PLACE I! RETURN BOX to remove this checkout from your record. TO AVOID FINES mum on or before die duo. l-DATE DUE DATE DUE DATE DUE L_J = - LACIE fiEr—w l—Tl L MSU In An Afflrmdivo ActioNEqual Opportumy Imlon CWM' THE RELATIONSHIP BETWEEN FAMILIES AND SCHOOLS IN INTERVENTIONS TO NITIGATE THE IMPACT OF FAMILY BACKGROUND ON CHILD ACHIEVEMENT: Lessons From a Pilot Intervention BY Margaret L. Palmiter A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY College of Social Science] School of Social Work 1990 0?) Gym U3 5.3 ABSTRACT THE RELATIONSHIP BETWEEN FAMILIES AND SCHOOLS IN INTERVENTIONS TO HITIGATE THE IMPACT OF EAHILY BACKGROUND ON CHILD ACHIEVEMENT: Lessons from a Pilot Intervention BY Margaret L. Palmiter This study begins with the theoretical issues related to the impact of family background on child achievement, identifies the programmatic efforts that have developed in response to concerns of distributive justice and social equity, and outlines the important issues in evaluating programmatic effects. An evaluation of a pilot intervention is presented. Children in kindergarten through second grade were referred by teachers for a twelve month family intervention consisting of child task-socialization groups, parent education groups, and family support services. The evaluation used a quasi-experimental design, with control (N = 12) and intervention (N = 26) groups to test the impact on several outcome measures. The variables measured were child behavior at home, child behavior at school, child self-esteem and adult self- esteem. The Child Behavior Checklist was utilized for the behavioral measures, the Harter Self-Perception Scales for the child self-esteem and the Rosenberg Self-esteem scale for the adult self-esteem. Also included is a detailed analysis of the political and process issues in the development of the intervention and evaluation designs. This is particularly significant, as the intervention is the result of an interagency effort, involving judicial, educational, social welfare, private non-profit and university components. Problems associated with the multiple constituents of the various agency stakeholders is discussed, with implications for future development of similar programmatic efforts. The results revealed no statistically significant differences between the control and experimental groups on any of the outcome measures. In part this may reflect both the small sample size and measurement problems resulting from interagency negotiations which delayed the timetable for program implementation and data collection. However, the conclusions provide several important observations for future development of programs to mitigate the impact of family background on child achievement. Included is a discussion of a model of family and school relationships, highlighting the importance of the interactions between families and schools in future program development of this type. DEDICATION This dissertation is dedicated to Dr. Linda L. Loewenstein, whose love, encouragement and unswerving faith in my abilities allowed me to discover them for myself. iv ACKNOWLEDGEMENTS I would like to thank my dissertation committee members for their support and leadership through this endeavor: Dr. Paul Freddolino for his advice, counsel and commitment to me and my successful completion of the program. Dr. Rena Harold for her support and caring during the most difficult times in the process. Dr. Melissa Barker for her ability to help me keep sight of what is truly important and never let me forget our radical roots. Dr. Roseline Ekpenyong for her willingness to share experiences and frustrations in ways that made me feel like I wasn't alone. I would like to acknowledge the Michigan Department of Social Services for allowing me to utilize the data from the Early-School-Aged Delinquency Prevention Project for this dissertation. I would also like to thank the agencies and individuals who participated in the Early-School-Aged Delinquency Prevention Project, for without their efforts this dissertation would not have been possible. And most importantly, I would like to express my appreciation to the children and families who were a part of this effort for sharing their ideas, feelings and participation in the program and in our joint struggle to make a better world. TABLE OF CONTENTS LIST or TABLES O O C O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O 0 Vi i 1 LIST or FIGURES O O C O O O O O O O O O O C C O C C O O O O O O O O C O O O O O O O O O O O O O Vix CHAPTER I - STATEMENT OF THE PROBLEM .... ..... ... ...... .. 1 INTRODUQION OOOOOOOOOOOOOOOOOOO......OOOOOOOOOO0.... 1 OUTLINE OF THE PRESENT STUDY ........................ 6 cmrn "o-LITMTWBRWI" .........OOOOOOOOOOOOOOOO 9 INTRODUCTION ........................................ 9 STATUS ATTAINMENT .................................. 13 Theoretical Overview ........................... 13 Family Background Variables .................... 17 School Environment Variables ................... 20 Social-Psychological Variables ................. 24 Interaction of Variables ....................... 28 Expectancies ................................... 32 INTERVENTION STRATEGIES ............................ 35 Home-based Early Intervention .................. 38 Early Child-Care and Education ................. 39 Parent Training and Education .................. 40 Family Support Services ........................ 43 PROGRAM EVALUATION ................................. 43 CHAPTER THREE - DEVELOPMENT OF THE INTERVENTION ........ 51 INTRODUCTION ....................................... 51 PRESENTING PROBLEM ................................. 53 AGENCY PERSPECTIVES ................................ 54 THE INTERVENTION DESIGN ............................ 61 Prevention Focus ............................... 62 Deficiency Model of Intervention ............... 63 Cognitive and Education FOrmat ................. 64 Additional Services ............................ 66 CHAPTER POUR - EVALUATION METHODOLOGY .................. 67 EVALUATION METHODOLOGY ............................. 67 INTRODUCTION ....................................... 67 THE SOCIAL CONTEXT OF THE EVALUATION ............... 67 Political Climate .............................. 67 Contract Negotiations .......................... 77 DEVELOPMENT OF THE EVALUATION DESIGN ............... 80 Goals .......................................... 80 The Evaluation Design .......................... 82 Internal Validity Concerns ..................... 82 DEFINITION OF CONSTRUCTS ........................... 84 Impulse Control and Behavior Management ........ 84 Social Skills .................................. 86 vi Child Self-esteem .............................. 87 Adult Self Esteem .............................. 91 OPERATIONALIZATION OF CONSTRUCTS ................... 92 Child Behavior Checklist (CBCL) and Teacher Reporting Form (TRF) ................... 93 Child Self-Perception Scales ...................100 Adult Self-Esteem Scale ....................... 105 Family Observation Form ....................... 106 Process Measures .............................. 108 CHAPTER FIVE - DATA COLLECTION, ANALYSIS AND RESULTS .. 109 SAMPLE AND DATA COLLECTION ........................ 109 ANALYSIS .......................................... 119 Reliability and Norms ......................... 119 Control and Intervention Comparisons .......... 121 Analysis of Group Attendance .................. 130 Analysis of Academic Performance .............. 132 Analysis of Additional Data ................... 135 DISCUSSION ........................................ 137 Sample Size ................................... 137 Intensity of the Intervention ................. 138 Components of the Design ...................... 139 CHAPTER SIX - SUMMARY AND CONCLUSIONS ................. 143 INTRODUCTION 0......0.0.0.0...0.000.000.0000....... 143 IMPLICATIONS FOR THEORETICAL APPROACHES .......... 145 IMPLICATIONS FOR APPLIED APPROACHES ............... 152 IMPLICATIONS FOR EVALUATION ....................... 161 LIST 0’ REFERENCES 0..........OOCOOCOCOOOOOOOOOO...0.0.0169 APPENDIX A - GOALS OF THE INTERVENTION DESIGN ..........180 ”PmIXB-CHILDRBFM you ......OOOOOOOOOOO00.0.0181 APPENDIX C - EVALUATION PLOW CHART .....................183 vii Table Table Table Table Table Table Table Table Table Table Table Table 1. 2. 3. 4. 8. 9. 10. 11. 12. LIST OF TABLES Demographic Characteristics of the Sample.....113 Sample Sizes on Instruments at Three Time PeriOdSoo0.0000000000000000.000000000000118 CBCL Scores by Group at Three Time Periods....122 TRF Scores by Group at the Three Time PeriOdSO......OOOOOOOOOOOOOOO0.00.00.00.124 Cautela SES Scores by group at the Three Time periOdSOOOOOOOOOOOOO0....0.00.00.00.0000126 Barter SPS For Grades 1 and 2 at Time 2 and Time BOOOOOOOOOOOOOOOOOOOO00.00.000.0000127 Harter SP8 for Grades 3 through 6 at Time 2 andTime BOOOOOOOOOOOOO0.00.000.000.0000128 RSE Scores by Group at the Three Time PeriOdSOOOOOOOOOOOO0.0.00.00000000000000129 Correlation between group attendance (ATTGRPS) and TOTCBCL, TOTTRF and CSES............131 Correlation between group attendance (ATTGRPS) and Harter Self-perception Scale for Third to Sixth Graders..................132 TRF Math Performance by Group at the Three Time Periods............................133 TRF Reading Performance by Group at the Three Time PeriOdSOOOOOOOOOOOOO0.0.0.0.0134 viii Figure Figure Figure Figure Figure Figure 2. 3. 4. LIST OF FIGURES The socioeconomic life cycle: schematic representationOOOOOOOO......0......0.0.0.13 The Wisconsin Model of Status Attainment......15 AgenCies in the EDPPOOOOOOOOOCOOOO0.0.0.00000052 The Multiple Constituents of agencies in the EDPP.........OOOOOOOOOOOOOOO0.000.000.00074 Operationalizing Constructs And Instrument selectionOOOOOOOOOO0.0.0....0.0.0.0....0.93 Model of Overlapping Spheres of Influence in Families and Schools................ 149 ix CHAPTER I STATEMENT OP THE PROBLEM INTRODUCTION Social stratification in the United States has remained virtually constant since the 19608, and the resulting social inequality has persisted from generation to generation (Bell, 1989). Acceptance of this demographic reality is in conflict with the values of equality and opportunity so much a part of the American culture, but at the same time, attempts to ameliorate this inequality come into conflict with other values. The belief that one "earns one's status" is an important value within American Society. Much of this sense of "equity" and "distributive justice" is represented in the just world hypothesis-the belief that the world is fair and rewards one properly (Tropman, 1989, p. 7-8). While equality is important to us as a society, equity is also important (Tropman, 1989). Our ideal society is one in which the rules are fair and individuals achieve based on their own efforts and desires. But when some individuals do not achieve and the gaps between social classes increase, we share a collective concern and even guilt (Tropman, 1989, p. 48). These conflicting values are the impetus behind our attempts to understand the nature of social inequity. What characteristics of individuals or groups influence the achievement of success? Why is socio- economic status such a strong predictor of lack of success? What factors affect the relationships between socio- economic status and achievement of success? Social work has been concerned about issues of distributive justice and social equity throughout its history (Reynolds, 1963; Pumphrey and Pumphrey, 1961; Chambers, 1966). Using the foundation of ecological systems theory, the social work profession is a theoretical and practice model which attempts to intervene at the interface of the person and the environment and has always seen the importance of the societal context of individual functioning (Hartman, 1970; Hepworth & Larson, 1986; Johnson, 1986; Munson, 1980; Perlman, 1959). This framework, while referred to by many names over the years, has consistently placed the profession in a unique position to make a contribution to debates regarding social mobility and social stratification. These debates have taken two directions. The first is a theoretical perspective, including research to delineate the process of social mobility. Sociologists, as well as others, have examined the process of social mobility and status attainment, working to develop models that more clearly explain what factors influence the process. The second perspective is an applied one, where social workers, as well as others, have struggled with the types of intervention and programmatic elements that could impact the process of social mobility or status attainment. Early theoretical discussion of this issue lead to an awareness of the importance of education in this process. In 1927, Sorokin's Social Mgbility first outlined the role of education in social stratification systems (Sewell & Hauser, 1976). His work, and the additional work of Talcott Parsons in 1959, assured that no students of social stratification could ignore the role of education in social mobility (Sewell, & Hauser, 1976). Through the work of these authors as well as additional research (Blau & Duncan, 1967; Duncan, 1967; Hauser, Sewell & Alwin, 1976; Sewell, Haller & Portes, 1969), the relationship between education and social stratification has become a major concern. The publication, "Equality of Educational Opportunity," also known as the "Coleman Report," written in 1966, found that although there are few substantial differences between the resources and general educational outcomes of schools, there "remain unacceptably large differentials among the achievement of major racial and socio-economic groups" (Hauser, Sewell & Alwin, 1974, p. 309). Researchers have found a significant relationship between school achievement and socio-economic status (Sewell, Hauser & Featherman, 1976). Examination of this relationship revealed the close connection between schooling and familial resources and suggested this connection influences both present and future successes. The "Wisconsin Model of Status Attainment" introduced by Sewell, Haller and Portes in 1969, confirmed the intractable relationship between socio-economic status and achievement (Sewell, Haller & Portes, 1969). This theory found socio-economic origins to be the key variable in predicting educational achievement and, to a lesser extent, later occupational and economic achievement (Sewell & Hauser, 1976). These differentials in academic achievement are in conflict with our ideal society, where success in school is supposed to be the key to economic and occupational success. "Schooling mitigates gender, class and racial barriers to success" (MacLeod, 1987, p. 1). While successful achievements in school appear to influence later socio-economic status, present socio-economic status influences the child's ability to achieve in school. The lack of achievement of children from lower socio-economic backgrounds is becoming a critical concern. According to the United States Department of Education, in 1983, one third of the adult population of the United States at that time was illiterate or read at a level which barely enabled them to perform basic tasks necessary for survival in our society (Kozol, 1985, p. 4, 8). According to a 1983 study on Adult and Continuing Education, fifteen percent of recent graduates of urban high schools read at less than a sixth grade level (Kozol, 1985, p.4). In 1982, the United States ranked forty-ninth out of 158 members of the United Nations in literacy rates (Kozol, 1985, p. 5). High school drop-out rates continue to add to the problem. Hunter and Harman reported in 1979 that 64 million persons aged sixteen and over had not completed high school and were not presently in high school, and a recent estimate by these authors for 1984 is in excess of 60 million (Kozol, 1985, p. 9). The relationship between socio-economic status and educational achievement has become a major topic of research, theory development, and program planning. If we as a society wish education to mitigate the impact of socio-economic status on life success, then solutions must be found to enable children of lower socio-economic backgrounds to have successful school experiences. The desire to mitigate the impact of socioeconomic status on achievement has lead to the applied perspective of status attainment. These efforts have frequently taken the form of intervention programs to impact on the supposed relationship between socio-economic status and school achievement or status attainment and they assume the theoretical connection described above. The thrust of the programs has been to combat that relationship or compensate for it. But theoretical work has suggested that the relationship between socio-economic status and achievement is complex, with many interacting variables, making the development of intervention strategies difficult and complicated. Along with the development of intervention strategies have come the development of evaluation strategies. Since interventions in this area are typically framed as attempts to correct a problem, the next step has been to develop evaluation models that will determine if the intervention succeeded. Again, the process of social mobility remains a complex one and both theoretical and applied efforts reaffirm that complexity. Therefore, meaningful evaluation models, like meaningful intervention models, have been difficult to develop. OUTLINE OF THE PRESENT STUDY This study includes a number of perspectives on the relationship between socio-economic status and school achievement. Chapter Two examines the literature regarding social mobility and status attainment. This literature is concerned with the impact of variables on achievement and success, whether those variables are part of the environment and familial characteristics, or whether those variables are interventions imposed on the environment. The review begins with a theoretical overview of some status attainments models, as well as research that has further delineated the variables and their effect on achievement, success and status attainment. The second section of the review is a discussion of the applied approach to status attainment issues, including intervention literature and the various practical approaches that have been taken in response to the identified social problems. The third and final section describes the issues in evaluation of these intervention strategies and the relationship between research and intervention in programs to improve social mobility. Chapter Three is a discussion of the development of The Early-School-Aged Delinquency Prevention Program (EDPP), which was a pilot effort. Its purpose was to provide services to children in grades kindergarten through third who were having academic or behavioral problems in school, and it is an example of the applied approach to problems of status attainment. The intervention was designed to offer services to improve the children's functioning in school and to increase their likelihood of greater achievement later in life. This chapter describes the intervention, its theoretical foundation, the process of its development and the participants. Chapter Four is a description of the development and implementation of the evaluation of the Early-School-Aged Delinquency Prevention Program. The first section of this chapter describes the social context of the evaluation, including the political climate and contract issues, which had a significant impact on the shape of the intervention and its evaluation. Following this is a discussion of the evaluation design itself, beginning with program goals and issues of validity and evaluability, and ending with a definition and operationalization of the important constructs. Chapter Five is a description of the results of the evaluation and Chapter Six is a discussion of the implications of this study for future theoretical and empirical research, program development and evaluation, and the role of the social work profession. CHAPTER TWO LITERATURE REVIEW INTRODUCTION The person-environment connection is a critical element in social work theory and practice literature. Social work is a synthesis of knowledge from other disciplines, values and skills that result in a problem-solving process that actively recognizes the importance and impact of interaction among the systems of society at both a micro and macro level (Hartman, 1970; Johnson, 1986; Perlman, 1959). This "intervention into human transactions is the contemporary perspective of social work practice" (Johnson, 1986, p. 93). Transactions is a way of conceptualizing the fact that interaction with the environment and others in the environment is "not just a simple interaction but an interaction influenced by other interactions in the situation" (Johnson, 1986, p. 93). This theoretical perspective places social work in a unique position in the debates regarding intervention strategies to mitigate the impact of family background. While social workers can assess the strengths and weaknesses of individuals or families, they can also see the impact of the interaction of the various systems. Social work is in an excellent position to draw connections 10 between intervention designs and the resulting impact on families. Social work's commitment to interaction can also work to undermine a belief that it is possible to determine causal relationships between various strategies and desired outcomes. We see the complexity of human interaction and have developed the skill to consider the numerous environmental and individual influences effecting the lives of people. This has caused us to have little faith in causal models of intervention outcomes and has encouraged us to focus on the "art" of practice as opposed to defining and advocating specific skills that are effective in accomplishing certain goals (Aldeman, 1986; Fanshel, 1980; Karger, 1983). The process of status attainment and the relationships therein are complex, and models used to understand these relationships must reflect that complexity. Intervention models are frequently designed from practice knowledge and wisdom, with little information from the findings of research on status attainment. Consequently, there is typically not a clear, causal model behind an intervention strategy. Rather, and this is particularly true for social work, there is a professional, clinical respect for the complexity of human interactions and a belief that causal models do not provide an accurate picture of the human change process. In the midst of all this is the evaluator, 11 who is charged with the responsibility of determining the effect of the intervention so that policy-makers can decide how to spend their money. The evaluator attempts to make her or his way through this myriad of conflicting information and attitudes. In view of these professional issues, this literature review will be organized around three important themes. They are 1)theories or explanations of social mobility and status attainment processes; 2)development of intervention strategies to impact on that relationship; and 3)evaluation methods to determine if the interventions we offer improve the social mobility of persons of lower socioeconomic status. The first section, status attainment, is an overview of literature on the developmental process of social mobility, including prominent models of status attainment. Within this section is a description of research to define various elements of those models more completely, particularly family background, the school environment, and social- psychological influences and expectancies. Also included in this section is a discussion of the interaction of variables and the additional contribution or impact that variable interaction has on the status attainment process. The second section describes intervention strategies as a response to the lagging achievement and successes of persons of lower socioeconomic status. Also focusing on 12 the role of education in achieving success, these interventions reflect a commitment to strategies designed to mitigate the negative impact of lower socioeconomic status. This section is a review of the theoretical foundations for many of these interventions. While some program planners began with a deficiency model approach to intervention, others approached these programs from an empowering perspective. The arguments regarding these two perspectives contribute some of the most critical debates today regarding intervention strategies (Washington & Oyemade, 1987). The third section of the literature review is a description of the issues in the theme: how do we know what works? Included in this discussion are the major debates and dilemmas in evaluation research, particularly relating to interventions designed to compensate for the negative impact of socioeconomic status on achievement. While status attainment researchers described the impact of more accurately defined variables, intervention strategists described the impact of various forms of intervention. Descriptions of family background variables in the status attainment literature strongly resemble discussions of environmental influences in program design literature. While the connections between these two bodies of literature can be seen in an overview of both, it is not clear that there has been any systematic integration of the 13 knowledge gained by one into the thinking of the other. This lack of integration, among other factors, has lead to problems in developing causal models to evaluate interventions, which will be discussed in more detail in later chapters. STATUS ATTAINMENT Theoretical Overview Research into the relationship between SES and achievement got its first major start in Blau and Duncan's classic 1967 study - The American Occupational Structure (Sewell & Hauser, 1976). In this work, Duncan's "Schema of the Socioeconomic Life Cycle" was first applied (Sewell & Hauser, 1976). Family Background ————a Life Chances Schooling ————9 Level of Living }:b -——+’ Health, Welfare Income -———§ Status, Acceptance Expenditures-———§ Satisfaction,Morale Figure 1. The eocioeconouic life cycle: schematic representation. 14 The significance of this work is found in its statistical processes and the development of the socioeconomic life cycle model. It provides a context for understanding the relationships between socioeconomic status and achievement. This model enables researchers to examine social mobility as a process of status attainment over the life cycle, and stimulates a discussion of the relationships among components of the model. While family background is an exogenous variable, the other variables reflect an interaction with the social environment during various developmental stages. The process and outcome of that interaction during one stage can strongly influence other stages. In their original study, Blau and Duncan (1967) examined the educational and occupational status of fathers and the influence of that status on the later successes of their sons. Results supported the model, suggesting a strong relationship between these two variables (Blau & Duncan, 1967). Using Duncan's model, other researchers expanded the definitions of the variables and increased our understanding of their complexity (Alwin 8 Thornton, 1984; Davis, 1982; Mercy & Steelman, 1982; Sewell & Hauser, 1976). For example, Blau and Duncan originally defined family background as the educational and occupational status of the father. More recent researchers expanded the 15 definitions of family background, leading to better operationalization of the variable (Sewell & Hauser, 1976). When Sewell and Hauser (1976) developed the "Wisconsin Model of Status Attainment," they used the model to examine the central role of schooling, particularly in decisions and actions regarding post secondary education. Sewell and Hauser examined the antecedents of educational attainment. Their description reflects the complexity of the variables as well as adding a number of social- psychological influences into the process of status mobility, as seen in Figure 2. FAMILY g > MITIGATING 11 > DEPENDENT BACKGROUND VARIABLES VARIABLES father's education son's grades sons's post in high school high school fathers’ educational occupation significant attainment . others index mother's SES level education son's aspirations of son's occupation parental income Figure 2. The Wisconsin Model of Status Attainment In their 1976 study Sewell & Hauser looked at adolescents finishing high school, focusing on factors that influenced their decision to go to college and the impact of that decision on later occupational and economic outcomes. In this work, they differentiated patterns of 16 males and females and demonstrated that socioeconomic status of the adolescents’ families played a major role in educational, occupational and earnings attainment (Sewell and Hauser, 1976). Other researchers have further delineated family background and explored the effect of this variable on status attainment outcomes (Alwin & Thornton, 1984; Blechman & McEnroe, 1985; Davis, 1982; Mercy & Steelman, 1982; Patterson & Stouthamer-Loeber, 1984). Still other efforts have added the school environment as a variable, and examined the role the school plays in achievement outcomes (Epps & Jackson, 1985; Hamilton, 1983; Farnworth, Schweinhart & Berrueta-Clement, 1985; Fraser & O'Brien, 1985; Werthamer-Larsson, Kellam, Dolan, Brown, & Wheeler, 1988). The expansion of variables has included discussions of social-psychological variables. These discussions have been important not only for the additional information they provided on the variables, but also for the descriptions of the interaction of social-psychological variables with the other variables in the achievement equation (deKanter, Ginsburg, & Milne, 1986; Eccles, 1983; Entwisle, Alexander, Pallas and Cadigan, 1984; Harter & Connell, 1981). Recent research on status attainment and school achievement has more consistently evaluated the interactions among variables, suggesting that interaction % E 17 affects outcomes in ways individual variables do not (Alexander, Entwisle, Cadigan, Pallas, 1984; Becker & Epstein, 1981; Chandler, Argyris, Barnes, Goodman & Snow, 1986; Epstein, 1982; Epstein, in press; Garbarino, 1982; Kagan & Schraft, 1982; Epstein, 1983; Pallas, Entwisle, Alexander, Cadigan, 1984; Slaughter & Kuehne, 1988). The response to issues regarding interaction of social- psychological variables, family/ school characteristics and family/ school interaction has led, in part, to research on expectancies. This literature has provided a comprehensive examination of the interrelationships of the variables (Chandler et al., 1986; Corcoran, Duncan, Gurin & Gurin, 1985; Eccles, 1983; Kagan & Schraft, 1982; Lowry, 1983; Miller, Manhal, & Mee, 1989; Scott-Jones, 1982; Slaughter & Kuehne, 1988;). This section of the literature review will outline research addressing some of the variables in the status attainment equation, including family background variables, school environment variables, social-psychological variables, the interaction of variables, and expectancies. Family Background variables Researchers Alwin and Thornton (1984) examined the effects of home environment on school achievement, with special attention to the influence of the family in eeyly 18 development. They specifically added to the operationalization of family background as a "home environment," including size of family, relationship to siblings and interactions in the home (Alwin & Thornton, 1984). The premise of these researchers was that the influence of family background would vary at different times in the developmental life cycle and that family size and relationships to siblings would also play a role. They found that while the relationship between socioeconomic status and school achievement were quite similar across the major developmental time periods, family size was negatively related to school achievement and maternal employment bore no significant association to school achievement. While the results of their study did not, in fact, uphold their initial hypothesis, the efforts they made are reflective of the interest in explicating family background. Mercy and Steelman (1982) further defined family background by delineating family structure (birth order and size) and patterns of childhood interaction (pre-school education, time reading, time with friends and activities). They analyzed the effect of these variables on the verbal and nonverbal ability of their sample of children using the Wechsler Intelligence test and found support for the position that "effect of socioeconomic status on ability is 19 mediated by the child's environment" (Mercy 8 Steelman, 1982,-p. 540). The influences of family environment were also explored in the work of Patterson and Stouthamer-Loeber (1984) and Blechman and McEnroe(l985). In both of these studies, the patterns of interaction within families were defined and used as variables, measuring their effects on the achievement and success of children (Blechman and McEnroe, 1985; Patterson and Stouthamer-Loeber, 1984). Patterson and Stouthamer-Loeber attempted to determine if family management practices (activities to monitor the child's whereabouts, effective discipline, effective problem- solving skills, and activities to support pro-social skill development) have an impact on delinquent behavior, (Patterson 8 Stouthamer-Loeber, 1984, p. 1299), while Blechman and McEnroe examine family problem solving techniques in an attempt to develop effective measures of this family characteristic. Patterson and Stouthamer-Loeber found that certain family management practices were highly correlated with delinquent behavior (Patterson 8 Stouthamer-Loeber, 1984). In a laboratory setting, Blechman and McEnroe (1985) provided guidance toward the development of measures of family problem-solving techniques and found these techniques to be present across socioeconomic status (p. 436). Both efforts empirically supported the hypothesis 20 that family interaction was a major influence on child competency, while also providing a further delineation of family interactions. In these discussions of family environmental influences, theorists and researchers have taken the perspective that families of lower socioeconomic status are deficient in certain qualities that would allow them to function more effectively. These deficiencies result in less competent children and later less successful adults. For example, Garbarino and Sherman (1980), in their discussion of ecological perspectives on families, described families of lower socioeconomic status as "socially impoverished families." These families were considered high-risk families, with members who were at a greater risk of having psychological problems and greater incidence of child abuse and neglect (Garbarino 8 Sherman, 1980). While these authors maintained there are societal and community variables that heighten risks of serious problems for these families, their initial vulnerability is by virtue of their inadequate resources. School Environment Variables Recent research efforts have examined the role of the school environment in the relationship between family background and child success in school, and they have 21 placed school environment into the model as a variable influencing status attainment (Alwin and Thornton, 1984; Davis, 1982; Epps 8 Jackson, 1985; Mercy 8 Steelman, 1982). Epps and Jackson (1985) added the school environment as an additional variable influencing the achievement of children. They were particularly concerned with achievement among the African-American population, and postulated that the school dimension would increase the ability to explain achievement outcomes (Epps 8 Jackson, 1985). They added school factors (such as racial composition of the school, mean social class, curriculum information, and the average ability of the student body) into the equation, as well as the social-psychological elements already discussed by other authors (Epps 8 Jackson, 1985). They were not able to measure any significant effect of the school environment, but other results led them to the conclusion that even with the inclusion of a school environment variable, the Wisconsin Model Of Status Attainment is not useful for understanding the occupational attainment process of African-Americans. Their modification of the model explained only 17% of the variance in occupational attainment for males and only 14% of the variance in occupational attainment for females. They advocated the development of a model which includes other "structural” or environmental variables (such as 22 racial discrimination or labor market segregation) that might better capture the environmental realities of blacks, instead of the "Wisconsin Model which is focused so strongly on individual merit" (Epps 8 Jackson, 1985, p. 42). Farnworth, Schweinhart and Berrueta-Clement (1985) reported on a study of academic competence and its relationship to delinquency. Using the words of Hirschi, they indicated that "academic competence is linked to delinquency by way of success in and attachment to school" (Farnworth, Schweinhart and Berrueta-Clement, 1985, p. 445). These authors suggest that success in school must be measured as more than achievement on standardized tests, but also as success in relationships in the school environment. They claim the school must be seen as a "socializing agency" (Farnworth, Schweinhart, 8 Berrueta- Clement, 1985, p. 461). This role of the school as a socializing agency is important in examining the role of the school environment in achievement outcomes, and ecological research regarding classroom environments has made an important contribution to this effort. Based on his review of ecological studies of classrooms and schools, Hamilton concluded that "it is the balance between the socialization and academic instructional functions of schools that is at issue“ (Hamilton, 1983, p. 319). Hamilton suggested that 23 interactions and relationships at school play as significant a role in child competence as academic achievement. Both aspects of the educational environment must be equally strong (Hamilton, 1983, p. 319). The school environment is another important element in discussions of children’s ability to achieve success in certain academic environments. Werthamer-Larsson, Kellam, Dolan, Brown and Wheeler (1988) studied the impact of schools and classrooms on the maladaptive behavior of children (including inappropriate social contact, disobedience and aggressiveness and learning problems). Taking an epidemiological perspective, these authors studied a stratified sample of 1043 first grade children attending the East Baltimore Public Schools. Using several assessment instruments, they analyzed the relationship between child characteristics, family characteristics, classroom environment and child performance. They concluded that the maladaptive behavior of children was not totally explained by child characteristics (such as family background, etc.), but was also affected by the school environment within which the child functioned (Werthamer- Larsson, Kellam, Dolan, Brown and Wheeler, 1988). 24 Social-Psychological Variables The other aspect of achievement and success that has become a major element of all research in this area involves social-psychological dimensions. From the beginning of this discussion, when Sewell and Hauser included "a significant others index and occupational aspirations as intervening variables," psycho-sociological characteristics became part of the debate (Sewell 8 Hauser, 1976). The result has been further study of the internal, motivational aspects of child achievement, with attention paid to the impact, as well as the origins, of these internal motivations. The most common area of exploration has been the variable of self-image. This discussion has examined the extent to which the evaluation of significant others or self-evaluations are influential in achievement outcomes (Entwisle, Alexander, Pallas, 8 Cadigan, 1984). Without exception these variables have been linked, but there is still confusion about the ways they are linked and in what directions. The discussion of internal motivation (which self-esteem, self-concept, significant others index, and self perception are all attempting to address) inevitably leads to the proverbial chicken or egg questions. Did self-concept influence academic achievement, or did 25 academic achievement influence self-esteem? (Entwisle et al., 1984; Harter 8 Connell, 1981). Barter and Connell tackled this question in their research on intrinsic motivation. They identified perceived competence and perceived control as correlates of a child's motivational orientation (Harter 8 Connell, 1981). They then used four different models to analyze their data, each placing a different variable at the beginning of the causal chain. Their results suggest that perceived control, "defined as the amount of knowledge children claim to know about who or what is responsible for their academic successes and failures" is the critical causal variable (Barter 8 Connell, 1981, p. 45). The importance of their work for the purposes of this discussion is that it reflects the dilemma faced by researchers in understanding the direction of influence of social-psychological variables. While these researchers added to our knowledge of the components of self-esteem, they also recognized the difficulty in establishing which influences come first. Entwisle et al. (1984) joined in the effort to determine causal direction in these relationships in their study on the academic self-image of first graders. Using longitudinal data on 825 first graders in Boston schools, these researchers found that cognitive outcomes and self- image were linked, but that the direction of the linkage 26 varied depending on the developmental stage of the child. At the start of schooling, affective status seems to shape cognitive status more than the reverse, but this begins to change at the end of the first grade for some groups (Entwisle et al., 1984). This addition of social-psychological variables has helped us to recognize that the relationships between family background and child achievement are not simple or linear. However, an additional danger presents itself. The addition of this material into the discussion can result in stereotypical conclusions about the social- psychological characteristics of families of lower socioeconomic status and the impact of those characteristics on their ability to provide an environment that fosters achievement and success. This danger can be seen in the attitudes and responses of the people responsible for education policy at the national level. For example, representatives from the United States Department of Education, in a presentation at the 1986 conference on Designs for Compensatory Education, discussed issues of parent involvement in children's academic activity. The clear message in the presentation was that lower income parents whose children were not doing well in school needed to impart appropriate values to their children to enhance their ability to learn (deKanter, Ginsburg 8 Milne, 1986). After asserting that "research 27 has consistently shown family background to be among the most important influences predicting a child's performance in school," these authors suggested that families take "more responsibility for their children’s improper conduct or failure to otherwise meet their obligations" and promote more "positive values" in the children (deKanter et al., 1986, p. 4, 11, 20). Many believe that this "culture of poverty" philosophy has led to an "underclass" stereotype that is not useful in our attempts to discover the linkages between socioeconomic status and achievement. This consequence becomes clear as intervention strategies have been developed and these programs have been focused on "fixing" the weaknesses of the family as opposed to finding and building on strengths (Corcoran et al., 1985). Corcoran et al. suggest that this theory of a "culture of poverty" has created stereotypes that cannot be supported by quantitative data. ...evidence from the Panel Study of Income Dynamics undercuts the proposition that the primary causes of persistent poverty are the types of motivational and personal qualities that supposedly characterize the ”culture of poverty". Poor people differ from others on some motivational and personality measures, but the differences appear to have been caused by the events they have experienced. There is virtually no consistent evidence that the motivational and psychological characteristics measured by the study affect subsequent achievement, either within or across generations. Nor is there substantial evidence to support the proposition that poverty or welfare dependency is usually passed on from one generation to the next (Corcoran et al., 1985, p. 531-532). 28 Similarly, Davis examined the relationship between family background, schooling, and occupation on attitudes and behaviors as measures in the cumulative General Social Survey and found that the "notion of class cultures receives little support" (Davis, 1982, p. 586). Attitudes and corresponding behaviors vary significantly across social classes. The concept of a "culture of poverty" limits our understanding of the factors affecting achievement and significantly affects the direction of our solutions. While it is critically important to include the social-psychological aspects of achievement and status attainment in the models we use to understand the process, it is equally as important that we guard against the isolation of emphasis on those characteristics in the development of intervention strategies. Interaction of variables Recent research has examined the interactions of the variables in the socioeconomic life cycle model and several themes have emerged. Some authors have explored the consistency between the home and school environments and the impact that consistency (or lack of consistency) has had on child achievement in school. Using survey responses from teachers and parents gathered during their previously mentioned longitudinal study on first graders in Boston, 29 authors Alexander et al. used regression analysis to look at standards of deportment in the home and the school, and evaluated the impact on child achievement of consistency between these standards (Alexander et al., 1984). These authors found several important results. First, teachers' standards affect student learning, but not as much as techniques of classroom management and how the evaluation of the child’s performance is completed (Alexander et al., 1984, p. 13). In fact, variations in teacher attitudes were more significant in affecting child performance than parent/ teacher agreement on standards of deportment (Alexander et al., 1984, p. 14). Others have looked at home/ school discontinuities, and their impact on child performance. Chandler, Argyris, Barnes, Goodman and Snow (1986) examined whether or not home/ school discontinuity was an explanation of inability of children from lower class backgrounds to achieve adequate literacy levels. Their study was ethnographic, consisting of teacher and parent interviews, and observations of the families in completing an assigned homework-like task. Reading, writing and language test scores were the dependent variables. The results of this study reflect a finding similar to that of the Boston First Grade study. "Their observations did not substantiate home/ school discontinuity as a cause" of poor academic performance (Chandler et al., 1986, p. 6). 30 In fact, they found more continuity than expected between the home and school environments, particularly around the home-work like tasks they were observing. "There were many parallels between the dyadic interactions over homework like tasks observed at home and those which we saw in the classrooms" (Chandler et al., 1986, p. 12). Another theme that has emerged in discussions of the interaction of variables is the issue of parent involvement in their child’s education and there has been substantial theoretical and empirical work developed on this topic. Becker and Epstein (1981) conducted a survey of teachers to define various types of parent involvement, and to ascertain the attitudes of teachers and parents regarding the involvement of parents in the educational process. While three-fourths of the teachers agreed that parent involvement was a good idea, about one-half of them had serious doubts about the success of practical efforts to involve parents in learning activities at home, let alone at school (Becker 8 Epstein, 1981, p. 7). Such findings have been supported by other research. Stallworth (1982), in a national survey of educators for the Southwest Educational Development Laboratory in Austin Texas, found that educators had positive feelings about parents in general, but saw their roles as limited to family problems and when children had particular problems with the educational system. While virtually all of the 31 teachers and most of the literature supported the positive impacts of parent involvement, it was clear from the surveys that it was not a widely accepted practice in the schools (Stallworth, 1982, p. 2). Williams and Stallworth (1982) did a follow up survey to learn what the barriers were to parent involvement in schools, and found that much of the problem was one of definition. While teachers and other school personnel all agreed that prospective teachers need to be prepared to involve parents in the education of their children, there was variation in the definition and understanding of parent involvement (Williams 8 Stallworth, 1982, p. 8, 11). It was also apparent that there was little understanding on the part of teachers about how to incorporate parent involvement effectively as part of their teaching style (Williams 8 Stallworth, 1982, p. 13). Researchers have attempted to demonstrate the impact of various forms of parent involvement on child achievement. Epstein (1982), in a study of 613 fifth graders, measured the impact on students' performance in school of teachers' parent-involvement techniques. The evidence is consistent that parents' encouragement, activities, and interests at home, and their participation in schools and classrooms affect their children's achievement, even after the student’s ability and family 32 socioeconomic status is taken into account (Epstein, 1982, p. 1). Expectancies Corcoran et al. (1985) define expectancies as "the individual's assessment that his or her performance will lead to the desired outcome" (p. 519). This concept speaks to the effect of social and social-psychological events on the individual. If your experience, and the experience of those around you, is that no amount of performance results in a change in outcome, it is difficult to believe that increased performance will now make a difference. This experience can be translated to various forms of motivational and self-esteem issues. In addition, it is also true that the beliefs or expectancies of significant others can directly affect the beliefs and expectancies of children (Eccles, 1983; Corcoran et al., 1985; Scott-Jones, 1982). Several authors have looked at the expectancies of parents and the impact those expectancies have on the performance of children. Scott-Jones (1982), in a study of kindergarteners and first graders and their mothers, found that expectations had a dramatic positive effect on child performance, if the expectations were developmentally and individually realistic. She also found that parents used 33 information regarding tasks and previous performance in setting expectancy levels, suggesting that accurate data from the schools to the parents can be effectively incorporated into parental expectancies to the benefit of child performance (Scott-Jones, 1982). Miller, Manhal and Mee (1989) confirm the Scott-Jones hypothesis in their research. In a study of 2nd and 5th graders and their mothers and fathers, these reachers found that parents do not accurately estimate their child's abilities, as indicated by discrepancies between parental expectations and their child's performance on certain tasks. However, they also concluded from their results that parents’ beliefs affect their behavior to their child, which affects the child's development (Miller, Manhal 8 Mee, 1989). These authors implied, and others indicated directly, that one effect of parent involvement in their children’s schooling is the changes that can occur in parental expectancies. Epstein (1983) found this in her study of the effects of teacher practices of parental involvement. In her survey of the parents' of 1269 students, she learned that parents attitudes toward involvement in their children’s education were essentially positive, and that teacher practices can affect parent responses (Epstein, 1983). 34 Kagan and Schraft confirm this finding in their study of different forms of parental involvement. They compared two urban elementary schools, which were selected specifically because of very different perceptions regarding parent involvement in education (Kagan 8 Schraft, 1982). One elementary school had parents involved in very traditional ways, such as school conferences and PTA, while the other school used parental representation in their governance structure (Kagan 8 Schraft, 1982). While the outcomes they found were not influenced solely by involvement level of the parents, it was true that for lower income parents there was a strong correlation between the level of involvement and changes in personal aspirations and their perceptions of power (Kagan 8 Schraft, 1982, p. 4). Eccles (1983) describes a study of 668 first and second graders that is an example of an attempt to bring together many of these variables and to examine their interrelationships. This study examines the developmental and causal links among cultural factors, historical events, expectancies, values and academic behavior with a planned approach to examine the complex interrelationships among these variables (Eccles, 1983). The variables in this model were explicitly defined. Eccles included information about the cultural milieu (such as labor market sex divisions and cultural stereotypes), socializers' behaviors 35 and beliefs about their ability, the child's goals and the child's perceptions (Eccles, 1983, p. 80). In testing the causal model of the determinants of achievement-related behaviors, Eccles (1983) found that students' "interpretations of reality," such as self- concept or perceptions of the expectancies of significant others, were more influential determinants than past, objective reality (p. 137). Parents had a stronger affect on children's achievement-related beliefs than teachers, while parents had little influence as role models. "It was their role as direct socializers of achievement beliefs and attitudes that was important" (Eccles, 1983, p. 137). INTERVENTION STRATEGIES All the previous literature suggested that the relationships between socioeconomic status and achievement and success is at best, complex. Consequently, the best intervention strategies to improve school achievement for lower socioeconomic children are difficult to ascertain. There have, however, been a number of programmatic efforts over the years to provide children or families of lower socioeconomic status with whatever assistance the designers believed might allow them to succeed in spite of the odds. All of the various intervention strategies appear to have certain elements in common. First, they begin from 36 the premise that the development of human behavior is influenced by the environment (zigler 8 Anderson, 1979). While some refer to this as "naive environmentalism," (Washington 8 Oyemade, 1987) there is a strong perception in these programs that the influences of environment are primary in the social mobility process (Halpern, 1987; Newberger, Melnicoe and Newberger, 1986; zigler 8 Anderson, 1979). Second, descriptions of structural influences in the program design literature strongly resemble discussions of family background variables in the status attainment literature. Intervention authors list those characteristics of families which place them in "high-risk" categories and use those characteristics as justification and direction for intervention. For example, Halpern (1987), in his discussions key social and demographic trends affecting young families, lists trends that he claims place families in jeopardy: increased poverty in families with young children, shifts in social resource allocation away from these families, and the growing perceived costs of child-rearing. He recommends society respond to these risk factors with "practical, informational and emotional support to young families" such as pre-school education, child, and parental education (Halpern, 1987, p. 10). 37 Seitz, Rosenbaum and Apfel (1985) support Halpern's premises in their discussion of family support intervention. These authors maintain that poverty leads to chronic stress which is an impediment to family functioning. The program they describe offers a cognitive program for the children, intensive health services and day care as the family support package. The other element that appears to be common among intervention strategies is some degree of adherence to the "culture-of-poverty" notion. Washington and Oyemade (1987) suggest that this was the major philosophy of the Head Start Program at its inception and continues to be a major influence. Head Start was established to ...interrupt the sequence of poor parenting which leads to children with social and intellectual deficits which in turn leads to poor school performance, joblessness, and poverty leading again to high risk births, inappropriate parenting and a continuation of what has been termed the poverty cycle (Washington 8 Oyemade, 1987, p. 48). Halpern (1986) confirms this aspect of intervention strategies. In his article describing community—based support programs, he suggests that "chronic family stress and inattentive parenting deny young children the emotional security, role models, and skills they will need to compete in an increasingly complex world" (Halpern, 1986, p. 18). While the adherence to a "culture-of-poverty" perspective has received a great deal of criticism, particularly recently, there remains a degree of this philosophy in many 38 of the intervention programs developed. The degree of adherence to that philosophy is reflected in the design of the intervention, particularly in the roles and responsibilities of the professionals and the participants. For instance, during the last two decades Head Start has changed many of its programs to provide a more active and empowering role for parents in response to the criticism of the "deficiency model" (Washington 8 Oyemade, 1987). While all of the intervention programs provide some concrete services to families, there is great variation in the types and provision of the services. Head Start models, for instance, have great variation from program to program due to a large amount of local control (Washington 8 Oyemade, 1987). The following discussion will review four major types of intervention models: home-based early intervention, early child care/ child education programs, parent training and education, and family support services. While many programs have an overlap of one or more of these services, it is helpful to look at them individually. Home-based Early Intervention Home-based early intervention programs target "at-risk" infants and their parents with services in the home, focusing on the care-giver. Such programs continue to be popular, according to Halpern, even though there is little 39 evidence or theoretical foundation for their approach (Halpern, 1984, p. 387). These programs provide assessment of a target population, treatment of the population, and then attempt to assess the effects. The most common example has been the development of infant mental-health programs in Community Mental Health Centers across the country. These treatments typically consist of personal relationship building, sustained support for the mother, work to strengthen parental coping skills, information sharing, observation-surveillance and case-management (Halpern, 1986). Many of the Head Start Programs that were developed around the country also incorporate some aspects of a home- based early intervention strategy. Goodson and Hess described a Head Start program in which mothers are trained on a one-on-one basis to teach school-related skills to their young children (Goodson 8 Hess, 1987). In this program, mothers receive in-home training to improve the cognitive growth of their children. Early Child-Care and Education Early child care and/or child education programs have a focus similar to that of the home-based early intervention programs, but are less comprehensive. These programs offer services to children in day care or pre-school settings and 40 are designed to encourage cognitive and emotional development. While the issues of day care and child development have now become a concern of many socioeconomic groups (Schweinhart, 1985), the debate about compensatory effects continues to be targeted on children from lower socioeconomic families (Brown, 1987). One major focus of Head Start programs was an educational pre-school program for lower socioeconomic children to provide them with the tools they would need to compete in the academic environment and to compensate for the lack of those skills in their family environment (Brown, 1987; Washington 8 Oyemade, 1987). Parent Training and Education Parent training and education is the third service frequently offered by intervention programs addressing these concerns. As Meier notes in his introduction to Brown’ 8 book W IDEEIXQDEIQD, "the evidence to date indicates that knowledgeable and skillful parenting is the most effective and economical means for fostering the optimum development of the child" (Meier, 1987, p. 5). While there are many different models of parent education, each of the intervention programs described here incorporates some version into their services. 41 Variations in parent education and training models include differential social reinforcement, parent effectiveness training, developmental system models and parent and teacher models, just as a few examples (Holden, Lavigne 8 Pameron, 1986; Pakizegi, 1985; Strom, Griswold 8 Slaughter, 1979). All of these various approaches provide parents with information and understanding of realistic expectations of their children at particular developmental stages, as well as skill development to improve their ability to respond to their children more effectively. One major debate in the area of parent education has been whether or not various models are more or less effective with certain groups of parents. Strom, Griswold and Slaughter (1979) when using the PAAT (Parent As A Teacher) Inventory, found that there were differences between Anglo and Black mothers in four of the five PAAT subsets. These differences were statistically significant, with Anglo mothers frequently obtaining the highest mean scores and Black mothers obtaining the lowest scores (Strom et al., 1979). Class differences were also noted in this study, with the differences between upper and lower class mothers being the major sources of variance in all areas. The reasons for this variance suggest the class bias of the educational design of the program. "Upper class responses tended to be more consistent with child development 42 research than those of lower-class" (Strom et al., 1979, p. 19). Holden et al.(1986) found other problems associated with differences in the socioeconomic status of participants in parental training. In their study, parents of lower socioeconomic status, younger age, and less education tended to drop out of the program more than other participants (Holden et al, 1986). This caused the researchers to raise the "goodness of fit" question, suggesting that parental training programs must be designed to fit the particular needs and characteristics of the parents involved (Holden et al., 1986, p. 6). Parent education and training was also part of many Head Start Programs. While these programs had a variety of models for parent education, their primary focus was on providing parents with the tools to improve their children’s academic performance. These programs were also more frequently designed to incorporate elements of individualized instruction as opposed to a group format (Brown, 1987; Washington 8 Oyemade, 1987). "Those studies which emphasize individualized instruction in the home or in a Center seem to show effects more consistently than those in which intervention occurs in small groups" (Palmer, 1987). 43 Family Support Services Almost all of the programs that have been developed to assist families of lower socioeconomic status include a social service element. This takes many forms, including family support services, health services, case management and informal support networks. These services are usually offered as an addition to other more formalized aspects of a program. There has been discussion about the difficulties in identifying the components of this particular area of service, as there is so much variation (Gladow 8 Ray, 1986; Halpern, 1984; Meier, 1978; Palmer, 1987). The next section on program evaluation will elaborate on the impact of these difficulties. PROGRAM EVALUATION The effectiveness of intervention programs for families of lower socioeconomic status is a hotly debated topic. The results are inconclusive at this point, predominantly because a wide variety of services has been offered and differential effects have been hard to pinpoint. Intervention programs are designed quite differently from each other in goals, content and method. These differences make comparative analyses difficult, if not impossible. It is also frequently the case that while improvements in 44 children's school performance or family interactions can be observed, these improvements cannot be attributed to specific intervention activities. As Goodson and Hess point out, "It is easier to produce effects in intervention programs than it is to identify the specific factors which contribute to the success" (Goodson 8 Hess, 1987, p. 75). Limitations in measurement and sampling have made analysis of the effectiveness of intervention programs suspect (Halpern, 1984). ' This inability to identify specific intervention activities that result in the desired effect is thought to represent a lack of causal models in these programs (Halpern, 1984). This omission creates significant problems for an evaluator, problems that become obvious during an evaluability assessment at the beginning stage of evaluation design (Rossi 8 Freeman, 1985). In order to assure that an evaluation "fits" the program the evaluator begins by assessing the relationships among the stakeholders' interests, the goals of the program, the actual process of the intervention, and the expected effects or hypotheses (Rossi and Freeman, 1985). Ideally, this assessment occurs throughout the planning of the innovative intervention to assure that the intervention consists of operationalized goals that will accomplish the desired effect egg maximize rigorous evaluation. Unfortunately, this is frequently not the case. 45 wisdom and experience, with little formal attention given to this careful delineation of goals and strategies (Freeman 8 Rossi, 1985; Halpern, 1984; Fanshel, 1980; Karger, 1983). Consequently, many interventions do not use a well- delineated causal or impact model, and quantitative outcome evaluations (which some call rigorous) are extremely difficult to accomplish (Rossi 8 Freeman, 1985; Halpern, 1984). Others would suggest that the reason for this lack of causal models is because we are not able to differentiate what aspects of the complex human interactions involved in an intervention actually result in the hoped for outcome. These authors would advocate that while intervention must make itself more amenable to evaluation, evaluation methods must become more responsive to the realities and complexities of the world (Halpern, 1986; Karger, 1983; Schuerman, 1983). In view of these arguments, it should not be surprising that intervention egg evaluation strategies that see the importance of interactions show the most promise. Programs that seem to have a significant impact on individual children are those which take the perspective that the tasks of parents and schools are shared tasks, and offer opportunities to foster the relationships between parents and school personnel (Epstein 8 Dauber, 1988; Epstein, in press; Lightfoot, 1975; Lusterman, 1986; McGuire, 1985). 46 press; Lightfoot, 1975; Lusterman, 1986; McGuire, 1985). Furthermore, evaluation strategies that allow for evaluation and intervention to interact and respond to each other appear to show the most promise in effectively informing our efforts to design programs that make a difference in the problems we face (Bond 8 Halpern, 1986; Washington 8 Oyemade, 1987). An example of a study that examined the relationships of parents and schools was a national study of inner-city private elementary schools for the 1978-79 school year conducted by Cibulka, O’Brien and Zewe (1982). The schools selected were required to be 70% minority and eligible for Title I funds for underachieving poor children (p. 18). Eight cities were selected: New York, Newark, Washington, D.C., Detroit, Chicago, Milwaukee, New Orleans and Los Angeles. The instrument used to gather information on the schools selected from these cities were questionnaires administered to a sample of parents and teachers and to all principals, interviews with all principals and personal observation and interviews in each school. These researchers came to several conclusions upon analysis of the data. They make a distinction between an outcome model of school effectiveness and a preference model, and provide an analysis of their data from both perspectives. In regard to an outcome model, they examined student social_and academic behavior and its relationship 47 to school characteristics. In interpreting the school's impact on academic and social behavior, they found that "school factors appear to compensate for family background deficits" (Cibulka et al., 1982, p. 167). These students exhibit fewer attendance and discipline problems and math and reading scores seem to indicate that the most successful schools seem to be overriding the biases imposed by the unfavorable effects of family background (Cibulka et al., 1982, p. 158). The authors attempted to isolate those characteristics of schools which seem to account for their demonstrated effectiveness regarding the outcome measures. These traits included strong instructional leadership, a concept of shared work, a safe school environment, and clarity of mission and shared purpose (Cibulka et al., 1982, p. 22). The preference model of school effectiveness, discussed by these authors, provided an added dimension that could be helpful in understanding the impact of the relationships between parents and schools. The preference model is one where school effectiveness is evaluated by the choices parents make and their satisfaction with their children's school (Cibulka et al., 1982, p. 184). These authors suggest severe limitations to the outcome model and that a preference model can be more useful in understanding what it is about private schools that result in parents choosing them for their children even in the 48 face of great financial hardship (Cibulka et al., 1982). Such an understanding could move us closer to learning what in the parent school relationship contributed to the improved outcome of certain schools. This study suggested there is a connection between parents and schools that affects outcome measures of school effectiveness, but they also advocated looking more closely, through the preference model, at the process and dimensions of the connection between parents and schools in order to understand its potential impact. The interaction of the social institutions of school and family can also been seen in recent analyses of the Head Start Program. Head Start began with the culture of poverty notion as its foundation, has moved to an increasing notion of the "social structure theory of poverty" (Washington 8 Oyemade, 1987). This approach maintains that poverty is influenced more by structural limitations, such as discrimination, than by personal characteristics, such as motivation. The result has been a movement to emphasize parent-involvement in decision-making within the Head Start program...parents also became involved in policy councils and committees...and in influencing non-Head Start institutions (Washington 8 Oyemade, 1937, p. 60). Washington and Oyemade (1987) suggest that the approach of Head Start must now focus more closely on empowering the 49 family and increasing the role families play in the program and their communities. Bond and Halpern's (1986) discussion of "action research" describes the positive impact of the interaction between evaluation and intervention. Because the evaluation informed the intervention throughout its development and implementation, a specific intervention was made more effective. Such an interactive approach between evaluator and intervention has been the subject of on-going debate (Bond 8 Halpern, 1988; Fanshel, 1980; Heineman- Pieper, 1985; Patton, 1982; Rossi 8 Freeman, 1985). These debates suggest a direction for future early intervention programs closely associated with the theoretical and practice commitment of the social work profession. The next three chapters are a description of the development and evaluation of an intervention program that reflects the issues outlined in this review. Chapter Three is a description of the development of the intervention, Chapter Four a description of the development of the evaluation, including the political influences that played such a major role in both stages. Chapter Five is a discussion of the evaluation results. This program serves as a sample of the programs and processes that have been designed, implemented and evaluated to increase social mobility. The final chapter, Chapter Six, will be a 50 discussion of the implications of both this programmatic example and the theoretical discussions begun here. CHAPTER THREE DEVELOPMENT OF THE INTERVENTION INTRODUCTION This chapter describes the initial development of the Early-School-Aged Delinquency Prevention Program (EDPP). The impetus for developing the program came from staff at a number of local, public and private agencies, concerned about the problems faced by children and families in their community. Personnel from these agencies met together for almost two years, discussed the problems they faced and worked out a strategy to respond. This process had a significant impact on the actual design of the program and it influenced the implementation and evaluation of the program as well. This chapter outlines the development of the program and the impact of the players and the process on the intervention design. It is important to identify the participants in the project and the clearest way of doing so is by the following chart. 51 52 AOEWCY ACRONTW ROLES City Government CG Developer; funder Local Probate Court LPC Developer; contributor Local County Department LCDSS Developer; funder of Social Services City School District CSD Developer; in-kind supporter Michigan State Univ MSU Developer; designer; School of Social Work Evaluator Local Elementary School LES Referrals; Intervention site Local Private Agency LPA Program Provider Figure 3. Agencies in the EDP? The agencies involved in this project had a variety of roles. These roles changed during the various stages of program development, some agencies were only involved at specific stages, and in two instances the individuals representing the agencies changed at various stages. The LCDSS was the primary agency responsible for the project at all stages, while the other agencies assumed various roles at different points in time. Some agencies, specifically the local County Probate Court and the City Government were not involved through-out the length of the program. MSU was involved in both the development of the intervention and the evaluation, but the personnel involved 53 in each stage were different. They will be referred to as MSU intervention staff and MSU evaluation staff. The staff of the local private agency (LPA) will be referred to as program staff. PRESENTING PROBLEM In July of 1985, the Local County Probate Court (LPC) and the local County Department of Social Services (LCDSS) began meeting regarding the increase in delinquent acts committed by children aged 10 to 14 in the county. In their review of the juvenile home caseload, the LCDSS noted that 68% of the youth petitioned to the local County Probate Court were from the school district of a moderate sized city in the county, with a majority of those petitioned from two particular schools in the city (DSS Officer, 1985, p.1). These geographic areas also contained a large number of families receiving public assistance. Representative from the City School District (CSD) were invited to participate in these early discussions with the Local Probate Court and D88, and they reported an increase in unmanageable behavior in classrooms and non-involvement of parents as major problems in these same schools. They also indicated these problems were a concern in the elementary grades as well. 54 These agencies joined in an informal commitment to address these problems. The Local Probate Court, LCDSS and the City School District personnel envisioned a prevention program that would provide services to children and their families. These children would be in the early elementary grades, particularly in the schools which were feeder schools for the aforementioned high-delinquency schools. City School District personnel were convinced that children likely to have later problems in school could be identified in early elementary school (DSS Officer, 1985). With this assurance and a strong commitment to early intervention, the interagency committee decided to contract with Michigan State University School of Social Work (MSU) to develop an early-school-aged intervention to address these issues, with the goal of directing these children into more positive academic and personal experiences. AGENCY PERSPECTIVES The perspectives and experiences of each of the agencies influenced the design of the intervention. The LCDSS was the funding source for the intervention, and in the joint venture DSS would purchase the services and monitor their delivery. This had a major influence on the program from several vantage points. First, all of the 55 participant families were required to be public assistance recipients. Stereotypical labeling of these families as inadequate was much more likely in view of the fact that the entire sample would consist of families of lower socioeconomic status. This set the tone for the design of the intervention and the attitude of many of the participants. The influence of the LCDSS also affected the design of the intervention in other ways. The LCDSS personnel were very committed to including two major elements in the EDPP. First they were firm that the program have a family focus, and that services to the families would be paramount. In one of the initial documents, for inStance, the LCDSS personnel referred to "aggressive family-based prevention oriented protective services....Services will emphasize involvement of the family no matter how resistive they might be initially" (DSS Officer, 1985). The staff from the LCDSS was also committed to intensive and comprehensive services being made available to families through the program. They frequently referred to this as "good old-fashioned casework", and their model placed great importance on the case-management function of the social worker. Another comment in one of the initial documents was that "efforts will be made to connect each of the families with needed community based services. In the 56 absence of such services the families needs will be met to the extent possible by the primary worker" (DSS Officer, 1985). These comments point out 2 important aspects of the LCDSS philosophy regarding this program. First, the families that agreed to participate in the program yeglg be totally involved in all aspects. These comments seem to be in contrast to the voluntary nature of the program. Parents were to be reassured that they were free to participate or not participate, as they saw fit, and this choice was present at any time in their participation in the program. LCDSS staff were convinced that the services were so obviously needed and could so effectively be offered, that the commitment and competence of the staff was the primary concern, not the interest or commitment of the families. This brings up the second important aspect of these comments. At no stage in the planning of this program were consumers of the services ever consulted or involved. The identified needs of the families and the program design were all developed from a professional perspective on the problem and one that was greatly influenced by agency attitudes and experiences. This left out important information regarding the needs of the families and the services they felt would be most helpful in meeting those needs. 57 The City School District envisioned a program that would be similar to one they had offered from 1977 to 1980 in several of the elementary schools being suggested for this study. It was referred to as the ...Elementary Counseling Grant and provided counseling services to children identified as having "high socio-emotional needs" in eight elementary schools in the area (City School District, 1980, p. 3). The program included classroom, individual and small group services, efforts to improve student's in class behavior, peer interactions, and improve and increase communications between the families and the school. Services were offered to parents in the districts in a parent education format. This school district program was also intended to modify classroom curricula and the approach to students by offering resources to classroom teachers. These resources came via school counselors in the form of useful affective material and techniques and appropriate services (City School District, 1980, p. 4). The local elementary school (LES) also influenced the intervention design. In conversations with personnel from the school where the EDPP was to be located, two issues were raised consistently. First, there was concern on the part of school staff about the lack of involvement on the part of parents in their children's educational problems. 58 They asked several times whether or not the staff of the program would actively seek to engage the parents of the children referred. Second, teachers were concerned that the behavior of the children in school seemed to be worsening, and they were somewhat defensive about the efforts they had already made to address this problem. They were interested in the techniques that would be employed by the staff of the program and wanted to be assured that there would be regular and consistent communication between the program staff and the teachers. The local Probate Court’s influence on the intervention design stemmed from a program they were offering at the time. Intensive Neglect Services (INS) is a program which provides services to wards of the court who are abused or neglected. Services are provided to children in their homes, as a means of preventing out-of-home placements, or to children who have been removed from the homes of their parents. The premise of the program is that intensive, family-based services could allow for an early return of children to their parents, or prevent the removal of children in the first place (Probate Court Officer, 1985, p. 2). The services offered in INS include a complete assessment, voluntary participation (which for the court is unusual), parenting classes, a parent support group, a 59 support group for the minors, individual or family therapy, if needed (Probate Court Officer, 1985, p. 4). The program also plans regular recreational outings as a "means of improving interaction between family members and developing a positive socialization skills and interpersonal informal support networks" (Probate Court Officer, 1985, p. 4) The following quote from the Annual Report of the program in September of 1986 describes the Services most completely. The I.N.S. program is a preventive, treatment oriented, interdisciplinary and comprehensive outpatient therapy program which is coordinated with intensive casework Services, community mental and public health Services, parenting classes, and parent-child recreational] activity functions (Licensed Clinical Psychologist, Local County Probate Court, 1986, p. 1). The City Government became involved in this project because of political agendas and the need to involve appropriate political personnel. The City had a stake in the discussions of rising delinquency rates, as juvenile crime had become an important political issue. The city government role, however, became more critical as the development of the program unfolded. Because the LCDSS and the Michigan Department of Social Services were funding sources, there would be no money available for evaluation of the intervention. The Department restricts the use of these particular funds to programmatic efforts only. Therefore, it was the city who provided $10,000 the first year of the project for the development of the 60 intervention and then $10,000.00 the second year for the development of the evaluation of the EDPP. Following this stage other money was located to continue the evaluation, but this initial role of the city was significant in gaining political and economic support for the project. Michigan State University staff were concerned about two aspects of the program. First, the intervention staff was committed to a competent, theoretically appropriate base for the intervention. They utilized literature and experiences that supported the orientations of the various players, but they were also concerned that the intervention that was developed be implemented competently. Secondly, Michigan State University School of Social Work personnel were adamant about the need for an evaluation component of the program. The various players were not particularly concerned about the evaluation component of the design, and in fact, when there was difficulty in funding that piece of the project, most of the participants would have simply left it out. However, the Director of the School of Social Work was insistent that there be some evaluation built into the program, in order to have a way of indicating the impact of the program when it was completed. 61 THE INTERVENTION DESIGN The development of the intervention proceeded from July of 1985 to July of 1986. This period included discussions between personnel from Michigan State University School of Social Work and representatives of the various agencies, as well as efforts on the part of all parties to secure funding for the intervention. The resulting intervention design had the imprint of all of the participating agencies. However, it is important to note that evaluation personnel from Michigan State University did not participate in this phase of program development (Fitzpatrick, 1988). Several elements of the Intervention Design emerged from all the processes that have been outlined above. First, the agencies saw the program as preventive, in that they perceived the purpose of this type of intervention as preventing more serious behavioral and academic problems in adolescence. Second, the agencies shared a deficiency perspective regarding the families who would be referred to the program. And third, they supported a skill development, educational model for the intervention. 62 Prevention Focus Early in the development of the EDPP, the program was described in a local newspaper article highlighting issues regarding delinquency prevention. The following excerpts from the article are comments of the Director of Children's Services of the Local County Probate Court regarding the EDPP. What we see happening is they start having problems in kindergarten or first grade, but they’re still little. It might be possible to steer youngsters onto a different path by working with them at the first signs of problems that could develop into delinquent behavior. The ... School District and the ... County Department of Social Services will test that theory with a new program. The two-year program involves 40 students and their families. ...If the program demonstrates prevention programs work, it might lead to better ways of dealing with the youth crime problem (Leach, Hugh, 1987, 4A). These comments are the best example of how the participants saw the EDPP as having a preventive orientation. This orientation is the underlying perspective that runs throughout the design. It is also true, however, that while the agencies saw the focus as delinquency prevention, there was little discussion about specific pre-delinquent behaviors or characteristics to be addressed. The major factor stressed in the documents and later discussions was the inability of delinquents to achieve in school and the 63 fact that this inability could be documented at an early age (DSS Officer, 1985). Deficiency Model of Intervention All of the participating agencies operated under certain assumptions regarding the needs of the children and families who would be referred to the EDPP. They saw the families as lacking in certain values and skills that the program could provide. The lack of these skills were seen as decreasing the children's chances of doing well in school and thereby achieving in life. The design of this intervention reflected the assumed inadequacies of families of lower socioeconomic status, supported by the theoretical and empirical evidence that was discussed earlier in the literature review (Alwin 8 Thornton, 1984; Blechman 8 McEnroe, 1985; Davis, 1982; deKanter et al., 1986; Mercy 8 Steelman, 1982; Patterson 8 Stouthamer-Loeber, 1984; Williams, 1985). The MSU intervention staff, encouraged by the attitudes of the agency representative as well as their own practice training, designed an intervention to address the competency levels of the families in problem-solving techniques, parenting skills and communication of feelings and values. The following quotes from the rationale for the intervention design reflect this perspective. 64 Studies have shown that children who are socialized aggressively will tend to use aggression to meet their needs (Perry, Herr and Rasmussen, 1986). They have also revealed that social comparison processes increase a child's conformity to resolve peer disputes (Aboud, 1981). The literature further indicates that children who suffer social, emotional or cultural deprivation/ neglect early in life will have difficulty in these areas as they grow older (Wilking, 1979; Leif 8 Zarin- Ackerman, 1980; and Newman, Schwam 8 Schwam, 1979). (MSU School of Social Work Intervention Staff, 1986a, p. 4). Many mental health prefessionals maintain that young children cannot sustain emotional and intellectual gains in development unless their parents concurrently undergo attitudinal, behavioral, and cognitive reorientation (Haizlip, et. al, 1975)(MSU School of Social Work Intervention Staff, 1986a, p. 3). Cognitive and Education Format In keeping with the previous assumptions, the intervention design took a cognitive and educational format. It was assumed that the families were lacking in skills, resulting in anti-social behavior on the part of the children and a resulting inability to achieve in school. The children would be provided with a group setting to learn appropriate behavior and positive peer relations and the parents would receive parenting classes to improve their skills in helping their children. MSU Intervention staff developed child/task socialization groups for the children, to provide an environment that would compensate for the inadequate 65 provision of these skills at home. The children's groups had six topical modules: families, friends, feelings, self-awareness, fantasy and values. These modules ran for one month, and would be repeated twice. The group would meet twice a week for 12 months. At the same time, the MSU intervention staff developed an educationally-based group model of parent education that would improve the parenting skills of the parents and thereby increase the school performance and social skills of the children. The parents' groups were organized in modules, including child behavior management techniques, relationship building, self-awareness/ self-esteem and community concerns. These groups would meet once a week for six months. The following quote from the rationale for the parent groups provides a theoretical base for this design. Literature addressing the need for effective parent education programming is abundant (Dreikurs, 1964; Auerback, 1968; Gordon, 1970; Pickarts 8 Fargo, 1971; and Strom, Griswold 8 Slaughter, 1985). Within the area of parent education programming, it is generally accepted that what children learn is grounded in what parents teach. Thus parents must recognize and accept their role as teacher and be helped in developing child-rearing skills that will bring about healthy physical, social, and emotional and intellectual development in the children (Pickarts 8 Fargo, 1971). (MSU School of Social Work Intervention Staff, 1986b, p. 4). 66 Additional Services The other important aspect of this intervention design was the family service component. Regular home visits and individual contact with the families were built into the design, including case management and referral services. These contacts were initially referred to as integrative home visits, the intent of which was to further reinforce the skills and discussions of both groups. In keeping with social work values and skills described earlier, the home visits were also a recognition of the numerous other environmental influences and other systems that played a role in the lives of these families. While the LCDSS staff and MSU intervention staff were very committed to this level of service, it was also the least defined. As we will see in the discussion of the evaluation results, this area of service became much more significant in the implementation of the intervention than originally anticipated. CHAPTER POUR EVALUATION METHODOLOGY INTRODUCTION This chapter describes the development of the evaluation design. As described in the previous chapter, many agencies with their particular perspectives played a role in various stages of the program's development. When MSU evaluation staff began the process of designing the evaluation, all of the factors that have been discussed became part of that process. This discussion focuses on the impact of these factors on the evaluation and then describes the actual evaluation design that was selected. Included in the discussion are the influence of the social context of the program on the evaluation itself, goal and construct identification, evaluability, and instrument selection. THE SOCIAL CONTEXT OF THE EVALUATION Political Climate Every evaluation has a social context within which it must function and this evaluation was no exception. The Early School-aged Delinquency Prevention Program (EDPP) had 67 68 a number of participants with interest in the outcome for very different reasons. The LCDSS (the funding agency), the school district (the referral agency), the private agency (the provider agency), Michigan State University School of Social Work (providing the intervention design and the evaluation), the local elementary school (the referral and site agency) and the city school district (developer and contributor) all had their own investment in and perspective on the program. Agencies make policies and decisions in an environment of "multiple-constituencies" (Martin, 1980). In this model, described by Patricia Yancey Martin, outside constituent and interest groups influence the internal decisions of organizations. Each of the agencies involved had their own collection of interest and constituent groups which influenced its functioning. These constituent groups influenced the agency and that influence had an impact on the perspective of each agency regarding the design and the evaluation of the EDPP. One example of the multiple-constituencies phenomenon is the LCDSS, which was responsible for funding and monitoring the project. Because the funding source was federal delinquency prevention dollars, both state and local agency personnel were involved in monitoring the program. There have been on-going debates in the 69 legislature regarding the funding levels for the Michigan Department of Social Services, and frequent discussions about the inability of the department to provide services that were effective in preventing the problems faced by DSS clients or restoring them to self-sufficiency. These concerns appeared to increase the investment of the LCDSS in the outcome of the EDPP and lead the LCDSS to assume a strong sense of ownership of the program, even beyond its customary control and responsibility as the funding agency. LCDSS staff frequently commented on how few prevention programs existed in the state and how significant this opportunity was. The city school district (CSD) and the local elementary school (LES) both have multiple constituent groups influencing their functioning. Because of constant political battles regarding the funding of schools, resources are always a concern. Both the CSD and the LES were having problems finding enough resources to serve the children and families in their district and social work services in the district were limited. These problems were discussed by the school district participants in the development of the project. Increasing drop-out rates have been raised as a district concern by many community and government groups. Schools in the district have been responding to these concerns with pleas for more resources and more space. 70 While neither the city school district nor the local elementary school contributed money directly to the program, they both contributed space and in-kind services. This meant the program became a competitor for scarce resources. Teachers, school social workers and administrators, who had all been under-fire for lack of results, were extremely interested in the ability of an "out-side" program to accomplish what they had not. In meetings with the teachers, their initial questions ranged from concerns about demands on their time and space to curiosity about the intervention methods being employed and opinions about their likely success. The Michigan State University School of Social Work also had its interest groups. Research and grant money are important expectations of an academic unit, as the profession of social work attempts to increase its role in program evaluation, in an effort to increase its influence on policy making (Fanshel, 1980; Heineman-Pieper, 1985; Karger, 1983). Social work has a strong commitment to its values and methods, coupled with the belief that such values and methods can be effective. These opinions have come under attack from conservative analysts of the social service system in the recent past (Gilder, 1981; Anderson, 1979). Consequently, MSU intervention and evaluation staff were committed to the need for an evaluation of the 71 intervention. Their academic and political orientation led them to a quantitatively based out-come evaluation, in an effect to draw conclusions about the intervention that would be taken seriously in political arenas. Also, the values and methods of the profession were clearly the moving force behind the design of the intervention. As is commonly the case, the contradictions between these two elements of the profession would create problems in the evaluation, to be discussed later in the chapter. The agency providing the intervention was a local office of a state-wide, private, sectarian social service agency (LPA). This agency had its constituency and interest groups as well. The LPA is involved in a number of services under purchase of service contracts with a variety of public organizations at multiple levels. The services offered at the local office are also offered in a variety of other cities in Michigan, which have concerns about the issues raised in the development of the EDPP. The central office of the agency has an office of governmental affairs and is very concerned about the political impetus and ramifications of projects such as these. The EDPP provided some additional resources to the LPA, while also offering the opportunity to be involved in an area of service delivery where there was political 72 interest. The LCDSS was the funding agency and contract monitor for the EDPP, but was also responsible for other programs in the LPA that accounted for virtually all of the office's revenue. The LPA also had a client constituency that influenced their role in the evaluation. As mentioned earlier, the clients were not involved in any of the planning or design of the intervention, an important omission. However, after the program began the agency staff developed a relationship with the clients. The clients had their own perceptions about the school system and the problems facing their families. These perceptions greatly affected the role of the LPA in the implementation and evaluation of the intervention. Frequently, program staff found themselves advocating for the client perspective on issues, at odds with the intervention and evaluation design and the school. Every interest group influencing an organization has its own agenda and goals that must be recognized (Martin, 1980). The previous discussion provides a picture of the multiple-constituencies effecting each of these agencies. Then each agency became one of the "multiple stakeholders" in the evaluation (Rossi 8 Freeman, 1985, p. 175). Recognizing the multiple constituents each stakeholder brought to its role in the program, their roles as multiple stakeholders became even more complex. Rossi and Freeman 73 describe multiple stakeholders as significant players who have a stake in the outcome of the evaluation (Rossi 8 Freeman, 1985). That was certainly the case here. The following diagram uses the model of Martin (1980) in depicting the influences of these "multiple stakeholders" and their own "multiple constituencies" (Rossi 8 Freeman, 1985). The diagram depicts the relationships between the various participants in the EDPP, as well as the interest groups that impact each participant. UNIONS AND \ OFFICE LEGI SLATURE POLITICAL INFLUENCES 74 THE .PP THELOCALCOUNTY DEPAM'MENT OF SOCIAL SERVICES MICHIGAN DEPAHMENT OF SOCIAL SERVICES l THE LOCAL PRIVATE AGENCY MICHIGAN STATE UNIVERSITY SCHOOL OF SOCIAL WORK THE CITY SCHOOL DI STRICT l THELOCAL ELEMENTARY SCHOOL CLIENTS K C.S.W.E. COISIUNITY Figure 4. The Multiple Constituents of agencies in the EDPP 75 Frequently the reality of multiple stakeholders or constituents results in difficulty for the evaluator to decide the perspective for the evaluation (Rossi 8 Freeman, 1985). The result can be an unclear evaluation and difficulties in establishing proper communication between the various participants in the program and evaluation process (Martin, 1980; Rossi 8 Freeman, 1985). In this instance, the agencies involved came into the project with strong perceptions of the problem, the intervention and its potential impact, as well as the expectations of their own various interest groups. The result was a highly politicized process in terms of the development of the design of the intervention, the development and design of the evaluation, and the reactions of the various players to the results. One example where these factors played a role was in regard to the decision on the location of the intervention. Both the local Department of Social Services, and the local school district, wanted to locate the program as a pilot project in one elementary school. They believed that this would provide an opportunity to test the intervention strategy, and they were also limited in the amount of money available for the intervention. Consequently, they set criteria to determine which elementary school would be most appropriate. 76 These criteria reflected many of the assumptions with which the agencies came into the development effort. The first requirement was to locate the pilot project in an elementary school in the geographic area where the high delinquency rate was first discovered. Several elementary schools were located in the area or were feeder schools for the high schools in that area. The second consideration in where to place the intervention reflected the requirement on the part of the Department of Social Services that anyone receiving services must be receiving public assistance. Therefore, the school selected had to have a large enough public- assistance-receiving population to assure an adequate pool of children to be referred to the program. An additional consideration was one of control and responsibility for the program, as well as space. Although the school personnel would be referring the children to the intervention, and providing the space for the workers and the groups, DSS staff were clear that the responsibility for monitoring and controlling the intervention remained with them. Several initial contacts with schools were unsuccessful because there was disagreement between DSS staff and school personnel about ultimate control of the workers and the intervention. After several contacts, the program was located in an elementary school where thirty-eight percent of the 77 school's population was receiving some form of public assistance, a percentage that placed the school in the top five of schools in the district in regard to numbers receiving public assistance. The principal reported a high mobility rate, high single-parent family rate, and extensive drug abuse and violence in the neighborhood. This area of the city had a high concentration of Hispanic families, with a Catholic, Hispanic Community Center located nearby. Contract Negotiations The second area where the multiple stakeholders influenced the process of the program and evaluation was in regard to contract negotiations. The agency to provide the intervention was chosen through a Request For Proposal (RFP) process executed by the LCDSS. The agency would be under contract with the department to provide the intervention that had been developed between DSS, the city, the school district, the Probate Court and Michigan State University School of Social Work staff. In the summer of 1987, the County Department of Social Services sent out an RFP to all potential agencies in the area, outline the project and asking for bids. The RFP, following DSS procedures, was extremely detailed, specifically outlining 78 all of the work that had been done previously on the intervention. There was only one response to the this RFP. The group reviewing the RFPs consisted of representatives from all of the agencies who had participated in the development of the intervention. This group recommended that the RFP be re-submitted in an effort to elicit more response to the project. There were five responses to the second RFP, received in the Fall of 1987. The interagency panel reviewed the RFPs and recommended to DSS the agency that would provide the intervention. A contract was to be negotiated between the agency and the county Department of Social Services. The process of contract negotiation was arduous from the perspective of all of the various parties. Separate contracts were to be signed between the intervention agency and the Department of Social Services, between the City of Lansing and Michigan State University, and between the Department of Social Services and Michigan State University. Every constituent, or stakeholder, had certain criteria that had to be in the contract and every agency was interested in maintaining the appropriate levels of control over that portion of the program for which they saw themselves responsible. The result was that the process of contract negotiation delayed the start of the program from September, 1987 to March, 1988. This presented all of the 79 participants with very real difficulties, including 1)the program would run over two academic years, and therefore two teachers; 2)it would require a summer session in the middle of the program, rather than at the end; and 3)the fiscal years for funding were now not in coordination with the time-table for the intervention itself. Because the funds were available for a limited time, the option of waiting until the end of the 1987/1988 school year or the beginning of the 1988/1989 school year was not possible. The social context of the evaluation also influenced the perspective taken by the evaluation team. Because there was a concern on the part of funding agencies regarding outcome of the intervention, there was a need to provide specific information about its impact. This required a particular type of design and additional efforts to "separate the effect," particularly when random assignment would not be possible because of other requirements (Rossi 8 Freeman, 1985). These issues will be discussed in more detail when describing the actual design of the evaluation. 80 DEVELOPMENT OF THE EVALUATION DESIGN Goals The first step in designing the evaluation was conversations with the designers of the intervention. The evaluation Staff began with efforts to determine the problematic condition that the intervention was to address, the conditions under which it would operate, the goals it hoped to accomplish, the strategy for accomplishing those goals, and the outline of the actual intervention that would be provided (Rossi 8 Freeman, 1985). These steps quickly became problematic. As Rossi and Freeman indicate, "although a deficiency in existing conditions may be easy to recognize, a precise assessment of the empirical situation is usually required before one can formulate realistic goals and objectives and plan programs to achieve them" (Rossi 8 Freeman, 1985, p. 67). The designers of this intervention had identified a problem they were concerned about and had used theoretical and conceptual material to design the intervention they believed addressed that problem. However, there was little information that would make a conceptual connection between the specifics of the intervention and the problem they had identified. 81 The initial work of the evaluation staff was to identify, with the designers, the goals of the intervention, which are attached in Appendix A. These goals were to enable evaluation staff to identify constructs that the intervention was to impact, allowing the operationalization of those constructs. The goals, which were behavioral in nature, were to be accomplished through the group intervention format already outlined in the intervention design. Intervention strategies are frequently designed based on the experiences and practice wisdom of the designers, and such was the case here (Fanshel, 1980). It is also common for intervention strategies to be seen as effective and helpful, while there is little ability to demonstrate concretely what aspects of the intervention actually provided the needed assistance or made the hoped for impact (Halpern, 1986; Washington 8 Oyemade, 1987). As a result, the evaluation staff had difficulty in isolating the treatment and its effects on the problem. The evaluability assessment did not result in a clear connection from problem identification to intervention strategies directed at the problem to desired outcome (Rossi 8 Freeman, 1985). 82 The Evaluation Design Because random assignment was not possible and outcome information seemed important to the stakeholders, a quasi- experimental design was chosen for the evaluation. This required an identification of the treatment, outcome measures, and control/ intervention comparisons (Cook 8 Campbell, 1979). But the design also required that the evaluation staff pay attention to certain threats to validity. Since there was no attempt to generalize the information gathered to a population, internal validity was the most critical issue. A modified pre and post-test format was chosen, with a control and intervention group. Constructs would be measured prior to any participation in the program, at the six month point in the program and approximately 3-4 weeks following the completion of the program. Research staff worked to explain the various concerns to the school, LCDSS and program staff in an effort to control as many of these problems as possible. Internal validity Concerns A quasi-experimental design reflects dissatisfaction with the ability to utilize experimental techniques in the 83 real world (Cook 8 Campbell, 1979). But the inverse of this problem is the dissatisfaction in non-experimental designs to infer causation. The evaluation staff attempted to locate and address threats to internal validity, in order to maximize our ability to infer causation from the outcome results of the evaluation. Several threats to validity were immediately apparent in this particular intervention design. Because the pilot project was in one school, there would be several occasions where the children in both control and intervention groups would be in the same classroom. Diffusion or imitation of treatments, compensatory equalization of treatment, and compensatory rivalry by control group members were all potential threats to internal validity (Cook 8 Campbell, 1979). Numerous meetings were held with the personnel from the school to discuss these problems, with evaluation staff encouraging them to address the needs of the children in the same way they would if there were no project. While it was not feasible to expect that children with identified problems would receive no other services, we encouraged them to follow regular and customary practices, as opposed to attempts to compensate. While these conversations could certainly not guarantee results to our liking, it made staff aware of the issues. The threat of maturation is always a problem, particularly when dealing with children of such young ages 84 (Cook 8 Campbell, 1979). We choose the control/ intervention group design in an effort to control for the effect of maturation. We were also concerned about the high mobility of the population and the subsequent threat of subject mortality. We planned to monitored the characteristics of the participants who dropped-out of the program, and examine those differences at the time of data analysis. DEFINITION OF CONSTRUCTS Using the goals developed in consultation with the designers of the intervention, the following constructs were identified: impulse control and behavior management, social skills, and self-esteem. The desired outcome was that the children and families participating would see an improvement in these three areas. Impulse Control and Behavior Management The literature used for this aspect of the intervention suggests that there is an important connection between anti-social or oppositional behavior in young children and the development of later delinquent behavior. In recent decades a number of investigators have reported evidence that antisocial acts in childhood or 85 early adolescence are related to antisocial acts later in life. (Loeber, p. 1432) This literature consistently supported the concept that efforts must be made to teach children to exhibit socially acceptable behavior in order to foster a positive and healthy experience, particularly in school. The method to teach these skills to children adopted by the intervention designers was a cognitive-behavioral approach. This choice is particularly logical in view of the perspectives of the various participants in the initial development of the project. Literature exploring cognitive development, cognitive-behavioral treatment, and limit- setting, was the foundation of the approach developed in the intervention (Carrol and Steward, 1984; Kettlewell and Kausch, 1983; Hoorwitz, 1985; Patterson 8 Stouthamer- Loeber, 1984). ’ While some of the assumptions underlying the program related to the ability to assist children in developing impulse control, others related to the importance of teaching parents how to control their children's behavior as a way of developing internal controls. Setting limits and behavior management is an important part of parental responsibility. Disruptions or omissions in the parents’ application of family-management practices have consistently been found to correlate with preadolescents anti-social behavior (Patterson 8 Stouthamer-Loeber, 1984). 86 Studies have shown that children who are socialized aggressively will tend to use aggression to meet their needs (Perry, Perry 8 Rasmussen, 1986). Clearly the concept of impulse-control and behavior management reflect, for the purposes of this project, the degree to which children are able to exhibit socially acceptable and appropriate behaviors in particular kinds of settings. The evaluation had to define those behaviors and determine a way to measure them. Social Skills The focus in this portion of the intervention design was on a concept of "sociocultural deprivation" (Leif 8 Zarin-Ackerman, 1980). In contrast to the socially rich family environment stands the "socially impoverished" one, in which the parent-child relationship is denuded of enduring supportive relationships and protective behaviors, deprived of both nurturance and feedback, the essential elements of support systems (Garbarino 8 Sherman, 1980). The children and families referred to this program were considered to be in this category, and the effects of their "socially impoverished" state were poor social skills, few supportive peer relationships, and isolation. One area of particular interest in the literature cited for the intervention is the ability of children to establish positive peer relationships in school and the impact that 87 ability or lack of ability has on the child’s performance (Asher, Hymerl, 8 Renshaw, 1984). Again, the designers of the intervention choose a cognitive-educational approach to address this particular concept. Using the concepts of learning to talk about problems and the ability to problem-solve, they developed an intervention that provided children and their parents with opportunities to learn the cognitive skills to improve their ability to perform these tasks (Bretherton, Fritz, Zahn-Waxler, 8 Ridgeway, 1986; Rickel, Eshelman, 8 Loigman, 1983). Child Self-esteem Although the concept of self-esteem is consistently mentioned in all of the documents used to develop the intervention, there is virtually no supporting literature regarding the intended conceptualization on the part of the designers. They used the term as if it was a given, with no definition needed. In the process of Operationalizing the concept and attempting to locate appropriate measures, the research staff found a great deal of confusion and difficulty in adequately defining the construct. Many questions emerged in the process of reviewing the literature on self-esteem. First, was self-esteem an 88 appropriate outcome measure, or was it a determinant for other outcome measures? The causal direction is still a question (Eccles, 1983). It also became clear that there is a global concept of self-worth, but there are also other aspects of self-concept that had been developed. There is a personal self-concept seen from one's own perspective that could contain any number of aspects or domains, such as physical self-concept, behavioral self-concept, internal self-concept, gender identity, racial identity, or socio- economic identity. Also there are other types of self- concept that are not seen as individual or personal. These are related to the perception of self in terms of others or other concepts that should be equally important. For instance, social self-concepts, ideal self-concept (or what you wish to be), and the perception of significant others in regard to self-concept (Harter, 1981). All of these possibilities made it clear that the concept of self-esteem or self-concept had not been sufficiently developed to allow for appropriate operationalization and measurement. At this point, however, the intervention design staff was no longer involved and the private agency intervention staff was beginning to implement the program. Attempts to clarify the concept further were met with frustration, as the program staff assumed these issues were already decided, 89 and other participants did not understand the significance of further clarification. Consequently, the research staff went about the process of defining the concept while the implementation of the intervention got under way. It was not possible to delay the implementation, as delay had already created a problem. The program proceeded with an informal, unstandardized instrument measuring self-concept, while the research staff pursued the development of the construct and the location of other, more rigorous measures. In this process there emerged several frameworks that were very applicable to this project. First, there were several authors who identified a specific academic self- image. Entwisle suggested that younger school-aged children are "forging their image of an academic self" (Entwistle, 1983, p. 1) Her interest was in the types of information they take into account in this process. Entwistle also raised other issues that were particularly pertinent to this effort. What were the influences of the evaluations of others, and how did this developing academic self image affect school achievement. Children’s development of an academic self-image impinges on the transition to full-t1me schooling for two main reasons. First,...a child's notion of "self" can be seen as one outcome of schooling. Second, how the child conceptualized "self" influences other outcomes (Entwisle, et al., 1983, p. 1-2). 90 The other issue that became paramount in our exploration of self-esteem is the importance of the variety of ages included in this project. There were a number of standardized instruments to measure the self-esteem of pre- adolescents or adolescents, but very few to measure the self-esteem of younger children. Even when measures were available for younger children, they were frequently not appropriate for use with kindergarteners or first graders. Several authors have taken a "life-span perspective" on self-worth (Harter, 1988, p. 1). In their discussions, they maintain that self-worth changes across various developmental stages. They differentiate between the concept of global self-worth and what is referred to as "domain specific evaluations of one's competence or adequacy" (Harter, 1988, p. 1). According to Harter, self-concept is multidimensional as opposed to unidimensional, consisting cf various domains of self evaluation (Harter, 1988 p. 3). Harter sees the development of these domain specific evaluations as precursors to the development of global self-worth. Consequently, she developed self-esteem scales, or self- perception scales as she calls them, which measure the child's perception in these various domains. Younger children (pre-school through the 2nd grade) are not able to verbalize or report their concept of self-worth. "Young 91 children to not have a verbalized concept of their self- worth, as tapped by self-report measures" (Harter, 1988, p. 6). Research staff decided to utilize the Harter concept of self-worth, including the various domains, as well as her instruments. This more multidimensional perspective of child self-esteem, as well as a consideration of developmental concerns, made this the best choice. These were introduced into the project at the time of the second data collection in the first year, and would be used at all three data collections in the second year. Adult Self Esteem The intervention design included a goal of improving the self-esteem of adults. There are many more standardized instruments in this area, although there continues to be much debate regarding the definitions of the concept (Damon 8 Hart, 1982; Harter, 1982, 1988; Rosenberg, 1979). The design of the intervention included the expectation that the parents would benefit from the parent groups in more ways than improved parenting skills. It was also hoped that the group interaction would increase the parents' self-esteem and self-confidence (Auerback, 1968; Blechman 8 McEnroe, 1985; Gordon, 1970). Evaluation 92 staff proceeded to review a number of instruments to measure this construct. It was decided that the Rosenberg Adult Self-Esteem Scale would be used in the project in view of the fact that the instrument has been standardized across many populations of varying characteristics. OPERATIONALIZATION OF CONSTRUCTS The goals and constructs outlined, the next step was to operationalize the constructs. This involved an identification of appropriate measures of the identified areas, as well as a plan for administration of these instruments. The following chart provides an outline of the instruments and collection process, which the following narrative describes. 93 CONSTRUCT Impulse control and behavior management Social Skills Child Self- Esteem Adult Self- Esteem MEASURE CBCL TRF CBCL TRF Cautela Self- Esteem Scale Child Self- Perception Scale Rosenberg Self- esteem Scale CITATION Achenback 8 Edelbrock, 1983 Achenback 8 Edelbrock, 1986 Achenback 8 Edelbrock, 1983 Achenback 8 Edelbrock, 1986 Cautela, Cautela, 8 Esonia, 1983 Harter, 1982 Rosenberg, 1979 COLLECTED Wave 1, 2, 3 Wave 1, 2, 3 Wave 1, 2, 3 Wave 1, 2, 3 Wave 1, 2, 3 Wave 2, 3 Intervention - Wave 1, 2, 3; Control - Wave 2, 3 Figure 5. Operationalizing Constructs And Instrument Selection Child Behavior Checklist (CBCL) and Teacher Reporting Form (TRF) The CBCL and TRF were designed by Thomas Achenback and Craig Edelbrock. designed to measure a variety of behaviors. These are variations on an instrument The forms were developed initially because "the study and treatment of child and adolescent psychopathology have been handicapped 94 by a lack of standardized procedures for describing the relevant behavior" (Achenback 8 Edelbrock, 1983, p. 1). While it is true that this particular study is not dealing with issues of child psychopathology, there are several characteristics of the instrument that make is very appropriate for this situation. First the instrument has been standardized, including reliability and validity information, as well as norms for children from certain age groupings. The instrument is also concerned about children's behavior in both the home and school environment, a critical advantage from the perspective of the EDPP. Also, this is a relatively simple instrument to administer, making the cost and ease of its use a strong positive. Finally, the itemization of behaviors was thorough and appropriate for our sample. The initial development of the CBCL began with a collection of behaviors in their children that parents report as a problem. These behaviors were derived from parents, earlier research by the authors, clinical and research literature, and consultation with clinical and developmental psychologists, child psychiatrists, and psychiatric social workers. After numerous testings, the list of 118 behaviors presently in the CBCL were the result. Behaviors are ranked on a scale of 0 to 2, with 0 meaning the item is not true of the child, 1 meaning it is 95 sometimes true and 2 meaning it is very true or often true. The higher the score on the CBCL, the more problematic the child's behavior. In addition to the behavioral section, the CBCL includes a section on social competence. This included information from the parents about the child's social activities, such as clubs, sports and friends, as well as chores and relationships within the family. Parents indicate the clubs, sports or chores their child participates in - to a maximum of four. Then they indicate the frequency of their child’s participation in the activity, as well as their assessment of the child's competence in comparison to other children his/ her age. This was a compromise to attempt to compensate for the problems in other approaches to measuring social competence. Frequency, for instance, does not necessarily address issues of skill, and vice versa (Achenback 8 Edelbrock, 1983, p. 6). Because the original intent of the instruments is to develop behavioral information around psychopathology in children, some of the behaviors in the 118 were not appropriate for this sample, resulting in frequent 0 scores on those items. But many of the behaviors were consistent with ones the parents and teachers had already identified on the initial referral form, to be discussed later. 96 The Teacher Reporting Form (TRF) was developed using the CBCL as a base. Behaviors in both the behavioral section and the social competence section that were not appropriate for school were substituted with behaviors that are more traditionally concerns of teachers and other school personnel. The remainder of the instrument has been kept the same. The instrument is scored identically to the CBCL, and a higher score is reflective of more problematic behaviors. Behavior Problem Scales were developed from the 118 behavioral items in both the CBCL and the TRF. These scales are a further delineation of certain psychopathological characteristics in children, developed and supported by clinical samples. The initial sample used to develop the scales was 42 mental health intakes in a variety of private and public, inpatient and outpatient settings. Using this sample, factor analyses were made of the 118 items on the CBCLs that were completed. The result was the development of 9 behavioral problem scales, with slight differences between boys and girls. This sample was predominantly white (81.2%) and of average SES (4.1 on Hollingshead’s seven categories of socioeconomic stratification) (Achenback 8 Edelbrock, 1983, p. 12). A similar process was developed in relationship to the TRF. In this instance, scales were determined through a 97 sample of TRFs for children referred for behavioral and social-emotional problems in 1979-1984 school years in 29 school and mental health services located in the eastern, southern and midwestern United States (Achenback 8 Edelbrock, 1986, p. 13). The result was 8 TRF behavioral problem scales for boys and 9 for girls. { Both these sets of problem behavior scales were normed. For the CBCL interviewers went to randomly selected homes in Washington, D. C., Maryland, and Northern Virginia. Interviewees were randomly selected using census track data to obtain a sample that approximated the SES and racial distribution of the clinical sample. For the TRF, the norm scores were determined through a sample of 100 non-referred children, randomly selected for teacher completion of the TRF in schools in Nebraska, Tennessee, and Pennsylvania. While the behavioral problem scales and their norm scores are not applicable to the EDPP sample, the design collection of behavioral information is very appropriate and should provide information important in ascertaining the impact of this intervention. The "Total Behavior Problem Score," which is the sum of the 1s and 2s on the completed form, could be helpful, and normed scores are available for that item. The total behavior problem score for the TRF is computed the same way and has the same value. 98 The total behavior problem score can be viewed as representing a dimension of behavior problems analogous to the construct of general ability represented by total scores on intelligence tests (Achenbach 8 Edelbrock, 1983, p. 70). The CBCL and the TRF have both been tested for several forms of reliability. Individual items on the CBCL had short-term test-retest reliability, more long term test- retest reliability (stability), interparent agreement and inter-interviewer reliability scores of over .90. The short-term test-retest reliability of scale scores was .89, and the interparent agreement was .66. Test-retest correlations for inpatients' scores over a 3 month period, or stability, averaged .74, and test-retest correlations for outpatients' scores over a 6 month period were in the .608. The TRF showed test-retest reliability over a relatively short period at .90. Over 15 days, the correlation dropped to .84. Over longer periods, i.e. 2 month intervals, the reliability was .68. Both the TRF and the CBCL have also been tested in a variety of ways for validity. The CBCL has been evaluated for content validity by examining the relationship of the scores of individual children to their perceived clinical status by parents and mental health workers. Construct validity addresses the concerns raised by the lack of constructs regarding behavior indicators of certain 99 psychopathological problems. However, the CBCL scores were compared to other instruments which appear similar to the scales used in this instrument. In these comparisons, scores comparing the CBCL and other widely used parent rating forms are as high as those typically found between tests of general intelligence (Achenbach 8 Edelbrock, 1983, p. 70). Criterion-related validity, which could be particularly important for our sample, showed no major differences between the samples used in designing the instrument and demographically matched referred and non- referred children. There was, however, a more significant effect for SES than any other demographic aspect, but even this was small (Achenbach 8 Edelbrock, 1983, p. 70). Most of the items in the TRF were derived from the CBCL, so the same issues of content validity would apply. Correlations between the TRF and similar scales of other instruments, particularly the Conners Revised Teacher Rating Scale, scores ranged from .62 to .90. The CBCL and the TRF are important instruments in measuring changes in the behaviors of the children, as perceived by the parents and the teachers, and also in looking at the social skills of the children. 100 Child Self-Perception Scales The program began with the Cautela Self-Perception Scale. This is an informal, unstandardized instrument developed by the authors for use in clinical settings. It is meant to be interactive between the clinician and the client, in an effort to see the clinical progress being ; made by the client during treatment. It consists of 56 items, all positively phrased, for which the child is to answer if the statement is not at all true of them, true a little or very much true. Some examples of statements are "I am good," " I am smart," or People can learn things from me" (Cautela, J. R., Cautela, J., and Esonia, S., 1983). A higher score reflects a higher self-perception. For the purpose of the project, the scale was abbreviated to 26 items and administered to all of the children participating in the program - intervention and control. It became immediately apparent that the children were providing socially desirable responses, as all of the responses were the highest score. Little analysis has been done of this instrument, as the research staff immediately set about locating another one. However, this is the only measure of self-esteem for the entire first year. Consequently, there will be some discussion of the results for the first year. 101 As has been previously discussed, after a review of the literature, the Harter Child Self-Perception Scales were chosen to measure child self-esteem. This scale has five versions that would be used for this population. First there is the pre-school/ kindergarten version - one for boys and one for girls. Then there is the lst and 2nd grade version, one for boys and one for girls. Then there is another version for 3rd through 6th graders (both girls and boys) that is written and not pictorial. All of the scales contain domain areas. This model maintains that children do not feel equally competent in all domain areas. In developing the 3rd through 6th grade scale, the items represent those behaviors most relevant for elementary school children. These were then expanded through interviews with children. Three separate domains emerged; cognitive competence, social competence and physical competence. The items were analyzed in a factorial analysis and four subscales were established. The first three were directly related to the domains already identified. The fourth subscale is that of global self-worth. Subscale reliability was assessed by employing coefficient alpha, which provides an index of internal consistency. These values ranged from .75 to .83, .75 to .84, .77 to .86, and .73 to .82 for the cognitive, social, 102 physical and general subscales. Test-retest reliability over a 9 month period showed correlations of .78, .80, .87, and .70 for one sample and .78, .75, .80, and .69 for another across the four subscales. The question format of the instrument makes statements about children in general and asks the child which characteristic is more or less like him or her. For example, the child is presented with the statements "some kid often forget what they learn but other kids can remember things easily." They are asked which kid is the most like them. Following their answer to this question, they are asked if that is really true for them or sort of true for them (Harter, 1982, p. 89). The responses are given a one through four score and a higher score reflects higher self-perception. This question format is a "structure alternative format" attempting to deal with the effect of socially desirable responses (Harter, 1982, p. 89). This format legitimizes either choice, as 50% of the children (theoretically) could be in either grouping. "Our confidence in this format is bolstered by the fact that children's verbal elaborations on the reasons for their choice indicate that they are giving accurate self- perceptions rather than social desirable responses" (Harter, 1982, p. 2). 103 Following the development of the instrument for 3rd. through 6th graders, the author began the downward extension of the scale. It was at this point that a pictorial scale was developed. The question format has remained the same as the one previously identified, but the picture format "allows us to depict skills and specific activities concretely" (Harter 8 Pike, 1984, p. 1970). It also engages the young child's interest better than a written scale (Harter 8 Pike, 1984, p. 1970). There is no global self-worth scale on the younger version of the instrument, since the author maintains that younger children do not have the ability to verbalize a concept of global self-worth, but rather have ideal self- images at that age (Harter 8 Pike, 1984, p. 1970). There are two general constructs, perceived competence and perceived social acceptance, with two domains within each of the constructs. The perceived competence construct consists of the cognitive competence and physical competence domains, and the perceived social acceptance construct consists of the peer acceptance and maternal acceptance domains (Harter 8 Pike, 1984, p. 1970). The author is clear that the younger scale should be viewed as a pre-cursor to self-esteem as opposed to an index of self-esteem (Harter 8 Pike, 1984, p. 1971). Subscales were determined in the same manner as the self- 104 perception scale for older children. Individual subscale reliabilities ranged from .50 to .85. Combined in their particular construct areas, the reliabilities increase substantially to .75 to .89. The reliability of the total scale is in the mid to high .805 (Harter 8 Pike, 1984, p. 1971.). In testing the validity of all sets of instruments, the author looked at a variety of types. In testing convergent validity, the children were asked to give reasons for their answers in various areas. In both the cognitive and physical domains, for the younger children, they were readily able to supply answers. In relationship to discriminant validity, children with certain characteristics that should result in lower scores were compared with children without those characteristics. For instance, children who were preterm infants should score lower in the physical domain. Such differences were found in both teacher and child responses to the instrument (Harter 8 Pike, 1984, p. 1978). In the older children, there were variations in validity for the various domains. For instance, in the cognitive domain, convergent validity was tested by comparing the children's perception and the teachers. The correlations improved significantly as the age of the child got older. The same results were seen when the scores of 105 perceived competence were compared to standardized achievement test scores (Harter, 1982, p. 92). In the social domain, the subscale score was compared to the Roster and Rating Scale developed by Roitascher. The correlation was found to be .59 (Harter, 1982, p. 94). The Harter Self-Perception Scales are an appropriate choice to measure the self-esteem of the 3rd graders, and demonstrating the pre-cursors to self-esteem of the kindergarten through 2nd graders. It also provides information regarding the social skills of the children and how they relate to their peers and their family. Adult Self-Esteem Scale The Rosenberg Self-Esteem Scale (RSE) is a ten item scale used to measure an individual's self-esteem. The concept of self-esteem used here is "general favorable or unfavorable global self-attitude" and would be comparable to Harter's concept of global self-worth (Harter, 1981; Rosenberg, 1979). The scale was originally developed on high school seniors, but has been used extensively with adults of all ages (Rosenberg, 1979). The RSE has been tested for internal reliability. Rosenberg’s original reliability coefficient score was .92 (Rosenberg, 1965). Subsequent reliability testing by Silber and Tippett (1965) illicited a reliability score of .85. 106 Construct validity was tested by comparing the RSE with a 6 item Guttman scale of depressive effect and a clear relationship between the two instruments was found (Rosenberg, 1965). The other comparable test of construct validity was to compare the ratings of peers to the RSE scores. "Those with high self-esteem were more likely to obtain high sociometric ratings from their peers (Rosenberg, 1965, p. 26).. The RSE also shows convergent validity with measures of the same construct using different methods. The RSE was correlated with the self-ideal discrepancy score (r =.67), the self-image questionnaire (r =.83), and to the psychiatrists rating (r =.56) (Rosenberg, 1965). The instrument is extremely easy to complete and the readability is eight grade (Rosenberg, 1965). A four point likert scale of strongly disagree, disagree, agree and strongly agree was used for this sample. The higher the score, the more positive the client's self-esteem. Family Observation Form Initially, concern was expressed about capturing information regarding the interaction of the family during the project. Research staff explored several instruments that would allow an opportunity to gather information about 107 the impact of the intervention of the interaction of the families in the home environment. While there are several good instruments available, all of them required extensive observation and transcription, which required additional personnel and resources. Since resources were limited, research staff, in consultation with the program staff, designed a reporting form called the Family Observation Form. This form provides quantified information regarding the interaction between parents and children during a quarterly home visit of the worker/ researcher. It was the intent of this form' to look at the positive and negative interaction patterns between children and parents and mothers and fathers. It was also intended to capture the relationship between the worker/ researcher and the client. Interviewers were asked to evaluate these areas and then give their evaluation a number value on the form. Initially the form was completed by the program staff, as they were the collectors of the first wave of data. Following that, research staff collected all data, including the Family Observation From. There have been frequent problems in getting information requested on the form, as it is common for only the mother to be present when research staff goes to the home to collect the information. While the original intent of the form was 108 important, the actual data available has proven to be very limited. Process Measures In addition to the quantitative outcome measures mentioned above, there were several process measures developed. Because the research staff had no supervisory responsibility for the intervention staff, a form was developed to provide the research staff with information about the process of implementation of the intervention. Research staff received an outline of each group session, an attendance list (to enable participation rates to be determined), a group summary form regarding interaction within the group, and a group summary form on each child participating in that group session. The same is received for the parent groups. This information allows the research staff to glean an understanding of the various subjects that are discussed and provides a frame of reference for certain problematic behavior. The final data collection item was a family interview that was completed during the second wave of data collection. This interview attempted to gather information from the family regarding their feelings about the child, the child's school and its personnel, the community and CHAPTER FIVE DATA COLLECTION, ANALYSIS AND RESULTS SAMPLE AND DATA COLLECTION Referrals to the first year of the intervention program were made by the children's teachers. Teachers were generally considered by the professional stakeholders to be the most appropriate referral source, able to identify children having problems, and to describe those problems behaviorally. Evaluation staff first drew up an initial list of behaviors created from information in school documents and evaluation forms, as well as from the previous project of the City School District. Then they met with the teachers and modified the list of behaviors to include all those that were recognizable and appropriate to use in referring the children. In a meeting with all of the teachers who would be referring children, evaluation staff facilitated development of concrete definitions for all of the behavioral terms. For instance, the description "inability to pay attention" was defined in terms of the length of attention span a child should be expected to have for her or his age. These types of definitions were an attempt to standardize the teachers' evaluations and remove as much subjectivity as possible. These definitions were attached to the Initial Referral Form together with the directions and guidelines (see 109 110 Appendix B). Teachers were encouraged to refer to the definitions in order to determine how to rate a particular student. Evaluation staff encouraged teachers to refer students who were in initial stages of difficulty, as well as those with problem behaviors of longer duration. This was done to reinforce for the teachers the emphasis on prevention, and therefore the desire to identify children before the problems had developed too far. In addition it was hoped that this would result in a balanced pool of subjects. Because of contract delays that postponed the start of the program, referrals were submitted in June of 1987, September of 1987, and finally, in January of 1988. During these three periods combined, teachers submitted 78 referrals. Of those, 40 had to be dropped for one or more reasons: seven were too old by the time the program began, 18 were not receiving public assistance and were therefore not eligible, 14 moved out of the district before the program could begin and the parents of three children were not interested in participating. The parents of the remaining 38 children were contacted about interest in the program, and when they agreed to participate, they were asked to sign a consent form. Among the children whose parents consented, 11 were female and 27 were male. Thirteen were third graders, nine were second 111 graders, six were first graders, and 10 were in kindergarten. Because of a high mobility rate in the school and the resulting threat to sample size, the participating agencies agreed to have 26 children in the experimental group. Families were then informed that 26 children would be selected to receive the intervention immediately, and that the remaining 12 would be placed on a waiting-list to receive the intervention the following year (i.e. control group). If enough children left the program to bring the total number below the 20 contractually required, children would be randomly selected from the wait-list group to return the number to 20. Otherwise, wait list children who did not receive the services of the program in Year I were promised placement in the program in Year II if they were still interested in participating. The Initial Referral Forms were scored as a way of ranking the children prior to group selection in order to increase the likelihood of equivalence between the experimental and control groups relative to behavioral problems. The scores of the 38 children ranged from 32 to 81. The higher the score the worse the behavior and a 105 was the highest score possible. The mean score of the children referred was 65. The intervention and control groups were selected as randomly as possible. After the children were ranked by 112 score, every third child was selected out for the control group. Fortunately, this process resulted in no split of sibling groups. Table 1 describes the sample. 113 Table 1. Demographic Characteristics of the Sample Characteristics Intervention Control Total N a 26 N = 12 N = 38 N % N % N % GENDER Male 20 74% 7 26% 27 71% Female 6 23% 5 42% 11 29% ETHNICITY White 17 65% 7 58% 24 63% Black 5 19% 4 33% 9 24% Hispanic 4 15% 1 8% 5 3% AGE 5 years old 4 15% 0 0% 4 11% 6 years old 6 23% 3 25% 9 24% 7 years old 4 15% 2 16% 6 16% 8 years old 3 12% 7 58% 10 26% 9 years old 8 31% 0 0% 8 21% 10 years old 0 0% 0 0% 0 0% 11 years old 1 4% 0 0% 1 3% GRADE Kindergarten 8 31% 2 17% 10 26% First Grade 5 19% 1 8% 6 16% Second Grade 6 23% 3 25% 9 24% Third Grade 7 27% 6 50% 13 34% 114 Although all parents were included in one parent group, program staff divided the intervention children into two groups. Group I was for the kindergarten and first graders and Group II was for the second and third graders. During the course of the year, Group I lost one child from the program. Group II lost four children. These children were not replaced as the total number of children in the intervention did not drop below 20. The control group lost two children by the end of Year I. Data from the Local Private Agency Program Staff indicate there were 86 sessions during an eleven-month period for Group I. This younger group had an average participation rate of 69%, with the summer months being the lowest attendance period. Group II had 84 sessions during an eleven-month period, with an average participation rate of 89%. The parents' group met for a total of 26 sessions, with a participation rate of 34%. The attendance in the children's groups and parent group was somewhat lower than had been hoped. The documentation provided by the program staff outlined the content and activities of the various groups in the intervention. Review of this material indicates that the content of the groups was consistent with the intervention design as described in Chapter Three. The evaluation design called for data to be collected at three time periods, once prior to the start of the 115 groups (Time 1), once at the six month point (Time 2) and a final time one month following the end of the groups (Time 3). The attached flow chart describes the data to be collected and the time frames for their collection. (Appendix C) Through interagency discussion, the decision was made that the program staff would collect the first set of data rather than evaluation staff. Program staff were required to make at least one home visit, meet each child and speak to the teachers prior to the start of the groups. In an effort to keep the number of people involved and contacts to a minimum, program staff completed the Time 1 instruments during these contacts. Evaluation staff gathered the data at Time 2 and Time 3. Ten undergraduate and graduate students were recruited to collect data. These students met with each family to complete the CBCL, the Rosenberg Self-Esteem Scale and the Family Interview. Following the interview with each family, the family observation form was completed by the researchers. Training for the these students consisted of one two-hour group meeting and individual meetings with the evaluation staff to review the material and answer any questions. The family interviews took an average of one hour each, depending on the number of children in the family and the 116 family situation. In many instances, staff had difficulty locating and making appointments with the families. Several families did not have phones and several trips were necessary to find the family at home. At each of the data collection times the process took four to six weeks instead of the two to three weeks planned. At Time 2, for reasons that have been discussed in Chapter Three, a new self-esteem scale (Harter Self- Perception Scale) was added to the data collection. Because the original scale (Cautela Self-Esteem Scale) was used at Time 1, it was continued for Time 2 and Time 3 to provide a base of comparison. Two child development students with expertise in early school-aged children were hired to administer both self- esteem instruments. These students went through a two-hour training, administered the test on volunteers for practice, and then completed the self-esteem scales on all of the children in the control and intervention groups. Both self-esteem instruments took a total of approximately 20 to 30 minutes per child. The original plan for data preparation was to hand- score the instruments, due to the small sample. The complexity of the instruments soon made it apparent that this would be too time consuming, so the decision was made to enter all data in its raw from, and then calculate the scores and analyze the results by computer. 117 Staff built an appropriate data base for each instrument using SPSS (Statistical Package for the Social Sciences, 1988). Data were then entered for each child on each instrument, examined for errors or inconsistencies, and prepared for analysis. Table 2 describes the data collected for the three time periods, with the numbers indicating the number of children for whom data were collected on that instrument at each time period. These numbers thus define the relevant sample sizes for the group comparisons presented in Tables 3 through 12 below. 118 Table 2. Sample Sises on Instruments at Three Time periods Time Period Instruments 1 2 3 Child Behavior Checklist (CBCL) Intervention 26 22 16 Control 12 10 6 Teacher Reporting Form (TRF) Intervention 25 26 19 Control 12 12 7 Cautela Self-Esteem Scale Intervention 24 19 18 Control 11 10 10 Harter Self-Perception Scale Intervention - 19 18 Control - 10 10 Rosenberg Self-Esteem Scale Intervention 14 16 12 Control 1 10 7 P—u ...—a.m 1 .— 119 Variations among the sample sizes across instruments in the same time period are due to two factors. In some cases the data could not be collected because families could not be located or did not wish to participate. In other cases, the data collected were incomplete (some items were missing) and these cases could not be included in the analysis. Also, there was a problem with collection of the Rosenberg during the first data collection, as program staff did not understand that the instrument was to be administered to the control families as well as the intervention families. By the time this error was discovered, the groups were well underway and it was determined that it was not appropriate to collect the data at that time. ANALYSIS Reliability and Horns Data analysis began with an examination of the internal reliability of the instruments for this sample. Cronbach's Coefficient Alpha was utilized to measure internal consistency. The CBCL and the TRF were analyzed in terms of the total scores. The subscales are groupings of behaviors 120 reflecting certain psychopathology. Because our sample was referred as a preventive measure and children in this program were not considered seriously mentally ill, the total score was a more appropriate behavioral measure for this sample. The internal reliability of these scales was very high for all three time periods, ranging from .87 to .96. The Cautela Self-Esteem Scale had an internal reliability of .78 to .88 and the Rosenberg Self-esteem scale ranged from .59 to .81. All of these figures indicate that the instruments used for this analysis were reliable for this sample within acceptable limits. The CBCL and the TRF Total Behavior Problem Scores were normed with a non-clinical sample. On the CBCL, 40 points is the upper limit of the normal range for boys and 42 for girls. The mean score of our total sample was 47 at the time the groups began. Scores were not computed separately for boys and girls because there were so few girls in the sample. The TRF Total Behavior Problem Score was also normed, and in this instance the upper limit of the normal range was 49 for boys and 42 for girls. The mean score for our sample was 50 at the time the groups began. Because our sample was not a clinical mental health sample, one would expect the scores of these children to be close to the normal range scored by researchers during the 121 norming process. While the average scores were slightly higher than the upper limit of the normal range, our sample is clearly not as disturbed as the children in the samples used to design the instrument. This affirms the decision to utilize the Total Behavior Problem Scores as the indicators of behavioral difficulties as opposed to the scales measuring psychopathological behavior or other elements of the instrument. Control and Intervention Comparisons The analysis most statistically relevant in the quasi- experimental design used in this evaluation is the comparison of the control and intervention groups across the three time periods. T-tests comparing the control and intervention group mean scores at each time period are reported in the following tables. For all of the behavioral comparisons, the hypothesis is that the intervention group Scores at Time 2 and Time 3 will be significantly lower (indicating fewer behavioral problems) than the control group scores at Time 2 and Time 3. MW). Data from the CBCL administered at the three reporting periods are presented in Table 3 below. 122 Table 3. CBCL Scores by Group at Three Time Periods Intervention Control T-test 2-Tail Mean Score Mean Score Statistic Prob Time 1 43 58 -1.98 .056 | Time 2 41 51 -1.44 .151 i” These results seem to indicate that despite random assignment from a list of children ranked by scores on a comparable instrument, results approached a statistically- significant difference between the control and intervention groups at the time the groups began. The control group had much higher scores on the CBCL than the intervention group when the groups began. This raises questions about the design's ability to find a statistically significant difference between control and intervention groups as a result of the intervention because the intervention group had a mean score (43) only slightly above the means for "normal" children (40 for boys and 42 for girls). This gave what might be considered an unfair disadvantage to the experimental group. The intervention would have had to be very potent to have an effect on children whose behavior, as measured by this instrument, was virtually "normal" to 123 begin with, whereas the control group in one sense had "no where to go but up!" Otherwise known as regression to the mean, it is not surprising that the differences between the groups actually lessened as the intervention proceeded. The differences in CBCL scores at Time 1 could be attributed to the fact that the control group was older than the intervention group and older children are more readily identified as having behavioral problems that would be measured on the CBCL. The control group had more 2nd and 3rd graders and fewer kindergarten and first graders. The median age in the control group was eight years old, while it was seven years old in the intervention group. While the behaviors of the younger children might have been interpreted as problematic in school, as evidenced by the teachers' evaluations on the Initial Referral Form used to identify potential participants in the program, they were not yet particularly troublesome to the parents as measured by the CBCL. Other issues regarding outcomes will be discussed following a presentation of all of the results. WW Data from the TRF administered at the three reporting periods are presented in Table 4 below. 124 Table 4. TRF Scores by Group at the Three Time Periods Intervention Control T-test 2-Tail Mean Score Mean Score Statistic Prob Time 1 50 49 .12 .906 Time 2 40 58 -l.92 .063 Time 3 48 61 -l.46 .158 In regard to this instrument, there was no statistically significant difference between the control and intervention groups from the teacher's perspectives at the time the groups began. The Intervention group appeared to make significant progress between Time 1 and Time 2, and the t-test value at Time 2 comes close to the .05 level of statistical significance. However, the score for the intervention group did not continue its progress from Time 2 to Time 3, and the differences between the control and intervention groups at Time 3 measurement dropped back to levels not considered statistically significant. When analyzing these results it is important to remember that the ratings completed between Time 1 and Time 2 were done by two different teachers, and across two academic years, due to the delay in contract negotiations. Time 1 data were collected in February, prior to the groups starting in March. Time 2 data were collected in October, 125 one month after the start of a new academic year. Time 3 data were collected in April of the second academic year, shortly before school let out in June. This schedule, in and of itself, has potentially had a major impact on the data collected, and consequently on the ability of the research design to detect improvements in the children assigned to the intervention group. The gelf-gseeem Scales Data from the Cautela Self-Esteem Scale (CSES) and the Harter Self-Perception Scales (SPS) used with the children are presented in this section in Tables 5 through 7. Data from the Cautela Scale for all three reporting periods are presented in Table 5 below. The hypothesis in these comparisons is that the intervention group scores at Time 2 and Time 3 will be significantly higher (indicating greater self-esteem) than the control group scores at Time 2 and Time 3. 126 Table 5. Cautela SES Scores by group at the Three Time periods Intervention Control T-test 2-Tail Mean Score Mean Score Statistic Prob Time 1 64 68 -1.42 .166 Time 2 68 7o — .57 .572 L Time 3 65 68 - .94 .356 i 5‘ Data reveal little difference between the control and ' intervention groups at any of the time periods, and also not much difference for either control or intervention across the three time periods. As discussed in detail above, it is suspected that the instrument was not sensitive enough to measure any change in the children's self-esteem. At Time 2, an additional measure of self-esteem, the Harter Self-Perception Scale, was added in an effort to increase our ability to measure the effect of the intervention on the children's perception of self. As was discussed above, the Harter Self-Perception Scale has three versions: pre-school] kindergarten, first and second grade and third through sixth grade. Because of the low number of kindergarteners, we were unable to statistically analyze that instrument. Data from the Harter Self-Perception Scales (SPS) are presented in Tables 6 (for grades 1 and 2) and Table 7 (for grades 3 through 6) below. 127 Table 6. Barter SPS For Grades 1 and 2 at Time 2 and Time 3 Intervention Control T-test 2-Tail Mean Score Mean Score Statistic Prob Time 2 *Cognitive Competence 21 21 .17 .871 Peer Acceptance 17 20 -.91 .385 Physical Competence 21 23 -1.21 .253 Maternal Acceptance 15 17 -.82 .433 Time 3 Cognitive Competence 22 22 0.0 1.0 Peer Acceptance 17 18 -.30 .769 Physical Competence 22 21 .12 .905 *Maternal Acceptance 15 15 .05 .965 *These values are not in fact equivalent. Therefore the T-test is not - to 0. They are rounded. _‘—.———‘ ....-i- I ‘ I 128 Table 7. Barter SP8 for Grades 3 through 6 at Time 2 and Time 3 Intervention Control T-test 2-Tail Mean Score Mean Score Statistic Prob Time 2 Scholastic Competence 16 17 -.80 .439 Social Acceptance l9 17 .74 .471 Athletic Competence 2O 16 1.49 .158 Physical Appearance 21 17 1.81 .090 Behavioral Conduct 16 13 .93 .368 Global Self-Worth 18 17 .46 .653 Time 3 Scholastic Competence 16 15 .22 .828 Social Acceptance 18 17 .52 .607 Athletic Competence 18 16 1.03 .317 Physical Appearance 18 20 -.90 .379 Behavioral Conduct 17 15 1.01 .328 Global Self-Worth 20 19 .84 .415 129 As can be seen from the results, even on this instrument there were no statistically significant differences between the control and intervention groups on any of the domains. This instrument did seem to provide more information about the self-esteem of the children in the sample from a wider range of perspectives, but we were L still not able to measure any significant differences between the control and intervention groups in regard to this construct. ' Adult golf-Esteem The final construct that was operationalized was adult self-esteem. The Rosenberg Self-esteem Scale (RSE) was the instrument chosen to measure this construct. Data from the Rosenberg Self-Esteem Scale (RSE) are presented in Table 8 below. The hypothesis for this analysis is that the intervention group scores for Time 2 and Time 3 will be higher than the control group scores for Time 2 and Time 3. Table 8. RSE Scores by Group at the Three Time Periods Intervention Control T—test 2-tail Mean Score Mean Score Statistic Prob Time 1 20.5 - - - Time 2 18.9 16.3 1.64 .115 Time 3 17.9 17.4 .24 .813 130 As discussed previously, this scale was not administered to the control group at Time 1, due to a missed communication between the MSU evaluation staff and the Agency Program Staff. Comparisons between control and intervention for Time 2 and 3 are shown, however, and there are no significant differences between control and intervention groups on this measure. Analysis of Group Attendance Additional analyses were done to determine if there was a correlation between attendance at the children's groups and the child behavior outcome measures and self-esteem measures at Time 3. The hypothesis for this analysis is that higher attendance will be correlated with lower scores on the behavioral measures and higher scores on the self- esteem measures. A Pearson Product-Moment Correlation was calculated to determine if the variable of attendance at the children's groups (ATTGRPS) was correlated with the four child outcome measures: Parent Child Behavior Checklist, Teacher Reporting Form, the Cautela Self-Esteem Scale, and the Harter Self-Perception Scale. The results of this analysis are presented in Tables 9 and 10 below. 131 Table 9. Correlation between group attendance (ATTGRPS) and TOTCBCL, TOTTRF AND CSES Variable Correlated Correlation Significance with ATTGRPS TOTCBCL -.2643 NS TOTTRF -.1378 NS CSES -.1113 NS The analysis summarized in Table 9 shows no significant relationship between group attendance and these three outcome measures at Time 3. However, while the correlations between the total CBCL and the total TRF are not statistically significant, both are in the hypothesized direction. This is not the case with the Cautela Self- esteem Scale, which potentially confirms the opinion of the research staff that this measure was not an appropriate measure of self-esteem for this sample. A correlation matrix was constructed to determine if the variable of attendance at the children's groups (ATTGRPS) was correlated to any of the various domains on the Harter Self-Perception Scale. The domains for this scale are as follows: scholastic competence (SCHCOM), social acceptance (SOCACC), athletic competence (ATHCOM), physical appearance (PHYAPP), behavior conduct (BEHCON), and global self-worth (GSFWTH). The data from this analysis are presented in Table 10 below. fifiimfid E'imu‘ga‘ , 132 Table 10. Correlation between group attendance (ATTGRPS) and Barter Self-perception Scale for Third to Sixth Graders Domain Correlated Correlation Significance with ATTGRPS scacou .1146 NS socncc .2896 ms L ATHCOM .3622 NS 3 PHYAPP .1390 NS T sancon .4441 NS GSFWTH .4441 ms In this analysis there is no significant correlation between group attendance and the scores on the subscales of the instrument. However, it is again true that the relationship between attendance and the self-esteem domains go in the hypothesized direction. Analysis of Academic Performance The impact of the intervention on the children’s academic performance was an important external outcome measure for the evaluation. However, the school does not give grades in kindergarten through second grade and therefore this measure was not available. The Teacher Reporting Form did include an evaluation by the teacher of 133 the child's academic performance in certain subjects. The teacher was to indicate whether the child was far below grade level(1), somewhat below grade level(2), at grade level(3), above grade level(4) or far above grade level(5) in a number of subjects. The two subjects for which data were consistently available for all age ranges were math and reading. as .n‘; “1 These rating by teachers were used instead of grades as 1 the literature suggests these types of teacher ratings are good predictors of grades. Also, math and reading performance are two subjects frequently used as indicators of school performance (Eccles, 1983; Entwisle et al., 1984; Hogg 8 Coladarci, 1989). Consequently, these teacher ratings were used as outcome measures and control/ intervention comparisons were made for all three time periods using a t-test. The results of these comparisons are in Tables 11 and 12 below. Table 11. TRF Math Performance by Group at the Three Time Periods Intervention I Control T-test 2-tail Mean Score Mean Score Statistic Prob Time 1 2.0 2e4 -1e25 e221 Time 2 2.5 2.2 1.13 .266 Time 3 2.3 2.1 .41 .685 134 Table 12. TRF Reading Performance by Group at the Three Time Periods Intervention Control T-test 2-tail Mean Score Mean Score Statistic Prob Time 1 107 2e? -3e28 e002 Time 2 1e8 leg -e43 e670 *Time 3 1.8 2.0 - - * No variance in the control group mean. The results show several interesting contrasts. First, the only statistically significant difference between the control and intervention groups is in reading performance at Time 1. This difference is probably reflective of the older ages and higher grades in the control group discussed earlier in this chapter. While the difference in math performance at Time 1 is not statistically significant, the mean for the control group is higher than the intervention at Time 1 in this performance area as well, suggesting a similar phenomenon. The other difficulty with this data is the change in teachers between Time 1 and Time 2. The fact that the ranking is by a new teacher, evaluating the student at the beginning of a school year, and not a ranking by the teacher who referred the child, confounds the data, making any conclusions about the impact of the intervention based on this measure difficult to justify. 135 Analysis of Additional Data There are several results of this intervention that cannot and should not be measured through quantitative statistical analysis. These results are seen in the discussions and relationships that have evolved over the 1.... ..Lar. q . course of the intervention and are heard in the comments and feedback of the various participants. All of the teachers who had children involved in the intervention have communicated to evaluation staff that they saw improvements and positive changes in many of the children involved in the program. In a feedback session the evaluation staff had with the teachers, the support of the project was without exception. In fact, the one issue that the teachers complained about was that the children and families were not in the program longer than 12 months. There was also verbal and written feedback from the parents who participated in the intervention. Many of the parents thought the program was helping their child in terms of behavior and that they as parents were benefiting from participation. Even intervention group parents who did not attend the group meetings gave these assessments. Clearly the services offered to the families outside of the group model had as great an impact on these parents as the 136 group meetings (an issue to be discussed at more length later in this chapter). Relationships between the City School District, the Local County Department of Social Services, Michigan State University, and the Local Private Agency are certainly more active now than they were in the past. While there continue to be disagreements in direction and philosophy, the dialogue begun in conjunction with this program is a positive result of the effort. As is the case with the quantitative data, not all of the qualitative feedback is positive. There were several parents who, although they thought they and their children benefited from the program, felt they had been singled out as inferior or inadequate in some way, an assessment with which they strongly disagreed. This reaction should tell us something about how to proceed in future programs. If the programs were open to public assistance and een_petlie eeeietegee families, it would be more possible to combat these stereotypical responses. This would also provide the opportunity to look at broader issues regarding the impact of socioeconomic status on school achievement, an important topic that could not be addressed in this study. 137 DISCUSSION Clearly, the hypothesis was that the evaluation would reveal a significant difference between the control and the intervention groups on the various measures at Time 3. Three possible reasons for the absence of significant results will be discussed here: 1) the size of the sample; 2) the intensity of the intervention; and 3) additional components in the design that should be evaluated prior to replication or repetition of the program. Sample Size The small size of this sample limits the ability to find real differences between control and intervention groups, even if they might be there. This is simply an issue of statistical power. With the sample size this small, only a very strong effect would be detected by the measures used. Thus the absence of statistically significant results may simply indicate that the sample was too small to detect a real effect. The issue of small sample size is a common problem for pilot projects and one that must be considered when evaluating the lack of statistically significant results. Given the need to demonstrate significant results, many 138 studies use this type of experimental or quasi-experimental design even when the total sample size is small. The environment of new program funding almost necessitates this approach, and in this case the cooperating agencies agreed that this approach should be tried, but not in isolation. Other criteria and information about the implementation and impact of the intervention must be considered along with the statistical analyses of the various quantitative instruments in arriving at a valid assessment of the impact “in—...- mu. 1 A-a .' of the program. Intensity of the Intervention The second issue regarding the lack of significant results is the intensity of the intervention. There was concern on the part of the participants that the families did not receive a sufficiently intense "dosage" of the intervention as reflected in the lower-than-desired attendance rates. The children's groups had different attendance rates. Group I, for the younger children, had an average attendance rate of 69%. Group II, for the older children, had an average attendance rate of 89%. While both are relatively good, the attendance rate for the older children was substantially better than the younger children' group. 139 Because a larger percentage of the experimental subjects were younger, this lower attendance may have negatively affected the likelihood of the experimental intervention having an impact. The parents' group attendance was very low, with the average attendance at the parent groups being 34%. While attendance is frequently a problem with group activities for parents of lower-socioeconomic backgrounds (Chandler et al., 1986; deKanter et al., 1986; Gladow 8 Ray, 1986; Goodson 8 Hess, 1987; Holden et al., 1986), this was a particularly important concern for the participants in this program. The lower attendance rate for the parents could also have contributed to the lower attendance rate for the younger children, who frequently need the involvement and reminders of their parents in order to participate in after-school activities. Components of the Design The third issue regarding absence of significant findings is related to components of the intervention design that should be examined in future programs of this nature. It was clear that the intervention focused almost totally on the families, and left the school virtually 140 untouched in either participation in the intervention or collection of data regarding the atmosphere of the school and classrooms. To return to the Epstein model, this lack of information meant that the entire section dealing with the school system was missing from this analysis. Such a gap also made it difficult, if not impossible, to gain any r.— insight into the relationships between the family and the R school, an aspect of child achievement that we have seen to 9 be significant in other studies. [ The most important design issue that potentially affected the results of the evaluation is the group model of the intervention. While the attendance of the parents at the group meetings was quite low, there was still a great deal of contact between the program staff and the families in their homes. Services were offered to the intervention families that had an effect on parents' relationship with their children and the school, as reported anecdotally by all participants. Give the low attendance in the parent group, an important question is whether or not the group format of the intervention is the best format for providing the information contained in the parent group curriculum. The instruments and the evaluation design were attempting to measure change in behaviors that were specifically targeted in the greep focus of the 141 intervention. Since the attendance in the groups (particularly the parents) was so low, the impact was not found. But perhaps the program staff had an impact on other aspects of the families' functioning that the instruments and design of the evaluation were not set up to measure. This question raises some interesting issues regarding what form of intervention is most effective with the population targeted here. In other intervention designs, while the educational model is followed and the focus is on improving the behavior in school, and academic achievement of the children, the contact with the families and parents is more directly one-on-one and the intervention and training that occurs is much more directed toward the particular family situation (Brown, 1987; Gladow 8 Ray, 1986; Halpern, 1986; Washington 8 Oyemade, 1987). Parents and staff work together to assist the parents in developing the particular skills they need to improve the school performance of their children. These skills may specifically relate to the academic aspects of the child's progress, but may also relate to other social services that the family needs in particular circumstances (Brown, 1987; Holden et al., 1986; Levenstein, 1988; Washington and Oyemade, 1987). Although it is distinctly possible these types of services were offered to this group of 142 intervention families, they were not the services on which the evaluation focused. This points to the need for a more focused intervention model, in which there is more clarity about exactly the problem areas that are being addressed and the intervention strategy that is being used to address them. CHAPTER SIX SUMMARY AND CONCLUSIONS INTRODUCTION The process of designing and implementing an evaluation of the EDPP (Early School Aged Delinquency Prevention Program) led to an exploration of broader and perhaps more compelling questions. The development of this evaluation required an understanding of the context of prevention programs such as this, as well as the most pressing issues effecting such programs. This inquiry led to investigation of three areas or themes. First, how was the relationship between socioeconomic status and achievement defined in American society from a theoretical perspective? Second, what were the applied aspects of this social problem, in the form of intervention strategies and programmatic responses? And third, what criteria and methods should we be using to measure the effectiveness of these programmatic efforts? It became clear that there was conceptual similarity between the three bodies of literature and that they were dealing with many of the same issues and concerns. However, there was no indication that any of these perspectives had incorporated the ideas and experiences of the others. This dissertation has discussed these issues on two levels. First is an example of one evaluation of an early 143 144 school-aged prevention project, including the development, implementation and results of an evaluation design. This discussion has included concrete information about the program, the sample, and the implementation of the evaluation, as well as process information about the development of the intervention and evaluation model. The second area of this dissertation has been the examination of the macro context of this particular social problem. This has included the social context of theory, practice and evaluation regarding programs to mitigate the impact of socioeconomic status on achievement. The EDPP was placed within that context. The focus has been on the processes that have affected the development and design of both the EDPP intervention and its evaluation, and the implications these processes have for future efforts in this area. The real message of the dissertation is in the second area of discussion. The implications for the profession of social work and this area of service delivery can be found by understanding and incorporating the analysis of the macro elements of this evaluation, more than the specific, micro level analysis of the particular data. Implications of this inquiry can be seen in all three of the thematic areas that have been addressed, and the interesting aspect is that the implications for the three 145 areas focus on the same two significant recommendations. In order to address this particular social problem more effectively, there must be more focus on the connections and relationships between various elements, and there must be a recognition of the multiple levels of these ' L interactions. ' __._ IMPLICATIONS FOR THEORETICAL APPROACHES This study began with a description of the conflict in American society between the values of equity and equality and the importance of education in overcoming that conflict. Education is seen as the key to social mobility and status attainment, and therefore the ability of children from lower socioeconomic backgrounds to achieve is critical to the goal of building an equal and equitable society. Theoretical and empirical work in this area has attempted to define the variables affecting the relationships and has begun to look at the interaction of those relationships as complex and multi-directional. The implications of this theoretical and empirical work are significant for the social work profession. The most promising work has been that in which the process of status attainment, and its precursor, child achievement, are viewed as outcomes of an interacting environment. Instead 146 of looking at the school as an institution that is not successfully educating young people, or the family as an institution that is not successfully imparting motivation and values conducive to success in school, we must look at these institutions as participating in a shared task - the socialization of children. Such a perspective takes a position very compatible with the knowledge base of the social work profession - “the intervention into human transactions" (Johnson, 1986, p. 93). Consequently, theoretical models that view the relationships as interactive provide an excellent opportunity upon which the social work profession can build. McGuire and Lyons (1985) describe the "transcontextual model" in which the home and the school are seen as "holons," or a "separate entity in and of itself, but also as part of an interconnected whole" (p. 39). This perspective is shared by other theorists who suggest that the school and the home are involved in a "complementary sociocultural task," shared roles, or that they are part of an ecosystem (Epstein, 1986; Lusterman, 1985; Lightfoot, 1975). In each of these instances, language which is very much a part of the theoretical framework of social work is used to describe the relationships between the family and the academic environment. As we have seen, the appreciation of 147 interactions, context, complexity and ecological frameworks has been part of the social work knowledge base and skill development for some time (Hartman, 1970; Johnson, 1986; Perlman, 1959). Joyce Epstein, in her article ”Toward An Integrated Theory of School and Family Connections," has provided an excellent perspective for understanding the relationships between families and schools. Epstein suggests that the roles of families and schools in society can be seen in one “...—...... ““"5? of three ways. First they can function separately, from a sociological or role theory perspective, with a completely different and specialized division of labor, each with control and respect in their particular setting or site. Second, they could be perceived as sequential, which takes a psychological perspective and provides that there is still a special division of labor and distinction of skills, but the family prepares children through certain developmental stages, and then the schools take over from there. In this example, the power and respect of each institution is stage specific. Third, the roles of families and schools can be perceived as shared, with a more generalized division of labor and overlap between the functions of the two institutions, with control over the settings shared, and with their respect and power student specific, through an understanding of the specific needs and abilities of each student (Epstein, 1986). 148 Epstein (1986), as well as other authors (Halpern, 1987; Newberger et al., 1986), describes the changes and stresses that are being placed on the American family. Epstein points to the increased education of parents, the improved knowledge of baby and child care programs, federal programs for children and families of lower socio-economic r status, and changing family structures as four reasons why the roles of families and schools can no longer be seen as = separate and competing, but must now be seen as shared and I cooperative (Epstein, 1986). Halpern joins in this opinion, describing today's young family as facing many stresses and needing additional supports and services (Halpern, 1987). The following diagram (Figure 6) is a representation of Epstein's theoretical perspective from her article on the connection between families and schools. It provides an excellent framework for further discussion about future work in this area and the potential role for the profession of social work. The model provides a variety of entry points for social work conceptualization and intervention, and could suggest a framework for evaluating intervention models. 149 Seesaw use 3358b 3 cocoa? no noeonmm Enema—25 uo "one: .m 086E 8:83.78 usuamum ego-.0 Economum Seamus .33 8398425... < 3.8..— A Loonom mo m332:6 530393 b33239 \oosotomum \oocotomua O couch m couch Allv Allv Sonom 555E 150 Epstein’s model provides us with a picture for many of the discussions included in this study. First, the interaction of schools and families are obviously located most significantly in the overlapping sections of the two circles. It is here that the interactions between families and schools most directly affect each institution and the children who function within them. It is here that Epstein suggests the greatest impact can be made on child achievement and motivation to achieve in school. By defining these internal relationships more concretely, and evaluating the effect the other factors in the model have on these relationships (such as Force A, B or C), then it is possible to develop and expand the shared tasks of families and schools to benefit the achievement of children. In recent studies, Epstein and other researchers have attempted to define further the concrete nature of the shared tasks that families and schools provide to children (Epstein 8 Dauber, 1988; Dauber 8 Epstein, 1989; Epstein, in press,a). Using a model called TARGET (task, authority, reward, grouping, evaluation and time), these authors have demonstrated the links between the functioning of families and schools, and they have shown that communication between these institutions can improve children's motivation to learn (Epstein 8 Dauber, 1988; Epstein, in press,a; Epstein, in press,b). 151 The other strength of the Epstein model is the ability to see and operationalize the many levels of interaction within this complex set of relationships. While there is clearly a role for children and parents, there is also recognition of the influences of families and schools. This model allows for the operationalization and definition of institutional variables, such as the social context of the school setting, the teaching practices of the teachers and the relationships between principals and teachers, just as a few examples. The social context of the institution of family, such as economic status of working mothers, societal expectations of family care-giving, and the attitudes toward various family forms, could also be included in an operationalization of this model. The presence of this component of the model reaffirms the fact that the factors affecting school achievement function at many levels, not just the individual child, parent, teacher or principal. An understanding of the process of status attainment, as this particular model demonstrates, must include the many interactions between variables and participants. It should also include the opportunity to examine those interactions at the many social levels within which they occur. While Epstein is particularly addressing relationships between families and schools and the effect 152 of those relationships on child achievement, the implications for this model can be carried far beyond to more global issues of social mobility and status attainment in society in general. These issues become even more important when looking at applied intervention strategies to address the lack of social mobility among lower F socioeconomic groups. IMPLICATIONS FOR APPLIED APPROACHES The importance of interaction and multiple levels is also clear in the discussion of applied strategies regarding status attainment. Epstein's model of families and schools provides a context for examining intervention strategies developed in response to the lack of achievement of children from lower socio-economic backgrounds. Most interventions, including the pilot intervention described in this study, focus their efforts on the family side of the interaction. Parent education, child management techniques and child education are interventions that will potentially impact the interactions within the family circle. Clearly, the intention is that the impact on these interactions will be felt within the overlap between the circles. However, there appear to be few intervention models that are specifically targeted to this area of overlap between schools and families. 153 Other researchers would express concern that there are not enough intervention models to deal with the school side of the circle. Hamilton (1983) proposed that the school environment itself and its impact on the overlap between the two circles must be examined. But frequently these efforts focus only on the school side of the model, L assuming that changes in this environment and this set of interactions will have the desired effect on the f interactions occurring within the overlapping circles. I The EDPP, studied in the present project, was an intervention designed to impact on the family side of the Epstein model. In this regard, much of the design followed the deficiency model previously outlined. The intervention designers utilized the literature that reflected a "culture of poverty" framework, suggesting that an improvement in skills and abilities on the part of the families would improve the child's functioning and, therefore, the family/ school relationship. That portion of social work knowledge and skills that works to improve the functioning of individuals and families became the major element in the intervention design. Social work knowledge and practice also had an impact on the imptementetien of the intervention. When the LPA (Local Private Agency) program staff began working with these families, their respect for a systems approach to 154 intervention caused them to look outside of the family for other factors and influences affecting the child's achievement in school. The program staff had frequent disagreements with the local elementary school personnel about the problems or direction of particular cases. Several meetings were held in an attempt to smooth over adversarial relationships between the LPA program staff and the school personnel regarding the implementation of the program in the school. The group intervention followed a deficiency model and provided an educational] cognitive based intervention for the families. Through discussions with the program staff, however, several things became evident. First, the children's groups began to provide an opportunity to deal with emotional issues and feelings about school and home, and moved away from a cognitive] behavioral format. Second, with attendance at the parents groups so low, much of the intervention with families was occurring at home, taking the form of helping families deal more effectively with all of their important systems. And third, upon gaining information about the families' perspectives of their situations, program staff became more convinced of the need to change systems outside of the families as well as the families themselves. The effects of the lack of involvement of the consumers of the service during the intervention design process became painfully clear. I _ 155 Even though the intervention model itself had focused on particular, individualized intervention techniques, and had not taken into consideration the broader systems’ influences on the problem, the program staff continued to provide services consistent with other aspects of social work training. In several specific instances, program staff reported spending a great deal of time with family members and school personnel to improve their communication. This work, however, was seen as secondary to the model employed. During meetings with the staff and debriefing from the first year, it became apparent that the program staff was looking more broadly at the societal and institutional issues facing the families with whom they were working. They began to develop strong feelings about the lack of response from institutions they felt should be more responsive to the needs of their clients. Because the design of the intervention only specifically dealt with one player in a complex set of interactions, the program staff were soon alienated from the intervention design itself. The result was a sense of anger and animosity toward the other players in these interactions, represented by the other components of Epstein’s model. The program staff began to identify a need for the intervention to address the school side of the 156 interaction. By the end of the second year of the program, the LCDSS (Local County Department of Social Services) and the LPA (Local Private Agency) were vehement about the need for the school system to take responsibility for its role in the problems facing these children and families. The local elementary school and the school district began to focus on the family side of the interaction. They expressed concern that the EDPP and the local DSS were not providing the degree of services that these families needed to overcome their problems. The awareness of the impact of interactions was also accompanied by a growing awareness of the multiple levels of that interaction. While the LPA and the LCDSS were concerned about the lack of a role for the school in these efforts, the local school and the district office were reticent about the increased control of the LCDSS. On several occasions, the two institutions (the school and the DSS) were each pointing the finger at the other regarding who needed to play a more active role in addressing the needs of these families. Clearly the political influences that were discussed earlier in this document came into play again, and the relationships between institutions were as significant an influence on the realities of these families as the work of each individual worker with each individual parent or child. 157 The problem reflected in these developments is that nothing in the intervention, or the response to it, focused on the interaction portion of the Epstein model. There was no specific theoretical base or intervention strategy that caused the program to focus on those aspects. One of the contributing factors to this development is that there is . no institutional structure or mechanism to focus on the interaction of the players in this social problem. The 1 school system has advocates and critics and the social service system has advocates and critics. But there is no clear, identifiable place or group of persons who are concerned about advocating for better interaction between these institutions so important in the lives of children. Consequently, it becomes easy to fall into blaming behaviors, with each side of the interaction pointing at the other. There are programs where the interaction of these elements and the multiple levels of that interaction are being directly recognized and addressed. With the increasing attention being paid to Head Start programming on a national level, there have been several recommendations made regarding changes in that program. The discussion of the recommendations of Washington and Oyemade (1987) demonstrate the efforts to include recognition of the interaction of various systems, as well 158 as the individual and institutional elements of that interaction. They suggest that Head Start must 1)address the feminization of poverty by promoting programs that encourage parent employment, that portray a non-sexist model, and that educate children about male/ female relationships; 2)focus on adolescent parents, including strategies to prevent adolescent pregnancy; 3)serve and meet the needs of employed mothers; and 4)increase and improve levels of and opportunities for parent involvement (Washington 8 Oyemade, 1987). These recommendations provide an example of how the inclusion of institutional and societal influences on the problem have resulted in concrete programmatic recommendations. The inclusion of curriculum regarding the issues of parent employment and the feminization of poverty is one good example. But the authors also reflect the importance of the interaction of elements in this process. When they speak of parent involvement, they reflect a commitment to involve parents in all levels of the functioning of the program, from paid staff to membership on the program’s governance bodies. Other programs have been developed to focus specifically on the relationship between families and schools in an attempt to improve the achievement of children. Comer (1988) describes a program in two inner- 159 city schools in New Haven, Connecticut. This program asserted that there was a "sociocultural misalignment" between the home and the school in inner-city schools in that community (Comer, 1988, p. 44). In Comer's description, the cultural and ethnic differences between the children’s family and their school provided an atmosphere of mutual distrust and competition. The result was that these children would not achieve in school, and their one hope of access to mainstream society would be blocked (Comer, 1988, p. 46). But a significant element of this particular project was that the intervention did not plan to teach the children and families how to change their environments to fit the schools. Rather, the program attempted to "promote psychological development in students, which encourages bonding to the school....[by] fostering positive interaction between parents and school staff" (Comer, 1988, p. 46). Each school had a governance team made up of principals, elected parents and teachers, nonprofessional support staff, and mental-health specialists. This body made major decisions in the lives and development of the schools and the programs. Parents were also involved in other activities in the school, such as attending school events and working as classroom assistants. This intervention focused specifically on the interaction between the families and the schools on an 160 institutional level through the development of governance and administrative structures that recognized a shared relationship. It also focused on the interaction between parents and teachers] principals, by developing mechanisms and programs for individual parents to deal more effectively with individual teachers and principals. The program worked to affect the relationship between the ‘7'”"77 participants in this effort and to address those relationships at multiple levels. The result was a program to build a school and community environment that would improve the success and achievement of the children who functioned within it. Comer does not suggest that all parents participated at equal levels, or that the development of the program was smooth or easy. What he does report, however, is that there was a significant improvement in the academic performance of the children. The schools ranked lowest in achievement among the 33 elementary schools in the city in 1968; by 1979 the fourth grade had caught up to their grade level and by 1984, pupils in the fourth grade in the two schools ranked third and fourth highest on the Iowa Test of Basic Skills (Comer, 1988, p. 48). All intervention programs cannot do all things for all people. The program that Comer describes went on for 12 years and had significant support. It is important to 161 recognize that the resources available for intervention strategies typically limit the scope of the intervention. However, even limited intervention designs must take into consideration all of the elements of the Epstein model. It may be that the particular intervention will only impact one aspect of the model, such as family interaction. However, recognition and understanding of the other components should effect the type of intervention strategy developed. Such an awareness can also assist the intervention designers in being realistic about what elements of the model the intervention is likely to impact, and it may provide researchers with guidelines for the evaluation strategies to measure that impact. IMPLICATIONS FOR EVALUATION The recommendation for more attention to interactions has particular importance in regard to evaluation at two levels. The first is in regard to the design of the intervention and evaluation models themselves, and the second is in regard to the interactions between the intervention processes and the evaluation processes on an on-going basis. Just as the intervention design must focus on the interactions of systems and individuals, evaluation designs 162 must develop ways to define and operationalize those interactions, as well as to measure them. Typically, the introduction of an evaluation component into an intervention design results in a narrowing of the definitions of intervention strategies and outcome measures. Because evaluation has become connected to a positivist framework, the "goodness of fit" question has typically been framed as a struggle to fit a complex and interactive intervention strategy into a goal oriented, outcome evaluation model (Schuerman, 1983; Schwandt, 1989). The result is an interaction between the intervention and evaluation designs that narrows the approach to one involving only specific, measurable and quantifiable data. Consequently, when evaluators develop the model of the evaluation, it too reflects this same narrow focus and specific goals. As the intervention unfolds and the actual process is much more complex and interactive than the intervention and evaluation designs reflect, the evaluator may become as alienated from the evaluation model as the program staff is from the intervention model. Because of the complexity of interactions that are actually occurring in the process of the intervention, the ability to draw conclusions or "separate the effects" of the components that have been measured in the evaluation design becomes more and more difficult. 163 Methodological problems in evaluation design have reflected this inability to incorporate the variety and complexity of intervention designs in typical evaluation models. Problems with sampling procedures, limitations in measurement, and political and economic limitations are but a few examples. Instead of trying to fit complex intervention models into narrow evaluation designs, the interaction between the intervention and evaluation design processes should be on-going during the development of the intervention. This would enable a more comprehensive and inclusive design of both, which can take into consideration the importance of interactions and multiple levels within the intervention (Bond 8 Halpern, 1988; Fitzpatrick, 1988). One of the most important aspects of this process should be the inclusion of consumers in the planning stages. This inclusion would begin the project with a commitment to the importance of interaction and empowerment of the consumers of the service. It would also improve the ability of professionals to understand what families mean when they look at the impact of family background on child achievement. It would appear that this understanding is critical in being able to develop intervention strategies to improve the ability of children to achieve. The second area in which interactions are important in evaluation is in regard to the on-going relationship 164 between the evaluation staff and the program staff during the implementation of an intervention. Whether at the initial planning stages or during the life of a program evaluation, the interactions between the evaluation staff and intervention staff are critically important. Many authors have indicated that the evaluator should have a clear and complete picture of the intervention before they can competently offer an evaluation (Bond 8 Halpern, 1988). Rossi and Freeman (1985) refer to this as having a grasp of the "social reality" of the program. They recommend that evaluators of any program, new or on-going, do what they call an "evaluability assessment" (Rossi 8 Freeman, 1985). In this process the researcher spends time in the program in ways similar to a field researcher, attempting to learn as much as possible about the meanings of the programs to all of the participants. Again, the perceptions of consumers of services are as important as the perceptions of the providers of services. In this way, according to Rossi and Freeman, the likelihood is much greater that the evaluator will design and implement an evaluation having meaning to the stakeholders and value to the program development process. Bond and Halpern describe an approach to evaluation research of Prevention Programs which they call "action research" (Bond 8 Halpern, 1988). This is research as the 165 process of collaboration between researcher and practitioner in a recurring cycle of basic research, consisting of program development, program evaluation and program modification (Bond 8 Halpern, 1988, p. 11). In Bond and Halpern’s model, the researcher enters the life of the program, and has an ongoing role in the r~ in“ implementation of the pilot intervention. Only with this kind of involvement can evaluators capture the aspects of an intervention that are having an impact on clients and staff alike and that have the greatest likelihood of providing feedback that will improve the program for future development. Bond and Halpern’s model would openly define the role of the researcher in the implementation of the evaluation and encourage a closer relationship than now generally exists between those implementing the intervention and those evaluating it. A concern for attention to multiple levels is also a critical recommendation in regard to program evaluation. The impact of the political processes on this particular intervention (the EDPP) was a critical part of understanding the nature of the intervention design and the elements of its implementation. These same political concerns also played a major role in the development of the evaluation. And yet there is nothing in the intervention or evaluation designs that reflect the institutional level’s involvement in this project. 166 In all of the data that was gathered, the only level that was addressed was the individual, micro level of child and parent behavior. There were several important characteristics of the environment of the home and school that we believe played an important part in the outcome of this intervention model. These were not included in the H 0...); -‘ 3": 3 evaluation design, predominantly because they were not included in the intervention design. However, if the evaluation process was closer to the one described here, then the elements of institutionalized influence could be discussed, defined and included in the analysis of the impact of the intervention. One of the ways that evaluation could play a major role in this process is in assisting the intervention staff during the development of the intervention to define and include institutional variables. While program staff are frequently aware of the impact of these variables in their client’s lives, they are less likely to be familiar with the ways that these variables can be operationalized, measured, and analyzed in relationship to the desired programmatic outcome. This is particularly important in an age of diminishing resources. As more and more stakeholders are interested in justification for the expenditure of resources on particular programmatic efforts, the results of program 167 evaluations are more and more important. At the same time, the pressures of this development have an impact on the intervention and evaluation designs, and the services that are offered. Evaluation needs to be seen as an on-going part of the process of intervention development at all levels, not just a process that is instituted after the fact to "test" the theory of the intervention. These issues in Program Evaluation hold great promise for the profession of social work. With our ability to analyze situations from a micro and macro level at the same time, because of our commitment to interactions, and with the skills that social workers learn in regard to systems analysis, we have the ability to make the connections between intervention development and evaluation development that could result in more meaningful evaluations of programs. In programs to mitigate the impact of socioeconomic status on child achievement we have the ability to emphasize the importance of the interactions of the participants in the process, and to emphasize continually the importance of the various levels of that interaction. All programs that are developed cannot be as comprehensive as the one that Comer described. However, the important elements that have been discussed here are the elements important to social work practice. We must 168 take a leadership role in developing interventions and evaluations, and the interaction between the two, that will make a difference in regard to this social problem on all levels of society. LIST OF REFERENCES LIST OF REFERENCES Aboud, F. (1981). Egocentrism, conformity, and agreeing to disagree. Qevelopmental Psychology, i1, 791-799. Achenbach, T. M., 8 Edelbrock, C. (1983). Manual fiet_the ehilg hehavio; ehecklist and revised child behavie; 2tefiil_. Burlington, VT: University of Vermont. Achenbach, T. M., 8 Edelbrock, C. (1986). anua f e teachets' teport {gym ang teaeher versioh ot the chilg hehavie; ptetile. Burlington, VT: University of Vermont. Adelman, H. (1986). Intervention theory and evaluating efficacy. Eveleetioh Review l_: 65-83. Alexander, R., Entwisle, D., Cadigan, D., 8 Pallas, A. (1984). Getting reagy fie; fitst gtade: Stahgatgs of geportmeht in home end school (Grant No. 1 R01 HD16302). Washington, DC: National Institute of Child Health and Human Development. Alwin, D., 8 Thornton, A. (1984). Family origins and the schooling process: Early versus late influence of parental characteristics. So ' ' Review, fig, 784- 802. Anderson, M. (1979). We e. New York: Hoover and Stanford. Asher, S., Hymel, S., 8 Renshaw, P. (1984). Loneliness in children. Child_nexslcnment. 55. 1456-1464. Auerback. A. (1968). 2arsnts_learn_thrsugh_dissussign- New York: John Wiley and Sons, Inc. Becker, H., 8 Epstein, J. (1981). a vo e ° Ieechez pteetices ehg jtdgmehts (Report No. 305). Baltimore, MD: John Hopkins University Center for the Social Organization of Schools. Bell. W- (1987)- Q9ntempsrarx_sgsial_selfare (2nd ed-)- New York: Macmillan Publishing Company. Blau, P. M., 8 Duncan, 0. D. (1967). The_hhetieen escapatignal_§trusturs- New York: John Wiley and Sons. Blechman, E. 8 McEnroe, M. (1985). Effective family problem-solving- shild_nszslgnment. 59. 429-437- 169 170 Bond, J., 8 Halpern, R. (1988). The cross-project evaluation of the child survival/fair start initiative: A case study of action research. In Weiss 8 Jacobs (Eds.), Eyeluating family ptegtams. New York: A. deGruyter. Bretherton, 1., Fritz, J., Zahn-Waxler, C., 8 Ridgeway, D. (1986). Learning to talk about emotions: A functionalist perspective. f f ' , 2;, 528-535. Brown. 8- (Ed-)- (1987). MW Eetly_1htetyehtieh. Boulder, CO: Westview Press. Caroll, J., 8 Steward, M. (1984). The role of cognitive development in children’s understandings of their own feelings. Child Develepmeht, 55, 1486-1492. Cautela, J., Cautela, J. C., 8 Esonia, S. (1983). Eetme fot heheviot ahalysie yith children. Champaign, IL: Research Press. Chambers, C. (1966). Seegtihe_fie;_3eteth. Minneapolis, MN: University of Minnesota Press. Chandler, J., Argyris, D., Barnes, W., Goodman, 1., 8 Snow, C. (1986). s c e s: O s s e: loy-ihceme petehts and ehilgten in a homework-lite tesk (Grant No. NIE-G-80-0086). Washington, DC: National Institute of Education. (ERIC Document Reproduction Service No. ED 231 812) Cibulka, J., O’Brien, T., 8 Zewe, D. (1982). lhhetgeity ' s u . Milwaukee, WI: Marquette University Press. [City] School District. (1980). ° W {City} 1 Michigan: Author - Comer, J. (1988). Educating poor minority children. e ' , 252, 42-48. Cook, T. 8 Campbell, D. (1979). s - e ' t o . Boston, MA: Houghton Mufflin. Corcoran, M., Duncan, G., Durin, G., 8 Durin, P. (1985). Myth and reality: The causes and persistence of poverty- WW. .4. 516-536. 171 Damon, W., 8 Hart, D. (1982). The development of self- understanding from infancy through adolescence. ghilg 29291998989. 93. 841-864. Davis, J. (1982). Achievement variables and class cultures: Family, schooling, job and 49 dependent variables in the cumulative GSS. e ica 8 'ca Reyiew, A_, 569- 586. Dauber, S. 8 Epstein, J. (1989). Peteht ettit s a act'ces o are t volveme 'n inner-cgty elementaty end middle schools. (Report No. 33). Baltimore, MD: The Johns Hopkins University Center for Research on Elementary 8 Middle Schools. -..-i . ....‘n‘fiw deKanter, A., Ginsburg, A., 8 Milne, A. (1986). Parent involvement strategies: A new emphasis on traditional parent roles. Washington, DC: United States Department of Education. (ERIC Document Reproduction Services No. ED 293 919) Dreikurs. R.. & Grey. L- (1964). Par9ntsi_gnid9_99_9hild gieeiplih_. New York: Meredith Press. DSS Officer. (1985). v' s e- e e- . Michigan Department of Social Services: Author. Duncan, 0. (1967). Discrimination against Negroes. LII: t o 1‘ :u r a! ; {Q‘U 0 -. _ g _,o e . 9919899. 311. 85-103- Eccles, J. (1983). Expectancies, values, and academic behaviors. In R. T. Spence (Ed. ), Aehieyemeht_eh§ ashi9x9m9nt_motix9s (pp- 75-146) San FranciSCO. CA: W. H. Freeman and Company. Entwisle, D., Alexander, A., Pallas, M., 8 Cadigan, D. (1984). The emergent academic self-image of first graders: Its response to social structure (Grant No. 1 R01 HD16302). Washington, DC: National Institute of Child Heath and Human Development. Epps, E., 8 Jackson, K. (1985). '0 an eecupational aspitetiehs and eezly gtteinmeht e: hlaeh 89199_and_f9mal99 (Final Report) Atlanta. GA: Southern Education Foundation. Epstein. J- (1982. March) Stsd9ntsi_r9astions_to t9a9h9r_9_9r99ti999.9f_nar9nt_in_9129n9nt v . Paper presented at the annual meeting of the AERA, New York. 172 Epstein, J. (1983). Effects on patents of teechet pr99ti999.9f_99r9nt_inxolz9m9nt (Report NO- 346)- Baltimore MD: John Hopkins University Center for the Social Organization of Schools. Epstein, J. (1986). o a n ' te ted t eo o eehoel ehd femily eohneetiohs. (Report No. 3). Baltimore, MD: The Johns Hopkins University Center for Research on Elementary 8 Middle Schools. Epstein, J. (In Press,a). Single parents and the schools: Effects of marital status on parent and teacher interactions. In M. Hallinam, (Ed.), Chahge ih seeietal ihstitutions. New York: Plenum Publishing. Epstein, J. (in press,b). Family structures and student motivation: A developmental perspective. In C. Ames 8 R. Ames (Eds.), es c o o 'va '0 c (Vol 13). New York: Academic Press. Epstein, J. 8 Dauber, S. (1988, August). Ieaehe: t itudes and r c 'ces of ar nt ' veme ' ' ner- c' e eme and iddle scho 8. Presented at the annual meeting of the American Sociological Association, Atlanta, Georgia. Fanshel. D- (1980)- Ih9_fatur9.9f_999191_gerk_r9999298. Washington, DC: National Association of Social Workers. Farnworth, M., Schweinhart, L., 8 Berrueta-Clement, J. (1985). Preschool intervention, school success and delinquency in a high-risk sample of youth. Ametieeh EQ_EESIQD_B§§§§IQD_QQBIBQI. 22. 445-454- Fitzpatrick, J. L. (1988). Roles of the evaluator in innovative programs: A formative evaluation. , 12, 449-461. Fraser, B., 8 O’Brien, P. (1985). Student and teacher perceptions of the environment of elementary school classrooms. El9a9ntarx_§98991_lournal. 95. 567-580. GarbarinO. J- (1982)- Qhildr9n_and_f9mili99_in_th9 eeeiel_ehyitehmeht. New York: Aldine Publishing. Garbarino, J., 8 Sherman, D. (1980). High-risk neighborhoods and high-risk families: The human ecology of child maltreatment. QRilQ.D§Y§lQEE§B§..§l. 188-198. 173 Gilder, G. (1981). We ove . New York: Banton Books. Gladow, N., 8 Ray, M. (1986). The impact of informal support systems on the well being of low income single parents. 298111.391991989. 1.. 113-123- Goodson, B., 8 Hess, R. (1987). The effects of parent training programs on child performance and parent behavior. In B. Brown (Ed.), Found: Long Ieth gains item Eatly lnterventieh. Boulder, CO: Westview Press. Gordon, T. (1970). ve ss a ' . New York: New American Library. Haizlip, T. (1975L Parallel groups: Psycho-therapy for preschool children and their parents. An9risan_199rnal Q£_E§Y2hisLIY 1.2.1951 1063- Halpern, R. (1984). Lack of effects for home based early intervention? Some possible explanations. Ahezieeh , £1, 33- -42. Halpern, R. (1986a). Home-based early intervention: Dimensions of current practice. thlg_flelfiete, Lzy, 387-397. Halpern, R. (1986b). Community-based support for high risk young families. geeiel_£eliey, 17-22. Halpern, R. (1987). eye social and demographic trends affecting young famil es Implications for early childhood care and education. XQBBQ_thl§I§n- Hamilton, S. (1983). The social side of schooling: Ecological studies of classrooms and schools. The El9m9ntarx_§98991_299rnal. fill 313'334- Harter, S. (1981). A new self-report scale of intrinsic versus extrinsic orientation in the classroom. D929lops9ntal_2929891991. 11. 300-312. Harter, S. (1982). The perceived competence scale for children- shild_n929lspm9nt. 51. 87-97- Harter, S. (1988). Causes, correlates and the functional role of global self-worth: A life-span perspective. In J. Kolligian 8 R. Sternberg (Eds. ) Bezeeptiehe_et com - . New Haven, CT: Yale University Press. 174 Harter, S., 8 Connell, J. (1981). A model of the relationship among children’s academic achievement and their self-perceptions of competence, control, and motivational orientation. In J. Nicholls (Ed. ), The d2xel22ment_2f_agh1§_§mgn__m2ti_ation (pp- 1- 49) Greenwich, CT: JAI Press. Harter, S. 8 Pike, R. (1984). The pictorial scale of perceived competence and social acceptance of young children- 9n11g_n2221221221. éé. 1969-1989- Hartman, A. (1970). To think about the unthinkable. Social Qasewggh, gl, 467-474. Hauser, R., Sewell, W., 8 Alwin, D. (1976). High school effects on achievement. In W. Sewell, R. Hauser, 8 D. Featherman (Eds. ), 8 oo ev _9h__ling_and_agh1__sment_in AEQIiQQn_§Qgi§tx (pp. 309- 342). New York: Academic Press . “l‘ ‘ ;"'nl_" fiat—-.. l Heineman-Pieper, H. (1985). The future of social work research. S221al_E2rL_Bsseersh_and_Ab§tragt§. 21. 3- 11. Hepworth, D., 8 Larsen, J. (1986). Dirs2t_§ogial_gork nra2tise___1hs2:x_and_§kill_ (2nd Edition) Chicago. IL: The Dorsey Press. Hogg, R. 8 Coladarci, T. (1989). Teacher based judgments of academic achievement: A review of the literature. Bexieg.of.Eduoational.Be§eargh. 22. 297-313- Holden, G., Lavigne, V., 8 Pameron, N. (1986, August). Effiegtivehess :L t c s pagents and chilgzgh. Paper presented at the annual meeting of the American Psychological Association, Washington, DC. Hoorwitz, A. (1985). Establishing partial limits on children' s behavior: Use of strategic, structural, and cognitive-behavioral approaches. Ihg_Amg;ig§n_Jggzh§1 2f_£am_lx_1n§ranx. A. 56-64- Johnson, L. (1986). ' o ' ° gppggggh. Boston, MA: Allyn and Bacon, Inc. Kagan, S., 8 Schraft, C. (1982). flhgn_p§:eh;§_ghd : 109 = OH‘ 0°? 9 ' |. 7‘ ‘nt 2 0- CON-z - 91 ‘! 1nx2lxemen§_1n_urban_§shool§ (Report NO- 5)- Boston. 175 HA: Institute for Responsive Education. (ERIC document Reproduction Service No. ED 281 950) Karger, H. (1983). Science, research and social work: Who controls the profession? Soc'al Wo k, 28, 200-205. Kettlewell, P., 8 Kausch, D. (1983). The generalization of the effects of a cognitive-behavioral treatment program for aggressive children. J na 0 n a Child Psyphology, ;_, 101-114. Kozol, J. (1985). 'ter ' a. New York: Anchor Press. Leach, Hugh. (1987, May 25). Lack of attention the biggest crime. 111_§£§£§_122:n§l. p- 4a- Leif, N., 8 Zarin-Ackerman, J. (1980). Sociocultural deprivation and its effect on the development of the child- In J- Bemporad (Ed ). _hild_dexelopment_1n n2rm11111_end_2§xohonethglogx. (pp- 362-391)- New York: Runner and Haxel. Levenstein, P. (1988). 5 ° - hild nd ' s gigggyghppgg. Columbus, OH: Ohio State University Press. Licensed Clinical Psychologist, Local County Probate Court (1985) Annu11_Be22rtL_E29l211123_of_11u1§1_§1121211 Sepyices. [City], Michigan: Author. Lightfoot. 8- (1975). E2rld§_aeart1__3elat12n§n1p§ hetgggh fpmilieg gpg sphopls. New York: Basic Books. Loeber, R. (1983). The stability of antisocial and delinquent child behavior: A review. Child ngelgpmeng, Q}, 1431- 1446. Lowry. M. (1983). s c s t2.211en1_1nxolxement1__b gtpdy of happiezs to participatign in phe edpgapiphgl ppocgss faced by black, low income, inner-city papeptg pf handicapped phildrep (Grant No. G008101535). Washington, DC: S ecial Education Programs. (ERIC Document Reproduction Service No. ED 244 487) Lusterman, D. An ecosystemic approach to family-school problemS- The_Ameri2an_lournal_2f_£sm1lx_1nereex. 1.. 22-30. 176 MacLeod, J. (1987). Ain’t no makin' i . Boulder, CO: Westview Press. Martin, P. (1980). Multiple constituencies, dominant societal values, and the human service administrator: Implications for service delivery. Adm1p1§ppdpiph_1n Wor , A, 15- 27. McGuire, D., 8 Lyons, J. (1985). A transcontextual model for intervention with problems of school underachievement. The_A1111212.1221211_of_211111 Thersnx. 11. 37-45- Meier, J. (1987). Introduction. In B. Brown (Ed.), ° 0 E ve ' . Boulder, CO: Westview Press. Mercy, J. 8 Steelman, L. (1982). Familial influence on the intellectual attainment of children. Apgp1pdp §O diplpgical Revigw, A_, 532- 542. Michigan State University School of Social Work, Intervention Staff (1986a). Qh11§rean.§roun_£ropo§11o East Lansing, Michigan: Author. Michigan State University School of Social Work Intervention Staff (1986b). r o o . East Lansing, Michigan: Author. Miller, S., Manhal, M., 8 Mee, L. (1989, April). enta e i renta u ac and c ' I development: A search for causal rglations. Paper presented at the bi-annual meeting of the Society for Research in Child Development, Kansas City, MO. Munson, C., (Ed.). (1980L figgjgl wgrh wjgh figijjggo Ihgp:y_dhd_ppdpp1pg1 New York: The Free Press. Newberger, C., Melnicoe, L., 8 Newberger, E. (1986). The American family in crises: Implications for children. QHIIED§_EIQDl§E__iD_E§§i§§Ii§§ §§4 559'739- Newman, C., Schwam, J., 8 Schwam, Jeffrey. (1979). The fatherless child. In C. Justin (Ed.), Bg§19_hdpdhpph of phild psych1gtpy, (pp. 357-371). New York: Basic Books. Pakizegi, B. (1985). The developmental dialectical approach to adaptive and maladaptive parenting. Washington, DC: The United State Department of 177 Education. (ERIC Document Reproduction Service No. ED 261 797) Pallas, A., Entwisle, D., Alexander, R., 8 Cadigan, D. (1986). thldgah gho do exceptipnally well in f1rst grade (Grant No. 1 R01 HDl6302). Washington, DC: National Institute of Child Health and Human Development. Palmer, F. (1987). The effects of early childhood intervention. In B. Brown (Ed.), Epppd; Long pepm gains at early intervent1ph. Boulder CO: Westview Press 0 Patterson, G., 8 Stouthamer-Loeber, M. (1984L Correlation of family management practices and delinquency. Qn111.2.xe122.ent.1.. 1299-1307. Perlman, H. (1959). c'a ca w ° A o -s v’ p;ppa_§. Chicago, IL: The University of Chicago Press. Perry, D. G., Perry, L. C., 8 Rasmussen, P. (1986). Cognitive social learning mediators of aggression. 91111.22211221121. 11. 700-711- Pickarts, E., 8 Fargo, J. (1971). Eareht edugatipp: t2ward_parental.someetense- New York: Meredith Press. Probate Court Officer (1985). 'v Se 'c 8. [City], Michigan: Author. Pumphrey, R. 8 Pumphrey, M. (1961). Iha_ha;1page at lama:uzu11a11EuL1muIa._Bead1ng§_1n.ph112§22h1211.and ' Ltution11.dexeloement- New York: Columbia University Press. Reynolds, B. (1963). ' ' t of_growth_1n.§221al.work- New York: Citadel Prese- Rickel, A., Eshelman, A., 8 Loigman, G. Social problems solving training: A follow-up study of cognitive and behaVioral affeCtS- IQB£n§l_QI_AQDQIEél_thl§ E§¥Eh2l29¥. 11. 15-23- Rosenberq. M- (1965)- 1221etx.and.the.ado1essent.§elf: . Princeton, NJ: Princeton University Press. Rosenberg, M. (1979). Qphpaiy1hg_pha_§a1£. New York: Basic Books. 178 Rossi, P., 8 Freeman, H. (1985). Evaluation: A gratemat1o.aenroash- Beverly H1118. CA: Sage Publications. Schuerman, P. (1983). esearc nd evalu o the hhmah_aapy1aa_. New York: The Free Press. Schwandt, T. (1989L Recapturing moral discourse in evaluation. 1122111.nal.ae§eargher. 1.. 11-16- Schweinhart, L. 1985. Early childhaod degaxppment s ' e' 'es' ' 'ggpga, Ypsilanti, Michigan: High/ Scope. Scott-Jones, D. (1982, March). The role of performance feedback in parents’ and children's expectations. Paper presented as part of a symposium entitled Tha fprmation of expectat1ops fog achievement tashs. Symposium conducted at the annual meeting of the American Educational Research Association, Austin, TX. Seitz, V., Rosenbaum, L., 8 Apfel, N. (1985). Effects of family support intervention: A ten year follow-up. v , §§, 376- 391. Sewell, W., Haller, A., 8 Portes, A. (1969L The educational and early occupational status attainment proceSS- Amer1212.122121231211.Bexiew.11. 82-92. Sewell, W., 8 Hauser, R. (1976). Causes and consequences of higher education: Models of the status attainment process. In W. Sewell, R. Hauser, 8 D. Featherman (Eds.) ' ' 1 (pp. 9127). New York: Academic Press. Sewell, W., Hauser, R., 8 Featherman, D. (1976). Introductory notes. In W. Sewell, R. Hauser, 8 D. Featherman (Eds.), oo ' c ° ve e Apapipah_§pp1apy (pp. 3-8). New York: Academic Press. Slaughter, D., 8 Kuehne, V. (1988). Improving black education: Perspectives on parent involvement. QIQQD ev w, 11, 59-75. Stallworth, J. (1982, March). rent vo vement n t e ools: A surve o add_appp_. Paper presented at the annual meeting of the American Educational Research Association, Austin, TX. 179 Strom, R., Griswold, D., 8 Slaughter, H. (1979). Parenpai hachggound: Does it matper in parent aducatipn? Washington, DC: U.S. Department of Education. (ERIC document Reproduction Service No. ED 241 154) Tropman, J. (1989). Anepican values and social weifape: Qplturai contradictions in the welfape state. Engelwood Cliffs, NJ: Prentice Hall. Washington, V. 8 Oyemade, U. (1987). Erojagp head start; t ese a utu e t e ds ° he conte o fam11y_naad§. New York: Garland Publishing Company. Werthamer-Larsson, L., Kellam, S., Dolan, L., Brown, H., 8 Wheeler, L. (1988, March). The epidemiology of maladaptive behavior in first grade. Paper presented at the meeting of The Johns Hopkins Prevention Research Center National Conference, Washington, DC. Wilking, V. (1979). The street child. In C. Justin (Ed.). WWW. (pp- 301- 308). New York: Basic Books. Williams, R. (1985). Eaziy dapecpipn fiaators pannan po popential school dpopouts, 1985-128 . Madison, WI: Boone County Board of Education. (ERIC document Reproduction Service No. ED 277 114) Williams, D., 8 Stallworth, J. (1982, March). A survey of educators pegarding papant 1nvoivemenp in aducation: Implications for teacher tpaining. Presented at the annual meeting of the American Educational Research Association, Austin, TX. Zigler, E., 8 Anderson, K. (1979). An idea whose time had come: The intellectual and political climate for head start. In E. Zigler and K. Anderson (Eds.), Eppjeap head Spa; . New York: The Free Press. APPENDIX A GOALS FOR THE INTERVENTION DESIGN Appendix A - Goals for the Intervention Design 0.3.3.. 9.9.6 000: c.7231 3.3.0.9.)». 9.7.301 00.210020... ”30...... 00.3.»... 009330.203 30...... .ungzi 00 03:02 2.0.0:... 30...... 2.1.5.3 5 00.3.2... «I... «30...... 0100—0! .2330 .5 u u. 50...... .033: 00 00.239: :9an 86 non..:..n.03 $60.... igllnao o. 3......— 02.30.30 «301.... 2.3.5. ”00010 3g 83% 930.2... «flora. 00.: «.6103 ~30... noon;— . 0035.53.20: “30...... .025: no c.0033. ...—n3. 0.72.01 :56... .023: no 09.3.: 3.... “and... 3:: . ”30...... .228 no 0.020.220: ...-.03. .0 col..— 0:30:03. .81.... .033: no 23.30 mail—«.303 "2.4.... ...—.0... no .3040 no 007.0 0320...: 50...... 0080.320: in: 03.1 03:01.: 00.: 5010.... MOS: .00 “30.1.9.0...” «5:. 007.30.. 0093.30.30: ”301.... .033: no .00....) 7.7.30.1 $.09 n7. 0: 7.. 520...": .033: no c.1025. ...-2.6. $.06... .033: no "301.... .0n.¢.\ .30: 0350.. ......030... 030.210 3.4.326 .52. 00.: .46—...... ..u. tc.1.3n..\ .3711"... $03.50... nos-£30.20: .30.... ...—In... no can... 3.: 30.5.9... 5610:. “3001.620: .30 .00... «0 10.06.00. 30.0... .033: no .9005. 00.. no: no 0.0:». 3.: nine. «00.23.20: “:02... Canaan—203.. 00.3.8 2.0.0.030: “301.... .023... .01 3.2.20 86.03303. man—.....So... “301.... .5003 .c.n... .30 20:01:30 30...... .... 00:20.30. “30...... calinldpao 0.. .3.» 2.33.333... ” 86.95.03 0. .... 00.: ”.6100. M00..- .2. 3.80.020... ”3:. 0:53.20: “30...... .033: 05.30.. 090...... 89.0.20: in: 2.... no .... .... .30. 0.08... 4.3.3: .303... 2:06.20: in: .5301 nc .3 0.05... Inn—Sn... $01.... .033}. no 100091.. 024.... «00:30. ”:03... .nnSaIfl. 0.01.... nag—o... osu}%cac no.3 .3019... ..p. 0.0... $.01 0032.33.30: 26...: .2ch no 7.2:. .50... .3. .3... .3 00:2..— .301.... 00.37.. 3.0.330. ...vacn. ...» “30...... .223 no .006... no.3»... ”30...... . 39.00.12... 0.7.32. «30...: .023... no 00090;... ...—.30. ~30}... ...... ..nnn mail—5:03 :66... .330: no .3... $5 003.... 50...... .0331. no 0000...». in: 02...... 50...... uaoouoaoaucpao 7.01.... 01.30.33. mon..3..§.3 7.01.... .223 no 900.. 601001.». 0.3.5... «301.... .032“... . no 7.... 03... .30.... .0320: no u... 9.06 180 APPENDIX B CHILD REFERRAL FORM Appendix B - Child Referral Form N 000. .540004050000004 00:0 :0 0.4000 54.040 00 0000000 00 .004.0: 0.0 50:0 ..c..... .00 ......»o. 00 0.0.0 0.0 .0 .0.400. .:0 00 ...... .... 54.004.00.000 00.4400. 000.000 00 04:: .00000uouu0a 0.044:0 0:0 0:40.000. 0.0.0.040 0:0 :4 .0400. 5: 0.04.0.0 .040404.0 0.4000000000 00 0.0u0. 04:h 54.004.00.000 004.404.. 00.000: ....000. .30.. . .. .0.00.a. .0 040.0 .00 .0 .0q4... 0:0 0:40.000. 004000.0004 00440000 0.00 00 0.40 00004 h0.004.00 04:» .00000.40 00:0.00 04 .00.. 0:0 000 .0 00:00:. 00 4400 00 .0000... 04 00:0 0040.400 0:0 540000000000 0:. .00... 0:0 ho 0040 0:0 0040.000. 00.044:0 000:0 v0 50.. .0M 00.0.0004. 0 04 000:0 00:0 .0.c0.000 0:0 .0 000.00: .0.0:0000 0:0 :04. 00040000040 0.400440. 0.000 0.0. 000.0 0 04 0043.00 00 0040040. 00004 0:h .00.- 0 04 0048.03 .c0uv44:0 .0:00 :04. 0.4:00040.40. 00: 0:0 .0 0: .4 004000 4440 00: .00004.V.Iddfldl 54400000 044:0 0:0 000 00: 50. 005 54400405» .000000040 0:0 04 00:04.. ...: 00 ..00000 044:0 0:0 000 .0 00:00:. 00 0.0u0. 04:h 00004.. 004:0. 504.040.4‘ 00: .00. 00 .005 .00 0003040 000 04 0.0000. 4.04550 08h 0000 004000000 0004.00.00. 00 008 .00.:0000. ‘0 0.400.000. 0004.44 0.0. 0 :04. 00040000. 000 .104:0 50:0 0.040: 000 00 00:00 40.0000 0.40004 .00. :04: 044:0 0 .0004000040 00 0000.00. :4 004440: 00 0.04.0:0: ‘00 00000000. 00 0.400.000. 00.04 0 0.0: 0.40 50:k .:04) 04000 54.004.00.000 0: .00 50:0 440:: 00044000 .0 000400... 04.0000 00 04:0 .4 40.0000 0.40004 0000 4040 044:0 c 000 .0. 40.0000 00400.4 0000 00000000000: .0x.00 4.000.004.000 00.4.000000 000 .0 0.00000 0:0 04 00:0000 0: 0450:. 0.40.0.0 000:0 .0 44¢ .004000uu0 000 0040000 .0 05040040 0004.00.00.04 0004004 0.40 500 04 .00400045400ub . 0050400 .0 00400.00... 5:. 000.0400 000000 .0 .00.044:0 .000505 :04) 0504. 050040 044:0 0 00:. 0.0 04:0 ‘0 004.000l 000 .0. 0.000004 5440004000u .0440000 .0 00050.0 00 0000000 00 .00440030 0:. 000004000 .04.. 00.00000 .004 04:h .0.00. ‘0 0440. 0:0 00 00404.) .0 .0800: 0000 00 :03. .00440030 .0n0. .0 000004300 .0 004000.000. .0 000004004 04 004000. 544000: 000000. :00. .4“ .04 .04 oh 4 0000 004:004 .000040000 0.0 0040404000 00:. 00004. .0.0.. .40:0 00 000.000. 0.. 50:0 00:0 0.30 0.48.. .00440000 400:0. 0:. 000004000 .0400 .0 0.00 0:00 0400. 50.0.... 400:0. 0000.00. 0:: 04440 0:h 50.0.0.0 400:0. 0000000: .0 . .000 .00400000.:0 .0044400 000: .00000004 4000 0.0 004.00.... 400.0. .0 0040.008 .000 .00404: .004:00. .004004u0 0.0 004000.000 40040.:. 00 00400.00 .0.000 :04) 0000000 00400000 000040404 144:0 0:0 0.0:. 000000.04 00 0040.000. 0.. 00 000000004 000:0 .4 ...... :04. 004000.... 54400405:. 04 .0 ...... 0040 0.4000000. 04400.0. 04 .0 ..0400000000 0040.000. 000: m4 00:0 000.40..0 00 04:.0040040. 04 0000 044004.50 04 0.0.0.0 0000 0:h 50.00 54000000.. 04 .« .004:0.¢ 0000000000 00 .0 5000000.. 0:0 00404000 00:) 0000.000. .0 0.0.0 0 :04) 005 00400.. 0000 04 0:: .0000... 04:0 0:050: 0..50:0 003000: 540.40 ...-0.. 0:0 00 044:0 0 00.0. 000 0:040 00» .0005 0 0500 0:4: .0 .:0¢00 0 500 0:0 04 400:00 00.. 0040 0400:. 044:0 0 00:0 0500 .0 .0100. 04:00.0000 0.0.0.0 0:h 54.04000. 400:0. 000000: .4 .0400! non o 1.0) 40400” H0 400:0» 0:0 00 00 4400 .0 ..0040l .08 .000:0000 .0:00 :04) 1400 00 00.u 400» .00004 0:0 .0 500 000:0 0040:0000 500 0.0: :05 ‘4 .5440u0u00 0.0. 00004. 00 .00:0.00000 0.40.000 000 00404000 :00: .0 00:04.: 0:0 .0 00.00 :4 001.000. :00: 0.0: .0004 0:0 ho 000m .50000004000om400 00. 04 00:0 00) 000.. 0:0 mo 000000000 0:0 002000: 04 04 .004u4u040 00: 04 0004 :0 V4 .0040004u4u040 0000 00000: 00004 040‘ 030.. 0:0 100 .00080000 :04) 0040000 0 04 00.40.0000 0..) 000000.... 000:» .000004..0 0: 00 0.0 00:0 0.04.0:0: 0:0 0:40.000. 0.000....0 0:0 .0v4>0.0 000:0 004000.0004 00:00000 0:h .0000. .0 0000.004 00 000400.. .0. 4040:0000 0:0 04 0.0:0 00:) .0» 0:: .5400000ua 0.0.0. 00 0: 00¢ 50. 00045000 000:) 00.044:0 .0» 0048004 0.0 00 .540003000000 .004000040 0:0 .0 00400.04.0000 0:0 000.... 00 0004.00. 0040400.. no 004000004 0:0 :04) .000000 54..0 0:0 :4 000000040 0:0 :4 504:04uu4v 0:40.: 0.. 0:) 00.044:0 000:0 5.400004 00 04 00000.0 0:0 .0 4000 0:h .0:0¢00 0.40.0 0004 0:0 .0.0 00.0.». 000 00:4.0000 000: 00: 00:0 .0040... 004000.... 0.4000004 0:0 00.» 04.000: 0:04. 104:0 :05 0:3 0:40.000! 00 000.00040 .005 04 00.044:0 0:0 00004000 00 005 000.0000. 03 .0004... 004000.... 000 0:0 00 00.044:0 0.000. 0.0» 04:b rack J<fl¢flhm¢ 944:0 nah usnhmdaxoo COL na°~hvaflhhxn 181 1132 "0000.000 4000404000 10¢. 20 unluhzou 000:) .00 4000400.. 0.00 .... 040:0 04:0 .0: «40.04 000.0 004.0 «4.0.4 000.0 0.00. «4.0.4 0.0.0 00 040000000. 044:0 04:0 .4 ”00000.0...0 00000.00 .NN 54.004.00.000 00044... 000.00. 00 .400 .4N 004000..0 00.00. 00 040.09 .ON a 0 N N 4 0004:0 0.0 5.0 00 00.04.03 .04 .00.: 05.040 0.0040 000040 N .00. .00....0! n v n u 4 04.0.0.0..00. ...... .00.... ... n v n N 4 00.4040... .40. .000 0000.0000006 .04 0000...! 00800.0 a v a n 4 .0 .40. 0:... . .0 44.0 .3... .04 n v n u 4 0:... ....4 . a. 44.: .0... .04 n v n u 4 0:... 44... . .0 44.. .1... ..4 n v N N 4 .0. .0. 54.000 000.00.... 0004400 .04 n v a a 4 sauuoa0a. ... 00.0..» .3... .«4 n 0 a N 4 400000 .0 00.044:0 .0000 00.. 04.000 .44 n v 0 N 4 00.040:0 .0800 00.. 000.0104) .4 .64 a v n a 4 .00.... .00... 004004.... ... .0 n v n N 4 . 0.00 00000.00. 0004.00.00. .00 0. 0.8 .0 n v N N 4 40.0000 0.40004 .000 000.00.80.06 .h n 0 a N 4 .00 .0. 0.00000. 5440004000U .6 n v a a 4 00.00... 4.000. 000.0... .n n v n N 4 0.0.0 :04. 0.4.00.000 5440000500 04 .v n v n N 4 0.0.0 :04) 0.00.0.0. 544.0... .4 .M n v N N 4 50.00 5400.00... .4 .N n v N N 4 54.04000. 400:0. 000.000 .4 ....z 05.000 000040 0.0040 ....0 0.0.4. u n .0040. a 0 00000000. I n 5400000... u N 4 05.04. 00004. n ".4000 00400440. .:0 00000 0.00 5.... 00.. 00.000 .004.0 0.0004 ..00..:00 0:0 .0 :00. 0.04:4:00 0.: 044:0 04:0 5400.00... 00: 00004.00 000.40 .00040.0 00000....0 0:0 00 00...... 00000 0. 0.0:. 000004 04 00.0 000:0 040:0 0:0 .UIHO uh.‘0..h ".000 "0‘0! I.‘dd£o rack J<¢¢m0m¢ 044:0 APPENDIX C PLO. CHART FOR TH! EVALUATION DESIGN Appendix C - Evaluation Flow Chart .8. JII . III ’93:! '33-‘16 o on; ...-I! . o s! I: 183 "Iiflufliflfljfliflil'lmflfllflflfiflmflfl“