquent families. (Spontaneous Agreement reflected the number of times one family member's questionnaire choice matched that of another family member. Decision time reflected the number of 16 minutes it took the family to discuss their consensus answers to the questionnaire. Choice Fulfillment reflected the number of times an individual family member's most wanted choice was what the entire family chose.) It was also shown that in normal families Spontaneous Agreement scores were positively correlated to the age of the child, whereas in delinquent families they were not. This suggested that, over time, normal families increasingly shared a frame of reference and delinquent families did not. Also, normal families had fewer chaotic responses than delinquent families. Overall, the data indi- cated that normal families functioned better in decision making than did delinquent families. Mead and Campbell (1972) studied triadic family interaction using a design similar to Ferreira and Winter's. Their families included either a normal adolescent child or a drug-abusing adolescent child. Mead and Campbell also used the Unrevealed Difference Technique to assess family interaction. The researchers discovered that Spontaneous Agreement differentiated normal families from abnormal families, with normal families showing greater Spontaneous Agreement in triads. They also found a positive correlation between Choice Fulfillment and Spon- taneous Agreement scores. Both the Ferreira and Winter study and the Mead and Campbell study offered support for the hypothesis that normal families are more effec- tive decision makers than abnormal families. The decision-making effectiveness of normal families compared to drug-abusing or delin- quent families seems to stem most strongly from their shared prefer- ences for certain outcomes. Neither of these studies explored the 17 mechanisms by which families reach decisions in areas of prior dis- agreement. Hetherington, Stouwie, and Ridberg (1971) also studied triadic family interaction. Their index child was either normal (NL) or diagnosed as (1) unsocialized-psychopathic delinquent (PD), (2) neurotic-disturbed delinquent (ND), or (3) socialized-subcultural delinquent (SD). They used the Structured Family Interaction Task and the Stanford Parent Questionnaire to evaluate family interaction and attitudes. Post hoc examination of their data supported the hypothesis that each of the four diagnostic groups was different from the others. For boys, 87 of the 119 tests of the effects of group membership resulted in significant differences (73%). For girls, 49 of the 118 tests of the effects of group membership resulted in sig- nificant differences (42%). In addition, the differences among the three delinquency groups were as large as the differences between any one group and the normal controls. The four groups differed in the way they expressed conflict and affect and on the dominance patterns in the family. Nondelinquent families demonstrated fewer negative expectations and less anxious emotional involvement than did delin- quent families. They also had a restrictiveness-permissiveness level very close to the mean of all families in the study, whereas the various categories of delinquent families were on either end of the restrictiveness-permissiveness distribution. Hetherington et al. found that the differences among families with female delinquents were not as strong as those for male delinquents, but were generally in the same direction. They also discovered that little correlation existed between the behavior measure (a structured interaction task) and the attitude questionnaire. Alexander (1973a, 1973b) reported results of two family inter- action studies. The first study compared normal and delinquent family interaction, while the second study examined normal families only, relating their interaction patterns to teacher estimates of the index child's aggressiveness. In the first study, Alexander rated each individual communication as either defensive or supportive. His data supported the thesis that normal families have different and more effective communication patterns than delinquent families. Normal families demonstrated a lower rate of defensive communication across all directional diads, and children in delinquent families had lower rates of supportive communication toward their parents than did chil- dren in normal families. Alexander also demonstrated the reciprocity of supportive and defensive communication in families; the more a family member gave supportive or defensive communications to another person in the family, the more he/she received the same kind of com- munication from that person. In normal families, this reciprocity was significantly demonstrated for supportive communication, but not for defensive communications. In delinquent families such reciprocity was significantly demonstrated for defensive communications, but not for supportive communications. Clearly, normal families behave dif- ferently from delinquent families, and the differences function to amplify further the family's "wellness" or "deviancy." The second study reported by Alexander examined only normal family interaction. The relationships between family communication 19 patterns and aggressiveness on the part of the child were explored. Alexander again demonstrated that normal families had higher rates of supportive communications than defensive communication, and that supportiveness was reciprocal in those families whereas defensiveness was not. He found that families with less aggressive sons demon- strated more parent supportive behavior and less mother to child defensive behavior. Even in normal families, levels of parental sup- portiveness and maternal defensiveness were related to the amount of child aggressiveness. Summar Looking at all family interaction studies reviewed, several com- parisons can be made. Ferreira and Winter's conclusion that normal families function better than delinquent families on decision-making tasks was based on their assumption that high Spontaneous Agreement scores, high Choice Fulfillment scores, and low Decision Time implied better family decision-making processes. Alexander's finding that normal families functioned better than disturbed families was based on his assumption that supportive communication was characteristic of well-functioning families whereas defensive communication was char- acteristic of poorly functioning (disturbed) families. Ferreira and Winter's finding that Spontaneous Agreement scores in normal families were correlated positively with the age of the child corroborated Alexander's finding that, in normal families, supportive communication was reciprocal (i.e., the more one member communicated supportively, the more supportive communications he/she received). Conversely, 20 Ferreira and Winter's finding that, in delinquent families, Spon- taneous Agreement scores were not correlated positively with the age of the child supported Alexander's finding that, in delinquent fami- lies, supportive communication was not reciprocal. Hetherington, Stouwie, and Ridberg found that normal families expressed less total conflict, less hostility, more warmth and that parents were less power assertive than families with delinquent chil- dren. Parents in the normal families studied were more willing to listen to their son than parents in delinquent families. Comparing these interaction dimensions to those used by Alexander, the normal families in the Hetherington et a1. sample exhibited greater supportive communication (warmth, listening) and less defensive communication (power assertiveness, hostility and conflict) than the delinquent families, lending further support to Alexander's similar findings. Also Hetherington et al. found that interaction in families with girls followed similar patterns as that in families with boys (except that in families with girls the patterns were weaker). This finding was somewhat supported by Alexander, who failed to find any difference on supportive or defensive communication due to sex, or to the interac- tion between sex and group membership. These studies support the idea that normal families function better than delinquent families. This was evidenced both by their differences on static agreement and satisfaction measures and their problem-solving procedures. On static agreement and satisfaction measures, normal families showed greater initial agreement before any family discussion took place, and a greater frequency of individual 21 choice fulfillment following family discussion. During the process of solving family problems, normal families demonstrated more sensi- tive, caring, warm, empathic-supportive communications and fewer hostile, power assertive, conflictual, judgmental-dogmatic, and indifferent-defensive communications than delinquent families. Theory Applied to Treatment The two counseling programs evaluated in this study are expected to alter communication patterns in families with acting-out adoles- cents to make them more similar to communication patterns in normal adolescent families. Both programs are also expected to improve ado- lescent self-concept and to reduce adolescent acting—out. This matching-to-sample philosophy is advocated by Parsons and Alexander (1973) and is used by many clinical researchers (Katz et a1., 1975; Stanley, 1978). To "match the sample" of normal functioning families, a counseling program for families with an acting-out adolescent should teach family decision-making strategies and increase the incidence of mutual positive reinforcement within the family. These changes are expected to alter the system to allow the natural internal resources of the family to take over and maintain a new, more functional balance. Overview of Remaining Chapters In Chapter 2, literature relevant to treatment programs for dysfunctional adolescent families is reviewed. The literature reviewed is in three main areas: (1) concurrent treatment of more than one family member, (2) treatment of the identified acting-out adolescent 22 only, and (3) treatment provided by paraprofessionals through a community-sponsored crisis center. Chapter 3 contains a review of measurement instruments designed to assess family interaction, adolescent self-concept and adolescent drug and delinquent behavior. Chapter 4 includes an outline of the treatment programs offered to adolescent families in this study, along with an evaluation of those programs. Examined are such details as a description of the sample, the measures used (including reliability data), design, hypotheses, and method of analysis. In Chapter 5, the outcome of the data analysis is presented. Chapter 6 contains a discussion of results, along with their implications for further research. CHAPTER 2 REVIEW OF THE TREATMENT LITERATURE This chapter includes a review of literature relevant to treat- ment in families with acting-out adolescents and treatment by para- professionals in a community crisis center. The review is organized into three sections: The first includes a summary of research in the area of family change as a result of whole family treatment; the second section includes a review of group treatment programs for the identified adolescent only; the third section is a review of the lit- erature related to treatment by paraprofessionals through a community- sponsored crisis center. Change as a Result of Family Treatment Only recently have researchers studied whether and how families change as a result of certain treatment programs. The goal of this kind of research is to discover whether a given treatment has resulted in the desired change in family behavior. Reliable instruments had to be developed to measure family behavior, and data on normal family interaction had to be gathered to provide a guide for setting treat- ment goals. Much of this work regarding families with acting-out adolescents was done in the late 19605 through early 19705, and is reported in Chapter 1 (normal family interaction) and in Chapter 3 (family interaction measures). 23 24 Evaluation of family treatment programs usually involves one subject population sorted into two or more groups, each receiving a different treatment, with change as a result of treatment being measured with a posttest. Often many variables are measured, and each variable is analyzed separately using either t_tests or analyses of variance. The problem of very large experimentwise error rates which result from this repeated univariate analytical approach greatly affects interpretation of research results. Experimentwise error rate can be estimated by using the formula presented by Hays (1963, p. 471): J (a) Z experimentwise error rate 2 l - (l — a)9 where j number tests performed a probability of the occurrence of a Type I error The upper limit of this range assumes perfect dependence among the tests. The lower limit of the range assumes the tests to be inde- pendent. For simplicity's sake, the experimentwise error rate reported for each of the studies reviewed in this chapter was figured using the upper limit of the range: experimentwise error rate = j (a). Therefore, the value given is the most conservative one. The experimentwise error rate is helpful in evaluating studies with many dependent variables, each examined by a constant level for a. As experimentwise error rate increases, more care must be taken in the interpretation of a few significant differences. Thus finding some significant differences when evaluating many variables is not, in itself, strong evidence of difference between the groups. 25 Parsons and Alexander Study A study evaluating the effects of a short-term treatment program for families with acting-out adolescents was carried out by Parsons and Alexander (1973). The researchers used a "matching-to-sample" philosophy. Knowing the interaction patterns of normal families, they designed a treatment program that would alter the interaction in delinquent families to approximate more closely that of normal families. Their sample comprised 40 mother-father-child triads. Each identified adolescent child had been arrested or detained by local juvenile court authorities for running away from home, being habitually truant from school, or being generally unmanageable. Families were randomly assigned to one of four groups. Using the Solomon Four-Group design, there were two treatment groups and two control groups. One treatment group and one control group received a pretest while the other two groups did not. Control-group families who received a pretest (PCP) received no treatment of any kind. Control-group families who did not receive a pretest (CP) received what the researchers called "a non-specific form of professional attention--in this case, group discussions focused on feeling expres- sion led by a professional psychologist" (p. 197). This was called an attention-placebo group, and was clearly a variant of the Solomon Four-Group design. The two treatment groups, PTP and TP, took part in identical programs. The treatment was administered to each family triad sep- arately, in one 2-hour session each week for 4 weeks. Participating families were also given a therapy manual that discussed the goals 26 of the program and the means used to meet those goals. Each family was to read and refer to the manual at home. Using the "matching-to- sample" philosophy referred to earlier, the researchers identified several "normal family" behaviors and structured their program to teach those behaviors to delinquent families. The sessions focused on three normal family behaviors: (l) to differentiate family rules from requests--behaviors that are expected from those that are optional, (2) to develop reciprocity in the exchange of positive responses through the development and use of a token economy system, and (3) to increase the variability of family communication patterns, thereby allowing more effective decision-making patterns to emerge and to ensure equality as well as clarity in communications. Six dependent measures were chosen to evaluate the impact of treatment interventions. These measures fell into two categories: accuracy-of-perception tasks and family interaction patterns. The accuracy-of—perception category was assessed using two measures--a "behavior specificity" task and a "vignette interpretation" task. The behavior specificity task called for each family member indepen- dently to list three behaviors he/she would like the other two family members to change, and then to list three behaviors the person thought each of the other two family members would want him/her to change. Each person listed 12 items; the total number of perfect matches, by pairs, was the family score. The vignette interpretation task pre- sented three child behavior situations, each calling for parental response. Independently, family members were to describe what action they expected both the parents and the child to take in each situation. 27 Each person listed a total of six actions; again, the total number of perfect matches, by pairs, was the family score. The family interaction category was assessed using four measures: (1) amount of silence, (2) frequency of simultaneous speech, (3) dura- tion of simultaneous Speech, and (4) equality of speech as measured by the variance around the mean number of words spoken by each family member. To assess these interaction variables, family members were given 20 minutes to discuss their independent responses to the two accuracy-of—perception tasks; this discussion was audiotaped, then rated. A separate 2 x 2 ANOVA was used to test for treatment, pretest, and interaction effects on each of the six dependent variables. In all, 18 statistical tests were performed, with an experimentwise error rate of .90. Parsons and Alexander reported no significant pretest effects. They reported three pretest by treatment interactions on three of the four family interaction variables assessed: (1) amount of silence, (2) duration of simultaneous speech, and (3) equality of speech. 0n the effects of treatment, Parsons and Alexander found significant differences on four variables: (1) amount of silence, (2) frequency of simultaneous speech, (3) duration of simultaneous speech, and (4) vignette interpretation. This last difference was in the direction opposite to that predicted. These findings, along with group means, are summarized in Table 2.1. 28 .xuwpmzcm egos .mw umcu .mucmwcm> cmppmsm mucmmmcamc mcoum gaze; Aomwuumoq co -mcav mappmmu .mo. v.ma .ucmEmmcmm mo aucmscmcu Focpcou acmapmmcu ll QI—Q (U .D "mug .Fom-mmp aflm .mem_ .smopogoawa _muccw,u use mecppsmcou to Faczzoa .Laucaxa.< .L .z vcm mcomcma .> .m x9 =xusum msoupao Xchmgh < "cowu:m>emucm AFPEmu Egmhuucogm= sock .muoz o.Fv twp.m N¢._ N.m “.4 N.N 4.0 snappmcmw> o._v o.Fv o.Fv ~.m _.m _.m o.m ammcmco cop>a3am «-.¢ mp.N o._v m_.mmap mp.~o_m mm.m¢m_ me.mm¢ agemaam to xgwpaaom ame.¢ o~.m_ o._v mm.o m~.mp mm.mm mm.m~ gummam mzomcappzswm to coppaeac mm.N om.mp o.Pv mm.p N~.m m¢.~ o~.m goamam maomcmppaswm to sucmaomca 4N¢.m amm.m mo.~ op.mFN mo.~o~ oF.N__ om.mmp mu=m_wm to pesos< mmwwww -umwumsmo WMWWmWW a azocw u azoco m azocu < aznmm “gate“ abate“ mu- mus ac- aha ms22.2.8: cowpuacmch AQM\F u.wmv.m .m mmeammmz pcmpcou new compomcmucH com mowumm.m use acme: ammupmoa aaocw u—.~ mpnmh 29 Parsons and Alexander stated, "these findings indicate that the treatment manipulation did in fact produce significant changes in the interaction patterns of the treated families" (p. 199). Because of the contamination of the control-group experience by the inclusion of a placebo treatment in the posttest-only condition, group means were closely examined to interpret the findings. On all measures showing a treatment main effect, the placebo-only group did less well than the pretest control group, indicating that the researchers were correct in asserting that the placebo treatment would not positively affect outcome scores. The placebo acted as a true control and hence did not alter the conclusions of the study. The treatment program evaluated by Parsons and Alexander did indeed improve family interaction patterns in the target population. Katz, Krasinski, Philip and Wieser Study Katz et a1. (1975) explored the effectiveness of short-term family therapy. Their sample comprised 17 families, each meeting the following eight criteria: (1) two parents and their natural or adopted children, intact for at least four years; (2) families no larger than six members; (3) all family members living at home; (4) parents between the ages of 25 and 50; (5) all children born in Canada and parents fluent in English; (6) children between the ages of 4 and 21; (7) gross yearly family income of at least $10,000; and (8) family agreement to participate in the research study. The six families in the experimental group and the five families in the ran- domly equivalent control group were selected from a larger group of 30 families seeking help at a local mental health center. Each family had a child identified as having problems. (No explanation of the types of problems was given.) In addition, these families all agreed to participate in therapy as a unit. A nonequivalent control group of six families was randomly selected from a list of 60 normal families with no known disabilities or disorders. The researcher included this nonequivalent control group to test for the influence of time alone. Each subject family was tested during an initial contact at the mental health center. During this first contact, each family was given an unspecified family planning task to work on for 5 minutes. This interaction was videotaped, and the tape was later transcribed. The transcription, along with the video and audio tapes, was rated using Riskin and Faunce's Family Interaction Scales (1970a, 1970b, 1970c). These scales are discussed in greater detail in Chapter 3. Each family in the treatment group then received four family therapy sessions over a 4- to 6-week period. The sessions were provided by a local psychiatrist specializing in child and family psychiatry. There was no indication of what these sessions included or how long each lasted. Following treatment, each family was pottested using the same procedure as the pretest. Families in the randomly equivalent control group and the nonequivalent control group received no therapy and were posttested after 4 to 6 weeks. Katz et al. had five hypotheses to test: (1) Collectively, all groups will change between the pre- and posttest. (2) The control group will change in a negative direction. (3) The nonequivalent control group will change (no specific kind of change predicted). 31 (4) The treatment group will change in a positive direction. (5) There will be a difference between the treatment group and both control groups on pre and post gain. The researchers defined positive change as movement toward Riskin and Faunce's (1975) description of adequately functioning families: "a tendency towards the middle range of clarity in expressing content material, an abundance of humor, less topic change, more information sharing rather than commitment demands, free- dom to express differences but a high amount of agreement, a wide range of affectivity and a high number of positive relationships" (p. 37). The Family Interaction Scales (FIS) used to assess family behavior patterns yield 29 variables; each was analyzed separately using two nonparametric statistics. The Mann-Whitney U Test was used for all pair comparisons and the Kruskal-Wallis Analysis of Variance was used for multiple group comparisons. In all, 145 statistical tests were performed, yielding an experimentwise error rate of 7.22. The high experimentwise error rate of 7.2+, and the very few sig- nificant differences found on the family interaction variables measured, combine to challenge the researchers' claim of a successful intervention program. Nothing in the data indicated that change occurred. Beal and Duckro Study An evaluation of a juvenile-court-sponsored short-term family counseling program was reported by Beal and Duckro in 1977. This program was aimed specifically at families with juvenile status offenders (JSO)--teenagers whose offenses would not be considered 32 criminal if they were committed by an adult. The most common status offenses are running away from home, school truancy and repeated dis- obedience to parental rules. The goal of the counseling program was to divert these families from the usual court system. The researchers felt such diversion was important because they believed that allowing the juvenile status offender to be treated in the same manner as the juvenile delinquent encourages the develOpment of true delinquent behavior, adds to the already existing problems of the family, and is costly to society in general. They suspected a mental health approach to the problem would be superior to a legal approach. Accordingly, the dependent variable for the evaluation of the JSO Diversion Program was the number of cases terminated without court action. In juvenile status offense cases, parents petition the court for help in managing their child. For the case to be terminated without court action, the parents need to withdraw their petition. This parental action--with- drawal of petition--was considered evidence of treatment program success. The treatment sample of 44 families was chosen randomly from all families participating in the voluntary JSO Diversion Program during a particular month. The researchers did not indicate the exact program acceptance rate of eligible 050 families; they merely indicated that a "large proportion of families accepted this alternative" (p. 78). The control group of 54 families was chosen randomly from all the 050 families referred to the court during the same month one year earlier, before the 050 Diversion Program had been in effect. In this way the 33 researchers controlled for seasonal variations in types and severity of offenses. Treatment consisted of six to eight family sessions focused directly on resolving the current problem without court action. Initially, crisis intervention techniques were used to work through anger and frustration and place the immediate crisis of the identified teen within a broader family perspective. Later in the counselling experience interventions were aimed at modifying unspecified familial interaction patterns. The control group experienced the traditional probation officer/identified teen contact, with little or no family participation. In both treatment and control groups, however, the goal was to keep the teen out of the court system. Data on the diversion of the treatment and control groups were analyzed using a chi-square test for independent samples. The treat- ment group had a significantly greater proportion of cases closed without court action (83%) than did the control group (65%) (X2 = 3.95, gf_= l, p_< .05). Beal and Duckro concluded that the JSO Diver- sion program had successfully achieved its purpose. A criticism of this research is directed at the equivalency of the treatment and control groups. The treatment group consisted wholly of volunteers, whereas the control group did not. Presumably, families who would have rejected the 050 Diversion Program had it been available remained in the control group, whereas similar families would not have been in the Diversion Program. This variable of acceptance, willingness to participate in a family treatment program, may account for some or all of the difference found between the two 34 groups on the successful diversion rate measure. (Although the authors indicated a high acceptance rate of families in the JSO pro- gram to be in the research study, they did not note the acceptance rate of the 050 program itself by the entire population of eligible court families.) An additional criticism of Beal and Duckro's research is the lack of information on the longer-term effects of this program. Immediately following treatment, many families withdrew their court petitions. Did these families later (1 month, 4 months, etc.) resubmit the petition? In the absence of any information on basic changes in family interaction patterns or attitudes, a single behavioral measure taken at one time offers little support for the hope of long-term change. Stanley Study A program for adolescent families, designed to teach methods of inductive discipline and problem solving, was conducted and evaluated by Stanley (1978). The program sought to affect the moral atmosphere or justice structure of the family, as evidenced by improved family communication, attitudes, and decision-making patterns. Stanley used a definition of moral atmosphere or justice structure developed by Kohlberg, Scharf and Hickey (1972): The justice structure is "the principles which govern the assignment of rights and duties and define the pr0per distribution of the benefits and burdens of social coop- eration" (p. 111). The families in Stanley's study were volunteers, all of whom responded to a letter sent by school authorities describing the 35 program. All families were Caucasian Christians and had at least one child in the ninth or tenth grade in the local public high school. Only families in which both parents and the adolescent child agreed to participate were accepted. They were predominantly lower-middle- class families, with parents holding jobs in the skilled trades or the nonmanagerial white-collar category. There was no mention of whether the adolescents in the family had been identified by sources outside or inside the family as having any particular kinds of prob- lems. Families were sorted into three groups, according to most con— venient meeting times: (1) Treatment Group I--A11 family members attend class sessions. Five families, including 10 parents and 7 adolescents, were in this group. (2) Treatment Group II--0nly parents attend class sessions. Six families, including 12 parents and 6 ado- lescents, were in this group. (3) Control Group--Five families, including 10 parents and 5 adolescents,werein this group and received no treatment. During the week preceding the program, all families were pre- tested. The week after the program ended, the families were post- tested, using the same procedure as in the pretest. Four tests were administered in this fashion, yielding nine variables at each of the two points in time. For the two treatment groups, five additional variables were assessed by means of tape recordings of the first and last family meeting. These family meetings occurred weekly from the fourth through the tenth week of the program. Also, both treatment groups participated in a follow-up testing on one variable 1 year after the program ended. In all, 14 variables were measured on the 36 two treatment groups and a subset of 9 variables were measured on the control group. These variables were analyzed using nine Analyses of Covariance and five Analyses of Variance for repeated measures. These 14 analyses yielded an experimentwise error rate of .62. The following four instruments were administered to all groups, before and after treatment: (1) the Ferriera-Winter Questionnaire for Unrevealed Differences (1965), using the positive choice fulfillment measure only; (2) the modified Parental Attitude Research Instrument, originally developed by Schaefer and Bell and modified by Cross and Kawash (1968); (3) the Family Decision-Making Attitude Scale, devel- oped by Stanley herself; and (4) the Kohlberg Moral Judgement Interview (1958). By means of tape-recorded family conferences, the two treat- ment groups were also assessed on five variables: (1) ratio of parent talking time to total conference time, (2) number of decisions reached by consensus, (3) number of parents' ordering responses, (4) number of parents' personal feeling or Opinion responses, and (5) number of parents' reflecting or summarizing responses. Families in the two treatment groups also were given the Kohlberg Moral Judgment Question- naire 1 year after the program ended. The two treatment groups received the same program; the differ- ence between the groups centered on who in the family participated. All treatments were given in a group. In Treatment 1, both parents and adolescent children participated in the group, and in Treatment 2, only the parents participated in the group. The program required 10 consecutive weekly group sessions, each lasting 2.5 hours, and led by two peOple--the experimenter and a specially trained school counselor, 37 both female. The curriculum design was influenced by an Adlerian approach to family education, as presented by Dreikurs and colleagues (1959) and by Gordon's Parent Effectiveness Training model (1970). The main emphasis of the curriculum was on resolving conflict and set- ting rules, using active listening skills as the major tool. The program was divided into four main phases. During Phase I the focus was on communication skills training; empathic listening and construc- tive confrontation were the major areas of emphasis. This phase took 3 weeks. Session 1 included an introduction to the program and a presentation of the rationale. Parents were encouraged to reflect on their own adolescence and the teenagers (if present) were invited to share their current experience as adolescents. Session 2 included a presentation of listening skills, based on a blend of Gordons' and Carkhuff's (1971) work. After the presentation, instructors modeled the desired behaviors and encouraged the participants to role play the skills. Home practice was also encouraged. Session 3 was devoted to confrontation techniques; the "I" message defined by Gordon was the major strategy taught. Phase II also lasted 3 weeks, and was devoted to exploring the weekly family meeting as a forum for family conflict resolution. Each family was encouraged to begin using the family meeting, and was instructed to audiotape the sessions for class discussion and instruc- tion. Session 4 (the first session in this phase) presented the family meeting procedures and encouraged discussion of family rules-- how they are made and the problems surrounding them. It was stressed that child participation in the rule-making procedure reduces the 38 incidence of disobedience. Sessions 5 and 6 consisted in discussing the on-going family meetings, using appropriate role-playing situa- tions to explore the rule-setting procedure and to generate alterna- tive procedures to increase the effectiveness of the process. Phase III was one session long (Session 7) and focused on conflict resolution. In response to the now-present frustration with existing family methods of conflict resolution, participants were taught a six-step method to resolve conflicts. They practiced the method in small groups, first using situations posed by the leaders and then using examples from their own family experiences. Phase IV focused on value clarification and took up the remain- ing three sessions. Participants were encouraged to explore and share their own values concerning a number of dimensions. The role that values play in determining behavior was discussed, as were methods of handling value conflicts in the family. Stanley tested five hypotheses across the three groups, and then explored five others comparing just the two treatment groups. The first five hypotheses are now listed, along with the results of their analysis. Hypothesis 1: Families who have taken the course will score significantly higher in conjoint decision-making effectiveness than families who did not take the course. This hypothesis was tested with an ANCOVA, using Ferreira and Winter's Unrevealed Differences measure, and was accepted, f_= 4.918, gj_= 2,12, p_< .05. A post hoc Scheffe examination showed no difference between the two treatment groups, but did show a significant difference by pairs between each treatment group and the control group. Hypothesis 11: Parents who 39 have taken the course will score significantly higher in nonauthori- tarian attitudes toward child rearing than parents who did not take the course. This hypothesis was tested with an ANCOVA using the PARI to examine fathers' nonauthoritarian behavior, mothers' nonauthori- tarian behavior, and combined parental nonauthoritarian behavior. The hypothesis was accepted, based on the difference shown by fathers between groups and the combined parental score between groups, Fathers: E_= 4.254, gf_= 2,12, p_< .05; Combined Parents: f_= 7.056, g:.= 2,25, p_< .01; Mothers: §_= 2.655, gf_= 2,12, n.s. On both variables showing a difference between groups, the post hoc Scheffe demonstrated significant differences only between the parents-only treatment group (Group 2) and the control group. Hypothesis lII: Parents who have taken the course will score significantly higher in equalitarian attitudes toward family decision making than parents who did not take the course. This hypothesis was tested with an ANCOVA using the Family Decision Making Attitude Scale to examine the father's attitude, mother's attitude, and combined parental attitude. The hypothesis was accepted, based on the differences shown by fathers and by combined parental attitudes, Fathers: F = 6.451, df = 2,12, p_< .05; Parents: E_= 7.289, gj_= 2,25, p_< .01; Mothers: f_= 2.302, gf_= 2,12, n.s. On both variables showing a difference between groups, the post hoc Scheffe failed to demonstrate a difference between the two treatment groups, but did show a significant difference between each treatment group and the control group. Hypothesis IV: Adolescents who have taken the course with their parents will score significantly higher in equalitarian attitudes toward family decision making than 4O adolescents in either of the other two groups. This hypothesis was tested with an ANCOVA using the Family Decision Making Attitude Scale to examine the adolescents' attitudes. The data failed to support the hypothesis, f_= 3.660, gf_= 2,11, n.s. Hypothesis V: Adolescents who have taken the course with their parents will score significantly higher in their level of moral reasoning than adolescents in either of the other two groups. This hypothesis, tested with an ANCOVA using the Kohlberg Moral Judgment Interview, was not supported, §_= 1.62, gf.= 2,14, n.s. The researcher then examined the data post hoc using a t_test for correlated data on the pre- and posttest scores for each group separately. The family treatment group (Group 1) showed sig- nificant change, t_= 2.92, p_< .025, but neither of the other groups demonstrated change. However, the ANCOVA demonstrated that when Group 1 was compared to Groups 2 and 3 there was no difference between groups. Following the comparison of treatment groups to control group, Stanley compared the two treatment groups to each other using data collected in the audio tape recordings of the first and last weekly family meetings. These data were analyzed with an analysis of vari- ance for repeated measures. For the measures of (1) parent talking time, (2) number of consensus decisions, (3) number of parent ordering responses, and (4) number of parent personal feeling or opinion responses,rw>differences were demonstrated between groups. Both groups changed equally (and significantly) on all variables from pre- test to posttest. 41 Summarizing the results, five of the nine comparisons between the three groups showed significant differences. In each case, the differences were shown on post hoc tests to be between one or both of the treatment groups and the control group. No differences were found between the two treatment groups on any of the variables. The five measures taken during weekly family meetings of the two treatment groups only, resulted in all comparisons showing time to be a signifi- cant variable for both treatments. However, no Time x Treatment interaction was discovered. Based on these data, Stanley concluded that both treatment pro- grams altered family communication patterns. She maintained that the total family group (Group 1) was more "effective in changing parental behavior, whereas the parent group had more of an impact on parental attitudes" (p. 116). This study by Stanley outlined a comprehensive family/parent edu- cation program addressing communication and problem-solving needs of the adolescent family. However, the evaluation of this program leaves much to be desired. There are two criticisms. First, in all analyses, the researcher used the individual as the unit of analysis, twice violating assumptions of independence without discussing the possible implications of this violation for the results. A set of parents was treated as two independent subjects, when in fact they were together for treatment. Likewise, families (or parental pairs) were grouped for part of the treatment and a portion of the treatment effect might properly result from the recurring interaction of the stable treat- ment group in addition to the specifics of the treatment program. 42 The violation of the assumption of independence probably resulted in a more liberal test than the reported a = .05 and a higher experiment- wise error rate than .62. As a second criticism, only 6 of the 14 variables were direct measures of family behavior. The remainder were questionnaire measures of parental perception of behavior and of family member attitudes. It is not known exactly how close the correlation between actual behavior and the perception of that behavior is. Indeed, in Chapter 1 it was reported that Hetherington et a1. (1971) found little corre- lation between actual behavior, as measured by a structured interac— tion task, and the seemingly related scales of a questionnaire measure. Therefore, Stanley's claim to have produced behavior change is some- what questionable in light of the instruments she used to measure that change. Parent Effectiveness Training Gordon (1970) developed and syndicated a relatively new parent education program through the Effectiveness Training Associates (ETA). The ETA trains certified instructors and controls the presentation of its programs by providing materials and ongoing supervision to prac- ticing instructors. The ETA programs have not been thoroughly evalu- ated; the research that has been done on these programs has been con- ducted only by persons apparently affiliated with the organization. Of the 17 studies cited by the ETA in a 1975 research summary paper (ETA, 1975), being evaluations of their programs, only one was pub- lished (Larson, 1972). The remainder were unpublished doctoral 43 dissertations, master's theses, or research studies. It is difficult to review the original research on effectiveness training programs. The one published research study is reviewed next. Larson (1972) reported an evaluation of several parent education programs offered by a junior high school in a midwestern state. All of the programs were aimed at improving family communication patterns and associated attitudes in both parent participants and their chil- dren. The sample comprised parents only; all parents had children at a local junior high school and had volunteered for one of the several treatment programs. A control group was made up of parents who could not get in the program of their choice because of course size limita- tions or scheduling conflicts. In all cases, these parents later attended a course and were counted a second time as experimental sub- jects. The report provided no demographic data on the sample. Each group was pre- and posttested, and raw scores were compared. No statistical tests were used to analyze the data. Seven different instruments were administered to participating parents to measure change; all were questionnaires or semi-structured written exercises. These measures were the following: (1) an adaptation of the Sears Self-Concept Inventory for Children, administered to parents; (2) an adaptation of the Goal Attainment Scaling System developed by Kiresuk and Sherman (1970); (3) a problem checklist referring specifically to concerns surrounding children and family atmosphere; (4) the Hereford Parent Attitude Scale; (5) self-report logs in which parents were asked to record anything they said, did, or handled differently as a result of group participation; (6) a parent concern survey containing typical 44 parental problem areas; and (7) a survey of parent written responses concerning their experiences in the group meetings. Each of the three treatment groups met for a 3-hour session each week for 8 weeks. Group 1 received the Achievement Motivation Program (AMP), which emphasized individual strengths and sought to help par- ticipants identify their strengths, resources, goals, and values. Individual conflicts were studied in the context of the individual's strengths. This program was aimed toward adding to each participant's strengths through individual discovery and group reinforcement. Group 2 received the Parent Effectiveness Training program, which helped parents explore the source of conflicts between them and their child, and redefined behaviors and interactions that were termed "problems." The program then taught listening and problem-solving skills, which enabled parents to work with their children to resolve the newly defined "problems" in a mutually agreeable way. Group 3 received a Discussion-Encounter Group (DEG) experience, which encour- aged parents directly and accurately to express their own feelings about such topics as dating, hours, study habits, dress, and manners. Structured exercises were provided to assist participants in this task. Larson examined the raw data and means derived from the data and concluded from the gain scores that both the Parent Effectiveness Training group and the Achievement Motivation group "showed substan- tial reduction in those areas of concern that led parents to join the groups" (p. 264). He further concluded that the motivation program was especially effective in improving parental self-concept and the PET program increased parents' confidence in their parental role. 45 The Discussion-Encounter Group produced slight negative changes on some measures and little or no change on the remainder. Larson's study is severely limited by the lack of a statistical analysis. Even the use of the mean as a group summary statistic is weakened by the researcher's failure to present the standard devia- tion or variance of the scores. There is no way of knowing whether these changes are significant. Also, the total lack of demographic data on the sample prohibits generalization to other groups. A second major flaw in the study is the absence of any direct behavioral measure. This is particularly serious because one of the major goals of these programs was to produce behavior change. As stated pre- viously, behavior change cannot be directly inferred from responses to attitude or behavior questionnaires. A third flaw, relatively minor in light of the already stated problems, is the use of some subjects as members of both the control group and one of the treatment groups. Some of these major criticisms also apply to other ETA program evaluations. None of the 16 studies summarized in the Effectiveness Training Associates' 1975 research bulletin used standardized direct behavioral observation methods. All relied on questionnaires measur- ing attitudes, perceived problems, perceived behavior of self or others, or experimenter perceptions of behavior that were not tested for reliability. Six of the 16 studies used a pre-experimental design, with all subjects entering a single treatment group. When statistical analyses were employed in studies with a multi-group design, a uni- variate statistic was used repeatedly to analyze multiple variables. 46 Discussion of Family Treatment Research Reviewing all the studies in this section, several conclusions can be drawn. In several cases, major design, measurement, or analy- sis deficiencies clouded the interpretation of research results. The findings of Katz et a1., Stanley, and Larson cannot be accepted with confidence due to flaws or limitations in their design and analysis. The Parsons and Alexander program as well as the Beal and Duckro pro- gram have strong research evidence to support their effectiveness. Three treatment programs reviewed dealt with each family sep- arately. Parsons and Alexander studied families from a juvenile court population. They provided a brief series of sessions that taught the family to (1) differentiate rules from requests, (2) develop reci- procity in the exchange of positive responses, and (3) increase the variability of family communication patterns. An interesting concept presented in this study is the "matching-to-sample" philosophy, which assumes that if a given delinquent family can change its interaction patterns to be more similar to those of normal families, the delinquent family will begin to behave more like normal families in other areas; i.e., the delinquent adolescent will stop the unacceptable behavior. Underlying this philosophy is a belief in the healing powers of the family system itself. Katz et a1. provided an unspecified form of brief family therapy and discovered improvement in treatment families not found in control families, as measured by numerous variables. Beal and Duckro measured only one variable, the rate of successful diversion of juvenile court families from the official adjudication procedure. They provided eight conjoint family therapy sessions aimed 47 at resolving, through interpersonal change, whatever crisis had sent the family to court. They demonstrated a diversion rate that was higher than that achieved by the traditional probation officer approach. The final two programs reviewed in this area of family change through treatment provided family group experiences. Stanley took normal volunteer families from a school and provided a lO-week struc- tured course addressing both family communication skills and attitudes. The program focused on conflict resolution and rule setting, using active listening skills as the major tool. The researcher found change on both communication and attitudinal variables. Larson also worked with volunteer parents from a school setting and worked toward changing both communication patterns and attitudinal variables. He provided three different treatment programs and compared them all to a no-treatment control group. Parent Effectiveness Training (PET) was offered to one group. This program taught the use of listening and problem-solving skills to resolve parent-child conflicts. A second group was offered an Achievement Motivation Program (AMP), which focused on identifying the participants' personal strengths and helped them to build on those strengths. The last treatment program was a Discussion-Encounter Group (DEG) experience; it focused on increasing the awareness and sharing of feelings about such controversial parent- child topics as dating, dress, and manners. Larson concluded that both PET and AMP were effective in producing desired change; however, his data were not analyzed statistically. 48 The greatest need at this time is for a treatment-evaluation study that combines strong research methodology with an appropriate analysis. The results of such an effort will be more easily inter- preted and, if significant, will indicate true effectiveness of the treatment program. Change as a Result of Adolescent Group Treatment Group counseling has long been recognized as an effective mode for the treatment of adolescents (Amini & Salasnak, 1975; Gilliland, 1968). Jones (1968) saw the effectiveness of the therapeutic group lying in its encouragement of social interaction within a secure setting which fosters the expression of feeling. Group counseling has been used with teenagers experiencing a wide range of problems includ- ing acting-out behavior, underachieving at school, drug abuse, and delinquency. Experimental research using behavioral instruments to measure the effects of group counseling is scarce, but there are many published accounts of group counseling programs evaluated with less rigorous methodology (Rachman, 1969; Barclay, 1969; Riegel, 1968; Rabinowitz, 1973; Truax, 1971; Woody & Woody, 1975). Reviewed in this section are adolescent group counseling programs evaluated with an experimental design. Gilliland Study A study evaluating the effects of a group counseling experience for black high school students was carried out by Gilliland (1968). His sample consisted of 30 black teenagers, randomly drawn from the high school population and randomly assigned to either a control or 49 a treatment group. Boys were kept separate from girls, so there were two control groups and two treatment groups. There was no attempt beyond random selection to equate the groups on factors of age, ability, school achievement, grade point average, grade level, or attendance percentages. The two treatment groups received 1 hour of group counseling each week for the entire academic year, for a total of 36 sessions. The counseling sessions were conducted using a group- centered approach with the leader exhibiting unconditional positive regard for the participants and striving to actively involve all members in the experience. The control group received no group coun— seling. Seven measures were selected to evaluate change as a result of group counseling. These measures were the vocabulary, reading, and English usage scales of the Cooperative English Achievement Tests; the Occupational Aspiration Scale; the Vocational Development Inven- tory; the Index of Adjustment and Values; and grade point averages. The data were analyzed using a "least-squares analysis of data, with unequal subclass numbers, . . . employing the matrix-inversion method of solving for the partial regression coefficients" (Gilliland, 1968, p. 148). This analysis yielded an analysis of variance table with main effects of group and sex, with any group by sex interaction not reported. Fourteen statistical tests were done, with an experi- mentwise error rate of .14. Gilliland reported significant positive findings on five of the seven tests of group effect. The three scales of the Cooperative English Achievement Tests, the Occupational Aspiration Scale, and the 50 Vocational Development Inventory all showed significant difference at the p_S .01 level. On the Index of Adjustment and Values measure, the experimental groups demonstrated a nonsignificant change in a negative direction, while on the grade point average measure change was in a positive, but not quite significant, direction. None of the tests of the effect of sex identity provided significant. In summary, Gilliland feels that the group counseling experience improved the participant's academic achievement and vocational aspira- tions and develOpment. The group also served to allow the partici- pants to acknowledge the reality "of the limited opportunities available to Negroes in their particular environmental setting, and no longer [seek] to deceive themselves" (Gilliland, 1968, p. 149). A major criticism of this study is the fact that subjects were treated in groups, but data were analyzed for individuals. As dis- cussed in the review of family research, the effect of this violation of independence is impossible to estimate, but most certainly func- tions to liberalize the statistical tests performed. Warner and Hansen Study The effectiveness of verbal-reinforcement and model-reinforcement group counseling with alienated high school students was studied by Warner and Hansen (1970). Their sample consisted of 180 students selected randomly from 238 11th grade students who scored at least one standard deviation above the mean on a test of alienation at three different high schools. The subjects were randomly assigned to one 51 of four experimental conditions. Group 1 received a verbal rein- forcement group experience, Group 2 received a model reinforcement group experience, Group 3 received a placebo group experience, and Group 4 was a no-treatment control group. All groups were pre- and posttested using Dean's (1961) scale of alienation. The two treatment groups were run by behaviorally trained coun- selors and had six 40-minute sessions, each focused on an aspect of alienation. The first and fourth sessions were focused on powerless- ness, the second and fifth sessions on normlessness, and the third and sixth sessions on social isolation. The counselor's aim was "to get the students to look at the positive aspects of their situation and to consider positive steps that the students could take to elimi- nate feelings of powerlessness, normlessness and social isolation" (Warner & Hansen, 1970, p. 170). The only difference between the two treatment groups was the mode of reinforcement. The counselor in the placebo group made no attempt to influence group discussion by focusing attention or by reinforcing statements of group participants. Dean's alienation scale was used to measure change. The scale yields an overall score and subscale scores on powerlessness, norm- lessness, and social isolation. A 3 x 4 x 2 univariate ANOVA was used to test for main and interaction effects of counselor, group, and sex on the overall alienation score while a 3 x 4 x 2 MANOVA was used to test for the same main and interaction effects on the three alienation subscales. Each of the 24 statistical tests was evaluated at the p_= .01 level, with an experimentwise error rate of .14. 52 The major criticism of this study is again the violation of the assumption of independence which is brought about by treating subjects in groups, and using the individual subject (rather than the group) as the unit of analysis. With the exception of this flaw, the study is well done, and the conclusions are supported by the data. Data from the ANOVA and MANOVA tests indicated that there was no Counselor x Treatment interaction or Sex x Treatment interaction. The compari- son using orthogonal contrasts of the two treatment groups to the two control groups resulted in significant differences at the p_= .01 level on all four dependent variables. The comparison of the model reinforcement group to the verbal reinforcement group yielded no sig— nificant differences on any variable and the comparison between the placebo group and the control group yielded no significant differences on any variable. Warner and Hansen concluded that both reinforcement styles of group counseling reduced the amount of student alienation, with no difference in effectiveness between the two different reinforcement styles, verbal-reinforcement and model-reinforcement. Sarason and Ganzer Study Using institutionalized delinquent boys, Sarason and Ganzer (1973) investigated the effectiveness of including active role playing in a structured group counseling program based on modeling theory. Their sample consisted of 192 male first offenders between 15; and 18 years of age. Subjects were randomly assigned to one of three groups, Treatment I, Treatment II, or Control. The three groups were comparable 53 in age, IQ, diagnostic classification, and type and severity of delinquent behavior. Treatment groups met for 16 sessions, each 1 hour long, over a 5-week period. Treatment I, the modeling condition, was given to groups of four or five subjects simultaneously, with each group having two leader/ models who were trained psychology graduate students. Each session had a specific topic, such as the job interview, that was thought to present particular problems to this population of adolescents. The topic was explored via a role-play scene which was to be enacted during the session. The participants were instructed about what points to pay special attention to, and they observed while the leader/ models carried out the scene. The group then discussed the content and outcome of the role play, followed by a re-enactment of the scene with members of the group taking different parts. Each participant was actively involved in the role play. Treatment 11, the discussion condition, maintained the same sequence and content as the modeling condition, but did not employ any role playing. All interaction was strictly verbal, with leaders making interventions analogous to those in the modeling condition. Additionally, half of the groups in each treatment condition were audio or video taped during parts of each session. The taped groups observed or heard their actions and again either copied the model or discussed the interaction, depending on which treatment condition they were experiencing. 54 The control group received no group treatment, but did receive the general program of the institution, which was also received by members of the two treatment groups. Describing the dependent measures for this study is difficult because the variables are not clearly or completely stated. Several of the instruments used to measure attitudes and behaviors are listed, but the multiple variables measured by the instruments are not enu- merated. Likewise, statistics used to evaluate the data are not presented, but are merely summarized. An especially disturbing feature of this research report is that there is no way to figure the experi- mentwise error rate because the tests of variables showing no change between groups are not reported. The researchers concluded that both modeling and structured discussion groups had a positive effect on participants, with no strong or consistent differences between the two treatments. Their strongest evidence in support of this statement is the 3-year recidivism rate of the experimental groups. The recidivism rate of the control group was similar to that of the general population for the institution studied, while the rates for the modeling and discussion treatment groups were much lower. A Chi Square test found the differences between groups significant at p_< .06. Block Study An evaluation of a rational-emotive group counseling program for minority high school students at risk for failure, misconduct, and dropping out was carried out by Block (1978). The subjects were black 55 and hispanic 10th and 11th grade students enrolled in an inner-city school having a high proportion of pupils from low-income families. All participants were promised one Social Studies credit for success- ful completion of the program, and each student met the following criteria for the previous year: (1) G.P.A. S .65, (2) 40 or more absences, (3) 25 or more times tardy, (4) 25 or more class cuts, and (5) fivemzmm ucmzccwpwo new mace «poem unaccouuwpmm mommmccmh mpmom cow>mgmm Empnoca meEmu xmmh cowuumcmch apmsmm Pretest Measures Posttest Measures mc_mcco_umm:o cow>mgmm ucmaocwymo new mace m_mum unmucou-$_mm mmmmmccmh m_mum cow>mgmm Empnoca xpwsmu xmmp cowuomcmch wasmu .m 51 52 510 51 52 510 51 52 le Group 1 Group 2 Group 3 124 Alternate Hypothesis 1: The average scores of the two treatment groups will not equal that of the control group. T + T C # 1 2 2 ”la: Where C = Control Group T1 Family Program T2 Adolescent Program Null Hypothesis 2 No difference will be found in averages of the dependent variable scores between the family treatment group and the adolescent treatment group. Symbolically: H02: T1 = T2 Where T1 = Family Program T 2 Adolescent Program Alternate Hypothesis: The average family treatment group scores on the dependent variables will not equal those of the adoles- cent treatment group. Hza‘ T1 I T2 Where T1 12 Adolescent Program Family Program 125 Data Analysis Two analyses were employed to determine if change took place as a result of treatment. A multivariate analysis of covariance (MANCOVA) was performed on the data obtained from the Family Inter- action Task, the Family Problem Behavior Scale, and the Tennessee Self-Concept Scale. The data obtained from the Drug and Delinquent Behavior Questionnaire were analyzed separately with a Multivariate Analysis of Covariance (MANCOVA) applied to the posttest scores, with pretest scores used as the covariates. There are three reasons the MANCOVA was chosen to test each of the two orthogonal contrasts and also the three scores from the Problem Behavior Questionnaire. First, the MANCOVA is a test of posttest dif- ferences between groups. This is the kind of test a treatment evalua- tion requires. Second, randomization could not be counted upon to provide equivalent groups because of the varying acceptance and dropout rates of each experimental group. Therefore, use of the pretest as a covariate was essential to control the confounding variables. Third, a univariate analysis repeated 12 times, once for each dependent variable, would result in a very high overall Type I error. A multi- variate analysis treats the collection of variables simultaneously and reduces the possibility that demonstrated differences are a result of Type I errors rather than treatment differences. This analysis does not consider the partial group treatment of families. The family is the unit of analysis, but obviously there is some family-by-group interaction that is not considered in this analy- sis. It was decided not to use the group as the unit of analysis for 126 two major reasons. First, only a small number of families could be treated at one time, and the evaluation needed to be completed within one year after the program's development. Second, the treatment groups were open-ended--families entered and left at any point. The attendance varied from week to week, reducing the effect of rela- tionships between participating families. However, analyzing families as units, rather than groups as units, will result in more liberal tests than indicated by the listed Type I error of .05 per test. In total, four statistical tests were planned for the posttest data, each at p_= .05, yielding an experimentwise error rate of .20. A summary of the analysis design can be found in Figure 4.3. Summar The effects of a short-term family treatment program for families with an acting-out adolescent were compared to the effects of an ado- lescent treatment program and a no-treatment control group. Thirty families, randomly assigned to three experimental groups using referral source as a blocking variable, were pretested on four measures, treated, and then posttested on the same measures. The measures used to evalu- ate change as a result oftreatment were: (1) the Family Interaction Task, (2) the Family Behavior Change Scale, (3) the Tennessee Self- Concept Scale, and (4) a Drug and Delinquent Behavior Questionnaire. The posttest data from the family and self-concept measures were analyzed with a multivariate analysis of covariance. The data from the Drug and Delinquent Behavior Questionnaire were also analyzed with a Multivariate Analysis of Covariance. 127 x2 auteuuonsano aoueuag Bums 'ua 'Avuaa 2 < 5 <2: (193 old) 4119;] 'AV 3? g (193 ‘OJd) uaai 'PI .; 3 .2 (193 old) 'aed 'Av “>3 O m a. OJ 5% (115) saetlnoad 1; E (113) suoudmsm 3 > *5 (113) aansolg <3 8 C Q) g: (5351) "4143-4165 C; (5351) °5°d '101 '; “x aaieuuonsano .10;/19an (198 'OJd) ms '41: ‘2 g (198 old) uaai “p1 '; _8, (193 'OJd) ’JPd 'AV «3 g E 3 (11:1) summed“: a. g; (m) suoudmsm 3 4..) E (11:!) aansow '2 '0 > 3 (5351) 'wa-llas s. (5351) 'SOd '10]. '; in o o O - I— N u— l- N r- v— N '- ml as us m m m in m m m Group 1 Group 2 Group 3 Data-analysis design. Figure 4.3. CHAPTER 5 RESULTS Two treatment programs for families with acting-out adolescents were described in Chapter 4. One program treated the entire family, whereas the other treated only the identified adolescent. Also described was a research strategy to evaluate the effects of each of the two programs by comparing them to each other and to a no-treatment control group. In this chapter the results of the evaluation are presented. Two hypotheses were tested initially, using three instruments to measure change. These instruments were: (1) the Family Inter- action Task, which yielded three variables; (2) the Family Behavior Change Scale, which yielded two variables; and (3) the Tennessee Self-Concept Scale, which yielded two variables. Each experimental hypothesis is presented next, followed by the analytic data used to evaluate the hypothesis. Following the presentation of the results of the first two hypothesis tests, data from the fourth measure, which was also given to subject families, are presented and explored. The Drug and Delin- quent Behavior Questionnaire, the fourth measure, was analyzed sepa- rately using a MANCOVA and is hence reported last. 128 129 Analysis of the Experimental Hypotheses Hypothesis 1 No difference will be found between the dependent variable scores of the control group and the average of the dependent variable scores of the two treatment groups. Stated symbolically: Data from the family and self-concept measures were used to evaluate this hypothesis. The mean and standard deviation of each experimental group on the nine family and self-concept posttest vari- ables are summarized in Table 5.1. Differences between groups were very small on all nine variables. The largest differences relative to the standard deviation occurred on the Tennessee Self-Concept Scale Total Positive score. 0n the Total Positive score the group means ranged from 34.67 for the control group adolescents to 41.0 for the family group adolescents. The posttest data were analyzed using a Multivariate Analysis of Covariance, MANCOVA, with eight pretest scores on these same variables used as covariates. There are eight covariates (pretest scores) and nine dependent variables (posttest scores) because one of the dependent variables, the FPB-Change Score, cannot be measured in a pretest. Figure 5.1 is a summary outline of the Multivariate Analysis of Covariance. 130 mcoum mmcmgouucow>mcmm Empnoca apwsmm mcoum apwsmm mmwcm>mgmm smpnoca apwsmm meoum compuueow>mcwm smpnoca »FFEwm mcoum ucmcma mmmcw>msmm empnoca armsmu acoum cmwpsowa--xmmp cowuomcmucH szEmu mcoum compazcmwonuxmmh cowuumcwpca xpwEmm ocoum ocamo—o-uxmmp cowuomcmucfi xmemm mgoum Empuwuwcuuepmmunmpmum unaccouucpmm mommmccwh mcoum m>_uwmoa Papopnumpmum uamucou-mpmm mommmccm» mum a>< mma came mm; a>< «ma some eHa taco PHE mopu 4H; um meme much mom» "sax em.o_ somm._ mmNP.F NmN.P homo. 4m__. wmao. soo.~ N4.o_ .m op mu - Eaemoca ooe.~ oo_.m oom.N oom.m NN_o. Pmmm. mmwo. oa.mm oo._e .M - s_csaa mmo.m macs. PNN._ 44mm. coco. meow. News. ask.“ Neo.m .m N swamoaa . . . . . . op w u pcmumwpou< mom.4 mNN.m Pmo.m com m coco comm ammo NA 4m mm mm .x mm._F _Kom. comm. mmmm. Laos. NNON. mmmo. em_.m moo.m .m o. _w u _oca=ou ooo.m Noo.m oco.m ooo.4 mmoc. mmmm. ammo. om.mm em.am .m a>< came a>< some amen mops um asap .1 mum ama mma «ma FHE PHL HHE mum» meme c azozw mpamwcm> :umu co azocw gumm we :oPumw>ma ugmucmum ecu cum: ammuumoa u_.m mpnmh .mucmPcm>ou co mwmxpmc< mpmwcc>_p~:z any we mcwppzo Newsaam ._.m mg=m_m 131 N -N-_ .N -Tlllv- L o u H h . o: o 1 NF + FF u 0: op . m . Emcmoca . a - s_PEaa N P op H N Emcmogm N o u pcmommpou< F op H Pu n Focucou N _ m> w> N> m> m> ¢> m> N> _> m> N> m> m> ¢> m> N> F> .M m “ummuumoav mmpnmwcm> “newsmamo Aummpmcav mmpmwcm>ou 132 The Null Hypothesis 1 (C = Il_;_I§.) could not be rejected, E_= .6450; df_= 9,11; p_< .7402. To gain an understanding of how dif- ferent the three experimental groups were on the pretest variables, and how these differences might have affected their outcome scores, an examination of the covariates was undertaken. This question is especially important because of the possible selection-by-treatment interaction described in Chapter 3 for the family treatment group. A regression analysis was used to discover the amount of the variance accounted for in each dependent variable by the covariates as a group. The highest association was found for variable 1, TSCS--Tota1 Positive Score. The collection of covariates accounted for 52.31% of the variance on this measure. The probability of this result occurring by chance was less than .0415. The second highest association was found for variable 4, FIT-—Disruption Score. The collection of covariates accounted for 43.03% of the variance on this measure. The probability of this result occurring by chance was less than .1411. The statistics for the regression analysis using all eight covari- ates simultaneously are shown in Table 5.2. (It is important to note that these figures are most likely inflated due to the high ratio of variables to subjects.) Very few of the covariates accounted for a significant amount of the variance. Therefore, it seems possible that using all nine covari- ates did not improve the sensitivity of the analysis to possible real differences between experimental groups. To test this possibility, an f_te5t of the hypothesis of no association between dependent 133 variables and covariates was carried out. The null hypothesis could not be rejected, f_= 1.2360; df.= 72,74; p_< .1829. Because the covariates as a group did not account for a significant amount of the variance in the dependent variables, they were dropped from the analysis. A multivariate analysis of variance (MANOVA) was then applied to the posttest scores in a retest of Hypothesis 1: Is there a difference between the posttest scores of the control group and the average posttest scores of the two treatment groups? Again, the Null Hypothesis 1 of no difference could not be rejected, E_= .7149; df_= 9,19; p_< .6896. Table 5.2: Statistics for Regression Analysis Using Eight Covariates Simultaneously Variable Mungglg 3’ Multiple R_ [_ p_less than TSCS-TotP .5231 .7233 2.6050 .0415 TSCS-SC .3431 .5858 1.2405 .3296 FIT-Clos .1924 .4386 .5658 .7927 FIT-Disr .4303 .6560 1.7937 .1411 FIT-Pecu .3580 .5983 1.3242 .2906 PBR-AvP .1529 .3910 .4285 .8894 PBR-Teen .3867 .6218 1.4972 .2231 PBR-AvF .1413 .3759 .3909 .9121 F05 .3400 .5831 1.2236 .3380 Key: TSCS-TotP Tennessee Self-Concept Scale--Total Positive Score TSCS-SC = Tennessee Self-Concept Scale--Self—Criticism Score FIT-C105 = Family Interaction Task--Closure Score FIT-Disr = Family Interaction Task--Disruption Score FIT-Pecu = Family Interaction Task--Peculiar Score PBR-AvP = Family Problem Behavior--Average Parent Score PBR-Teen = Family Problem Behavior--Teen Score PBR-AvF = Family Problem Behavior--Average Family Score FCS = Family Problem Behavior--Change Score 134 Consistent with the hypothesis of no association between the independent and dependent variables, the probability at which the reported posttest results would occur by chance dropped from .7402 found with the MANCOVA to .6896 with the MANOVA. However, the find- ings remained nonsignificant. To further an understanding of any group differences on the pre- test, pretest group means were examined. Only two of the eight vari- ables demonstrated differences between groups which were greater than 1 standard deviation. These two variables were the Disruption score of the Family Interaction Task and the Average Family score of the Problem Behavior Rating. Even on these two scales, the differences were quite small, indicating little or no systematic difference between the groups on the pretest. The means and standard devia- tions of each group on each pretest variable are summarized in Table 5.3. Additional exploration of the data. At this point, it was clear that a significant difference between the control group and the average of the two treatment groups could not be demonstrated using the nine outcome variables simultaneously. Perhaps the variables were clustering in such away as to obscure any real differences. To examine this possibility, correlations between all dependent variables for the total sample were computed using Pearson's [, These correlations are presented in Table 5.4. Next, using a procedure developed by McQuitty (1957), a linkage analysis was performed. The dependent variables clustered into three groups. The first cluster comprised the three Family Problem Behavior Rating scales. The 135 mgoum xmemm mmmcm>mgmm smpnocm apwsmu u m>< «ma mcoum cmmhuucow>mgwm sm_noca waEmm u :mwh mm; mcoum “coca; mmmcm>mgmm Empaoca apwEmu u a>< mma mcoum campsuma11xmmh cowuumcmuch APP5mm u some AH; mcoum comuazgmwouuxmmh cowpumemch »FPEmu u cm_o emu mcoom mesmopu--xmmh cowgomcmpcm zFPsmu u mopo Ham «Loom Emwuwpwcuue—mmuumpmum unaccoulmpmm mommmccmh u um mum» mcoum m>wuwmoa panop-1mpmum uamucouumpmm ommmmccmh muck womb "xox ”Rum. omem. momm. some. Nmmm. noko. op.“ ww.m .m ma - smcmoca . . . . . . . . o_ - spee~a com 4 oo_ 4 omm 4 mFFo comm memo m Pm m Am .m omen. Npmm. mmmm. muse. m___. mews. mm.o mm.m .m o_ Na - sacmoca . . . . . . . . - pcmummpoc< map a mom m moo a m_oo mmaP _mmo A mm on mm .m o . . . o o . . WI moon 44° F o_4a memo mm_P .mmo cm a me m o, _o u _ocp=ou mwm.m 4m4.m ~N~.4 m.Po. mmmw. some. o..m um._m .m a>< camp a>< some Lawn mo_u um asap 1. gas am; mma «ma PHG ecu cHa mumc meme c ca mpnmwcm> scan :0 azocw comm eo cowamw>mo ucmucm“m new cum: pmmuwga "m.m mpamh 136 mcoom mmcmsuu-com>mgmm Empnoca zF_Emm u mum wcoom xpwsmu mmmcm>msmm Em—noca zpwsmm u m>< mma mcoum compulcop>mswm Empaoca apwsmu u some mma mcoum pcmcua mmmgm><-ucom>m;mm Empaoca zp_smm u a>< mm; mcoum Law—:uwaunxmmh cowuomcmucH apPEmm u some emu mcoum cowpgacmwn--xmmh :owpumcmuca »P_Emm u amps AH; mcoum acamopu-1xmmh cowaumemucu APPEmm u mopu Hum mcoum Emwuwuweuuwpmmunw—oum unmucouucpwm mommmccmh u um mump mcoum m>wuwmom quopuuopmum aawucounwpmm mommmccmh u much womb "Aug ... memo.- owe..- emm_.- NNoo. Nenp. Nose. mpoo.- home. mom momo.u ... mews. comm. mmNo. oeeN. Noon. FRON.- mmu_.- . m>< «ma omo_.- mam“. ... Nmmm. msvo. m_N_. «mow. mmmN.- mmFN.- :mm» mm; emm~.- ommm. nomm. ... cepo.- NNFN. Pomp. emoo.- nmmp.- m>< mm; NNoo. mmNo. mNeo. wepo.- ... mmmp. mNm—. NmNo. Ncmp.u :uma PH; Nemp. oeVN. mFNF. Nm—N. Romp. ... come. memo.u mNFN. cmpn EH; Nose. Room. «mow. FOMF. mNoF. came. ... mmum.- mmwo. mopu pH; m_mo.- FRON.- mmmN.- emoo.- Ammo. memo.- mmnm.- ... oueo. om mum» Name. mmmp.- omFN.- Nmm_.- N¢m~.- mNFN. mmmo. oNeo. ... much mumh mum u>< cow» m>< some cmwo moFu um ape» mma mma mma hum pH; pH; mum» womb mmpnmwcm> ucmucmamo _P< cmmzpmm mcomumpmccou ue.m mpnmh 137 correlations in this cluster were .79 and .84. The second cluster comprised the three Family Interaction Task scales plus the Tennessee Self-Concept Scale--Self—Criticism Score. The correlations in this cluster ranged from .17 to .50. The third cluster comprised the Tennessee Self-Concept Scale--Tota1 Positive Score and the Family Change Score. This correlation was .48. The configuration of the clusters and their correlations are summarized in Figure 5.2. CIUStET‘ 1 PBR-AVP \/ Y‘ = ‘84 \) PBR-AvF V / ‘$ r = .79 PBR-Teen Cluster 2 FIT-Djsr <:' r = ’50 ‘3) FIT-Clos r = .17 $ $1 r = ”'37 FIT-Pecu TSCS-SC r = .48 912332541 TSCS-TotP (K IEE> FCS Key: PBR AvP Family Problem Behavior--Average Parent Score PBR AvF = Family Problem Behavior--Average Family Score PBR Teen = Family Problem Behavior--Teen core FIT Disr = Family Interaction Task--Disruption Score FIT Pecu = Family Interaction Task--Peculiar Score FIT Clos = Family Interaction Task--Closure Score TSCS SC = Tennessee Self-Concept Scale--Se1f-Criticism Score TSCS TotP = Tennessee Self-Concept Scale--Total Positive Score FCS = Family Problem Behavior--Change Score Figure 5.2. Dependent variable clusters and their correlations. 138 The question at this point was: Is there a difference between the posttest scores of the control group and the average posttest scores of the two treatment groups for each of the three groups of clustered variables? Because it was already known that the use of covariates might not improve the sensitivity of the analysis, their usefulness was tested for each cluster. The covariates used for each cluster were the pretest scores on the measures in that cluster, with the exception of the Family Change Score, for which there was no pre- test score. For example, Cluster 1 dependent variables were the three posttest scores derived from the Family Problem Behavior Rating Scale. Accordingly, the covariates tested with this cluster were the three pretest scores derived from the Family Problem Behavior Rating Scale. The covariates in Clusters l and 2 did not remove a significant amount of the variance of the dependent variables. Cluster 1 test of the significance of the three covariates yielded E_= .9490; df_= 9,53.7; p_< .4917. Cluster 2 test of the significance of the four covariates yielded f_= 1.1896; df_= 16,61.7; p_< .3017. For each of the first two clusters, a Multivariate Analysis of Variance (MANOVA) was per- formed on the null hypothesis of no difference between the control group score and the average score of the two treatment groups. For Cluster 1 (the three Family Problem Behavior Rating scores), the null hypothesis could not be rejected, §_= .2774; df.= 3,22; p_< .8412. For Cluster 2 (the three Family Interaction Task scores and the TSCS-- Self-Criticism score), the null hypothesis could not be rejected either, f_= .5959; dj_= 4,20; g_< .6698. 139 Cluster 3 consisted of the Tennessee Self-Concept Scale--Total Positive score and the Family Change Score. The one covariate in Cluster 3 (TSCS--Total Positive score) did remove a significant amount -of the variance in the dependent variable. The test of the signifi- cance of the covariate yielded E_= 7.3406; dj_= 2,25; p_< .0032. Accordingly,aaMultivariate Analysis of Covariance (MANCOVA) was per- formed on the dependent variables of Cluster 3 to test the null hypothesis. The null hypothesis could not be rejected, E_= .0926; df_= 2,25; p_< .9119. A summary of the results of the test of Hypothesis 1 using the clustered variables is presented in Table 5.5. None of the three clusters of variables detected any difference between the control group score and the average score of the two treatment groups. Hypothesis 2 No difference will be found on the dependent variable scores between the Family Treatment Group and the Adolescent Treatment Group. Stated symbolically: Ti‘Tz Data from the family and self-concept measures were used to evaluate this hypothesis. The mean and standard deviation of each experimental group on the nine family and self-concept posttest measures are summarized in Table 5.1. These posttest data were analyzed using a Multivariate Analysis of Covariance, MANCOVA, with eight pretest scores on these same vari- ables used as covariates. Figure 5.1 is a summary outline of the Table 5.5: Hypothesis 1 140 Summary of Results of the Clustered Dependent Variables, Cluster 1 Variables Included Test of Covariates Test of Hypotheses MANOVA Cluster 2 Variables Included Test of Covariates Test of Hypotheses MANOVA Cluster 3 Variables Included Test of Covariates Test of Hypotheses MANCOVA Problem Behavior Rating: Teen, and Average Family Average Parent, E = .9490; d: = 9, 53.6928; p_ < .4917 5 = .2774; g: 3, 22; p < .8412 Family Interaction Task: Closure, Peculiar, and Disruption; Tennessee Self-Concept Scale-~Self-Criticism _F_=1.1896;_Cfl=16, 61.7386; 2 < .3017 H0]: 5 = .5959; 511 = 4. 20; p < .6698 Tennessee Self-Concept Scale: Total Positive and Family Change Score E= 7.3406: g: = 2. 25; p_< .0032 Ho]: _F_= .0926; df_= 2.25; p< .9119 141 Multivariate Analysis of Covariance. The null hypothesis could not be rejected, [_= 2.3482; df_= 9,11; p_< .0917. Using the information generated when testing Hypothesis 1, that the covariates as a group did not account for a significant amount of the variance, a Multivariate Analysis of Variance (MANOVA) was used to retest Hypothesis 2. The null hypothesis of no difference between the scores of the two treatment groups could not be rejected, E_= 1.3340; df_= 9,19; p_< .2841. Contrary to the hypothesis of no association between the dependent and independent variables, the probability at which the reported posttest results would occur by - chance rose from .0917 found with the MANCOVA to .2841 with the MANOVA. It appears that the covariates did extract a small but important por- tion of the variance. However, the outcomes of both the MANCOVA and MANOVA tests failed to result in the rejection of the null hypothesis. Additional exploration of the data. Using the variable clusters discovered during the test of Hypothesis 1 and summarized in Figure 5.2, the hypothesis of no difference between the two treatment groups was tested using the scores on the variables in each of the three clusters. The hypothesis was first tested with a MANOVA using the three variables in Cluster 1, the Family Problem Behavior Rating Scores. The MANOVA was used because the covariates associated with these dependent variables did not remove a significant amount of the vari- ance. The hypothesis of no difference could not be rejected, 5= .9354; d_f_ = 3.22; p_ < .4404. 142 The hypothesis of no difference between treatment groups was then tested with a MANOVA using the variables in Cluster 2, the Family Interaction Task scores and the Tennessee Self-Concept Scale--Self— Criticism score. Again, the MANOVA was chosen because the covariates associated with these variables did not remove a significant amount of the variance. The hypothesis of no difference between treatment groups could not be rejected, [_= 1.6671; df_= 4,20; p_< .1970. The final test of the hypothesis was performed with a MANCOVA using the variables in Cluster 3, the Tennessee Self-Concept Scale—- Total Positive score and the Family Change score. The MANOCVA was chosen because the test of no association between the independent variables (the covariate) and the dependent variables (the posttest score) was rejected at the p_< .0032 level. The hypothesis of no difference between the two treatment gr0ups again could not be rejected, [_= .5122; df_= 2,25; p_< .6054. A summary of the results of the test of Hypothesis 2 using the clustered variables is presented in Table 5.6. None of the three clusters of variables detected any difference between the two treatment groups. Hypothesis 3 Hypothesis 3 tested the difference between the dependent variable scores of the control group and the average of the dependent variable scores of the two treatment groups. H03: C =.Il;:;BE 2 143 Table 5.6: Summary of Results of the Clustered Dependent Variables, Hypothesis 2 Cluster 1 Variables Included Test of Covariates Test of Hypotheses MANOVA Cluster 2 Variables Included Test of Covariates Test of Hypotheses MANOVA Cluster 3 Variables Included Test of Covariates Test of Hypotheses MANCOVA Problem Behavior Rating: Average Parent, Teen, and Average Family 5 = .9490; d: = 9, 53.6928; 2 < .4917 H02: 5 = .9354; gy= 3, 22; p< .4404 Family Interaction Task: Closure, Peculiar, and Disruption; Tennessee Self-Concept Scale--Se1f—Criticism _F_=1.1896;d_f=16, 61.7386; p< .3017 H02: _F_= 1.6671; 93:: 4, 20; p_< .1970 Tennessee Self-Concept Scale-~Tota1 Positive and Family Change Score f_ = 7.3406; df = 2, 25; p < .0032 H02: 5 = .5122;a_1== 2, 25; p_< .6054 144 The mean and standard deviation of each group on the posttest for each of the three drug and delinquent behavior scales are sum- marized in Table 5.7. Group means on the Drug Use scale varied from 38.6 to 45.1, with a standard deviation varying from 14.8 to 35.0. Group means on the Status Delinquent Behavior scale varied from 19.0 to 48.0, with the standard deviation varying from 14.8 to 35.0. Group means on the Non-Status Delinquent Behavior scale varied from 1.1 to 7.8, with the standard deviation varying from 2.4 to 13.8. Table 5.7: Mean and Standard Deviation of Each Group on the Scales of the Drug and Delinquent Behavior Questionnaire Status Non-Status 3- Drug Use Delinquency Delinquency 7' 40.6 19.0 7.8 Control = G] 10 §_ 43.4 22.5 13.8 Adolescent = G 10 X’ 45.1 48.0 1.6 Program 2 §_ 36.6 35.0 3.7 Family = G 10 )T 38.6 24.0 1.1 Program 3 §_ 20.8 14.8 2.4 The posttest data were analyzed using a Multivariate Analysis of Variance (MANCOVA) with the pretest data for the same variables used as covariates. The Null Hypothesis 3 could not be rejected, f_= 1.3672; fl= 3, 22; p < .2789, 145 Hypothesis 4 Hypothesis 4 tested the difference between the two treatment groups. The mean and standard deviation of each experimental group on the posttest for each of the three drug and delinquent behavior scales are summarized in Table 5.7. The posttest data were analyzed using a Multivariate Analysis of Covariance (MANCOVA), with the pretest data for the same variables used as covariates. The Null Hypothesis 4, T1 = 12, could not be rejected, E_= 1.5036; df_= 3,22; p_< .2414. Summary of Research Results Two research hypotheses were tested using a Multivariate Analysis of Covariance (MANCOVA) on the nine dependent variables derived from three family and self-concept measures. Null Hypothesis 1, n0 differ- ence will be found between the dependent variable scores of the con- trol group and the average of the dependent variable scores of the two treatment groups, could not be rejected, §_= .6450; df_= 9,11; p_< .7402. Null Hypothesis 2, no difference will be found on the dependent variable scores between the family treatment group and the adolescent treatment group, could not be rejected, [_= 2.3482; df.= 9,11; p_< .0917. Further exploration of the data was undertaken into the cluster- ing patterns of the dependent variables because of the possibility that the way the variables clustered was obscuring some real 146 differences between groups. The dependent variables fell into three nonoverlapping clusters. Cluster 1 was made up of the three Family Problem Behavior Rating scales. Cluster 2 comprised three Family Interaction Task scales and the Tennessee Self-Concept Scale--Se1f— Criticism score. Cluster 3 was made up of the Tennessee Self-Concept Scale--T0tal Positive score and the Family Change score. Multivariate analyses of the variables in each cluster were then carried out on both null hypotheses: (1) No difference between scores of the control group and the average of scores of the two treatment groups, (2) No difference between the scores of the two treatment groups. In all cases, the null hypotheses could not be rejected. All analyses failed to demonstrate any difference between the control group score and the average score of the two treatment groups. They also failed to demonstrate any difference between the two treatment groups. Furthermore, the nonsignificant differences found between groups were uniformly small, indicating that the prob- lem was not lack of power. The fourth and final measurement instrument used to assess change as a result of treatment was the Drug and Delinquent Behavior Ques- tionnaire. Responses from this questionnaire were also analyzed using a Multivariate Analysis of Covariance (MANCOVA). No differences were found between the three experimental groups on the posttest responses to the questionnaire. In summary, none of the data supported the claim that either of the two treatment programs produced positive change in the partici- pants. 147 Because of the additional statistical tests applied to the data following the failure to reject the null hypothesis, total experiment- wise error rate rose from the planned .20 to .65. At this rate, .65 of the 13 tests of significance performed would be expected to be significant by chance alone. None of the 13 tests showed any sig- nificant differences. CHAPTER 6 SUMMARY AND CONCLUSIONS The purpose of this study was to examine, by experimental design, the clinical effectiveness of two treatment programs offered by a paraprofessional community service agency to families with an acting- out adolescent. The results of this evaluation raised interesting questions regarding both program evaluation strategy and program effectiveness. This chapter includes a brief summary of literature relating to families with acting-out adolescents, a summary of this research project, and a discussion of issues raised by this study as well as implications for further research. Brief Literature Summary Adolescent Family Problems Many problems in adolescent families have been seen as the result of stress related to rapid social change. Increasingly, the family is required to take more responsibility for meeting the physical and emotional needs of its members; simultaneously, the family's resources with which to meet those needs are decreasing. Bronfenbrenner (1970) stated that the decline in intergenerational communication has been largely responsible for the failure of the modern family to meet the challenges of its new social world. 148 149 Family Interaction Studies Studies examining interaction differences in normal adolescent families and in families with a delinquent adolescent were reviewed. It was found that normal adolescent family interaction differs in systematic ways from dysfunctional adolescent family interaction. Ferreira and Winter (1965) found that normal families were better at making joint decisions than were delinquent families. Normal families evidenced more initial agreement on problem solutions than delinquent families and took less time to resolve their existing differences than did delinquent families. In a similar study, Mead and Campbell (1972) independently demonstrated the same phenomenon. Alexander (1973b) found that families with a delinquent adolescent demonstrate a higher rate of defensive communication than normal families and that defensive communication is reciproCal in delinquent families whereas it is not reciprocal in normal families. Alexander also found that delinquent families demonstrate a lower rate of supportive com- munication than normal families and that supportive communication is not reciprocal in delinquent families, whereas it is reciprocal in normal families. Hetherington, Stouwie, and Ridberg (1971) found that normal families express less total conflict, less hostility, more warmth, and that parents were less power assertive than families with delin- quent children. Parents in normal families were more willing to listen to their children than were parents in delinquent families. What emerged from these several studies of family interaction was the conclusion that normal families are different from delinquent 150 families on both static agreement and satisfaction measures, as well as on their methods of solving problems. Family Treatment Studies Studies evaluating the effectiveness of various adolescent family treatment programs were also reviewed. For three of the studies, major methodological and/0r statistical flaws challenged the positive evaluation that each researcher gave his/her treatment program(s). Three of the treatment programs reviewed treated each family separately from other families. Parsons and Alexander (1973) studied families from a juvenile court population. They provided a brief series of sessions in which the family was taught to (1) differen- tiate rules from requests, (2) develop reciprocity in the exchange of positive responses, and (3) increase the variability of family commu- nication patterns. An interesting concept presented in this study was the "matching-to-sample" philosophy, which assumes that if a given delinquent family can change its interaction patterns to be more simi- lar to those of normal families, the delinquent family will begin to behave more like normal families in other areas; i.e., the delinquent teenager will stop unacceptable behavior. Underlying this philosophy is a belief in the healing powers of the family system itself. Despite minor deviations from accepted research design, Parsons and Alexander were able to find differences between their treatment and control groups. Their treatment did produce change in the desired direction. Katz, Krasinski, Philip, and Wieser (1975) provided an unspecified form of brief family therapy and discovered improvement in treatment 151 families not found in control families, as measured by many variables. Again, the repeated use of a univariate statistic resulted in a very high error estimate (721%), which guaranteed that they would find "significant" differences. Beal and Duckro (1977) measured only one variable-—the rate of successful diversion of juvenile court families from the official adjudication procedure. They provided eight conjoint family therapy sessions aimed at resolving through interpersonal change whatever crisis had sent the family to court. They demonstrated a diversion rate that was higher than that achieved by the traditional probation officer approach (p < .05). The design and analysis of this study were appr0priate for its goal. However, the control group to which the treatment group was compared was not equivalent, and the results attributed by the researcher solely to treatment effect were possibly the result of an interaction between selection and treatment. The final two treatment programs reviewed provided family group experiences. Stanley (1978) took normal volunteer families from a school and provided a lO-week structured course addressing both family communication skills and attitudes. The program focused on conflict resolution and rule setting, using active listening skills as the major tool. She found change on both communication and attitudinal variables. As did most other treatment evaluaters, Stanley repeatedly used a univariate statistic to analyze a number of variables separately. Her high error estimate (62%) throws a considerable shadow on the interpretation of significant findings. 152 LEE§22.(1972) also worked with volunteer parents from a school setting and attempted to change both communication patterns and atti- tudinal variables. He provided three different treatment programs and compared them all to a no-treatment control group. Parent Effec- tiveness Training (PET) was offered to one group. This program teaches the use of listening and problem-solving skills to resolve parent-child conflicts. An Achievement Motivation Program (AMP) was offered to a second group. This program focused on identifying the participants' personal strengths, and helped them to build on those strengths. The last treatment program was a Discussion-Encounter Group (DEG) experience and consisted in increasing the awareness and the sharing of feelings about such controversial parent-child topics as dating, dress, and manners. Larsen presented group means (but not standard deviations) and concluded that both PET and AMP were effec- tive in producing desired change. His data were not analyzed statis- tically, thereby making it impossible to determine whether the differences found were significant or not. Current Research Study_and Results The effects of a short-term family treatment program for families with an acting-out adolescent were compared to the effects of an ado- lescent treatment program and a no-treatment control group. Thirty families, randomly assigned to three experimental groups using refer- ral source as a blocking variable, were pretested on four measures, treated, and then posttested on the same measures. The measures used to evaluate change as a result of treatment were (1) the Family 153 Interaction Task (FIT), (2) the Family Problem Behavior Scale (FPBS), (3) the Tennessee Self-Concept Scale (TSCS), and (4) a Drug and Delinquent Behavior Questionnaire. The posttest data from the family and self-concept measures were analyzed with a multivariate analysis of covariance. The data from the behavior questionnaire were also analyzed with a Multivariate Analysis of Covariance. Two research hypotheses were tested using a Multivariate Analysis of Covariance on the collection of nine dependent variables derived from three family and self-concept measures. Null Hypothesis 1, no difference will be found on the dependent variables between the con- trol group and the average of the two treatment groups, could not be rejected, f.= .6450; df_= 9,11; p_< .7402. Null Hypothesis 2, no difference will be found on the dependent variables between the two treatment groups, could not be rejected, [_= 2.3482; df_= 9,11; p_< .0917. Further exploration into the clustering patterns of the dependent variables was undertaken, because of the possibility that the way the variables clustered was obscuring some real differences between groups. The dependent variables fell into three nonoverlapping clusters: Cluster 1 was made up of the three Family Problem Behavior (FPB) rating scales (Average Parent, Teen, and Average Family). Cluster 2 comprised the three Family Interaction Task (FIT) scales (Closure, Disruption, and Peculiar) and the Tennessee Self-Concept Scale (TSCS)-- Self-Criticism score. Cluster 3 was made up of the TSCS--Tota1 Posi- tive score and the FPB-Change score. Multivariate analyses of the variables in each cluster were then carried out on both null hypotheses: 154 (1) no difference between scores of the control group and the average scores of the two treatment groups and (2) no difference between the scores of the two treatment groups. In all cases, the null hypotheses could not be rejected. All multivariate analyses failed to demonstrate any difference between the control group score and the average score of the two treatment groups. They also failed to demonstrate any difference between the two treatment groups. The fourth and final measurement instrument used to assess change as a result of treatment was the Drug and Delinquent Behavior Ques- tionnaire. Responses from this questionnaire were analyzed with a Multivariate Analysis of Covariance. No differences were found among the three experimental groups on posttest responses to the question- naire. In summary, none of the data supported the claim that either of the two treatment programs produced positive change in the partici- pants. Discussion During the course of the literature review, it became apparent that some of the reviewed clinical evaluation research had been done with poorly controlled methodology and improper statistical analyses. The goal of the present study was to evaluate two treatment programs using standard conventions of research methodology and appropriate analyses. It was hoped that whatever positive results were found 155 could be clearly attributed to the independent variables and not to statistical artifacts or sloppy design. During the course of this study, certain concessions were made to the reality imposed by research in a community (as opposed to a laboratory) setting; subject families were volunteers, making selec- tion an independent variable in the research design. The effects of selection and its impact on interpretation of results are discussed in this section. Next discussed is the importance of using a statistical analysis consistent with research design and some consequences of inappropriate analyses. The third area of discussion is a possible explanation of the cluster patterns of the dependent variable. Finally, the fact that no difference was found between the treatment groups is discussed. What was (or might have been) respon- sible for this finding? Selection as an Independent Variable Families who were referred to the experimental programs for treatment were not a random sample of families with an acting-out adolescent. They were families who, for unknown reasons, were referred by involved mental health professionals, or families who requested treatment themselves. Clearly, the discretionary nature of this process made selection a critical problem in this study which threatened external validity. A second selection factor was that not all referred families accepted the treatment program to which they 156 were assigned. The initial strategy developed to deal with the problem of families not accepting their treatment assignment was to pretest all families referred for treatment, whether they accepted treatment or not. This would have allowed any systematic difference between "not-accept" families and "accept" families to be explored. However, most families not accepting treatment also refused to be tested. Therefore, whatever differences existed between the two groups of families remain unknown, and the effects of treatment were contaminated in two ways by the selection process. Further adding to this confounding situation was the fact that one of the experimen- tal groups (the Family Treatment Program) had a higher "not-accept" rate than either of the other two groups. This situation might have been due to the parental participation requirement present in the Family Program and not present in the Adolescent Program and the Control Group. If this were 50, families in the Family Program might collectively be different from families in the other two groups. Chi Square analyses of the pretest measures failed to disclose any sig- nificant differences between groups. The extent of contribution of the selection process remains unanswered, and any significant find- ings would have been interpreted as resulting from a selection-by- treatment interaction. The original (but discarded) design for this study, which required pretests for all referred families, would have allowed some of the effects of the independent variable "selection" to be assessed. Selection, as represented by volunteer subjects, is a part of all human research in nonlaboratory settings, and is not explicitly 157 discussed in many research reports, leaving the reader with the impres- sion that whatever results were found were solely a result of the effects of treatment. This commonplace misrepresentation of find- ings can lead to problems. For instance, a probate court judge, unschooled in research methodology, might order a family into treat- ment assuming that a program of proven effectiveness will change the family. The absent factor of willing acceptance could alter the out- come. The findings of Beal and Duckro's study (1977), which was reviewed in Chapter 2, could lead a judge to believe that any family ordered to take this program would have the same chance to be suc- cessfully diverted from the legal system as did the volunteer subject families. That judge might be very disappointed should he/she write the order for treatment and find that families did not change as expected. This disappointment--that the results of therapy do not match the expectations-~frequently leads to disillusionment with the entire process of counseling. Researchers would do better to claim less, and be able to stand behind their claims. Consistency of Analysis and Design Also contributing to the inflated number of treatment programs improperly claiming treatment effectiveness was the practice of repeatedly using a univariate statistic to analyze many variables. The complexity of family interaction requires that many variables be examined in an attempt to understand behavior and behavior change. When analyzing these variables, a multivariate statistic is required. Such a statistic treats the many variables simultaneously, eliminating 158 the problem of very low confidence in resulting findings. Multi- variate statistics are easily applied to computer-analyzed data, and clinical researchers' continued use of repeated univariate analyses results in erroneous conclusions. Every treatment evaluation study reviewed in Chapter 2 claimed treatment effectiveness, but only five claims were supported by the results of the statistical analysis. Cluster Patterns of Dependent Variables Data derived from the application of McQuitty's (1957) linkage analysis to Pearson correlations between the dependent variables yielded three clusters of variables. Cluster 1 was made up of the three Family Problem Behavior (FPB) rating scales (Average Parent, Teen, and Average Family). Cluster 2 comprised the three Family Interaction Task (FIT) scales (Closure, Disruption, and Peculiar) and the Tennessee Self-Concept Scale (TSCS)--Self—Criticism score. Cluster 3 was made up of the TSCS-~Total Positive score and the FPB-- Change score. All of the largest correlations between measures were positive, with the exception of the correlation between the TSCS-- Self-Criticism scale and the FIT--Closure scale. It has been well established that an individual's scores on a number of variables obtained by a particular method of measurement are affected by his/her response to the characteristics of that method as well as by his/her place on the continuum the test intends to measure (Magnusson, 1967). Because of this dual determination of scores, the cluster patterns obtained on the dependent variables might be a result of underlying 159 differences in the constructs measured by each test or of differences in the method of testing. Because the highest correlations generally existed between dif- ferent scales of the same instrument, it is likely that the greatest amount of variance on each scale was attributable to measurement method. However, data supported the attribution of some of the vari- ance on each scale to the construct measured. The data from the Family Problem Behavior scale supported the notion that the greatest amount of variance could be attributed to the method of measurement. The FPB yielded four scores; three of the scores were derived by means of individually completed questionnaires, and the fourth resulted from a family consensus arrived at through family discussion. The three questionnaire scores clustered together, but the consensus score (FPB--Change score) was clustered with the TSCS--Tota1 Positive scale. The TSCS--Total Positive scale was correlated -.2196 with the FPB--Teen score. This was the second highest correlation for the TSCS--Total Positive scale, its highest correlation being with the FPB--Change score (y_= .4797). The FPB--Teen score was an estimate .of adolescent dissatisfaction with the current family situation rela- tive to the presenting problem. The TSCS--Tota1 Positive scale was a measure of self-esteem, and included (according to the TSCS manual) items that measured how the adolescent viewed and evaluated his/her ‘behavior in the context of the family. If the constructs measured by the two tests were related, as they seemed to be, the scales measuring them ought to have been more highly correlated than they 160 were with scales measuring entirely different constructs. In part, the data bore this out, since TSCS--Tota1 Positive scale had its second highest correlation with the FPB--Teen score (§_= -.2196). This last correlation was not higher than those discovered between the various questionnaire scales of the FPB (§_ranging from .5367 to .8390), so although some of the variance can be attributed to con- struct similarity, the majority of the variance must be attributed to the characteristic features of the measurement method. Another finding that supported the conclusion that method of measurement accounts for more variance than construct measured was the correlation between the TSCS--Self-Criticism scale and the FIT-- Closure scale. The TSCS--Se1f—Criticism scale purports to measure the degree to which the adolescent is willing to be critical of him/ herself. The FIT--Closure scale measures the degree to which family members interfere with each other's opportunity to bring ideas to successful closure. Included as closure problems are disqualifica- tions of self or other (disparaging criticism). The constructs pur- portedly measured by these two methods seem quite similar and ought to be positively related. The relationship found was confusing in that the correlation between the TSCS--Se1f-Criticism scale and the FIT--Closure scale was the highest correlation for the TSCS--Self- Criticism scale, but it was in the direction opposite to that expected ([_= -.3738). Clearly, there is some relationship between the con- structs measured by these two scales, but the relationship is differ- ent than hypothesized. Adding to the confusion was the predicted higher correlation between the Closure and Disruption scales of the 161 FIT, and the lower correlation between the FIT Closure scale and Peculiar scale, which was not predicted. The FIT scales did seem to cluster, but not as tightly as the FPB questionnaire scales. In conclusion, the method of measurement seemed to account for the largest amount of the variance between scores, but the construct measured did account for some portion of the variance. No Difference--Why? The most obvious possible reason for the failure to find differ- ence among the three experimental groups following treatment is that the treatment programs were not effective in producing desired changes. Other family treatment programs have demonstrated their ability to behaviorally change families. The program evaluated by Parsons and Alexander (1973) as well as the program evaluated by Beal and Duckro (1977) worked with each family separately, in contrast to the partial group treatment program evaluated in this study. Perhaps families are less effectively treated in groups. Stanley (1978) treated families in groups and demonstrated attitudinal change as a result of treatment. The program evaluated by Stanley required 10 weekly sessions, each 2.5 hours long. Additionally, the program :required that each family hold and audiotape seven 1-hour-long family meetings (without group leaders) related to topics discussed during group meetings. In total, each family received 35 hours of a struc- tured program, in contrast to the maximum of 20 hours and minimum of 12 hours offered by the programs evaluated in this study. Perhaps more time is needed to change families than was offered during the treatment programs evaluated in this study. 162 Subjects were randomly sorted into groups using referral source as a blocking variable. The blocking variable was not considered in the analysis of the data, and the analysis may be more conserva- tive than the a level indicated (p_= .05 per test). To estimate the extent of the error, the between-block differences on one randomly selected dependent variable were examined. No difference could be demonstrated, using Chi Square, between the different referral groups on the variable Total P of the Tennessee Self-Concept Scale. It is likely that the blocking variable contributed nothing to the analysis of the data, and hence any tendency toward conservatism was very small or nonexistent. Another possible explanation for the findings is that the instru- ments used to measure change were not sensitive enough to the feelings and behaviors assessed to pick up any change that might have occurred. The exploration into possible sources of variance among subject scores across dependent variables led to the conclusion that the largest amount of the variance on most scores was attributable to the method of measurement, with a lesser amount of the variance attribut- able to real difference on the concepts measured. It is this real difference portion of the variance that will assess any meaningful change as a result of treatment. If this portion of the variance was relatively small, the instruments may have been proportionately insen- sitive to any change. Related to the sensitivity of the instruments is the issue of sample size: Perhaps 10 families per experimental group was too 163 small a number for the sensitivity level of the measurement instru- ments. A final possible reason for the failure to find differences between the experimental groups may be the analysis used. A series of t_tests would have allowed the use of directional hypothesis, which would have increased the power of the analysis. With attention to limiting the experimentwise error rate by setting a = .01 for each of the nine tests of the dependent variables, the use of a univariate statistic is acceptable. Another method of analyzing the data could have been the use of gain scores rather than the Analysis of Covari- ance. More degrees of freedom would have been allowed using this method, possibly resulting in positive findings. However, consider- ing the small differences between groups on the posttest, it is extremely unlikely that any significant difference exists between them. Statistical analyses can only uncover what is already there, not create it. lmplications for Future Research There is a definite need for research into the effectiveness of family treatment programs. If an evaluation study is designed to evaluate behavior objectively, using a statistical analysis to test the truth about (or significance of) an observed change, then the project should include both conventions of acceptable experimental design and statistical analyses appropriate to the design. Any com- promises made relating to design ought to be clearly delineated, along with their associated impact on the results. Such rigorous 164 studies are needed. Also needed are evaluations based on the subjec- tive experience of the participants. As Dean and Whyte (1958) indi- cated, the objective "truthfulness" of a person's statement is not the only important issue; also important is what a person's statement reveals about his/her perceived reality. To alter a person's per- ception of reality is to change that person. This study attempted to address both observed behavior and subjective experience. Given the tremendous time investment required for treatment of families, evaluation of family treatment programs should be kept as simple as possible while still remaining meaningful. This might be accomplished by limiting the number of experimental groups to two: one treatment group and one control group. In this way more families would be in each group, thereby increasing the chance of recognizing any real difference. A second way to increase the chance of recognizing any real dif- ference resulting from treatment is to limit treatment goals to a few goals that are readily defined and reliably measured. A final suggestion is that researchers interpret their findings in a way consistent with the limits of their design and analysis. Selection as an independent variable, behavior change inferred from attitude questionnaires, and the effects of a low level of confidence in results due to repeated use of univariate statistics all require discussion of how they affect interpretation of results. Related to the need for treatment evaluation research is the need for further research into the nature of existing family 165 interaction measures and the development of newer, more valid, and more reliable measures. Most current family interaction measures were developed by clinicians, based on constructs emerging from various theoretical perspectives. Most measures have been validated by the developers with selected narrow populations, usually including fami- lies with a schizophrenic member. Family measures normed on larger populations, and on populations with other than families having schizophrenia and nonstatus delinquency diagnoses, are needed. Observations on Clinical Effectiveness Research Many summaries of research evaluating the effectiveness of clini- cal intervention report significant findings that are not supported by the data. In effect, researchers were misrepresenting their data. As professionals, Clinical researchers cannot continue to tolerate the use of such unacceptable research strategy. Like the previously dis- cussed issue of selection-by-treatment interaction, the use of inap- pr0priate analytical tools often leads to false claims of effectiveness and increased public mistrust of the counseling process. APPENDICES 166 APPENDIX A PROBLEM BEHAVIOR RATING FORM 167 APPENDIX A PROBLEM BEHAVIOR RATING FROM l. How disruptive is this problem in your family? 1 2 3 4 5 not at all a little bit moderately quite a bit extremely much 2. How much does this problem interfere with your enjoyment of the family? l 2 3 4 5 not at all a little bit moderately quite a bit extremely much 3. How strongly do you wish that something would be done about this problem? 1 2 3 4 5 not at all a little bit moderately quite a bit extremely much 4. How serious is this problem in your family? l 2 3 4 5 not at all a little bit moderately quite a bit extremely much 168 APPENDIX B DRUG AND DELINQUENT BEHAVIOR QUESTIONNAIRE 169 II. APPENDIX B DRUG AND DELINQUENT BEHAVIOR QUESTIONNAIRE Directions Across the top of the next section of next page you will see listed several different kinds of drugs. Down the side of the page are questions. Each question goes with each drug. If the answer to the question is yes, put a check in the square. If the answer is no, or does not apply to you, leave the square blank. Example: A o :: ..— 2 v 1,3 0: ca 23 c S- U'I ‘- 3 a c Q) o s. w -'-> 'r- .— S— Q) 3 Q) C w- " O M m s: o- n. 3 s. c: s. .c a) '0- 'F' D. o 'U W m C m 3 _J D D Z _l I H l. Have you ever used / / / This question is really eight questions, and this person has at least once used wine, uppers, and marijuana. He has never used beer, liquor, uppers, downers, hallucinogens, heroin, or inhalents. If you have any questions about this form, please ask. Uppers = amphetamines, methamphetamines, speed, benzedrine, dexedrine, and methedrine Downers == sedatives, barbiturates, seconal, and phenobarbital Inhalents = Pam, glue, deodorants, hair Spray, etc. This part deals with your non-drug behavior. Just answer as well as you can, after thinking about what you have done this very last week. 170 OKOUDVOSUl-wa-H _J 171 Beer Wine Liquor Uppers Downers Marijuana LSD, PCP (THC) Heroin Inhalents . Have you ever used this drug? Used it in the last 6 months? Used it in the last month? Use it once a week? Use it 2-3 times a week? . Use it every day? Do you use l-2 each time? Do you use 3-4 each time? 1!) ’AI. Do you use 5 or more each time? L) . Why do you use this drug? (Pick only one answer for each drug you use.) a. To relieve tension, pressure or nervousness. b. To join in a group, or most of my friends use it. c. To get rid of a down or to forget problems. d. For fun, enjoyment or celebration. e. To feel sophisticated, modern or adult f. To understand myself, to expand my mind. ll. 12. I3. 172 Where do you get this drug from? a. b. c. d. e. (Choose only l.) parents buy it steal it deal it other If you buy it, where do you get the money from? (Choose only l.) a. b. c. d. parents work steal it other How high do you usually get? a. b. O 0. not high at all a little high . smashed . ozone Beer Wine Liquor Uppers Downers Marijuana LSD, PCP (THC) Heroin Inhalents I73 PART II l. During the past week, have you shoplifted anything? No Yes If yes, how many articles? What do you guess was their total value? $ 2. During the past month, have you shoplifted anything? No Yes If yes, how many articles? What do you guess was their total value? S 3. Have you in other ways stolen anything during this last week? No Yes If yes, what were they? What do you guess was their total value? $ 4. Have you in other ways stolen anything during this last month? No Yes If yes, what were they? What do you guess was their total value? $ 5. Have you broken into any buildings this last week? No Yes If yes, how many? -6. Have you broken into any buildings this last month? No Yes If yes, how many? 7. Have you run away from home during this last month? No Yes If yes, how many times? If you have run away from home, how many times in your life have you done so? ID. 174 During this last week have you stayed out later than you were allowed to? If yes, how many times? Usually, how late were you? Have you gotten into any really big verbal hassles with your parents this last week? If yes, about how many times? Have you gotten into any physical fights this last week? If yes, how many? With whom? No No No (Ex. friends, brother, neighbor, etc.) Yes Yes Yes 175 The following questionnaire form has recorded on it the scale-- drug use (D), status delinquent behavior (S) and non-status delinquent behavior (N)-—that each question or answer belongs to. Additionally, the number of points given for each response is indicated. To score the questionnaire, the sum of the number of points earned for each question is placed in the correct column to the right of the form. A total for each page is at the page bottom, and grand totals are indicated at the upper left corner of page l. These scoring columns, point totals, and summary boxes do not appear on the questionnaire when given to each teenage subject. The scoring data are added later, during the scoring process. I76 For Scoring Only D l. Have you ever used this drug? D 2. Used it in the last 6 months? D 3. Used it in the last month? D 4. Use it once a week? D 5. Use it 2-3 times a week? D 6. Use it every day? D 7. Do you use l-2 each time? D 8. Do you use 3-4 each time? D 9. Do you use 5 or more each time? DlD. Why do you use this drug? (Pick only one answer for each drug you use.) a. To relieve tension, pressure or nervousness. b. To join in a group, or most of my friends use it. c. To get rid of a down or to forget problems. d. For fun, enjoyment or celebration. e. To feel sophisticated modern or adult. f. To understand myself, to expand my mind. C? I F- as v c 2 ,g‘ 13 'E >. gm‘figgc'éSS'E 05-0)"? '9-v-LL.U)O S.- Q) SQJC'I- “0'5 0) CC‘D-ZS-DS-S QJ'I- w-QOfUWQJ: m3..l:>o2_.l:1:o-aDSN I I 2 3 3 I 3 3 3 I I 2 3 3 I 3 3 3 I I 2 3 3 I 3 3 3 I I 2 3 3 I 3 3 3 I I 2 3 3 I 3 3 3 I I 2 3 3 I 3 3 3 I I I I I I I I I 3 3 3 3 3 3 3 5 5 5 5 5 5 5 5 5 5 points eakh ‘~\ '8 3 lpointkach 3 3. as 5 paints eakh «E l point each 3? S 3poimtsea)ch 93 8 W 3 poihts I77 23 I P- a: V C '6 .. W 'P c a “F, ‘- >5 (0 m U c L- O r— S_ W S- : Q. c a) o U C o S. a) "'7 ‘F‘ !— [LC/Io ‘- 0) 3 a) C 'I— a O m 0) s: O" D. 3 S- D S- .: £§S§8§B§SDSN ll. Where do you get this drug _5 from? (Choose only l.) V, ‘5’ s e a, parents 5 points each ; a» 5 See #12 $_ 3 b. buy it &3 " U? o 0 >’ c c. steal it l0 points dac 'E E N o S d. deal it l0 points alac g‘.’ D 8 3.’ S e. other 5 points each ”’ 3 12. If you buy it, where do you 1: get the money from? (Choose 3.: only 1.) avg; 'U I: U) 44 ‘ (D 3 C a. parents , 5 p01nts each ‘5 u, ‘5’ 5 ‘“ 3U» b. work 5 points. each .3” -’6 ‘3 c f— 0 S- N c. steal it ”I PGIMS 3CD cu 493’ ‘5 13: s d. other 5 points e ch 8 5% D l3. How high do you usually get? a. not high at all l point each b. a little high 2 points each c. smashed 4 points each d. ozone 5 points each Subtotal N l. N 2. N 3. N 4. N 5. N 6. S 7. I78 PART II During the past week, have you shoplifted anything? No_g_ Yes___ If yes, how many articles? # given O-lO = l pt l0-30 = 3 pts What do you guess was their total value? $ over 30 = 5 pts During the past month, have you shoplifted anything? No_9__ Yes___ If yes, how many articles? # given O-lO = l‘pt lO-3O = 3 pts What do you guess was their total value? $ over 30 = 5 pts Have you in other ways stolen anything during this last week? No_(_l_ Yes_l__ If yes, what were they? 0-l0 = l pt l0-3l = 3 pts What do you guess was their total value? 3 over 30 = 5¥pts Have you in other ways stolen anything during this last month? No_9__ Yes_l__ If yes, what were they? O-lO = l pt l0-3l = 3 pts What do you guess was their total value? $ over 30 = 5 pts Have you broken into any buildings this last week? No_Q__ Yes___ If yes, how many? # X IO Have you broken into any buildings this last month? No_9__ Yes____ If yes, how many? # X 10 Have you run away from home during this last month? No_Q__ Yes If yes, how many times? # x lo If you have run away from home, how many times in your life have you done so? S 8. S 9. NIO. I79 During this last week have you stayed out later than you were allowed to? If yes, how many times? # l hr = 2 Have you gotten into any really big verbal hassles with your parents this last week? If yes, about how many times? # Have you gotten into any physical fights this last week? If yes, how many? # x 5 With whom? (Ex. friends, brother, neighbor, etc?) ill No__(_)__ Yes_ No _9_ Yes_ No_g__ Yes___ Total BIBLIOGRAPHY I80 BIBLIOGRAPHY Ackerman, N. W. Treating‘the troubled family. New York: Basic Books, Inc., l966. Adler, D. L. Review of the "Test of Personality Adjustment." In 0. K. Buros (Ed.), Personality tests and reviews. Highland Park, N.J.: The Gryphon Press, l970. Alexander, J. F. Defensive and supportive communications in family systems. Journal of Marriage and the Family, l973, _Zfi, 6l3-6l7. (a) Alexander, J. F. Defensive and supportive communications in normal and deviant families. Journal of Consulting and Clinical Psychology, l973, 40, 223-23l. (b) Amini, F., & Salasnek, S. Adolescent drug abuse: Search for a treatment model. Comprehensive Psychiatry, l975, 1g, 379-389. Anderson, D. A. The family growth group: Guidelines for an emerging means of strengthening families. Family Coordinator, l974, 23, 7-l3. Annis, H. M. Patterns of intra-familial drug use. British Journal of Addiction, l974, g2, 36l-369. Anthony, E. J., & Koupernik, C. The child in his family. New York: Wiley Interscience, l970. Atchison, C. 0. A comparative study of the self-concept of behavior problem and non-behavior problem high school boys. Dissertation Abstracts, l958, 1g, part I, nos 5-7, lOlD. Atchison, C. A comparative study of the self-concept of behavior problem and non-behavior problem high school Boys. Unpublished doctoral dissertation, Indiana University, l958, as cited in W. H. Fitts, Tennessee Self-Concept Scale, manual. Nashville: Counselor Recordings and Tests, 1965. Bales, R. F. Interaction rocess analysis: A method for the study of small groups. Cambridge, Mass.: Addison-Wesley, l9Sl. Barcai, A. An adventure in multiple family therapy. Family Process, 1967, g, l85-l92. IBI I82 Barclay, L. E. A group approach to young unwed mothers. Social Casework, l969, 69, 379-384. Bartlett, D. The use of multiple family therapy groups with adoles- cent drug addicts. In M. Sugar (Ed.), The adolescent in group and family therapy. New York: Brunner/Magel, l975. Beal, D., & Duckro, P. Family counseling as an alternative to legal action for the juvenile status offender. Journal of Marriage and Family Counseling, l977, 3, 77-8l. Bentler, P. M. Review of the Tennessee Self-Concept Scale. In 0. K. Buros (Ed.), Personality tests and reviews II. Highland Park, N.J.: The Gryphon Press, l975. Berman, K. K. Multiple family therapy: Its possibilities in prevent- ing readmission. Mental Hygiene, l966, go, 363-370. Bing, E. The conjoint family drawing. Family Process, l970, 9, l73-l94. Blum, R., & associates. Students and drugs. San Francisco: Jossey-Bass, Inc., l969. Bradfield, R. H., et al. Project BEAM: An experiment in intervention. Journal of Negro Education, l975, 44, 34-4l. Bratter, T. E. Helping affluent families help their acting-out, alienated, drug abusing adolescent. Journal of Family Coun- seling, 1974, g, 22-3l. Braun, J. R., & Tinley, J. J. Faking study of scores on the Self- Perception Inventory, Psychological Reports, 26, ll8. Brehm, M. L., & Back, K. W. Self-image and attitude toward drugs. Journal of Personality, l968, 36, 349-354. Brill, H. Historical background of the therapeutic community. In H. C. B. Denber (Ed.), Research Conference on Therapeutic Community. Springfield: Thomas, l960. Bronfenbrenner, U. Two worlds of childhood: U.S. & U.S.S.R. New York: Russell Sage Foundation, l970. Buckland, C. M. Toward a theory of parent education: Family learning centers in the post-industrial society. The Family Coordinator, l972, gl, l5l-l62. Campbell, A., Converse, P. E., & Rodgers, W. L. The quality of American life. New York: Russell Sage Foundation, l976. I83 Campbell, D. T., & Stanley, J. C. Experimental and quasi-experimental designs for research. Chicago: Rand McNally, l963. Carkhuff, R. R. Training as a preferred mode of treatment. Journal of Counseling Psychology, l97l, 1g, l23-l3l. Carson, R. C. Interaction concepts of personality. Chicago: Aldine, 1969. Carter, W. W. Group counseling for adolescent foster children. Children, l968, 15, 22-27. Cattell, R. B. Review of the "California Test of Personality." In 0. K. Buros (Ed.), Personality tests and reviews. Highland Park, N.J.: The Gryphon Press, l970. Clark, J., & Kempler, H. Therapeutic family camping: A rationale. The Family Coordinator, l973, gg, 437-443. Cooper, D. The death of the family, New York: Vintage Books, 197l. Crime in the United States: Uniform crime report, l97l. Washington, D.C.: U.S. Government Printing Office, l972. Cross, H. J., & Kawash, G. F. A short form of PARI to assess authori- tarian attitudes toward child rearing. Psychological Reports, 1968, 23, 9l-98. Dean, D. G. Alienation: Its meaning and measurement. American Sociological Review, l96l, gg, 753-758. Dean, J. P., & Whyte, W. F. How do you know if the informant is tell- ing the truth. Human Organisation, I958, 11, 34-38. Dohner, A. Turned off drugs-~Turned on to life. Symposium presented at the meeting of the International Congress on Drug Education, Montreux, Switzerland, October 1973. Dreikuis, R., Corsini, R., Lowe, R., & Sonstegard, M. Adlerian family counseling. Eugene: University of Oregon Press, 1959. Duncan, P. Family interaction in parents of neurotic and social delinquent girls. Unpublished doctoral dissertation, University of Wisconsin, 1967. Edelson, M. Ego psychology, group dynamics and the therapeutic com- munity. New York: Grune & Stratton, l964. Effectiveness Training Associates. Summary of research of E.T.A. programs (Rev. ed.). Pasadena: ETA,—l975. l84 Epstein, J. Divorced in America. New York: E. P. Dutton & Co., l974. Farina, A. Patterns of role dominance and conflict in parents of schiZOphrenic patients. Journal of Abnormal and Social Psy- chology, 1960, 61, 3l-38. Farkas, M. 1. Students' view of school drug use. Journal of Drug Education, 1974, 4, 457-468. The farmer's life is the one he loves. The Detroit Free Press, June 27, l976, p. lD. Fejer, D., & Smart, R. G. The use of psychoactive drugs by adults. Substudy No. 46l. Toronto: Addiction Research Foundation, l972. Ferreira, A. J., & Winter, W. D. Family interaction and decision- making. Archives of General Psychiatry, l965, 46, 2l4-223. Fitts, W. H. Tennessee Self-Concept Scale, manual. Nashville: Counselor Recordings and Tests, I965. Fitzgibbons, D. J., & Cutler, R. The factor structure of the Tennessee Self-Concept Scale among lower-class urban psychiatric patients. Journal of Clinical Psychology, l972, 26, l84-186. Framo, J. L. Symptoms from a family transactional viewpoint. In C. J. Sager & H. S. Kaplan (Eds.), Pro ress ig_group and family therapy. New York: Brunner/Magel, l972. Freudenberger, H. Treatment of dynamics of the "disrelated" teenager and his parents in the American society. Psychotherapy: Theory, Research and Practice, l969, 6, 252-253. Gable, R. K., LaSalle, A. J., & Look, K. E. Dimentionality of self- perception: Tennessee Self-Concept Scale. Perceptual and Motor Skills, l973, 66, 551-560. Garfield, S. L., & Bergin, A. E. (Eds.). Handbook of psychotherapy_ and behavior change; An empirical analysis (2d ed.). New York: Wiley & Sons, l978. Geismar, L. L., & LaSorte, M. A. Understanding the multi-problem family: A conceptual analysis and exploration in early identi- fication. New York: Associated Press, 1964. Gibb, J. R. Defensive communications. Journal of Communications, l96l, 6, l4l-l48. Gibbs, D. N. Student failure and social maladjustment. Personnel and Guidance Journal, 1965, 46, 580-585. I85 Glass, G. V., & Stanley, J. C. Statistical methods in education and psychology. Englewood Cliffs, N.J.: Prentice-Hall, l970. Gordon, T. P.E.T.--Parent Effectiveness Training. New York: New American Library, I970. Gough, H. G. Review of the "Test of Personality Adjustment." In 0. K. Buros (Ed.), Personality tests and reviews. Highland Park, N.J.: The Gryphon Press, I970. Haberman, P. W., Josephson, E., Zanes, J., & Elinson, J. High school drug behavior: A methodological report on pilot studies. Pre- sented to the First International Conference on Student Drug Surveys, Newark, New Jersey, September l97l. Haley, J. Research on family patterns: An instrument measurement. Family Process, l964, 6, 4l-65. Hanawalt, N. G. Review of the Security-Insecurity Inventory. In 0. K. Buros (Ed.), Eersonality tests and reviews. Highland Park, N.J.: The Gryphon Press, 1970. Hays, W. L. Statistics: Probability,yinference, and decision. New York: Holt, Rinehart & Winston, 1963. Hetherington, E. M., & Deur, J. L. The effects of father absence on child development. Young Children, l97l, 66, 233-248. Hetherington, E. M., Stouwie, R. J., & Ridberg, E. H. Patterns of family interaction and child-rearing attitudes related to three dimensions of juvenile delinquency. Journal of Abnormal Psy- chology, l97l, Z6, l60-l76. Hinds, W., James, M., Gieszer, M., & Jacobs, B. A Survival Manual for the Drug Center Volunteer. Governor's Office of’Drug Abuse and AlcOhoTism, State of Michigan, August l972. Hollingshead, A. B., & Redlich, F. C. Social class and mental illness: A community study. New York: Wiley, 1958. Hoyt, C. Test reliability estimated by analysis of variance. Psychometrica, l94l, 6, 153-l60. Jackson, D. D. The study of the family. Family Process, l965, 4, l-20. Jackson, P. W. Verbal solutions to parent-child problems. Child Development, l956, 61, 339-349. I86 Jacob, T. Family interaction in disturbed and normal families: A methodological and substantive review. Psychological Bulletin, l975, 66, 33-65. Jenkins, R. L. Deprivation of parental care as a contributor to juvenile delinquency. In A. R. Roberts (Ed.), Childhood deprivation. Springfield, Ill.: Charles C. Thomas, l974. Jones, M. Beyond the therapeutic community. New Haven: Yale Univer- sity Press, l968. Juvenile court statistics l970. Washington, D.C.: U.S. Department of Health, Education & Welfare, l97l. Katz, A. J., Krasinski, M., Philip, E., & Wieser, C. Change in interactions as a measure of effectiveness in short-term family therapy. Familnyherapy. l975, 6, 3l-56. Kemp, C. G. Parents' and adolescents' perceptions of each other and adolescents' self-perception. Personnel and Guidance Journal, l965, 46, 58-62. King, F. W. Anonymous versus identifiable questionnaires in drug usage surveys. American Psychologist, 1972, 66, 982-985. Kiresuk, T., & Sherman, R. E. Prggram evaluation procedures manual. Prepared for use at Hennepin County Mental Health Center. Minneapolis, Minnesota, l970. (Mimeograph) Knaus, W. Rational emotive education: A manual for teachers. New York: Institute for Rational Living, l974. Kohlberg, L. The development of modes of moral thinkipg and choice in the years ten to sixteen. Unpublished doctoral dissertation, University of Chicago, l958. Kohlberg, L., Scharf, P., & Hickey, J. The justice structure of the prison--A theory and an intervention. The Prison Journal, 1972, 5] 9 3'14. Kratcoski, P. C., & Kratcoski, J. E. Changing patterns in the delinquent activities of boys and girls: A self-reported delin- quency analysis. Adolescence, l975, 46, 83-9l. Kreml, A. L. Understanding conflict in the normal family: An educa- tional model for family actualization. Unpublished M.A. thesis, Chicago Theological Seminary, June l970. Larson, R. Can parent classes affect family communications? School Counselor, March l972, 26l-270. I87 Lefeber, J. The delinquent's self-perception. Doctoral dissertation in preparation, University of Southern California, 1964, as cited in W. H. Fitts, Tennessee Self-Concept Scale, manual. Nashville: Counselor Recordings and Tests, 1965. Lewis, J. M., Beavers, W. R., Gossett, J. T., & Phillips, V. A. No single thread. New York: Bruner/Magel, l976. Lidy, T. The family as the developmental setting. In Anthony & Koupernik (Eds.), The child in his family (Vol. 1). New York: Wiley-Interscience, 1970. Light, P. K. Let the children speak: A psychological study of young teenagers and drpgs. Lexington, Mass.: D. C. Heath, 1975. Loughlin, F., & Wimberger, H. C. Group family therapy. Family Process, 1968, 6, 37-50. Loveland, N. The relation Rorschach: A technique for studying inter- action. The Journal of Nervous and Mental Disease, 1967, 146, 93-105. Magnusson, D. Test theory. Reading, Mass.: Addison-Wesley, 1967. Malouf, J. L., & Alexander, J. F. Family therapy research in applied community settings. Community Mental Health Journal, 1976, 46, 61-71. Martin, W. T. Self-Perception Inventory: A new test of personality. Psychological Reports, 1968, 66, 961-962. Maslow, A. H. Toward a psyphology of being, New York: Van Nostrand Reinhold, 1968. Maslow, A. H., Birsh, E., Honigman, I., McGrath, F., Plason, A., & Stein, M. The S-I Inventory, California: Consulting Psycholo- gists Press,il952. McCary, P. W. The effects of small self-understanding groups on the self-concept and anxiety level when group composition has been varied (Doctoral dissertation, Michigan State University, 1970). Dissertation Abstracts International, 1970, 64, 2112A. McQuitty, L. L. Elementary linkage analysis for isolating orthogonal and oblique types and typal relevancies. Educational and Psyeho- logical Measurement, 1957, 16, 207-229. McWhirter, J. J., & Kincaid, M. Family group consultation: Adjunct to a parent program. Journal of Family Counseling, 1974, 6, 45-48. I88 Mead, D. E., & Campbell, 5. S. Decision-Making and Interaction by families with and without a drug-abusing child. Family Process, 1972, 14, 487-498. Mehrens, W. A., & Lehmann, I. J. Measurement and evaluation in education and psychology, New York: Holt, Rinehart, & Winston, 973. Mendell, D. Combined family and group therapy for problems of ado- lescents: A synergistic approach. In M. Sugar (Ed.), The adoles- cent in group and family therapy. New York: Brunner/Magel, 1975. Minuchin, S. Families and family therapy, Cambridge, Mass.: Harvard University Press,_l974. Mishler, E. 6., & Waxler, N. E. Interaction in families: An experi- mental study of family processes and schizophrenia. New York: John Wiley & Sons, 1968. National Commission on Marijuana and Drug Abuse. Marijuana: A signal of misunderstandipg. The official report of the National Com- mission on Marijuana and Drug Abuse. New York: New American Library, 1972. Ottenberg, D. J. Addiction as metaphor. Alcohol Health and Research World, 1974, Exp. Issue, 18-20. Parry, H. J., Balter, M. B., & Cisin, I. H. Primary levels of under- reporting psychoactive drug use. The Public Opinion Quarterly, 1970, 64, 582-592. Parsons, B. V., & Alexander, J. F. Short-term family intervention: A therapy outcome study. Journal of Consulting and Clinical Psychology, 1973, 41, l95-201. Peabody, E. P. Record of Mr. Alcott's school, exemplifying the principles and methods of moral culture (3rd ed.). Boston: Roberts Brothers, 1888. Quay, H. C., & Peterson, D. R. The questionnaire measurement of personality dimensions associated with juvenile delinquency. Unpublished manuscript, University of Illinois, 1964. Rabinowitz, 0. Issues in the establishment of a community program for the early reachinglyfadolescents experimenting with drugs. Psychosocial Process, 1973, 6, 90-97. Rachman, A. W. Talking it out rather than fighting it out: Prevention of a delinquent gang war by group therapy intervention. Inter- national Journal of Group Psychothergpy, 1969, 16, 518-521. 189 Rentz, R. R., & White, W. F. Factors of self-perception in the Tennessee Self-Concept Scale. Perceptual and Motor Skills, 1967, 24, 118. Riegal, B. Group meetings with adolescents in child welfare. Child Welfare, 1968, 46, 417-418, 427. Riskin, J. M., & Faunce, E. E. An evaluative review of family inter- action research. Familerrocess, 1972, 41, 365-455. Riskin, J., & Faunce, E. Family Interaction Scales I: Theoretical framework and method. Archives of General Psyehiatry, 1970, 66, 504-512. (a) Riskin, J., & Faunce, E. Family Interaction Scales II: Data analysis and findings. Archives of General Psychiatry, 1970, 66, 513- 526. b Riskin, J., & Faunce, E. Family Interaction Scales III: Discussion of methodology and substantive findings. Archives of General Psy- chiatry, 1970, 66, 527-537. (c) Roberts, A. R. Childhood deprivation. Springfield, 111.: Charles C. Thomas, 1974. Roberts, A. R., & Roberts, B. J. Divorce and the child: A pyrrhic victory? In A. R. Roberts (Ed.), Childhood deprivation. Springfield, Ill.: Charles C. Thomas, T974. Rogers, C. Measuring personality adjustment in children nine to thirteen years of age. Contributions to Education [Teachers College, Columbia UniversitY], 1931, No. 458. Rubin, L. B. Worlds of pain. Lifein the working class family. New York: Basic Books, 1976. Rutter, M., & Brown, G. The reliability and validity of measures of family life and relationships in families containing a psychi- atric patient. Social Psychiatry, 1966, 4, 38-53. Samuels, D. J., & Samuels, M. Low self-concept as a cause of drug abuse. Journal of Drug Education, 1974, 4, 421-438. Saunders, A. L. Multiple family therapy in a community mental health center. Professional P§ychology, 1975, 6, 140-144. Silberman, C. E. Crisis in the classroom. New York: Vintage Books, 1971. I90 Silberberg, N. E., & Silverberg, M. C. School achievement and delinquency. Review of Educational Research, 1971, 41, 17-34. Singer, M. T., & Wynne, L. C. Principles for scoring communication defects and deviances in parents of schizophrenics: Rorschach and T.A.T. scoring manuals. Psychiatry, 1966, 66, 260-288. Singer, M. T., & Wynne, L. C. Thought disorder and family relations of schizophrenics III. Methodology using projective techniques. Archives of General Psychiatry, 1965, 16, 187-200. Smart, R. G., Fejer, D., & Alexander, E. Drug use among high school students and their parents in Lincoln and Welland Counties. Substudy 1-7 and Al-70. Toronto: Addiction Research Foundation, 970. Smart, R. G., & Jackson, D. A. A preliminary report on the attitudes and behavior of Toronto students in relation to drugs. Toronto: Addiction Research Foundation, 1969. Stabenau, J. R., Turpin, J., Weiner, M., & Pollin, W. A comparative study of families of schizophrenics, delinquents, and normals. ngchiatry, 1965, 66, 45. Stanley, S. Family education to enhance the moral atmosphere of the family and the moral development of adolescents. Journal of Counseling Psychology, 1978, 66, 110-118. Stoller, F. H. The intimate network of families as a new structure. In H. A. Otto (Ed.), The family in search of a future. New York: Appleton-Century-Crofts,T970. Stoller, F. H. The new groups. Psychology Today, 1967. Stoller, F. H., & Dreyfuss, A. The family workshop: A format for enhancing family experience. 1967.(Mimeograph) Stranahan, M., Schwartzman, C., & Alkin, E. Group treatment for. emotionally disturbed and potentially delinquent boys and girls. American Journal of Orthopsychiatry, 1957, 61, 518-527. Streit, F. The importance of significant others in youths' decision making about drug use and other deviant acts. Journal of Drug Education, 1974, 4, 409-419. Strodbeck, F. L. Husband-wife interaction over revealed differences. American Sociological Review, 1951, 16, 468-473. Suinn, R. M. Review of the Tennessee Self-Concept Scale. In 0. K. Buros (Ed.), Personality tests and reviews II. Highland Park, N.J.: The Gryphon Press, 1975. 191 Symonds, P. M. Review of the California Test of Personality. In 0. K. Buros (Ed.), Personality tests and reviews. Highland Park, N.J.: The Gryphon Press, 1970. Thorpe, L. P., Clarke, W. W., & Tiegs, E. W. Manual for the Cali- fornia Test of Personality(Rev. ed). Los Angeles: California Test Bureau, 1953. Tobias, J. J. Deprivation in the affluent suburbs. In A. R. Roberts (Ed.), Childhood deprivation. Springfield, Ill.: Charles C. Thomas, 1974. Truax, C. B. Degree of negative transference occurring in group psychotherapy and client outcome in juvenile delinquents. Journal of Clinical Psychology, 1971, 61, 132-136. Vacchiano, R. B., & Strauss, P. S. The construct validity of the Tennessee Self-Concept Scale. Journal of Clinical Psychology, 1968, 64, 323-326. Vernon, P. E. Review of the California Test of Personality. In 0. K. Buros (Ed.), Personaligy tests and reviews. Highland Park, N.J.: The Gryphon Press, 1970. Wahlroos, S. Family communication: A guide to emotional health. New York: Macmillan, 1975. Webster, H. Review of the Security-Insecurity Inventory.. In 0. K. Buros (Ed.), Personality tests and reViews. Highland Park, N.J.: The Gryphon Press, 1970. Weiss, W. U., & McFarland, A. W. Comparison of methadone clinic patients and methodone clinic drop-outs.using the Leary Inter- personal Check List. Proceedings of the Blst Annual Convention of the American Psychological Association, 1973, 6, 399-400. Winder, C. L., & Rau, L. Parental attitudes associated with social deviance in preadolescent boys. Journal of Abnormal and Soc1al Psychology, 1962, 64, 418-424. Winter, G. Love and conflict. Garden City, N.Y.: Dolphin Books, 1961. Winter, W. D., a Ferreira, A. J. (Eds.). Research in fami1y inter- action. Palo Alto: Science & Behavior Books, 1969. Wise, N. Juvenile delinquency among middle-class girls. In E. Vaz (Ed.), Middle-class juvenile delinqueney, New York: Harper & Row, 1967. I92 Woody, R. H., & Woody, J. D. Behavioral group counselling for ado- lescent females with behavior problems. Eeychological Reports, 1975, 66, 421-422. Worby, C., & Schachter, D. Family Interaction Task scoring manual. Rochester, N.Y.: University of Rochester, l972. (Mimeograph) MICHIGAN STATE UNIV. LIBRARIES IIIIWlWlIWlHIIIWWlUlIIIUIIIIWIHHIIWI 31293100642440