MSU RETURNING MATERIALS: P1ace in book drop to ”BRAKES remove this checkout from -,—. your record. FINES wil] be charged if book is returned after the date stamped below. C;3 1 r 1”"r ouvx g’ HWV} :- no lo V745 (fob? fig) (7)2Tr m3 9 3171931 EY213802696 (:23 l—-\ L‘ C I [-\F_ O: :3 3 Qt: :7 b.) THE INFLUENCES OF A COMMUNITY PREVENTION PROGRAM ON PARENTING BELIEFS AND PARENTAL COMPETENCE BY Sharon Williams Shay A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Family and Child Ecology 1989 If“ l" A! 54 ABSTRACT THE INFLUENCES OF A COMMUNITY PREVENTION PROGRAM ON PARENTING BELIEFS AND PARENTAL COMPETENCE bY Sharon Williams Shay During the past decade, a wide assortment of prevention programs have been developed with the goal of stemming the escalating tide of child maltreatment in the United States.— At this early juncture, however, most of these programs lack solid. evaluations. and information for replication” This investigation examined, the influences of a community prevention program, the Family Growth Center (FGC) in Lansing, Michigan, on parenting outcomes. Using a sample of thirty-one families, two one-way analyses of covariance clarified the extent to which utilization of FGC services and type of FGC service contributed to reported parenting beliefs and observed parental competence. A series of’ hierarchical regression analyses was performed to determine the influence of the Family Growth Center on parenting outcomes relative to other parent and family traits, and, to explore the degree to which parenting beliefs mediated the relationship between those traits and parental competence. Interviews with participant families provided qualitative data for case studies describing the subjective experiences of parent using the Family Growth Center. At the end of twelve weeks, mothers who were high utilizers of FGC services were significantly different than low utilizers in level of empathy, role reversal, and provision of appropriate play materials in the home. The type of service provided at the Center did not matter. Results of the regression analyses showed that while parent/family traits were the most powerful influences on parenting outcomes, utilization of the Family Growth Center was a significant factor in explaining variance in parenting beliefs, particularly role reversal, as well as observed maternal responsivity, avoidance of restriction and punish- ment, and provision of appropriate play materials. Certain dimensions of observed behavior were affected by parenting beliefs, particularly those that relate to discipline and to physical and temporal environments. For other aspects of parental competence, maternal education and/or childhood abuse were the most powerful influences, regardless of parenting beliefs. Data from the investigation were consistent with a proposed ecological model for prevention of child maltreatment, providing evidence that a proactive, enabling model of community support is a viable method for influencing parenting beliefs and parental competence, thereby reducing risk of child maltreatment. In loving memory of my father, Rufus Williams, whose encouraging words and gentle guidance live with me still, and for my mother, Aldena Williams, whose life reminds me daily of what one person can do to ease the suffering in this world. ACKNOWLEDGEMENTS This research would not have been possible without the cooperation and assistance of many individuals and organiza- tions. First, my deep appreciation to each of the parents and children who participated in the study, opening their homes and their hearts to us as we tried to answer these very important questions. My gratitude, also, to Toni Landick, Director, and all the staff and volunteers at the Family Growth Center for their hard work, commitment, creativity, and support. Their competence and professionalism were matched by warmth and empathy, creating a unique environment of acceptance and expertise. Thanks, too, to Ellen Behm, Executive Director, and the Board of Directors at Child Abuse Prevention Services, the administering agency of both the Family Growth Center and the contract which financed this project. Their desire to learn more about what helps families and prevents child maltreat- ment led to this research. Special thanks also go to Dianne May, Program Specialist, at Child Abuse Prevention Services, and to Nancy Chacon and Lori Rademaker, public health nurses, at the Ingham County Health Department, for the hours they spent conducting home visits and interviewing parents. Funding for this investigation was provided by the Michigan Department of Social Services, through its Office of Children and Youth Services. My gratitude to Dr. ‘Laura Daniel and others at OCYS who recognize the importance of both ii supporting innovations aimed at preventing child maltreatment and the need to evaluate the effectiveness of these efforts. I am also indebted to the College of Human Ecology at Michigan State University for the award of its prestigious Marie Dye Fellowship which allowed me to work full-time on this research. The support and assistance offered by friends and colleagues at the Institute for Family and Child Study was essential. Thanks especially to Dr. Robert Boger, my advisor, and to all the members of my committee: Dr. Margaret Bubolz, Dr. Esther Fergus, Dr. Robert Griffore, and Dr. Dennis Keefe. Also, I extend my deep appreciation to Dr. Lee Ann Roman for her consultation and listening ear; to Dr. Tom Luster, for sharing his time, his own dissertation, and his moral support; and, to Judy Pfaff for her expertise and hand- holding as we grappled with the computer analyses. I am also very grateful to Dr. Stephen Bavolek and Dr. James Garbarino, two geographically distant mentors who indirectly guided the development of this effort through their own published research and their willingness to share ideas and materials. Finally, to Wayne Workman, my husband and best friend, thank you for believing in me and giving me the emotional and financial resources I needed to complete this project; and, to our children, Christina, Stacy, and Ben, thank you for the profound learning which is only possible through being a parent, and for making us so proud to be yours. iii TABLE OF CONTENTS List Of Tables. 0 O O O O I O O O O O O O O O O O 0.. O O ..... O O O O 000000000000 OVi List Of Figures. 0 C O C O C O O O C O O O O O O O O O O OOOOOOOOOOOOOOO O ..... OVii I. INTRODUCTION.......... ............................ ....1 1 O 1 0 Purpose Of Study 0 O O O O O O O O O O O O O O O O O O O O ....... 4 1.2. Statement of the Problem and Research Questions...................................6 1.3. Conceptual Model for Investigation...........7 1.4. Assumptions.................................19 1.5. Conceptual and Operational Definitions......25 1.6. Theoretical Propostions and Hypotheses......28 II. LITERATURE REVIEW. 0 O O O O O O O O O O O O O O C O O O O ...... O O O O O O O O .33 2.1. The Ecology of Child Maltreatment...........33 2.2. The Mediating Influence of Social Support...79 III. METHODSOOOOOOOOOOOOOOOOOOOOOOOOOOOO ........ 00.0.000094 3.1. DeSignOOOOOOOOOOOOOOOOOOOOOOOO000......0.0.096 3.2. sampleOOOOOI...OOOOOOOOOOOOOOOOOOOOOO0.0.0.096 3.3. Dependent Variables........................101 3.4. Independent Variables......................105 3 O 5. Procedures. 0 O O O O O O O O O O O O O O ...... O O O O O I O O O O O 114 3.6. Data Analyses........... ................... 116 IV. RESULTSOOOIOO0.0000000000000000000000000... ......... 120 4.1. Association Between Parent/Family Traits and Parenting Outcomes....................121 4.2. Association Between Utilization of FGC Support and Parenting Outcomes............127 4.3. Association Between Parent/Family Traits and Utilization of FGC Support............129 iv 4.4. Differences in Parenting Outcomes Between High/Low Utilization and Between Nurturing and Comparison Groups...........130 4.5. Relative Influence of FGC on Parenting Beliefs and Parental Competence...........141 4.6. Mediating Influence of Parenting Beliefs on Parental Competence....................170 4.7. Case StUdieSOOOO0.00000000000000000000000.177 DISCUSSIONCOOOOOO0......OOOOOOOOOOOOOOOOOOOOOO0.0000184 5.1. Proposition 1..............................186 5.2. Proposition 2..............................191 5.3. Proposition 3..............................192 5.4. Proposition 4..............................194 5.5. Proposition 5..............................197 5.6. Proposition 6..............................199 5.7. Proposition 7..............................202 5.8. Assessing the Conceptual Model.............205 SUMY AND CONCLUSIONSOOOOOOOOOOOOOOOOOOOOO0.00.0.0209 6.1. Implications...............................211 6.2. Limitations................................214 6.3. Suggestions for Future Research............215 6.4. Conclusion.................................217 APPENDICES APPENDIX A: Participant Consent Form APPENDIX B: Instruments for Dependent Variables APPENDIX C: Questionnaires for Independent Variables APPENDIX D: Case Studies BIBLIOGRAPHY é-hh-bb-b hhpbuuuuuuw onooqoun bUNl-‘O‘U'Iwal-‘P LIST OF TABLES Factors Contributing to Child Maltreatment............77 Sample Characteristics...............................100 Childhood Abuse Experienced by Parent Participants...107 Components of Level of FGC Utilization Scores........109 FGC Utilization Sub-Sample Characteristics...........110 Summary of Measures..................................113 Data Collection......................................114 Correlations: Parent/Family Traits and AAPI Scores...122 Correlations: Parent/Family Traits and NCATS,HOME....124 Correlations: FGC Utilization and Parenting Outcomes.128 Correlations: FGC Utilization and Parent/Family Traits.........................................129 ANCOVA 1: Mean Scores and Standard Deviations........133 Summary of Analysis of Covariance 1..................135 ANCOVA 2: Mean Scores and Standard Deviations........138 Summary of Analysis of Covariance 2 .................140 Intercorrelations: Independent Variables............142 Regression Statistics Explaining Independent Variables................................ ...... .145 Preliminary Regression Statistics Explaining Parenting Outcomes..............................150 Regression Statistics Explaining AAPI Scores.........153 Regression Statistics Explaining NCATS Scores........158 Regression Statistics Explaining HOME Scores.........162 Correlations: Parenting Beliefs and Parental Competence......................................171 Regression Statistics: Mediating Influence of AAPI on NCATS................. ...... .................173 Regression Statistics: Mediating Influence of AAPI on HOME.........................................175 vi n4>nnntsh4>auuwsaraprd O O O O O O O O O O O O O O HFJ©¢D~JQCthdh3HLJNPJ 4.12 LIST OF FIGURES Ecological Framework: Causes of Child Maltreatment...11 Child Maltreatment Prevention Model...................18 Empirical Model for Investigation.....................19 Explaining Maternal Education........................146 Explaining Marital Status............................147 Explaining Number of Children........................147 The Interrelations Among Parent/Family Traits........148 Influences on Total AAPI Scores......................154 Influences on Reported Empathy.......................155 Influences on Reported Belief in Corporal Punishment.156 Influences on Reported Role Reversal.................157 Influences on Total NCATS Scores.....................159 Influences on Cognitive Growth Fostering.............161 Influences on Total HOME Scores......................164 Influences on Maternal Responsivity..................165 Influences on Physical and Temporal Environment......166 Influences on Maternal Involvement...................167 Influences on Variety in Daily Stimulation...........168 Influences on Avoidance of Restriction and Punishment......................................168 Influences on Provision of Appropriate Play Materials.......................................169 Influences on Parenting Beliefs......................205 Influences on Parental Competence: NCATS.............206 Influences on Parental Competence: HOME..............207 vii I. INT 0 ON No nation, and especially not this one at this stage in its history, can afford to neglect its children. Whatever importance we attach as a people to expenditure on armaments, to programs for older Americans, to maintaining high levels of consumption and to a hundred other purposes, the welfare of children has to be our highest priority... In the end the only thing we have is our young people. If we fail them, all else is in vain. Alan Pifer, Carnegie of New York Since the passage of Child Protection Laws in the mid-70’s, reports of child abuse and neglect have escalated dramatically. In 1986, 2.2 million children were officially reported as maltreated by child protective service agencies in the United States, an increase of 223% since 1976 (American..Association for Protecting' Children, 1987). Further, an incidence study conducted by the American Humane Society revealed that reported cases are only the tip of the iceberg; nearly two thirds of cases known to professionals are not reported to Children’ 5 Protective Service agencies (Russell and Trainor, 1984, p. 14). Research suggests that as abused and neglected children reach adolescence and adulthood, many become society's most disabled, dysfunctional and dependent individuals (Meyers and Bernier, 1987). Increasingly, child maltreatment appears to be a common denominator in our most serious social problems -- from delinquency and runaway behaviors of adolescents to violent and sexual crimes of adults. And, for many families, child maltreatment and family violence become patterns which are repeated in each new generation. (p. 25) Our society's approach to the problem of child maltreatment is built on a deficit model; it is largely legalistic, punitive, and often ineffective. Resources and interventions are not made available until a "case" is reported, investigated, and substantiated by legal evidence of abuse or neglect. Once this labeling occurs, the family has access to rehabilitative services. But in becoming eligible for service, troubled families are segregated and relegated an underclass status in the community. Family stress frequently escalates due to the intrusion of public, accusatory agents into a domain generally considered private and protected. Parents respond defensively, becoming uncooperative and resistant to any "help" offered by their accusers. To obtain cooperation, legal action may be taken, with petitions filed to the juvenile court for state jurisdiction of the child. Or, if there is insufficient evidence for court action, the case is closed without service. Nearly half of the cases opened by Children's Protective Service agencies are repeat referrals (Michigan Department of Social Services, 1987). Efforts to improve the treatment of child abuse continue, but the enormous human and economic costs incurred in after the fact intervention have led to the belief that child maltreat- ment can and should be prevented. While after the fact intervention can help prevent revictimi- zation in some families, its potential to change already established dysfunctional patterns of parenting is limited. Large-scale studies of child abuse treatment programs suggest that treatment may be effective in prevention of reoccurrence of maltreatment in fewer than half of participating families (Meyers and Bernier, 1987). The case for prevention is strengthened further by the enormous cost of after the fact treatment. Daro (1988) calculated some of the direct public expenditures associated with child maltreatment and conservatively estimated that medical, educational, and foster care services for child abuse and neglect cases cost approximately $500 million each year in the United States. This price tag does not include indirect costs stemming from the aftermath of child maltreat- ment, such as treatment for victims, long term mental health services, and public and private institutionalization. In addition, the economic costs reflect only one dimension of the problem; the lost human potential and untold suffering which accompanies child maltreatment are immeasurable, but pervasive losses. Further, it appears that the number of families at risk for child maltreatment is on the rise. A national profile of the maltreating population reveals that it is composed heavily of female-headed households, on public assistance, with younger caretakers and more children than families in the general population (Russell and Trainor, 1984). The proportion of families headed by single women and in poverty conditions has escalated dramatically in the past decade, raising serious concerns for the future of our nation’s children (Children’s Defense Fund, 1988). The 0.8. Census Bureau reports significant correlations between family type and poverty; a finding validated by the University of Michigan’s longitu- dinal study of income dynamics (Duncan, 1984) . Today, one- fifth of all families with children under 18 are headed by a woman, and, nearly one child in five lives in poverty. These facts suggest that the number of families at risk for child abuse and neglect is multiplying rapidly. Clearly, our society cannot hope to stem this rising tide by relying on after the fact intervention. During the past decade, a wide assortment of prevention programs and policies have been developed. A concensus about the central characteristics of prevention is emerging, emphasizing a comprehensive approach with multiple strategies, delivered through a variety of community institutions representing all sectors, and targeted to both the general population and to high risk groups (Meyers and Bernier, 1987). At this early juncture, however, most of these programs lack solid evalua- tions and information for replication. This investigation examined the influences of a community prevention program, the Family Growth Center in Lansing, Michigan, on parenting outcomes. 1.1- W 0—1" __D.¥STU Among those prevention programs which have been implemented, many are never formally evaluated or reported in the pro- fessional literature. Helfer's (1982) review of the prevention literature revealed that only three studies met his criteria for experimental research in preventing child maltreatment. The work of Boger et al. (1986, 1988) and Olds (1987) reflect changes in the pattern of primary prevention studies, but too few evaluation efforts are based upon sound empirical design. Without clear evidence of effectiveness, obtaining adequate commitment to and resources for prevention initiatives is difficult at best. In addition, the lack of evaluation data hampers diffusion efforts. As Krugman (1985) points out, the field must be willing to change if the data are convincing. But, when there are no data, the grounds for change remain theoretical and unconvincing. An ecological review of the causes of child maltreatment illustrates 'the,:multiplicity' of factors which combine to create role malfunction and pathological adaptation by care- giver and child. (An ecological model for the causes of child maltreatment is described later in this chapter, and a review of relevant literature is included in Chapter 2.) While economic hardship is an overwhelming characteristic of abusive and neglectful families, the inadequacy of social support and feedback to caregivers in our culture fuels isolation and stress, creating conditions under which maltreatment flourishes (Creighton, 1985; Polansky, et al., 1976; Gil, 1970; Gelles, 1973; Pelton, 1981; Garbarino, 1976; Egeland and Brunnquell, 1979). In Lansing, Michigan, Child Abuse Prevention Services has established several services aimed at strengthening families and reducing risk of child maltreatment. The Family Growth Center (FGC) program is based on the above research findings and is delivered at neighborhood sites. The Center provides both informal and formal social support, including respite child care and parent education, as well as linkage to other more formal community resources to self-referred and agency- referred families. The primary purpose of this research is to evaluate the effectiveness of the Family Growth Center, a community prevention program, in influencing parenting outcomes. 1.2. STATEMENT Q2 THE PROBLEM AflQ RESEARCH QUESTIONS This study will examine factors related to parenting beliefs and parental competence, focusing particularly on the role played by a community prevention program in influencing these parenting outcomes. The following research questions are addressed: 1) In what way are parent/family traits related to parent- ing beliefs and parental competence? 2) To *what degree is utilization of social support associated with parenting beliefs and with parental competence? 3) Are specific parent and family traits correlated with utilization of social support? 4) When confounding variables are controlled, is there a difference in parenting outcomes between high and low FGC utilization groups and between the FGC nurturing program group and a comparison FGC group? 5) What is the relative influence of a community prevention program (the Family Growth Center) on parenting beliefs and parental competence? 6) To *what extent. do jparenting beliefs mediate the relationship between parent/family traits and parental competence? 1.3. CONCEPTUAL MODEL FOR THE INEESTIQAIAQN Bronfenbrenner’s (1979) ecological model offers an ideal framework for building a comprehensive picture of child mal- treatment. Using a series of nested concentric circles, Bronfenbrenner defines the levels of influence which affect the living organism. The immediate and most powerful setting, the family, is embedded within and interacts with institutional systems such as the school, neighborhood, and church. These institutions are, in turn, contained within a broader, cultural context, which limits and shapes what occurs in the inner circles. This cultural context is influenced by the historical period and the prevailing technologies and values of the time. In Bronfenbrenner’s words, the ecology of human development is defined as: ...the scientific study of the progressive, mutual accommodation between an active, growing human being and the changing properties of the immediate settings in which the developing person lives, as this process is affected by relations between these settings, and by the larger contexts in which the settings are embedded (p. 21). As Garbarino (1977) points out, this ecological approach can "cope with the complexity of child maltreatment" (p. 722). Such a model focuses on the progressive mutual adaptation of organism and environment and recognizes the interdependent interaction of systems and the importance of social context. Belsky (1980) hypothesizes that child maltreatment is multiply determined, that these multiple determinants are ecologically nested within one another, and that much theoretical conflict characterizing the study of maltreatment is more apparent than real. He adds Tinbergen’s (1951) ontogenic level of development to Bronfenbrenner's ecological framework, differentiating those factors that the parent brings to the family setting from other microsystem influences. Viewing child maltreatment from an ecological perspective allows all the influences affecting parent-child interactions to become visible. Actions needed to prevent child abuse and neglect at each ecological level can then be proposed, discussed, implemented, and tested. If such activities occur simultaneously, a comprehensive, more efficient, approach is possible. An ecological review of the causes of child maltreatment provided in chapter two suggests that variables at each level of the ecosystem contribute to this outcome (Figure 1.1) . The fact that child abuse and neglect are recognized as issues of public concern is an encouraging first step. This chronosystem change began only a century ago and has grown significantly in intensity during the past two decades. Now, we struggle with the complexity and difficulty of the task before us. At the macrolevel, cultural ideologies which devalue children and view them as property of their parents make maltreatment possible. Garbarino (1977) views the cultural justification of the use of force against children as one of two necessary conditions causing child abuse. The fact that Western culture allows, and even encourages, the use of force against children makes the difference between "discipline" and abuse unclear. In addition, serious deficiencies abound in this nation's public policies and practices related to child nurturing. An example of this is the blatant lack of a national day care policy (Bronfenbrenner, 19844.. Cultural devaluing of children fosters inadequate public policies and practices, and both these factors contribute to insufficient social support. Frequently, this results in the isolation of families from potent social support systems, which, according to Garbarino, is the second "necessary condition" for child maltreatment. Child abuse can only occur when feedback and support are not adequately provided to persons in the caregiver role. Without support, stress becomes unmanageable, and without social sanctions precluding the use of force, unmanageable stress is fertile ground for child maltreatment. At the micro-level, isolation and dysfunctional parenting beliefs stemming from parent history and the immediate family situation contribute to parental role malfunction. Helfer (1980) describes the "World of Abnormal Rearing" or "WAR" cycle, in which early childhood deprivations result in developmental deficits, preventing the child from learning interpersonal skills necessary for competent and satisfactory adult roles. Polansky's findings (1981) support Helfer's model. Following fifteen years of research on child neglect, he suggests that two major character disorders differentiate neglectful parents: apathy-futility or impulsivity, and, infantilism. Polansky' concluded. that the .neglectful personality stems from intense, unresolved dependency needs. These personality deficits create barriers to effective peer communications, and, lead, in due course, to problems in parent-child interactions. They also contribute to family traits associated with child maltreatment, including early pregnancies, frequent separations, conflict ridden relationships, drug and alcohol abuse (Egeland and Brunnquell, 1979; Daly and Wilson 1985; Kaplan, et al., 1983). The resulting parental role malfunction is evidenced in a pattern of negative interactions with the developing child. Wasserman, Greene, and Allen (1983) found that abusive mothers used less verbal teaching, initiated fewer activities with their toddlers, and ignored them more than control mothers. They concluded that the abusive parent and her infant are locked into a mutually reinforcing negative spiral. The child's part in this transactive relationship is recognized by several researchers (Klein and Stern, 1971; Steele, 1980;-Hunter and Kilstrom, 1979; Glaser and Bentovim, 1979; Oldershaw, Walters, and Hall, 1986; Wasserman, Greene, 10 and Allen, 1983). This contribution may be biological (birth. weight, handicapping condition) or' behavioral. Wasserman, et al. (1983) found that infants who were abused more frequently ignored their mothers, while Oldershaw, et al. (1986) reported that children in an abusive group complied with their mother's commands less frequently than control children. FIGURE 1.1 Ecological Framework: Causes 9; Child Maltreatment Chrono/ Cultural Non-supportive Macro Devaluing -————-—> Public Policies of Children and Practices Exo/ Insufficient Meso Social Support and Feedback Micro: (Family Ecosystem) Individual Nonpreparation Parental Role 8 Family "———’ for Parenting —" Malfunction Deficits L T I I l (Child) : I Social- Child 7 Child Emotional F' """""" Maltreatment I Traits Deprivation (new cycle) 11 Parental role malfunction and the negative relations between parent and child, then, are only the last of a series of conditions which result in child maltreatment. They also form the initial conditions leading to the potential abusive behavior of the abused child as an adult. Virtually every retrospective study of characteristics of maltreating parents identifies an abusive/neglectful background as a significant factor (Blumberg, 1974; Egeland and Brunquell, 1979: Polansky, 1981). With each abusive or neglectful parent- child relationship, there is potential for a new generational cycle to begin. There is, however, no evidence to suggest that every adult who was abused or neglected as a child maltreats their own child. In fact, many survivors become healthy, well- functioning parents (Justice, Calvert, and Justice, 1985). Clearly, mediating factors can ensue to help break the cycle of child maltreatment. As noted earlier, ecological theory, as well as empirical evidence suggest that one important mediating variable in family functioning is the availability and enactment of social support. Social networks and the support that members provide both directly (meso-system) and indirectly (exo- system) affect the behavior, attitudes, and beliefs of individuals (Bronfenbrenner, 1979; Dunst, 1986). In 1960, Litwak reported that support to parents may be effective in buffering the stress associated with the birth of a child. 12 Since that time, a sizable body of literature has grown, suggesting that social support has powerful mediational influences on family functioning (Boger, et a1. 1983, 1986: Cochran and Brassard, 1979; Crockenberg, 1981; Hetherington, Cox, and Cox, 1976, 1978; McCubbin, et al., 1980; Mitchell and Trickett, 1980; Dunst, Vance and Cooper, 1986). Summarizing results of eight cross-sectional studies over ten years of work with families of developmentally impaired or at-risk children, Dunst and Trivette (1987) report that social support accounted for a significant amount of variance in personal well-being and family functioning. ...(the) stresses and strains associated with the rearing of a disabled or at-risk child are lessened by support available to individual family members, particularly support that matches family identified needs (p. 17). In a later publication based on the same studies, the researchers (Dunst and Trivette, 1988) state that parental attitudes toward and perception of their child are related to extrafamily support. They note that participant parents were less likely to portray their child’s behavior as troublesome or difficult when social support, well-being, and interactional patterns moderated their perceptions. Researchers have explored dimensions of social support and studied the relationship of these dimensions to various outcomes. Dunst (1985) reports that qualitative (satisfaction with support) rather than quantitative (number of sources of support) dimensions of support were the most important 13 mediators of family functioning. Clark (1983) found that help from esteemed network members who convey a sense of "we" rather than "you" and "I" in dealing with problems was more likely to produce beneficial effects. Vaux and Harrison (1985) found that satisfaction with support was related not only to the size of the support networks, but to closeness of network relationships, and composition of networks. The most important factors in satisfaction with social support were the existence of a marital or marital-like partner and the proportion of extended famiLy and close friends in the network. Similarly, Giovannoni and Billingsley (1970) found that relationships within the kinships systems were more important correlates of maternal adequacy than relations with friends or neighbors. On the other hand, both Dunst (1985) and Crnic et a1. (1983) found that extra-family or community support influenced parent perceptions about their children and buffered effects on parent-child interactions. Giovannoni and Billingsley (1970) reported that adequate mothers more frequently attended church and were much more often engaged in informal church activities than either potentially neglectful or neglectful mothers. These research findings, as well as others summarized in Chapter 2, suggest that. social support, particularly informal assistance provided by family, friends, and neighbors is a powerful factor influencing parent attitudes and family functioning. In their integration of network analysis and child develop- ment knowledge, Cochran and Brassard (1979) postulate both 14 direct and indirect influences of family support networks on the developing child. The child is influenced directly by contact with persons outside the immediate famiLy on a. recurring basis, and indirectly, through the mediating influence of the parent. Dunst and Trivett (1988) hypothesize that the mediating influence of social support on parenting attitudes is explained by three factors: 0 the assistance provided means that the burden of care doesn’t weigh solely with one person, excessive time demands are minimized, reducing the probability that well-being is negatively affected: 0 members of the parent’s social network serve as models: parents adopt or modify their parenting styles if esteemed network members demonstrate effective and nurturing behavior; 0 the opportunity to share "trials and tribulations" of childrearing with other parents helps individuals see that raising children is difficult and painful at times, and that no one person can do it alone. The importance of social support in rearing children is intuitively logical and practically evident to anyone who has parented young children. But barriers to such support have escalated during the past fifty years in this country. Experts in child development frequently speak of the disinte- gration of the extended family in our industrialized society and its implication for young parents. With the disintegration of traditional modes of family life and of the rearing of children, in the wake of our century’s massive urbanization and industrialization, we have lost the security people once derived from long-standing customs, from growing up as part of a large extended family, and from all other experiences these provided (Bettelheim, 1987, p. 8). 15 T. Berry Brazelton, an esteemed and frequently quoted expert on child-rearing, notes that when young mothers and fathers do not have another adult with whom to share negative, ambivalent feelings about their child, an isolated, frightening anxiety can build up. Brazelton links the increasing frequency of such isolation to our society’s escalating incidence of child maltreatment, particularly for children who are at provocative, difficult stages of development (Brazelton, 1974). He concludes that communication with other parents who are going through similar struggles takes on a "therapeutic aspect". Other parents reinforce the positive while allowing ventilation of the negative sides of parental ambivalence. Such sharing frees the parents of restrictive emotional baggage which can interfere with their ability to provide a caring atmosphere for their children. The difficulty of accessing adequate social support in this society is multiplied for parents who bring childhood deficits to their new role, as well as for those who are hampered by economic deprivation and isolating living condi- tions. Parents who were reared in dyfunctional families, where extended family members are not likely to communicate or model effective parenting, frequently lack the inter- personal skills required to build and maintain healthy support networks with partners or friends. In some cases, friends and family members may be available but their inter- action patterns tend to escalate rather than dissipate 16 parental pressure. Such relationships can be intrusive and draining rather than supportive and energizing. These same parents, as well as others who function at higher levels, may also be coping with single parenthood, insufficient economic resources, inadequate housing, poor health, and. other pressures which further erode their ability to be physically and emotionally available to their children. An ecological model for community prevention of child mal- treatment (Figure 1.2) reflects the complex etiology of the problem and recognizes the healing potential of positive social support systems. In this model, the neighborhood Family Growth Center supple- ments the participant family’s mesosystem, enhancing and expanding informal support networks and providing a bridge to formal support services. By proactively making available sufficient social support, the program enables and empowers parents to carry out their caregiving role» The increased level of social support and direct feedback influences parent ideology and competence, promotes effective child management and helps break the generational cycle of abuse and/or neglect. Because children are provided services at the center, it is hypothesized that the program also directly influences their social-emotional growth and future ability to parent. In these ways, the Family Growth Center program reduces risk of both current and future incidence of child maltreatment for participating families. 17 FIGURE 1.2 Child Meltreetment Prexentien Medel Exo/ Family Growth Center Meso Social Support Micro: (Family Eco-system) Parent Parent ‘ Parental & Family ’ Competence Traits ” A (Child) Social- Child 1 Child Emotional ““““““ Management Traits Growth (new cycle) In the current investigation, a portion of this model (Figure 1.3) will be tested in order to examine factors related to effective parenting beliefs and parental competence, particularly the relative influence of a planned community intervention on these outcomes. 18 FIGURE 1. 3 Empirical Model f9; lnvestigatlon Family Growth Center Social Support J, Parenting Beliefs \ Parent & Parental Family Competence Traits 1.4. S PTIO S 1.4.1. Ecological Assumptions Subjective Reality In ecological theory, reality is assumed to be both objective and subjective. But, as Urie Bronfenbrenner (1979) states, what matters for human behavior and development is environ— ment as it is penceivgd by the developing individual rather than as it may exist in objective reality. Reality is constructed by human beings (Bubolz, Eicher, and Sontag, 1979) . This constructed reality interacts with the subject in determining behavior. Kurt Lewin expressed this relationship symbolically: B=f(PE), that is, behavior evolves as a function of the interplay between person and environment (Bronfenbrenner, 1979) . This assumption of a subjective reality requires that research methodology and 19 program implementation incorporate both the human subject nng the environment in which s/he is embedded. Ecologists further assume that it is impossible to understand behavior solely from objective properties of an environment without reference to its meaning for the people in the setting (Bubolz, 1988; Bronfenbrenner, 1979; Hook and Paolucci, 1970; Bubolz, Eicher, Sontag, 1979; Westney, Edwards, Brabble, 1986). Practical intervention must therefore recognize the legitimacy of the family's perceptions with respect to their concerns and needs, as opposed to relying on an outside observer’s "objective" assessment of their reality. er e Since its earliest beginnings, the field of human ecology has emphasized the interdependence between humankind and environment, holding that: ...the structure and behavior of organisms are significantly affected by their living together with other organisms of the same and other species and by their habitat (Ernest Haeckel, 1870). Human development is thus a gynnnin process, involving mutual accommodation between the individual and the changing properties of the immediate setting, as this process is affected by relations between and beyond these settings (Bronfenbrenner, 1979). Explanations for human behavior are to be found in the interactions between characteristics of people and their environments, past and present (Bronfenbrenner, 1979). 20 Pragmatism Historically, human ecology has had the practical purpose of improving the well-being and life quality of humankind (Bubolz, 1988). Such a mission assumes application of theory and research to practice. The human ecologist conducts research in actual environments (both immediate and remote) in which human beings live. Human ecologists are defined as "change agents", aiding society in moving toward a better understanding of human-environmental interactions (Westney, Edwards, Brabble, 1986). It is there- fore fitting that ecological research and program implementa- tion focus on solutions to important human problems, such as child maltreatment. 1.4.2. Assunntlons about Child Abnse Preventign Human Potential for Child Maltreatment Child abuse prevention approaches assume that given sufficient environmental stress and accompanying inadequacy of personal and/or social resources, any adult is capable of maltreating a child (Gil, 1970; Gelles, 1973; Pelton, 1981). All families need information and social support in order to function adequately. Child maltreatment is therefore a symptom of family dysfunction; and, incidence of abuse and/or neglect is a social indicator reflecting the quality of life for families within a given society, community, or neighbor- hood (Garbarino, 1976; Garbarino, Sherman, and Crouter, 1979: Garbarino and Sherman, 1980; Steinberg et al., 1981). 21 EMEOEGEEGD E Child abuse prevention programs are generally based on a philosophy of empowerment, and aim to build on family strengths during the natural course of the life cycle. Although they may be targeted to certain high-risk groups, services are most often characterized by open eligibility and community-based.«delivery' through either informal networks (churches, neighborhood associations) or formal institutions (schools, hospitals, mental health, family and children's agencies). Highly dysfunctional families are best integrated into community education and support services which are available to anyone who wishes to enhance positive interpersonal patterns and strengthen parent-child relation- ships. This blending of families with varying abilities and needs allows mutual support, and "freedom from drain" within the social network (Garbarino and Sherman, 1980). Qomerebeol-‘d—veéeereeeh Finally, given the complexity of child maltreatment, it is assumed that a comprehensive approach is necessary, one that employs multiple strategies and involves a varied group of professionals and institutions representing all sectors of a community (Meyers and Bernier, 1987). While many programs are interdisciplinary in actual practice, they may be organized through the sector which takes a primary leadership role in initiating the particular effort, including: the medical/health care profession, community 22 support systems, the ‘workplace, social services, and educational institutions. 1.4.3. The Intezyennlen Hegel; Fenily finennn Center The Family Growth Center program model reflects what Carl Dunst (1988) calls an enabling model of intervention, which: creates opportunities for family members to become more competent, independent, and self-sustaining with respect to their ability to mobilize social networks to get needs met and attain desired goals (p. 88). Dunst outlines several principles which underly this type of approach: 1) The model assumes an ecological or social systems perspective on families (as described above). 2) The family, and not an individual member, is seen as the unit of intervention. 3) The program’s objective is empowerment of family members; interventions are carried out in a manner in which family members acgnire e sense 9; eontrol over their lives as a result of their efforts to meet their needs. 4) .A proactive stance is assumed in working with families; people are considered competent or to have the capacity to become competent. 5) Services focus on family“ identified. rather' than professionally identified needs and aspirations as primary targets. 6) Major emphasis is placed on identifying and building upon family capabilities in contrast to focusing on deficits and weaknesses. 7) A second major area of emphasis is placed on enhancing the family's social support network. 8) The model assumes a shift and expansion in roles ' played by professionals from an "expert" resource who decides what families need, to a highly responsive "partner" who creates opportunities for 23 family members to become more competent, independent, and self-sustaining with respect to their ability to mobilize social networks to get needs met and attain desired goals. In short, an enabling model of intervention places major emphasis on empowering families, strengthening their natural support networks, and enhancing their acquisition of a wide variety of competencies (Dunst, 1985: Dunst and Trivette, 1988). Such. programs de-emphasize the help-seeker’s responsibility for causing problems but emphasize his/her responsibility for acquisitions of competencies necessary to solve problems and meet needs. The focus is on growth- producing behaviors rather than treatment of problems, with attention paid to fostering acquisition of adaptive behaviors. The help-seeker plays a major role in deciding what is important while the help-giver supports, encourages, and creates opportunities for realization of identified goals. The Family Growth Center, in Lansing, Michigan, provides neighborhood-based support to families, offering drop-in child care, informal support groups, formal parent education sessions, information and referral, clothing and skill exchanges, social events, and volunteer training and place- ment. A small professional staff plans and coordinates delivery of these services in space donated by local churches, using 20 to 30 trained volunteers each month. Professionals from a variety of more formal helping agencies frequently offer services at the centers; for example, 24 public health nurses hold immunization clinics and mental health therapists may convene short-term therapy groups at the Center. Services are provided to any family with young children (2 weeks to 6 years) who chooses to come to the center. There is no cost for the service and no eligibility requirements. Parents are encouraged to donate goods and services as their resources allow and special eligibility programs have been developed for specific high-risk populations, such as adolescent parents. The centers are intentionally located in neighborhoods which are known to have high incidence of reported child maltreatment. 1.5. CONCEPTUAL MD OPERATIONAL DEFINITIONS The core concepts for this investigation are parenting beliefs and parental competence, which are the two outcomes of interest in testing the empirical model and evaluating the effectiveness of the Family Growth Center program. The independent variables of interest are parent/family traits and FGC social support. 1.5.1. erenden; Variables Parenting Beliefs A parenting belief is a mental position pertaining to children and childrearing which is reported by individuals legally responsible for dependent children. In this study, parenting beliefs associated with child mal- treatment are of particular concern. These include beliefs about use of physical punishment, roles of parents and 25 children, empathy for and expectations of children. Parenting beliefs associated with child maltreatment are operationalized by scores on the Adult-Adolescent Parenting Inventory (Bavolek, 1984). MW Parental competence is defined as the degree to which a person legally responsible for the nurturance of a dependent child demonstrates a capacity for appropriate social interactions with that child and provides a home environment which is safe and stimulating. Parental competence is operationalized by observations on the Nursing Child Assess- ment Teaching Scale (NCATS, Barnard, 1978) and the Home Observation for Measurement of the Environment (HOME, Caldwell and Bradley, 1979). 1.5.2. lndenengent Variables ParentlFanily Trelts Parent/family traits refer to those characteristics of parents and families which may influence parenting beliefs, parental competence, and utilization of social support. Two categories of parent/family traits are identified, those which parents bring to their caregiving role (parent history) and current situational characteristics of the family. Of particular interest in this study is the relationship of "high risk" parent and family traits, such as past childhood abuse to current parenting beliefs and competence. Other operationalized measures of the parent history dimension 26 include maternal education and maternal age at primiparity. Current family traits include family income, number of children, and marital status. This distinction between parent history and current family situation is important since historical factors may generally be assumed to precede current ideology and skills, while the direction of causation between current family situation and parenting outcomes is less validly presumed. In addition, current situational variables may be more malleable than factors experienced in the parent’s past, such as childhood abuse. Family Groytn genten Social Suppont Family Growth Center social support refers to the physical and instrumental assistance, attitude transmission, resource and information sharing, emotional and psychological support provided to families (Dunst, 1986) through the Family Growth Center program. In this study, two dimensions of social support are operationalized: FGC utilization and type of FGC service. FGC utilization refers to the degree to which families access support services. Operationally, level of FGC utilization is determined by an equation which combines the number of FGC services used, length of time the family has been involved, staff perception of parental involvement, and several other factors described in detail in chapter three. Type of FGC service refers to whether or not the family participated in a specific structured educational service 27 aimed at offsetting the generational perpetuation of dysfunctional parenting practices. This "Nurturing Program" (Bavolek and Comstock, 1985) entailed 13 weekly, 2 1/2 hour sessions in which parents and children learned, together and separately, new information and patterns of behavior. Constructs specifically addressed in the program included inappropriate parental expectations of children, parental lack of empathic awareness of child's needs, parental use of physical punishment, and parent-child role reversal; all of these factors have been associated with child maltreatment. 1.6. THEORETICAL PROPOSITIONS AND HYPOTHESES Several theoretical propositions and hypotheses are derived from the conceptual frameworks described above and are tested in the investigation. Eroposition l Parent/family traits will be systematically associated with parenting beliefs and levels of parental competence. HYP 1.1: Parental childhood. abuse ‘will be negatively correlated with scores on the AAPI, NCATS, and HOME inventories. HYP 1.2: Maternal age at primaparity will be positively correlated with scores on the AAPI, NCATS, and HOME inventories. HYP 1.3: Maternal education will be positively correlated with scores on the AAPI, NCATS, and HOME inventories. 28 HYP 1.4: Family income will be positively correlated with scores on the AAPI, NCATS, and HOME inventories. HYP 1.5: Number of children will be negatively correlated with scores on the AAPI, NCATS, and HOME inventories. Proposition 2 Family Growth Center (FGC) utilization will be positively related to non-abusive parenting beliefs and parental competence. Hyp 2.1: Level of FGC utilization will be positively correlated with scores on the Adult-Adolescent Parenting Inventory. HYP 2.2: Level of FGC utilization will be positively correlated with scores on the NCATS and HOME inventories. Proposition 1 Parent/family traits will be systematically associated with utilization of social support. HYP 3.1: Parental childhood abuse and number of children will be negatively correlated with level of FGC utilization. HYP 3.2: Maternal age at primiparity, maternal education, and family income will be positively correlated with level of FGC utilization. 29 Proposition 1 When confounding variables are controlled, parenting beliefs and parental competence will increase as a function of Family Growth Center utilization. HYP 4.1: Controlling for pretest scores, parents with a high level of utilization in the Family Growth Center will score higher on the AAPI than parents with a low level of utilization. HYP 4.2: Controlling for pretest scores, parents with a high level of utilization in the Family Growth Center will score higher on the NCATS and HOME inventories than parents with a low level of utilization. Proposition Q When confounding variables are controlled, parenting beliefs and parental competence will increase as a function of Family Growth Center type of service. HYP 5.1: Controlling for pretest scores, parents who participate in the Nurturing Program at the FGC will score higher on the AAPI than parents who receive other services only. HYP 5.2: Controlling for pretest scores, parents who participate in the Nurturing Program at the FGC will score higher on the NCATS and HOME inventories than parents who receive other services only. 30 Proposition g Social support provided by an enabling model of intervention will account for a significant amount of the variance in parenting beliefs and parental competence beyond that attributable to parent/family traits. HYP 6.1: Level of FGC utilization and/or type of service will account for a significant amount of the variance in AAPI scores after parent/family traits are accounted for in a hierarchical regression equation. HYP 6.2: 'Level of FGC utilization and/or type of service will account for a significant amount of the variance in NCATS and HOME scores after parent/ family traits are accounted for in a hierarchical regression equation. Proposition 1 Parenting beliefs mediate the relationship between parent/family traits and parental competence. HYP 7.1: Scores on the Adult-Adolescent Parenting Inventory will be positively correlated with scores on NCATS observations. HYP 7.2: Scores on the Adult-Adolescent Parenting Inventory will be positively correlated with scores on the HOME Inventory. 31 HYP 7.3: When AAPI scores are controlled, the amount of variance in NCATS and HOME scores explained by parent history and family situation will not be significant. The next chapter reviews the child maltreatment and social support literature which formed the basis for the conceptual model and the derived propositions described in this introduction. Chapter three summarizes the research methods and data analyses which were used in testing the stated hypotheses, Chapter four reports results of the study, chapter five discusses these results, and chapter six concludes with implications and limitations of the investigation. 32 II- LITERATHBE BEYIEH 2-1- THE EQQLQQX 9E QEILQ MALIBEAIMENI Pioneers in the emerging field of child maltreatment were medical professionals whose primary focus was, by nature of their function, on microsystem factors which contribute to the problem (Solnit, 1980). Through their direct care of individual families, these physicians and psychiatrists identified common patterns which differentiated maltreating parents. As a result, the earliest explanations for child abuse and neglect were based on retrospective clinical studies and case histories. The "psychiatric" model of child maltreatment was built from this base of knowledge and experience (Blumberg, 1974; Spinetta and Rigler, 1972). Others have since argued that a sociological model is more valid in explaining causes for child maltreatment, shifting the focus from the dysfunctional parent to a dysfunctional social environment (Gil, 1970; Gelles, 1973). Parke and Collmer (1975) describe yet a third approach, the social- situational, which explores how patterns of interaction between family members create and maintain an abusive cycle. These contrasting viewpoints have led to considerable debate among researchers and practitioners regarding whether the origin of child abuse is primarily based on a "sick parent" or a "sick society". Advocates for each model offer convincing arguments which lead to different implications concerning the appropriate intervention needed to successful- ly prevent child maltreatment. 33 The psychiatric model emphasizes therapy and education to parents, and future parents, which will enable them to more effectively carry out their caregiving role. Such programs are targeted particularly at families already identified as abusive or neglectful, but they are also advocated for those considered "high risk" due to characteristics which appear to be associated with maltreatment, and, for the general popula- tion during predictable times of family stress, such as the arrival of a new baby. The social-situational model expands this intervention to focus more broadly on changing inter- actional patterns which perpetuate the abusive/neglectful cycle. In contrast, the sociological model stresses sweeping societal changes which will address contextual factors associated with child maltreatment. Professionals from this perspective suggest major alterations in attitudes, values, economic policies, and social support systems so that healthier, more nurturing environments will be available to children and families. While proponents of these models tend to emphasize the differences in their views and methods, there is much evidence to suggest that all of these propositions are correct. Child maltreatment results when individuals with personality deficits, from dysfunctional family systems, living in non-supportive, stressful environments reach the end of their coping ability. Or, conversely, child maltreat- ment occurs when non-supportive, stressful environments foster dysfunctional family systems and produce individuals 34 with personality deficits. To argue which is the more fundamental cause of maltreatment is tantamount to debating which came first, the chicken or the egg. Using and ecological framework in discussing its etiology, Belsky (1980) argues that child maltreatment is multiply determined, that these multiple determinants are ecologically nested within one another, and that much of the theoretical conflict characterizing the study of maltreatment is more apparent than real. There is no one solution to child maltreatment, but many solutions. Our success in ameliorating the problem depends on the degree to which different knowledge bases are combined to reveal its complex etiology. Such a comprehen- sive perspective can lead to the understanding necessary for strategically targeting the multitude of talent and resources required to effectively address this tragic social problem. As noted in Chapter 1, Bronfenbrenner’s ecological model offers an ideal framework for building a comprehensive picture of child maltreatment (Bronfenbrenner, 1979) . In this literature review, factors associated with abuse and neglect are organized by the level of ecological influence, as defined by Bronfenbrenner. The review focuses primarily on physical abuse and neglect and does not incorporate a separate body of literature related to the etiology of sexual abuse. This limitation was necessary in order to accommodate the researcher’s self-imposed boundaries of time and subject matter. The following review should therefore be viewed as a first attempt in the development of an ecological framework 35 for studying the causes of child maltreatment (see Chapter 1, Figure 1.1). 2.1.1. ghronoeystem Effects en Qnild Maltreatmenn Some day, maybe, there will exist a well-informed, well-considered, and yet fervent public conviction that the most deadly of all possible sins is the mutilation of a child’s spirit; for such mutilation undercuts the life principle of trust, without which every human act, may it feel ever so good and seem ever so right, is prone to perversion by destructive forms of conscientiousness. (Erikson, 1972) Bronfenbrenner defines the chronosystem as historical context: the time in which we live and its implications for developmental experience (Luster, 1988). Historical Perspective Child maltreatment is not a new phenomenon. (Steele, 1980, Radbill, 1980, DeMause, 1975). DeMause postulates six evolu- tionary trends through which parent-child relations have progressed in Western civilization. Through this evolution, our culture has moved from the "infanticidal mode" (antiquity) in which parents resolved their anxieties and ambivalence about children by infanticidal acts, to the currently emerging "helping mode", where both parents participate fully in 'the child’s daily life, recognizing her/his need for protection and self-actualization. The stages described by DeMause reflect an ever increasing cultural empathy for children. Contemporary families may demonstrate any one of the six modes through which we have progressed. According to DeMause, parents who abuse or 36 neglect their children are operating at a less evolved level of caregiving. Similarly, cultures and sub-cultures which condone maltreatment reinforce and perpetuate less evolved modes of child rearing. In the late 19th Century, our nation was sufficiently "evolved" to recognize and respond to a case of child abuse (Solnit, 1980). In New York City, a tenement social worker took pity on "Mary Ellen", who was routinely chained to her bed and beaten by her mother. No laws existed to protect children from parental abuse at that time. :n: desperation, the social worker approached the American Society for the Prevention of Cruelty to Animals, which ultimately took legal action on Mary Ellen’s behalf. Noting that she was a member of the animal kingdom, the ASPCA invoked animal cruelty laws for Mary Ellen’s protection. The publicity surrounding this incident shocked the country, and resulted in the establish- ment of the New York Society for the Prevention of Cruelty to Children in 1874. By 1905, 400 such. societies existed, dedicated to the enforcement of existing laws that prohibited cruelty to human beings (Steele, 1980). After this flurry of organizational activity, child maltreatment gradually disappeared from public attention. A ”rediscovery" occurred in 1962, when Dr. C. Henry Kempe publicized the "The Battered Child Syndrome", originally reported by Dr. John Caffey in 1946 (Kempe, 1962). The syndrome referred to cases in which subdural hematomas in 37 infants were often associated with atypical fractures of the limb and ribs. Kempe's book prompted public outrage and a new movement to respond to vulnerable children, .After the publication of "The Battered Child Syndrome", new laws swept the country, delineating" societal responsibilities in 'the protection of vulnerable children. By the mid-70’s, a national center on child abuse and neglect was established and all fifty states had passed reporting laws mandating that certain professionals report suspected cases to child pro- tection units within the public welfare system. Since the passage of Child Protection Laws, reports of child abuse and neglect have escalated dramatically. In Michigan, for example, there were 42,000 reports in 1985, a 111% increase since 1975, when only 20,400 reports were received. The American Humane Association reports similar trends nationwide. This increase is explained, at least in part, by the fact that the public is now sensitized to child abuse and neglect, and consider it a social problem demanding action. In 1974, only ten percent of the nation’s population viewed child maltreatment as a prevalent social concern. By 1984, this proportion had increased to ninety percent (National Committee for Prevention of Child Abuse, 1985). ELLE—ls 0 'ca Memes in Definim Melee-2m Child maltreatment is typically viewed as one end of a continuum representing quality of parental caregiving: 38 If we were to draw a graph of all parents, ranging according to their ability, we would probably end up with a familiar bell-shaped curve. At one end would be a single dot, representing the only possible claimant to perfection as a mother, the Madonna, but let us not forget that Mary also had the perfect child. Most of us would fit into the large rounded part of the curve representing those who offer their children excellent, good, or good- enough parenting. At the other end of the spectrum, the curve would not descend steeply: rather it would slope very gradually and might cover some 20 to 30 percent of parents, all of whom have some difficulty in caring for their children adequately (Kempe and Kempe, 1978, p. 10). But where is the line drawn which separates an adequate parent from one who is abusive or neglectful? The above historical summary suggests that this line cannot be considered fixed. It is instead dynamic over time and space, influenced by attitudes, values, and standards predominant during a particular historical period, in a particular culture. Zigler and Aber (1981) assert that the basic nature of the phenomenon of child maltreatment is developmental, "a developing system of developing systems" (p.14). The parent, child, and environment are not static, but dynamic, ever- changing systems . Parke and Collmer (1975), in their interdisciplinary literature review of child abuse, present a definition which reflects the dynamic nature of the cultural context in which families are embedded. They state that an abuse victim is: any child who receives nonaccidental physical injury (or injuries) as a result of acts (or omissions) on the part of his parents or guardians the: yielaee .me eemnnm @3115 concerning the treatment of children (1975, p. 513: emphasis added). 39 Legal definitions of child abuse and neglect are stated in "Child Protection Laws" across the nation, incorporating the concept of "nonaccidental harm", but adding "threatened harm" and broadening the perpetrator to include any "person responsible for the child’s health and welfare" (Michigan Department of Social Services, 1985a). In addition, maltreatment is expanded beyond physical abuse, taking in negligent treatment and sexual abuse. While these statutes do not directly address the issue of community standards, enforcement of Child Protection Laws is carried out by individuals and institutions entrusted with the public protector role at a given point in time. Community standards regarding adequacy of parental (or substitute) care are pre- sumably reflected in the judgments made by these agents. 2.1.2. Meezosysten lnfluenees en Child Maltreatment ...acquiescence to the demands of industrialization can unleash social forces which, if left unbridled, can destroy the human ecology -- the social fabric that nurtures and sustains our capacity to work together effectively and to raise our children to become competent and compassionate members of society. (Bronfenbrenner, 1981) According to Bronfenbrenner (1979), the macrosystem refers to cultural or subcultural consistencies and the belief systems or ideologies which underly such consistencies. wwmm Definitions of child maltreatment depend not only on the time period in which the event occurs, but also the place. In her cross-cultural analysis of child abuse and neglect, Korbin 4O lie‘ 6 l A IN um I IAy .9. he: L. 4 54V“ 7'. ...E 4 '3 «R (I) O 2’ (1980) describes the complexity inherent in defining acceptable and unacceptable forms of caregiving across and within cultures. She distinguishes between the "emic" 1 perspective, the viewpoint of members within a given culture, and the "etic" perspective, the outsider's broader frame of reference. Cultural conflict in defining abuse and neglect occurs when there is disparity between emic and etic p e rspectives . For example, the Vietnamese practice of "cao gio", in which heated metal coins are forcefully pressed on the child's body, was first interpreted as abusive by Western observers. The practice is, however, a traditional Vietnamese technique for reducing fever, chills, and headaches. Conversely, many Western practices such as isolating children in beds or rooms of their own at night, forcing young children to sit all day in classrooms, or allowing children to cry themselves to S3~eep would be questioned by many cultures. The more valid definitions of child abuse and neglect must there fore incorporate emic perspectives and be limited to u idiosyncratic departure(s) from culturally and socially ac(zeptable standards that result(s) in harm to a child or QQInpromises his or her physical, emotional, cognitive, social, or cultural development" (1980, p. 22). This is not to say that one automatically excuses behavior which is rationalized by cultural upbringing. But, the context of a behavior must be viewed holistically and in light of the fact 41 that virtually all cultures hold the value that children may not be damaged. Polansky (1981) found that while people from different classes and sub-cultures differ in their definition of optimal care, they tend to agree when defining inadequate care. Ideelpeieel I.___nene__nfl es in Child M_l__e___ta tre tmen Korbin acknowledges the variations across cultures in their tolerance of social conditions which contribute powerfully to incidence of abuse and neglect of children. She notes that the idiosyncratic child abuse and neglect seen in the United States and other Western cultures are relatively rare in Small-scale, non-Western cultures. When child-rearing Practices around the world are viewed as a continuum, Western cultures tend to fall at the extreme end in regard to harsher expectations of compliance, earlier child-training parctices, and lower infant indulgence. In addition, children are more often raised in isolated nuclear and s single-parent households . A mother is more often isolated in her role as the exclusive caretaker. Other adults, such as grandmothers or more experienced kinswomen, are less often regular participants in child rearing. And there is less often an accepted folk wisdom or cultural blueprint for child-rearing strategies. Child rearing has moved more toward the exclusive domain of the biological parents rather than the larger community. (1980, p. 30) Such cultural conditions appear to increase risk for child maltreatment in Western society. There is cross-cultural evidence that parental rejection is associated with social 42 isolation while acceptance is increased by the involvement of additional adults in the day to day progress of parent-child relationships (Rohner, 1975; Rohner and Nielsen, 1978). Ong (1985) argues that motherhood in a capitalist society is oppressive and inevitably traps women by "individualizing their predicament". Because gender roles are established through the family, women are assigned tasks related to reproduction and nurturance. The majority of mothers and children are isolated, with women assuming power only in the private domain. The structural powerlessness of women in the public sphere violates women as persons, and results in Violence toward their less powerful children. In a qualitative study of eight abusive mothers, Ong found that most had their first baby in their teens and felt they were "trapped into motherhood". All lived in households with a strict division of labor, based on conventional gender 3:‘<>1es, and experienced financial constraints which precluded sllbstitute care. They were rarely able to satisfy their own aite the theoretical support for this model, its implementation has been slow in coming. A multidisciplinary approach is employed in only a few cases involving child abuse, and even more rarely in child neglect (Mouzakitis, 1985). In 1976, Helfer and Schmidt pointed out that successful multidisciplinary team services requires a blfoadening of the concept of protective services from that of a. "unidisciplinary social agency to a comprehensive team of Irrofessionals who work together as a single unit". A decade later, in 1987, Helfer once more challenged the field to eniminate "Pascal’s Method of Community Services", the random assignment of families to rigidly defined services, and advocated for coordinated community service teams fer assessment and treatment of cases. There is need within each community for an interagency structure for responding to child abuse and neglect, easing entry into needed services, monitoring progress, and establishing community standards for legal jurisdiction (Pelton, 1983). In addition, given the escalating numbers of cases, and the ineffectiveness of after the fact intervention, comprehensive prevention initiatives may be the only real solution to this tragic problem (Helfer, 1982; Daro, 1985; Meyers and Bernier, 1987). The effectiveness and cost-effectiveness of prevention has been well documented in other areas of child and family welfare (Children's Defense Fund, 1986; Berrueta- Clement et al., 1984). Evaluation of child abuse prevention programming is just beginning, but promising programs and 50 Policies have been developed. Experts in the child abuse field are now proposing multidisciplinary strategies for Prevention (Cohn, 1983; Helfer, 1982; Newberger and He“Therger, 1982) . These advocates conclude that prevention Planning, development, and field testing with evaluation are hllportant initiatives to be undertaken in every community. The above review reveals that malfunctioning within the Social service delivery system contributes to the continuing maltreatment of children. This is evidenced by the horror Stories periodically appearing in the media, in which Children suffer death at the hands of their parents after being returned home by authorities, or, are victimized by the "system", shuffled from one foster care home to the next until they reach late adolescence. These exosystem factors must be addressed to successfully reduce child maltreatment, whether it is perpetrated by parental or societal caregivers. 2.1.4. Mesosystem Influences 2g Child Maltreatment 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 feed- back, the essential elements of support systems (Garbarino and Sherman, 1980). A mesosystem comprises the interrelations among all those settings in which the individual actively participates, referring, for example, to the relationship between the school and the home, or the family and its neighborhood (Bronfenbrenner, 1979). 51 Matisse; fires SEVeral authors argue that the cause of child maltreatment is unmanageable stress, created by the family's social Situation. Gil’s (1970) analysis of the problem focuses on s°<.:io-economic pressures which weaken the caretaker's psycho- 1Ogical mechanisms of self-control. Such pressures include 10w socio-economic status and limited education. Ideological acceptance of "normal violence" within certain sub-cultures erls the situation. Gelles (1973) presents a social- psychological model for explaining child maltreatment which features situational stress as the precipitating factor in child maltreatment. In the "Myth of Classlessness", Pelton (1981) disputes the commonly presented assumption that child abuse and neglect occur without regard to socioeconomic class. He points out that while maltreatment can and does happen in high income, highly educated families, the preponderance of reports involves families from low socio- economic levels who live in poverty or near poverty circum- stances. Refuting the argument that "reported" cases do not reflect the true distribution of child abuse cases, Pelton notes that increased reporting over the years has not led to a change in the economic patterns of cases, and, the highest incidence occurs in groups living in the most extreme poverty. The vast majority of child abuse fatalities are children from very poor families. The influence of situational stress on child maltreatment rates is supported by numerous researchers. Straus (1980) 52 Studied a nationally representative sample (n=1,146) and found that parents who experienced none of the eighteen Stresses in an instrument patterned after the Holmes and Rahe Stress scale had the lowest rate of child abuse. As the rulmber of stressors increased, so did the rate of child allbuse. In their review of retrospective literature identi- fying etiological factors in child abuse and neglect, Egeland and Brunnquell (1979) consistently found that unusual amounts (Jf'social and economic stress were mentioned as characteris- tics of maltreating families. Using counties in New York as the units of analysis, Garbarino (1976) tested the hypothesis that socioeconomic support for the family is directly associated with the rate of child maltreatment. In the study, child maltreatment rates were based on the number of reports received by the local protective services unit from September 1, 1973 through January 31, 1974. U.S. Census data were used to generate indices of socio—economic and demographic characteristics of the: counties ‘under' study. These ‘variables reflected five dimensions: transience, economic development, educational development, rural-urban, and socio-economic situation of mothers. Using stepwise multiple regression, Garbarino found that five of the variables accounted for 36% of the variance in predicting child maltreatment rates; all twelve variables accounted for 41% of the variance. The factors most directly reflecting the socioeconomic support system for mothers were the most predictive: percentage of women in labor force who 53 $3! Lia to: have children under 18 years of age and median income of households headed by females. Steinberg and associates (1981) used two census areas (Los Angeles and Anaheim) to study the association between child maltreatment rates and community economic change. They hypo- tll'lesized that the net loss of jobs at the community level 31¢aads to actual or anticipated individual job loss and increased parental stress, which in turn leads to increased child abuse. They collected monthly child abuse reporting rates over a two year period (February, 1975 through February 1977), as well as changes in the size of the work force, and conducted a time series analysis based on monthly values. Results showed that abuse was inversely related to the size of the workforce in both counties. In a controlled study of mothers of maltreated children in suburban families, Salzinger and associates (1986) focused on the type of life stresses which may differentiate abusive and non-abusive families. The researchers interviewed 41 mothers from open children's protective services cases regarding life events, and compared their responses to 24 randomly selected control mothers. The sample was mainly white, middle to lower-middle class: both groups were matched on socio- economic status, age, and family size. Results showed that the groups were similar in the number of stressful life events experienced, but were different on the me. While illness and death were equally prevalent between the two 54 r‘ “J p «gee A Lens P 9r5n j groups, the frequency of separations experienced was greater for the maltreating families. The abusive group also reI><3rted more spousal discord, more frequent alcohol and drug altDuse in their immediate families, fewer positive recent e‘Ients, and more isolation. Family Support Systems The association between adequate family support and child maltreatment is documented by several researchers. In a follow-up study comparing high risk families who were later reported as abusive against those who were not, Hunter and Rilstrom (1979) identified differences which might explain how some parents overcome a family pattern of inadequate childrearing. They found that the social resources of non— repeating families included a richer network of social connections compared with the repeating families. Friends, relatives, churches, were mentioned more often. Non- repeating mothers who were not living with the father of the baby were likely to be living with their own family. By contrast, this option did not appear to be available to any of the mothers whose babies were subsequently reported to be maltreated. Egeland and Brunnquell (1979) report similar findings from their study of 275 low-income, primiparous women who were receiving prenatal care at a Minneapolis clinic. All were considered "high risk": 60% were unwed at the time of their pregnancy. When these subjects’ infants were 20 months old, 55 the researchers identified 26 whose children were receiving inadequate care (four of whom were open Children's Protective Services cases) and compared them to a group of 25 11"Others who were offering high quality care. Several factors differentiated the two groups, the most significant being the level of social support available to the mother. The majority of inadequate mothers (74%) were single with only 45% reporting that either the baby’ s father or their extended family was supportive. Only 32% of the adequate mothers were single, and 100% of them reported that either the infant's father or their extended family was supportive. Green (1979) did a qualitative study of eight abusive fathers and found that all reported non-supportive social environments. In his study of 1400 expectant mothers who were screened relative to abuse risk and then followed after pregnancy, Altemeier (1979) found a significant correlation (r=.332) between social support and later abuse and neglect. Though some of these studies have limitations in methodology which preclude causal inferences, the preponderance of evidence supports the theory that social support is an important mediating variable in the etiology of child maltreatment. Garbarino and associates conducted a series of studies investigating social support as a function of neighborhood context (Garbarino, 1976: Garbarino, Sherman, and Crouter, 1979: Garbarino and Sherman, 1980). These studies demonstrated that the major ecological feature of child maltreatment involves economically depressed mothers, alone 56 in the role of parent, attempting to cope in isolation without adequate facilities and resources for children. Using the predictive equation developed in New York (Garbarino, 1976), the researchers compared "high risk" neighborhoods (higher than predicted child abuse and neglect rates) with "low risk" neighborhoods (lower than predicted rates) in Lincoln, Nebraska. Criteria for selection included large discrepancies between actual and predicted rates, high child maltreatment rates, and similar socioeconomic and racial profiles. The researchers conducted interviews with "expert informants" and did on-site inspection of neighbor- hoods, as well as interviews with 46 resident families. They found a consistent pattern differentiating the two areas. In each of the eight measured aspects of neighborhood (public image, quality of life, informal suppports, etc.), interviewee comments painted a negative picture of the high risk area. In addition, there was less exchanging of child supervision in the high risk area, less self-sufficiency, and less adequate provision of child care. Overall, the high- risk area was more stressful and "run down". The authors conclude that those families who need the most social support tend to be clustered in areas that are least able to meet those needs. Under such circumstances strong support systems are most needed, but least likely to operate, of course. The high-risk area needs outside intervention to increase its capacity to fend for itself and to strengthen families as a way of reducing the demands they place on already tenuous informal helping networks (1980, p. 195). 57 The above review suggests that the mesosystem contributes to the family's vulnerability to maltreatment in two ways. First, the environmental context may directly contribute to the degree of stress experienced, and, secondly, support needed to counteract such pressure may not be made available to the family system. ‘Under these conditions, the stress experienced. by the parent is compounded rather than ameliorated and child maltreatment may result. 2.1.5. Microsystem Influences pp Child Maltreatmepp One of the most important skills that a child must learn, during the brief years of childhood, is how to get his or her needs met in an acceptable manner and when the most appropriate time is to seek this fulfillment...-All people have needs, and we all must fall back on the foundations laid in childhood as we develop ways to have these needs met. (Helfer, 1980, p. 39) Bronfenbrenner defines the microsystem as the "pattern of activities, roles, and interpersonal relations experienced" by an individual in a given physical and social setting (1979,p. 22). Charastsristiss.2f the Eamilx firstsm In addition to socio-economic stress, four major features of family systems appear to increase risk for child maltreat- ment: structure, size, social isolation, and alcohol/drug abuse. Incidence studies of reported child abuse cases in the United States and Britain demonstrate an over-representation of single parent families headed by women in the population 58 (American Humane Association, 1984: Creighton, 1985). In light of the above discussion regarding mesosystem influences, this finding is not surprising. As Garbarino (1976) points out, .the relation of women, particularly mothers of young children, to the world of work is problematic. Women who work outside the home consistently receive less income than do men, making them more likely to experience economic stress. In addition, there is a general inadequacy of child support services in low-income settings. Single women with children who choose not to work are likely to live in poverty conditions which contribute to situational stress and adverse effects on children. There is also evidence suggesting that stepparent families may be at increased risk for child maltreatment. Creighton (1985) found that less than one-half of abused and neglected children in Britain were living with both natural parents. She also noted a relationship between family structure and the type of abuse, finding that single mothers were more frequently neglectful and substitute fathers more frequently sexually abusive. Daly and Wilson (1985) compared household composition to child abuse reports in Ontario, Canada and found that all household types other than two natural parents are high-risk environments for child abuse. They report that the risk was greatest for young children in stepparent house- holds. Giles-Sims and Finkelhor (1984) summarized possible explanations for the association between child maltreatment and stepparent family structures. The social-evolutionary 59 theory postulates that parental investment is less because children do not carry on genes, making child abuse more likely. It is also theorized that stepfamilies experience more stress which contributes to child maltreatment; and, the legitimacy of the stepparent role may be questioned, leading to dynamics which block parental authority and lead to an abusive response. Finally, the effect may be the result of selection: that is, the same factors which make people prone to divorce and remarriage may make them prone to abuse or neglect children. Family size and child spacing have also been identified as covariates with child maltreatment. In his review of child neglect literature, Polansky (1981) found that the number of children in a family was one of the main factors associated with child neglect. Creighton (1985) also reports that abusive and neglectful families in Britain have larger than average numbers of children even when socio-economic class is controlled. Hunter and Kilstrom (1979) found that adequate child spacing was one of the three factors differentiating abusive from non-abusive at-risk families. Given the influence of social support on child maltreatment, it is not surprising that isolation shows up as a family characteristic in abuse and neglect literature. Polansky, et a1. (1981) studied factors, other than low-income status, which affect the level of care parents provide children and concluded that isolation from helping networks was a major 60 contributor to child neglect. He notes that the quality of child care is related inversely to poverty, but is separable. The number of things over which parents have power expands when others are available and willing to grant access to their material and social possessions. In her comparison of abusive and non-abusive mothers, Salzinger (1986) also found that maltreating families tended to be more isolated than the comparison group. Milner and Wimberley (1979) constructed an instrument aimed at assessing child abuse potential, then tested it on reported abusive and non-abusive groups. The twenty-five items accounted for 99% of the variance between the groups. The scale measured four factors, one of which was "loneliness", the parent's feeling of being alone, rejected and unloved. The abuse of alcohol and drugs appears to be more common in maltreating families than in the general population. Kaplan, et al. (1983) compared 76 parents of abused and neglected children to 38 control parents, both of whom were being treated at a New York hospital. The two groups were statistically equivalent on socio-economio status, racial and ethnic distribution, family structure, family size, parental age, and children's age. The abusive group was significantly higher on alcoholism (25% compared to 5% in the non-abusive group). In her comparative study, Salzinger (1986) also found that maltreating mothers were exposed to alcoholism and drug abuse in their immediate families more frequently than the non-abusive group. Altemeier (1979) reports that alcohol 61 and drugs were significantly correlated (r=.399) with reports of abuse and neglect in his comparison of high and low risk families. Webster-Stratton (1985) found that 48% of abusive families had problems with alcohol and drug use compared to 24% of non-maltreating families. Based on case studies and literature reviews, Blumberg (1974) and Green (1979) argue that alcohol is an important contributing factor to child maltreatment. Green postulates: ...alcohol is perceived as the ultimate source of dependency gratification which cannot be obtained from spouse and family, and it temporarily obliterates feelings of inadequacy, depression, and low self-esteem. (p. 275) BQEQQEQL QQ§£§Q£§£i§E12§ and §§2é212£§ Professionals focusing on microsystem causes for child mal— treatment offer convincing evidence which suggests a generational cycle of abusive behavior stemming from the parent's early childhood experiences (Helfer, 1980). Virtually every retrospective study of characteristics of maltreating parents identifies an abusive/neglectful back- ground as a significant factor (Blumberg, 1974: Egeland and Brunnquell, 1979: Polansky, 1981). Kaufman and Zigler (1987) caution, however, that many methodological problems accompany a large number of these studies, most notably failure to utilize representative samples, comparison subjects, observers blind to subjects’ maltreatment status, formal definitional criteria, and descriptive or inferential statistics in reporting results. After integrating the findings of different investigations, the authors estimate 62 that the rate of abuse among individuals with a history of abuse is approximately six times higher that the base rate for abuse in the general population. Kaufman and Zigler conclude that there are many factors that diminish the like- lihood of abuse being transmitted across generations. In his review of the literature, Spinetta (1972) acknowledges the limitations of such research, noting that most are broad studies which do not test specific hypotheses, the samples studied are not representative, and they are based on ex post facto analyses. However, the preponderance of such findings, across many places and subjects, offers a convincing argument for the validity of the aggregate evidence. He concludes that it is a defect in the character structure of the parent, stemming from his/her own childhood experience, which, in the presence of added stresses, results in child maltreatment. Belsky’s (1984) process model of determinants of competent parental functioning presumes that parental functioning is multiply determined, but hypothesizes that the parent's psychological resources are the mpg; influential determinant because of their direct effect on parent functioning, as well as the role such resources play in recruiting contextual supports. Webster-Stratton (1985) compared two groups of parents whose children were being treated for "child oppositional behavior". Both groups were low to low-middle socio- economic status. Half of the 40 families were open 63 protective services cases, while the other half were not considered abusive. She found significant differences between the abusive and non-abusive parents in their child- hood histories. Forty-six percent of the abusive group reported being abused as children compared to only six per- cent of the non-abusive parents. The results of this study, however, must be cautiously interpreted since the two groups did differ relative to specific income (more abusive families under $5,000) and family structure (more abusive families were single). These confounding variables may explain the difference in abusive behavior. Personality deficits present in maltreating parents have been enumerated by a variety of researchers, most of whom relied on case studies and retrospective analyses in drawing conclusions. Spinetta (1972) notes that while environmental stress certainly correlates with child maltreatment, most deprived families do not abuse their children. He portrays the abusive parent as an adult with frustrated dependency needs, an inability to empathize with the child, and disregard for the infant's or child’s own needs. Abusive parents also display a lack of impulse control, a general character defect which allows aggressive impulses to be expressed to freely. Green (1979) reports similar personality characteristics for the eight abusive fathers he studied, noting that they had impaired impulse control and disturbance in identity formation. 64 Blumberg (1974) lists, six: common, denominators of psycho- pathology in the maltreating parent: early deprivation or rejection poor self-image and low-self-esteem narcissistic, immature, poor ego-control marital difficulties, poor choice of mates unwanted pregnancy or disappointment in child role reversal Empirical tests of these premises have been conducted by several researchers. Kaplan, et a1. ( 1983) compared a control group with clinical parents referred for abuse and neglect and tested both on a variety of personality and social measures. The researchers found that the abusive group demonstrated significantly more mental disorders, par- ticularly depression. They conclude that while this in no way minimizes the validity of environmental stress factors as a contributor to child maltreatment, psychopathology contri- butes independently to its occurrence. Justice and Diamond (1985) tested 23 abusing couples in Texas and a matched. comparison group on a series of stress and personality tests. They report that the only significant interaction found was between stress and violence as a socially scripted response, and conclude that whether abuse becomes an expressed response to stress depends on the presence or absence of some characteristic of the parent. In his book, Lamaggg mpg, Polansky (1981) summarizes fifteen years of study of child neglect, offering a portrait of neglectful parents. His studies compared neglectful families to a comparison group from the same ethnic and 65 socio-economic population. In addition to social isolation, Polansky found two ‘major "character disorders" which differentiated. the neglectful parents: apathy-futility or impulsivity and infantilism. He concludes that the neglect- ful jpersonality' stems from intense, ‘unresolved. dependency needs. For the neglectful parent, the infant is an "adult pacifier", a bulwark against loneliness whose job it is to meet the adult's needs, rather than the reverse. Helfer (1980) pulls these findings together in his description of the "WAR Cycle" (World of Abnormal Rearing). He postulates that early childhood maltreatment and depriva- tion result in developmental deficits which prevent the child from learning interpersonal skills necessary for competent and satisfactory adult roles. The parent’s unrealistic expectations for the child result in unmet dependency needs and a "role reversal" in which the child takes care of the parent. The child becomes responsible for the parent’s actions, rather than his/her own and there is no boundary developed between parent and child. The child does not learn that there are predictable consequences for his/her behavior, and gains no sense of power over the environment. Feelings and actions become enmeshed, resulting in impulsive behavioral responses to emotional states. As the child matures, s/he feels little control over her actions or her fate, and does not gain experience in rational decision- making and problem solving. 66 In this process the child is usually "bombarded with negative sensory messages", that eventually force the senses to "shut down". Helfer maintains that children reared in such environments have difficulty using their senses for receiving or transmitting positive messages. This deficit creates barriers to effective peer communications, and, eventually leads to problems in future parent-child interactions. The negative images projected onto the child by the parent create low self-esteem and a conviction that "I’m no damn good". As the child reaches adolescence, these deficits preclude successful separation and self-identity. Having "missed childhood", the young adult now seeks someone to take care of him/her, to make decisions and solve problems, and to fill the awesome pit of loneliness stemming from unmet dependency needs. Children from such backgrounds frequently find each other, and the cycle begins again with an early, unplanned pregnancy and a shaky, temporary, alliance between two very needy people. The relationship between maternal age and child maltreatment, assumed in this model, has been confirmed by several authors. In their comparison of abusive and non-abusive high-risk mothers, Egeland and Brunnquell (1979) found that the mean maternal age was significantly lower for the abusive group (19.3 years compared to 24.5 years). Daly and Wilson (1985) also reported that the risk of abuse was maximal for children born to young mothers, and that the risk declined with maternal age at childbirth. In her epidemiological study of 67 abused and neglected children in the United Kingdom, Creighton (1985) found that early parenthood was far more prevalent in the abusive population (35.3%) than in the general population controlled for socio-economic class (10.7%). There is support for the premise that a lack of parenting skills influences maltreatment rates. Garbarino (1977) argues that the majority of child abuse stems. from role malfunction; which is generated by the combination of a low level of caregiver skill, and situational stress. He reviews literature suggesting that three major factors are needed to facilitate effective role adaptation: rehearsal of the role, clarity of expectation, and minimal normative change. As reflected in Helfer’s model, abusive parents have limited opportunity to gain knowledge about and clear expectations for the parenting role. In addition, they are not likely to have the models needed to rehearse a positive, nurturing role during childhood. Since parenting responsibilities require a maximum normative change, role malfunction results. Evidence for how such a process develops and is maintained within the family system is described in the next section. MW As Parke and Collmer (1975) point out, a number of clinical investigators have noted the selectivity of abuse; usually a single child within a family is the target for abusive treat-. ment. Parke and Collmer identify two ways in which the child 68 may stimulate maltreatment: by some genetically determined characteristic which evokes a negative response in the care- giver, and, by behavior patterns which are shaped in response to the maltreatment. This section will focus on genetic characteristics identified in the literature which tend to put the child at special risk for maltreatment. In a retrospective study which explored the impact of low birth weight on child maltreatment, Klein and Stern (1971) found that 23.5% of 51 child abuse cases in a Montreal hospital were low-weight at birth. This proportion compares with 9-10% in an indigent population hospital. Egeland and Brunnquell (1979) also report that low birth weights were more frequent in abusive than non-abusive high risk families. Creighton (1985) found that low birth weight infants were over-represented in the abuse/ neglect distribution when [compared with Britain's general population controlled for socio-economic status. Low birth weights do not operate in isolation, however, as Kleine and Sterne point out. Other confounding factors, related to both low birth weights and child maltreatment, may partially explain these findings. For example, forced maternal-child separation, due to the vulnerability of the infant, could interfere with maternal- infant bonding, increasing risk of child maltreatment (Hunter and Kilstrom, 1978). In addition, other social characteris- tics of the mother may predispose her to deliver a low weight infant, such as inadequate prenatal care or drug use (Kleine and Stern, 1971). However, as Parke and Collmer (1975) note, 69 the greater demands low birth weight infants place on parents may in themselves significantly increase risk for child mal- treatment. Steele (1980) describes the vulnerability of any infant considered "abnormal", including those with significant pre- maturity, congenital deficiencies, perinatal illnesses, and "difficult" temperaments. Disabled children, particularly those requiring hospitalization, appear to be at special risk for child maltreatment (Hunter and Kilstrom, 1979: Gil, 1970). Glaser and Bentovim (1979) reported that handicapped or chronically ill infants are more likeLy to be maltreated in the form of omission of care. Jaudes and Diamond (1985) studied 86 maltreated children with handicaps at LaRabida Children’s Hospital in Chicago and found that one-third had been abused or neglected afpg; the physical disability became apparent. Other child factors which are associated with maltreatment include illegitimacy and gender. In Britain, Creighton (1985) found that more boys than girls were maltreated for every abuse type, except sexual. Comparative data in the United States are not available since no national source currently breaks reports down by abuse type and gender (American Humane Association, 1984). Across all types of abuse, girls and boys are nearly evenly represented (49.7% male: 50.3% female) in this country. Given the frequency of single parent families represented in the abusive/neglectful 7O population, illegitimacy is a predictable factor, and is undoubtedly related to the degree of support available to the primary caregiver. Social Interactions ip Lhe Maltreatipg Family Several authors have studied the family interactional patterns, broadly described by Helfer and Garbarino, which maintain a abusive/neglectful cycle. Parke and Collmer (1975) define this approach as the social-situational model of child abuse. An analysis of interaction patterns between child and adult partners provides clues concerning the condi- tions under which abusive patterns develop and are maintained. In his process model for determinants of parenting, Belsky (1984) hypothesizes ‘that. the ‘marital relationship is the "first order support system", with inherent potential for exerting the most positive or negative effect on parental functioning. Adult partners in maltreating families do appear to have less stable and mutually supportive relation- ships than the general population. This is evidenced by the higher number of separations experienced (Salzinger, 1986) and the greater frequency of stepparent and single parent families (Daly and Wilson, 1985: Creighton, 1985). In addition, Webster-Stratton (1985) reported that 46% of her maltreating sample had abusive partners, compared to 17% of the non-abusive group. Salzinger (1986) also reported more spousal discord in a group of mothers with maltreated 71 children than occurred in a matched control group. Creighton (1985) found that marital discord was the most frequent stress factor quoted in the 4,679 abusive families studied in Britain. Most. social-situational studies Ihave focused, however; on parent-child interaction. 'The negativity of parental care- giving, posited by Helfer, is substantiated in this research. Burgess and Conger (1978) studied low income families from rural Pennsylvania, comparing three groups: abusive (n=17), neglectful (n=17), and. a 'matched control (n=19). Two observers saw each family for six hours during a given week, completing a questionnaire and using a computerized, taped coding system. Results indicated that higher rates of positive interactions occurred in the control group compared to either problem group. Abusive adults showed less overall interaction, while the neglect group demonstrated "extreme negativity" in their interactions. Adults in both problem groups were less compliant and less positive with their children. Wasserman, Greene, and Allen (1983) report similar results in their comparison of 12 abusing and 12 matched control dyads in Washington. Mothers and their infants (mean age = 14 months) were videotaped during free play, and a standardized coding system was used to measure maternal style and behavior. Results revealed two clusters of differences between the groups: level of activity and affective 72 behavior. Abusive mothers were less active and used less verbal teaching: they initiated fewer activities with their toddlers and ignored them more than control mothers. The abusive mothers also demonstrated more negative behavior and less positive affect. They did not appear to have a repertoire of positive behaviors available for playing with their children. The infants who were abused more frequently ignored their mothers and refused to be distracted. The researchers conclude that the abusive mother and her infant are locked into a mutually reinforcing negative spiral. Oldershaw, Walters, and Hall (1986) compared 10 abusive mother-child dyads with 10 matched controls in Toronto, Canada. During forty minute sessions, social interactions were videotaped in a simulated living room. Each session followed a sequence of activities, including mealtime, free play, clean-up, and a specific task completion. Compared to the control group, the abusive group used more commands in controlling their children and attached less affect of any kind to their initial commands: when they did display affect, it was more likely to be negative. Abusive mothers were also more intrusive and less consistent in their use of parenting strategies. Control mothers issued fewer commands and, when making them, more frequently used a pleasant tone of voice. Children in the abusive group complied with their mother's commands less frequently than the control children. The authors note that positive strategies were used by the abusive group, but not frequently. They conclude that 73 abusive mothers don't necessarily lack appropriate strategies, but do not possess the "functional knowledge necessary to appropriately execute strategies" (p. 731). Although samples sizes are small in the above studies, the consistency of their findings is impressive. The results support Helfer's WAR cycle model (1980), as well as the role malfunction hypothesis described by Garbarino (1977). This review of microsystem factors in child maltreatment suggests that social stress and personal characteristics are independent, interacting influences in the etiology of child maltreatment. Effective solutions to the problem must there- fore include carefully designed intervention programs which will: . assist parents, and future parents, in resolving early childhood deficits which interfere with parent-child relationships: 0 provide ample opportunities for "role rehearsal", skill-building, and information which will encourage successful adaptations to the parenting role; 0 help families break the negative cycle of behaviors and interactions which maintain destructive parent- child relationships and marital discord. 2-1-6- W :Leh Egelegxefghimnaltrsatmsat An ecological review of the causes of child maltreatment illustrates the multiplicity of factors which combine to create role malfunction and pathological adaptation by care- giver and child. Variables at each level of the ecosystem contribute to this outcome (see Table 1.1). The theoretical schema (Figure 1.1) included in Chapter 1 attempts to capture 74 this complex etiology. To effectively address the problem of child maltreatment, action must be taken at each ecologi- cal level. The fact that child abuse and neglect are recognized as issues of public concern is an encouraging first step. This chronosystem change began only a century ago and has grown significantly in intensity during the past two decades. Now, we struggle with the complexity and difficulty of the task before us. At the macrolevel, cultural ideologies which promote maltreatment must change. The fact that our Western culture allows, and even encourages, the use of force against children makes the difference between "discipline" and abuse unclear. In addition, the inadequacy of social support and feedback to caregivers in our culture creates isolation and stress, conditions under which maltreatment flourishes. Moreover, these conditions are increasing at alarming rates, raising serious concerns for the future. The proportion of children living in single parent families headed by women, and in poverty conditions has escalated dramatically in the past decade (Child Welfare League of America, 1988). At the exo- and mesosystem levels, social support systems and effective child protection mechanisms are discouragingly slow in coming. Families with the most pressing needs for social support are often clustered in "high-risk" neighborhoods; areas that have few human or physical resources available to bolster a faltering family system. Current interventions aimed at identifying and assisting abusive or neglectful 75 families frequently fail to provide enough, if any, support services. Instead, legal solutions are imposed and identified families are disrupted at high cost to parents, children, and society. Our greatest progress appears to be at the microsystem level, where understanding has grown regarding the dynamics of the maltreatment cycle, as experienced within the family system. This understanding provides valuable direction for actions which can be taken both to intervene in families where child abuse and neglect are occurring, and, even more promising, for preventative actions which may break the cycle before it begins. However, this optimistic outlook is dampened by the fact that until equal progress is made at other ecological levels, such interventions remain a drop in the bucket. The evidence suggests that the social context surrounding the parent-child relationship is a powerful contributor to dysfunction. Comprehensive prevention initiatives which impact on this social context and strengthen family relation- ships promise new hope for ameliorating this devastating human problem. The future of our society, of our species, depends on our ability to understand the complexity and urgency of child maltreatment, and to take action, at eggpy level. 76 TABLE 2.1 Eagppps Coptzibptipg pp Child Malpreatment Ecological Level Factor Primary Investigators(s) CHRONO Lack of cultural empathy toward Steele, 1980: children Radbill, 1980; DeMause, 1975: Solnit, 1980 MACRO Unrealistic role eXpectations Korbin, 1980: Kempe, 1973: Ong, 1985 Permissive attitudes toward Gil, 1970: violence Garbarino, 1977 Non-supportive public policies Bronfenbrenner, practices 1981 EXO Deficit model of intervention Selig, 1976: Bronfenbrenner, - 1981 Fragmented, ineffective response Helfer,1987; to maltreating families AAPC, 1987: Pelton, 1983 Inadequate prevention resources Helfer, 1982: Daro, 1985: Cohn, 1983: Newberger, 1982 MESO Situational stress Gil, 1970: Gelles, 1973: Pelton, 1981: Straus, 1980: Garbarino, 1976 Inadequate social support Rohner, 1975: Egeland & Brunquell, 1979: Garbarino & Sherman, 1980 MICRO Family Ipaips: Creighton, 1985: Single, female-headed & Daly/Wilson, 1985 stepparent Polansky,l981: Large number of children Hunter & Kilstrom, 1979 Social isolation Polansky,1981: Salzinger,1986 Alcohol/drug abuse Kaplan, 1983: Altemeier,1979 77 mm: . Abusive/neglectful childhood Psychological/ interpersonal deficits Immaturity/lack of parenting skills Role malfunction/reversal Child Ipaits: Low-weight birth Congenital deficiencies/ disabilities Difficult temperment Gender lnteractiop Patterns: Discordant adult relationships Negative parent-child interactions 78 Helfer, 1980: Kaufman & Zigler, 1987: Spinetta, 1972: Webster - Stratton, 1985 Blumberg,1974: Kaplan, 1983: Justice, 1985: Polansky, 1981 Egeland & Brunquell,l979: Daly & Wilson, 1985: Creighton, 1985 Helfer, 1980: Garbarino, 1977 Klein & Stern, 1971: Egeland & Brunquell, 1979; Creighton, 1985 Steele, 1980: Hunter & Gilstrom, 1979: Gil, 1970: Jaudes & Diamond, 1985 Steele, 1980 Creighton, 1985 Belsky, 1984: Salzinger,1986: Webster & Stratton, 1985 Burgess 8 Conger, 1978: Wasserman, 1983: Oldershaw,l986 2.2. 133 MEQLAILHQ INELUEHCE QB §QCIAL SUPPORI As revealed in the preceding review of the child maltreatment literature, there is substantial evidence to suggest that social support has positive, mediating influences on the capacity of parents to carry out their childrearing function. The primary tenet of ecological theory would predict such a relationship since it presumes that human development involves mutual accommodation between the individual and the changing properties of the immediate setting, as this process is affected by relations between and beyond these settings (Bronfenbrenner, 1979). On the basis of ecological theory, it is therefore probable that social networks and the support that members provide both directly and indirectly influence the behavior, attitudes, expectations, and knowledge of parents and their offspring (Dunst, Trivette, and Cross, 1986). Belsky (1984) proposes that social support functions in three general ways to influence parenting attitudes and behaviors by providing emotional support, instrumental assistance, and social expectations. While stated differently, Dunst and Trivette (1988) hypothesize the same three general functions, stating that social support lifts the burden from one person, minimizing excessive demands and reducing the probability that well-being is negatively affected (instrumental). In addition, members of the parent's social network serve as models, so that parents can adopt or modify their parenting 79 styles if esteemed network members demonstrate effective and nurturing behavior (expectations). And, finally, such support offers opportunity for parents to share the "trials and ‘tribulations" of 'childrearing with other parents (emotional). Empirical findings regarding the relationship of social support to parental functioning will be further explored in the remainder of this literature review. 2.2.1. Qgfiipipg Social Spppppp Tardy (1985) identifies five issues which, are implicitly involved, but generally not explicitly addressed, in research aimed at understanding the development and functioning of social support. These primary elements include direction, disposition, description/evaluation, content, and network. Direction refers to the reciprocal nature of support, it can be either given or received, or both. Disposition distinguishes the availability of support from its utiliza- tion or "enactment". Description/evaluation recognizes that measures of social support can either describe what resources are available and used by subjects, and/or evaluate the effect of that support. The content of social support can be categorized into four dimensions: emotional, instrumental, informational, and appraisal. Finally, network refers to the sources of support and may include family, close friends, neighbors, co-workers, the community, and professionals. 80 In reviewing instruments used in operationalizing social support, Tardy notes that the lack of agreement concerning its conceptualization impedes the production of valid generalizations related to the development and functioning of social support. He urges that researchers explicitly address these dimensions as they operationalize the concept in their investigations. Carl Dunst and Carol Trivette (1984) define social support as a multidimensional construct that includes physical and instrumental assistance, attitude transmission, resource and information sharing, and emotional and psychological empathy. The content of social support is explicitly recognized in this definition. Dunst, Trivette, and Deal (1988) also address four of Tardy’s five elements through their operationalization of the concept: 1) Family Support Scale - measures the helpfulness of sources of support to families rearing a young child. [description/evaluation] 2) Inventory of Social Support - determines the types of help and assistance that are provided to a respondent by different individuals, groups, and agencies that make up a persons’ personal social network. [disposition, content, network] 3) Personal Network Matrix - Assesses a number of aspects of needs, resources, and support, including frequency/ type of contact with different network members: which members help with specific needs: and extent to which these members can be depended upon when asked for help. [disposition, description/evaluation, content, network] This broad definition recognizes the multiple dimensions of social support, allowing investigation of both the cumulative 81 influences on outcomes, and the differential effects accounted for by utilization, satisfaction with, type, and source of support. 2.2.2. Mediating Influences pp Parenpipg In 1960, Litwak reported evidence that support to parents is effective in buffering the stress associated with the birth of a child. Since that time, an abundance of evidence indicates the powerful, mediational influences of social support on both psychological and physical health, as well as parental functioning (Belsky, 1984: Boger, et al., 1986, 1988; Dunst, 1986: Dunst, Vance, Cooper, 1986: Crockenberg, 1981, 1985, 1988; Cochran & Brassard, 1979; Hetherington, Cox, Cox, 1976, 1978; Coletta, 1979: Pascoe, et a1. 1981; Powell, 1980: Abernathy, 1973: McCubbin et al., 1980: Mitchell & Trickett, 1980: Giovanoni & Billingsley, 1970: Crnic, et al. 1983; Boger, et al., 1986). Summarizing results of eight cross-sectional studies over ten years of work with families of developmentally impaired or at-risk children, Dunst and Trivette (1987) reported that, in every study, social support (as defined in the preceding section) accounted for a significant amount of variance in personal well-being: and, in every instance except one, social support was related to family functioning. Similar results are reported by Dunst and associates in numerous other publications which delineate particular studies and their findings (Dunst, Trivette, and Cross, 1986: Dunst, 82 Vance, and Cooper, 1986: Dunst, 1985; Dunst and Trivette, 1988: Dunst, 1982: Dunst and Trivette, 1984). Dunst, Trivette, and Cross (1986) concluded that satisfaction with and number of sources of supports had main and/or interactive effects in all sets of outcome measures, including personal well-being, parental attitudes toward the child, family integrity, parental perceptions of child functioning, parent- child play opportunities, child behavior and development. The same researchers conducted multiple regression analyses to discern unique effects of six different sources of support and reported that parental attitudes toward their child and perceptions of child functioning are most related to extra- family support (Dunst, Trivette, and Cross, 1984). Crnic, et a1. (1983) examined the relationship of stress and social support to maternal attitudes and early mother-infant interactive behavior and found that stress and support signi- ficantly predicted maternal attitudes at one month and interactive behavior at four months. Using a sample of 105 mother-infant pairs (52 premature and 53 full-term) , case matched for family ethnicity and education, the researchers conducted a two year longitudinal study with home interviews, clinical visits and observations. Social support was operationalized by a scale which involved a series of questions regarding satisfaction with available support from three sources: intimate relationships, friendships, and neighborhood. or loommunity support. 'Using regression analyses, they found that stress and support significantly 83 predicted maternal attitudes at one month and interactive behavior at four months. Mothers with greater stress were less positive in their attitudes and behavior, while mothers with greater support were significantly more positive. Social support moderated the adverse effects of stress on mother’s life satisfaction and on infant interactive behavior. The authors concluded that social support from various sources facilitates more positive child rearing attitudes, as well as more positive behavioral interactions. Crockenberg (1981) investigated the influences of social support, infant irritability, and maternal responsiveness on the development of secure and anxious infant—mother attach- ments at one year. Using a sample of 48 mother and infant pairs with similar education, marital status and socio- economic status, she assessed infant temperament 5-10 days after birth, mother-infant interactions at three months of age, social support at three months, and administered Ainsworth’s strange situation measure when the infants were twelve months of age. Social support was operationalized through interviews which focused on the affective and material assistance experienced by the mother, relative to the stresses she experienced, thereby reflecting the functional adequacy of support. Three network sources were rated: father, older children in the family, and others. Employing hierarchical multiple regression analysis, Crockenberg found that the adequacy of the mother’s social support was clearly and consistently associated with the 84 security of the infant-mother attachment, and was the best predictor of secure attachment. Moreover, social support had its strongest effect on the irritable babies. Crockenberg concludes: It is reasonable, then, to propose that availabili- ty of social support will facilitate responsive mothering, particularly under stressful conditions, and thereby encourage secure infant-mother attachment... the network may also affect the child directly, through the contact of the child with members of the network (p. 558). In an extensive review of the literature, Crockenberg (1988) summarizes results of studies which explore the influences of social support on high-risk and low-risk families. She concludes that social support is associated with patterns of parenting generally considered appropriate and beneficial to children. 2.2.3. Difgepential Influences 91 Social Support As ' noted previously, social support is a multidimensional construct with at least five elements (Tardy, 1985). Several investigators have identified and operationalized two or more dimensions of social support in their research, revealing differential influences of these elements (Vaux and Harrison, 1985: Dunst, Trivette, and Cross, 1985: Dunst and Trivette, 1988: Dunst, 1982: Dunst and Trivette, 1984: Crnic, et al.). Dunst (1985) summarized results from several studies in which multiple regression was used in order to discern the unique contributions of various elements of social support. Dunst and his associates studied differential influences of 85 satisfaction with support, network size, and source of support: intrafamily (spouse or mate and other members of the :nuclear’ family), extrafamily (friends, relatives, and other acquaintances) and formal (professionals and professional agencies). These studies revealed the following differential influences of social support: 0 qualitative rather than quantitative dimensions of support are the most important mediators of family functioning: satisfaction with provision of support consistently emerged as more important than number of sources of support 0 personal well-being is generally related to intra- and extra-family and formal support: help from any level of ecological influence can have positive effects on emotional and physical well-being 0 family functioning is almost always influenced by support within the household and from friends, neighbors, and kin, and rarely influenced by formal provision of support 0 parental attitudes toward their children are most related to extrafamily support whereas parental overcommitment and overprotection are most related to intrafamily support 0 different dimensions of parent-child interactions are differentially related to different forms of support Vaux and Harrison (1985) found that the most important aspect of network composition in determining satisfaction with support, was the proportion of close friends and the presence of a spouse or marital-like partner. Crnic, et a1. (1983) found that support from intimate partners had the most power- ful effects on mother’s life satisfaction but m on parenting attitudes or behavior. Only community support showed buffer effects on mothers’ interactive behaviors with 86 infants. Crockenberg (1988) reports that it is unclear what kind or source of support is more likely to affect the ways parents care for their children: differential effects appear to depend on the specific needs of individual families. While social support from family members, particularly intimate partners, is generalLy a powerful influence, there are instances, such as the birth of a special needs child, where friends or professionals are better sources of support than relatives. The above findings suggest that community support can play a significant role in determining parental attitudes, perceptions of children, and interactive behaviors. Such evidence also offers hope that carefully planned, community interventions can positively influence parent-child relation- ships, particularly for families who are experiencing stress- ful situations. 2.2.4. Blames! W $1229.13. The intuitive logic and scientific evidence for' the importance of social support to families with young children are compelling. One strategy for improving parental competence and preventing child maltreatment is to enhance natural support networks, facilitating and enabling families to connect with resources they need, whether emotional, instrumental, or normative. Barriers to social support to families with young children have escalated during this century. The disintegration of 87 the extended family, increasing divorce rates, and expanding numbers of one-parent families and of two-parent, two wage- earner families have seriously jeopardized traditional sources of instrumental and emotional support to parents (Bronfenbrenner, 1984; Brazelton, 1974: Bettelheim, 1987). As a result, the human ecology for rearing children, that "social fabric that nurtures and sustains our capacity...to raise our children to become competent and compassionate members of society", is threatened (Bronfenbrenner, 1984, p. 38). In the United States, programs and policies to address this important issue lag far behind those of most industrialized, western nations. As seen in the review of child maltreat- ment literature, this lack of supportive policies and programs) impacts greatest on those with the least personal and material resources and the most situational stress. Parents who were reared in dysfunctional families, where extended family members are not likely to communicate or model effective parenting, frequently lack the interpersonal skills required to build and maintain healthy support net- works with partners and friends. In some cases, friends and family members may be available but their interaction patterns tend to escalate rather than dissipate parental pressure. Such relationships can be intrusive and draining rather than supportive and energizing. 88 The ecological model for community prevention of child mal- treatment described in Chapter 1 (Figure 1.2) recognizes the healing potential of positive social support systems. Programs like the "Family Growth Center", have developed at the grass roots level across the United States as one part of compre hensive community-based delivery of social support services to families with young children (Meyers and Bernier, 1987: NCCAN, 1988). The goals of such interventions are to proactively supplement and enhance the family’s informal support network and to provide a bridge to more formal support services, as needed. This intervention model is based on family strengths and capabilities, not deficits, and requires no identification of problems or proof of inadequacy before help is made available. 2.2.5. Eyaluating the Efiicacy pf Planned Commupipy Suppopg Boger, et al. (1986) measured the effects of an early inter- vention program (Perinatal Positive Parenting) in which trained volunteers provided information and emotional support to primiparous mothers during the perinatal period. Using a sub-sample of 24 mothers, 12 who participated in the program and 12 control subjects, the researchers conducted a fifteen month follow-up to measure differences in the two groups. Significant differences were found on the HOME inventory (Caldwell and Bradley, 1979): mothers who received perinatal support scored significantly higher on maternal involvement and appropriate play material scales. In a later replication (Boger, et al., 1988) a sample of families with infants in a 89 Neonatal Intensive Care Unit were randomly assigned to control and treatment groups. The treatment group received one-to-one support immediately following the infant’s birth, during the period of the infant’s hospitalization, and through in-home visits after the baby was discharged. A comparison of the two groups indicated that, at twelve months post-discharge, mothers in the treatment group scored significantly higher on several of the NCATS and HOME scales than control mothers. Citing the work of Unger and Wandersman (1985), Crnic, et al. (1986) and her own cross-national study (Crockenberg, 1985), Susan Crockenberg (1988) concludes that formal support can be effective in promoting sensitive and responsive parenting. Further, these planned interventions do not have to be delivered by professionals; trained paraprofessionals have been shown to be as effective as professionals in working with certain high-risk goups (Unger and Wandersman, 1985) . The important element for effectiveness appears to be an extensive outreach component through which continuity of services to the family is assured (Crockenbeg, 1988). In a randomized experiment, David Olds (1987) and associates tested a program of prenatal and nurse home visitation as a method of preventing health and developmental problems in children born to primiparas who were either teenagers, unmarried, or of low socioeconomic status. Four hundred dyads were randomly assigned to one of four treatments: 90 Group 1 (control) - no services other than screening and referral to a specialist as needed Group 2 - free transportation for regular clinical appointments, as well as screening Group 3 - a nurse home visitor during pregnancy in addition to the screening and transportation services (visits every 2 weeks, approximately 1 1/4 hours) Group 4 - same as group 3 except the nurse continued to visit until the child was 2 years of age. Following delivery, the nurses visited families every week, with decreasing frequency as the child grew older. Under predetermined crisis conditions, they visited weekly. Interviews and infant assessments were carried out at registration (prior to the 30th week of pregnancy) and at 6, 10, 12, 22, and 24 months of the infant’s life. Among the women at highest risk for caregiving dysfunction, those who were nurse-visited had fewer verified cases of child abuse and neglect during the first 2 years of their children’s lives. During the second year of life, the babies of all nurse-visited women, regardless of the families’ risk status, were seen in the emergency room fewer times, and they presented less frequently for accidents and ingestions than comparison-group babies. In contrast to women assigned to the comparison condition, nurse-visited women reported that their babies had more positive moods and were observed to punish and restrict their children less frequently than were their counterparts. Olds, et a1. conclude that this investi- gation provides coherent evidence that "nurse home visitors are capable of preventing a number of caregiving dysfunctions, including child abuse and neglect". 91 Controlled studies such as those conducted by Boger, et al., and by Olds and his colleagues are rare. As a result, evidence for the positive benefit of planned community interventions on parenting and family functioning is limited. Dunst (1986) contends that "we have failed to conceptualize and conduct experimental evaluations in a manner that has permitted us to document the efficacy of our intervention efforts" (p. 80). He argues that the assump- tions implicit in most investigations are not necessarily tenable. Specifically, these studies frequently assume that: e an intervention can somehow compensate for the lack of early intervention services prior to entry into a program 0 most children benefit equally from early intervention 0 the degree of involvement and intensity of intervention is the same for most program participants 0 duration and amount of treatment provided to subjects in intervention studies are of sufficient magnitude to demonstrate or refute the efficacy of early intervention 0 the early intervention program is the principal or only intervention provided the program participants 0 treatment and comparison groups differ primarily in terms of who does and does not receive intervention Dunst holds that attempts to evaluate the efficacy of community intervention programs that make any of the above assumptions are not likely to demonstrate positive effects. He offers an alternative evaluation model which is based on the broad definition of intervention as provision of support to families of young children from members of informal and formal social support networks. 92 When defined in this manner, a fundamental emphasis of program evaluation efforts is discerning both the cumulative (main effects) and interactive effects of the different forms of support. To the extent that different forms of support account for a significant proportion of the variance in the dependent measures, the relative impact of early intervention is substantiated (p. 122). Early intervention is therefore not considered to be services provided by a particular program, but rather the aggregation of sources of support by different individuals and groups. The research question, then, is not "Does early intervention work?", but "How much variance does early intervention account for beyond that attributable to other formal and informal treatments?" Dunst suggests that the hierarchical model of multiple regression be used to test conceptual models which reflect the researcher’s hypotheses regarding the :relative influence: of ‘various factors in determining outcomes. The total variance accounted for by different sets of independent variables is determined in a sequential manner, dictated by the theoretical framework underlying the study. Using the model proposed by Dunst, as well as other methods, this investigation examined factors related to effective parenting beliefs and parental competence, particularly, the relative influence of a planned community intervention (the Family Growth Center) on these outcomes. 93 III. M__ET_HQQ§ As summarized in the previous chapters, one strategy for improving parental competence and preventing child maltreat- ment is to enhance social support networks and enable families to connect with needed resources. Such interventions are especially important in today’s society, in which the disintegration of the extended family, increasing divorce rates, and escalating numbers of one-parent families and two-parent, two wage-earner families have seriously jeopardized traditional sources of instrumental and emotional support to parents (Bronfenbrenner, 1984: Brazelton, 1974: Bettelheim, 1987). And, they are most critical for parents who lack necessary interpersonal skills for connecting with and maintaining needed resources; those with the least social and material resources who are experiencing the most situational stress in their caregiving role.- An ecological model for community prevention of child mal- treatment reflects the complex etiology of the problem and the healing potential of social support systems (Figure 1.2). The Family Growth Center, in Lansing, Michigan, has implemented such a community intervention since 1977, supplementing the participant family’s mesosystem, enhancing and expanding informal support networks and providing a bridge to formal support services. It is hypothesized that this increased level of social support and direct feedback influences parenting beliefs and competence, promoting effective child management and helping break the generational 94 cycle of abuse and/or neglect. Further, because children are provided services at the center, it is likely that the program also directly influences their social-emotional growth and future ability to parent. In the current investigation, a portion of this prevention model was tested: FIGURE 1.3 Empirical Model fog Ipvestigation Family Growth Center Social Support l Parentingfi Beliefs Parent/Family :1 Parenta Traits _ Competence In the model, three broadly defined factors directly influence parental competence (observed parent behavior and home environment): parent/family traits, FGC social support, and parenting beliefs. Moreover, parent/family traits and FGC social support indirectly influence parental competence via their contribution to parenting beliefs. It is also hypothesized that parent/family traits are related to FGC social support, although in this study no prediction regarding the direction of this relationship is made. 95 3-1- m A multi-method design was used to test the empirical model and its accompanying propositions and hypotheses. Initial exploration of the associations among independent and dependent variables (Propositions 1, 2, and 3) was accomplished through correlational methods. Next, the sample was divided into comparison groups, and a quasi-experimental design was used to examine the differences between high and low FGC utilization groups and between the nurturing and comparision groups (Propositions 4 and 5). Finally, the variance in dependent measures within the total sample was examined using multiple regression methods, in order to clarify the relative influence of different sets of independent variables on parenting outcomes (Propositions 6 and 7). 3-2- em The sample was selected from a population of 262 families currently using Family Growth Center Services in Lansing, Michigan. Any parent who wishes may attend the Center’s programs, which are offered at three sites. Some families are referred by agencies, while others are self-referred. Many participating families are headed by single parent females with low incomes, while others are middle-class, two- parent families. Several families are open, or recently closed, Children’s Protective Services cases. 96 3.2.1. Becppitment Two groups of parents and children were desired for this study, one consisting of families completing the "Nurturing Program", a specific parent education program aimed at influencing parenting beliefs associated with child maltreat- ment (Bavolek and Comstock, 1985) and the other comprised of families involved in other types of education and support services available through the Family Growth Center. To this end, all 16 families enrolled in the summer Nurturing Program at the Mt. Hope Family Growth Center were asked to participate in the study. One hundred percent of those asked agreed to take part. Of these 16 participants, one mother had to drop out of the program after one session due to a shift change at her place of employment. Her family was therefore not included in the study, leaving 15 families in the "Nurturing Group" sample. The other group was formed from parents who utilized drop-in child care, educational sessions, or other support services during the past year. A demographic profile of the nurturing group guided selection of this second group. Thirty-four families who were similar in race, income, family type, family size, age of youngest child, and CPS status received letters describing the study, requesting their participation, and inviting them to attend an orientation and initial testing day at the Family Growth Center in late June, 1988. The mailing included a postcard for their response; FGC staff followed up with phone calls to answer questions and encourage participation. Two 97 families (6%) could not be reached, and thirteen (38%) were not able to participate due to conflicts with the testing day and/or previous commitments which precluded their involvement over the summer months. The remaining 19 (56%) agreed to participate and attended the initial testing day. Prior to the end of the investigation in September, 1988, two of these families moved out of the area and a third parent decided not to continue in the study, leaving a total of 16 in the comparison group and a total sample size of 31 families (11.8% of the 1988 FGC population). Consent forms which described research activities, ensured confidentiality, and formally documented the families’ willingness to participate were read and signed by parents prior to data collection (see Appendix A). A $20 honorarium was paid to each family who completed all phases of the assessment process. 3.2.2. Samele Wis The sample achieved for this investigation was diverse, but consisted primarily of white, low-income families. Since only two fathers participated in the nurturing program, all testing for this investigation was limited to mothers and their children, thereby controlling for gender. A summary of sample characteristics are presented in Table 3.1. The majority (56.7%) of participant families made $800 or less per month and had an average of 2.2 children to support: half (53.3%) depended on government assistance as their sole 98 source of income. Most (54.9%) were headed by single mothers, a small proportion of whom (9.7%) had live-in partners. Mothers in the sample appeared to be well educated, with over half (58.1%) having. taken some college coursework. Only about 16% of the sample had not completed high school. Two of the thirty-one (6.5%) mothers were of a minority race (hispanic) . The mean age of participant mothers was 29.8 years at the time of the study, and their average age at the birth of their first child was 24.1 years. One child (the youngest) in each family was selected as the referent child for observation. The referent children averaged 36 months of age, were most frequently (41.9%) the second born in their families? and were nearly equally divided between males (48.4%) and females (51.6%). Every effort was made to balance the two groups on key demo- graphic variables. Despite this attempt, the two sub-samples appear to differ on several important variables, though the sample size was too small to reveal statistical differences. Mean scores demonstrate that mothers in the nurturing group had less education, tended to be younger, had earlier child- births, and larger families. There were also more married couple families in the comparison group than in the nurturing groups. Given these apparent differences, one would expect that the nurturing group mean scores on parenting belief and parental competence measures would be lower than those of the comparison group. 99 TABLE 3.1 Sample'Chapagtepispigs N= Nurturing Group C-Comparison Group Total N C n=31 n=15 n=16 E22111 lessee es: M2229 Under $500 26.7% 26.7% 26.7% 5501-800 30.0% 26.7% 33.3% $801-1100 13.3% 13.3% 13.3% Over $1100 30.0% 33.3% 26.7% Warmest: Employed 43.3% 42.9% 43.8% Public Assistance 56.6% 57.1% 56.3% Nuppe; g; Children mean= 2.2 2.53 1.87 range= 1-7 1-7 1-3 Marital—Mat Single never married 9.7% 13.3% 6.3% Separated-divorced 35.5% 26.7% 43.8% Single, live-in partner 9.7% 20.0% -- Married couple 45.2% 40.0% 50.0% Mother’s Education Less than High School 16.1% 26.7% 6.3% High School Graduate 25.8% 33.3% 18.8% Attended College 58.1% 40.0% 75.0% Mophgp’s Age ‘ mean= 29.8 28.9 30.6 range= 22-38 24-35 22-38 mmamhmenrx mean= 24.1 23.1 25.1 14 to 19 years 19.4% 26.7% 12.5% 20 to 24 years 38.8% 40.0% 31.3% 25 to 30 years 35.6% 26.8% 43.8% Over 30 years 6.4% 6.7% 6.3% 100 3-3-MWW3 MOP TOAW Two outcomes were of interest in this study: parenting beliefs associated with child maltreatment and parental competence. Copies of the following operationalized measures are included in Appendix B. 3.3.1. recreating Bertie The Adult-Adolescent Parenting Inventory (AAPI), developed by Stephen Bavolek (1984), was used as the operational measure of this construct. The AAPI is based on identified parenting and child-rearing practices of abusive parents. Statements representing a synthesis of the known parenting and child rearing practices of abusive parents served as the foundation from which the item pool of the inventory was constructed. Information gathered from the literature and field testing ultimately resulted in the identification of four parenting and child rearing constructs most commonly associated with abusive parents: 1) lpapprgpriate Parental Expectagtions g tpa M: abusing parents tend to inaccurately perceive the skills and abilities of their child: 2) Parental Le__ck a Emeathis _a__Aw reness o_f. mile—C ’8 N_§_ee 8: maltreating parents demonstrate an inability to be empathically aware of and appropriately responsive to their Child's needs: 101 3) Parental 22122 9:; PM 5 c meal}: abusive parents frequently believe that babies should not be "given in to" and must be shown "who is boss". They strongly defend their right to use physical force. 4) Parent-Chilg Role Reversal: the child is expected to be sensitive to and responsible for much of the happiness of his/her parents. A "paper and pencil" inventory, the AAPI measures the degree to which parents hold the above beliefs. Initial field testing of the AAPI with adolescents occurred in 1978. Since then, an additional 3,939 non-abused adolescents and 214 abused adolescents throughout the country have participated in developing normative data. In 1980, the AAPI was field tested with adults in the Chicago area. Since that initial research, 782 adults with known histories as child abusers, and 1,045 adults from the general population around the country have participated in the development of the normative data. Item-construct correlations of the 32 items constituting the AAPI range from .53 to .75. Inter-item correlations within each construct range from .17 to .55 with the majority at .25. These data support the construct validity of the inventory. Internal reliability is also high, ranging from .70 to .86 for each construct: and the test-retest reliabili- ty of all items is .76. 102 Diagnostic and discriminatory validity of the inventory are supported by several studies conducted by the author and other researchers (Bavolek, 1984). 3-3-2- Parental Cemeetense Parental competence, the primary social outcome criteria selected for the evaluation, was operationalized by scores on two field-tested instruments: the Nursing Child Assessment Teaching Scales (Barnard, 1978) and the Home Observation for Measuring the Environment (Caldwell and Bradley, 1984). Hursipg Chilg Assessment Teaghipg Scale (NCAlg) The Nursing Child Assessment Teaching Scale was originated by Kathryn Barnard as part of the Nursing Child Assessment Project research in Seattle, Washington (Barnard, 1978). The purpose of the observational scale is to measure the signifi- cant. dimensions of interaction Ibetween. parent and child. During a teaching interaction, a trained observer rates parent and child interactive behaviors. For parents, the important adaptive behaviors are aapsitivity to the child’s cues, ability to alleviate tha chilg’s d'st ess, and the ability to mediate the environment for the child in ways that m waiters and easel/seem WV me (Barnard. 1978, p. 14). For children, the primary adaptive behaviors are the ability to pagapga alga: apaa for the caregiver and the ability to taapgpg tg tpa gatagiyar. In this study, the total NCATS score and scores on the parent behavior subscales were utilized as outcome measures for parental competence. 103 In testing for internal consistency of items and categories, data revealed that items in all categories were positively correlated. In test-retest reliability comparisons, the mothers’ behavior showed more consistency from one measure- ment period to another than that of the infants. This finding suggests that actual developmental changes may pre- clude significant positive correlations in test-retest comparisons. Outcome measures used in evaluating the scale’s validity include the Mental Developmental Index (MDI) and the Psychomotor Developmental Index (PDI, Bayley, 1977), the Pre- school Behavioral questionnaire, and others. The teaching scale score at one and four months correlated at r=.71 and .76, respectively, with the expressive language measure taken at 36 months. The eight month and twelve month teaching scale scores were significantly correlated with the MDI and PDI, at r=.66 and .67, respectively. These results suggest that the Barnard Teaching Scale measures of parent-child interaction are associated with later cognitive performance. HOME lnventoty The HOME Inventory is an observation/interview instrument developed to assess quantitative and qualitative aspects of the rearing environment (Caldwell and Bradley, 1979) . Each of the 45 items in the measure is scored yes or no and all items are weighted equally. A total score for the HOME is determined by counting the number of items scored "yes". In addition to the total score, the inventory is divided into six subscales: emotional and verbal responsivity of mother: 104 avoidance of restriction and punishment: organization of physical and temporal environment: provision of appropriate play materials; maternal involvement with the child: and opportunities for daily stimulation. Caldwell and Bradley (1979) reported six month stability estimates for the six subscales ranging from .24 to .77. The stability estimate from the total score over a one year period was r=.77. Several studies have indicated that the HOME is a. good predictor of various measures of cognitive ability assessed during the early childhood period (Luster, 1985). These include a correlation of .57 between 24 month HOME and.54 month Stanford-Binet IQ and .66 between 24 month HOME and the total score of the Illinois Test of Psycho- linguistic Abilities (ITPA) assessed at 37 months. The HOME is a better predictor of cognitive outcomes than social class variables. In this study, the total HOME score and all six of its subscales were used for measuring parental competence. 3.4. INDEPENDENT VARIABLES: OPEBAIIONAL MEASQBES Two independent variables were operationalized and measured for the investigation: parent/family traits and Family Growth Center social support. Questionnaires used for collecting the following data are included in Appendix C. 3.4.1. PatentZFamily Traits Several key characteristics of parents and families were of interest relative to their influence on dependent measures: these were sub-divided into past and present factors, in 105 order to better reflect the dynamic nature of family systems, and the transactional reality of parent-child relationships. gates; risen Factors stemming from past experience included parental childhood abuse, maternal age at primiparity, and maternal education. Information about age of the mother at the birth of her first child and level of education was collected through self-report questionnaires. As summarized in the introduction, the literature suggests that parents who experience abuse as children are more likely to have maltreating parenting beliefs and lower levels of parental competence. In addition, it is postulated that childhood deficits resulting from maltreatment contribute to early pregnancies and truncated education. To gather evidence pertaining to these hypotheses, a measure of child- hood abuse experienced by parent participants was created. During home interviews with the thirty-one participants, each parent was asked about the discipline and parenting they experienced as a child. An overwhelming majority (90%) reported that they had been spanked as children, and two- fifths said they received spankings regularly. As a result, spanking was not included as a measure of maltreatment. Table 3.2 summarizes the frequencies for the remaining variables which were used to measure the degree of childhood abuse experienced by parents in the study. An ordinal 106 variable was created which represented the degree of child- hood abuse experienced by participants in the study based on a formula which summed recoded frequency scores for injuries, verbal abuse, and rejection. TABLE 3.2 Childhoog appaa Experienceg by Parent Participants n=31 Frequency of Injuries None 66.7% 1-2 times 20.0%- Several times 3.3% Regularly 10.0% Verbal Abuse None 40.0% 1-2 times 16.7% Several times 6.7% Regularly 36.7% Rejection None 73.3% 1-2 times 13.3% Several times 10.0% Regularly 3.3% turret ___xFamil iterator Variables pertaining to the current family situation included marital status, number of children, and family income. These demographic data were collected through a self-report questionnaire during interviews with parents in their homes. 3.4.2. Pamily thytp Qentet Social Suppott Two dimensions of Family Growth Center support were measured for this study: level of utilization and type of service. 107 Legal 21: Diagram Families utilizing Family Growth Center Services vary in their length of involvement, the number and type of services they have used, and their motivation for involvement. During the past few years, the Family Growth Center has offered several educational and support opportunities, in addition to the Nurturing Program, which may influence parenting beliefs and parental competence. As stated in the hypotheses, it was expected that this variation in utilization would be related to outcome variables. To operationalize level of utilization, data were collected from parents and staff relative to the family’s participation at the Center. A formula which combined length of time involved, number of services utilized, type of services used, referral status, and staff perceptions of the family’s involvement was used to generate an ordinal score ranking participants on their level of utilization. Specific definitions and coding for these components are outlined in Table 3.3. Scores from these five factors were summed for each participant to create the level of utilization score. The mean for all 31 participants was 4.48, with a range of 1 to 8, and a standard deviation of 1.87. 108 TABLE 3.3 Componants at Level gt FGC Utilization Scores Factor Definition ‘ Coding Length: Number of months involved 0= less than 12 1= 12 or more Services: Respite Child Care 0= no use/only recent Social Support 1= used within past Parent Education (other year than Nurturing Program) Sessions: Number of nurturing 0=6 or less sessions attended 1=7 to 13 Referral: CPS or court involvement 0=yes 1=no Staff Rank: Rating by staff members Mean rating from 0-2 of parent’s involvement (0=low, 2=high) These scores were also used to divide the sample into high and low utilization groups for comparison of differences in parenting beliefs and parental competence over a twelve week interval. Persons were placed in the low utilization group if they scored below the median (4.0) on the level of utilization scale, and in the high group if they scored at the median or above. Table 3.4 illustrates the parent and family characteristics of these two samples. The groups differed significantly on number of children and level of maternal education. Participants in the low utilization group, on average, had significantly less education and more children than those in the high involve- ment group. 109 TABLE 3.4 PQQ Utilization Sub-Sampla Characteristics =High Utilization L= Low Utilization Total L H n=31 n=17 n=l4 Eamilx Income per Month Under $500 26.7% 18.8% 35.7% $501-800 30.0% 28.6% 30.0% $301-1100. 13.3% 7.1% 13.3% Over $1100 30.0% 31.3% 28.6% Sgurce at Income Employed 43.3% 37.5% 50.0% Public Assistance 56.6% 62.5% 50.0% Nunpet at thldren mean= 2.59 1.71* range= 1-7 1-3 Marital Status Single never married 9.7% 5.9% 14.3% Separated-divorced 35.5% 41.2% 28.6% Single, live-in partner 9.7% 5.9% -- Married couple 45.2% 35.3% 57.1% Mothet’a Education Less than High School 16.1% 29.4% -- * High School Graduate 25.8% 11.8% 42.9%* Attended College 58.1% 58.8% 57.1% Mothet’s Aga mean= 29.8 29.5 30.1 range= 22-38 22-31 22-32 Maternal Age at Primiearitx mean= 24.1 23.8 24.5 14 to 19 years 19.3% 29.4% 7.1% 20 to 24 years 35.6% 29.4% 42.9% 25 to 30 years 38.8% 29.4% 49.9% Over 30 years 6.4% 11.8% -- * p<.05 110 The low utilization group also tended to be younger at the birth of their first child, although this difference did not reach statistical significance. Because these factors are known to be associated with parenting patterns, it became important to statistically control for confounding variables when low and high utilization groups were compared. Typa gt lnvglvenant Of particular interest in this study was the relative influence of a specific program, Bavolek and Comstock’s Nurturing Program, on family characteristics associated with child maltreatment. It was hypothesized that the nurturing group would have less abusive parenting beliefs and greater parental competence at the end of the three month program than a similar Comparison group which did not participate in the Nurturing Program. To test this proposition, two sub-samples of families were obtained. The nurturing group (N) participated in a thirteen week program based on Bavolek and Comstock’s model, and a comparison group (C) participated in other FGC services during the same period. There were no statistically signifi- cant differences between the two groups, although the nurturing group was, on average, less likely to be married, had less education, larger families, and earlier childbirths. 3.4.3. Qualitative Data Qualitative data were collected through 1-2 hour interviews with five families in the sample, three from the nurturing 111 group and two from the comparison group. These data provided the basis for more detailed case studies which focused on family history, current situation, and the perceived effect of the Family Growth Center on families studied. See Table 3.5 for a summary of measures utilized in the study. 112 TABLE 3.5 Simmefums e Variable Operational Measure 1. Dependent A. Parental Beliefs AAPI B. Parental Competence NCATS, HOME mmmm A. Parent/Family Traits 1. Parent History * a. Childhood abuse Home interview b. Maternal education Questionnaire c. Age at primiparity Home interview 2. Current Family Situation * a. Income category/source Questionnaire b. Marital status Home interview c. No. of children Questionnaire 3. Parent a. Race FGC records b. Birthdate Home interview c. Parent age Questionnaire 4. Referent Child a. Age NCATS b. Parity NCATS c. Gender NCATS B. PGC Social Suppott 1. FGC Type of Service * Group Assignment 2. FGC Level of Utilization * Derived from: a. Length of involvement at FGC Home Interview b. Parent education Home Interview c. Child care services Home Interview d. Support group Home Interview e. Nurturing sessions Attendance Records f. CPS/Court involvement Home Interview g. Staff evaluation Ranking sheet 3. Qualitative Qata Case studies Interviews *Independent variables 113 3-5- BBQQEDQBflfi Several methods of data collection were utilized in the research. Paper and pencil tests were completed at the Family Growth Center, while observational and interview questionnaires were conducted in the family homes. Table 3.6 summarizes the data collection procedure. TABLE 3.6 Data Collection When/What/Where/Who Date 1. During Nurturing Parent Program 1st, last session by FGC staff AAPI - self report Participant Information - self report 2. On "testing day" (comparison group) June & Sept. by FGC staff AAPI - self report Participant Information -self report 3. Home visits (60-90 min) June & Sept. by trained public health nurses and CAPS program specialist Biographical questionnaire - parent(s) NCATS - mother and youngest child HOME - observation of environment 4. Qualitative interviews October (5 parents; 5 staff) by Principal Investigator 5. Research journal Throughout by Principal Investigator 114 Two Family Growth Center staff administered the AAPI and participant information forms. The same persons administered the tests at both data points, for both groups. For the Nurturing Group, the testing occurred during regular sessions at the beginning of the program and at the end of the program, twelve weeks later. The comparison group partici- pants came to the Family Growth Center for a one hour testing in late June, and again in late September. The: observational instruments required. more intensive training of unbiased raters. Two public health nurses from the Ingham County Health Department, and the program specialist at Child Abuse Prevention Services attended two day-long training sessions in use of the NCATS and HOME scales. The training was conducted by Lee Ann Roman, Ph.D., and Mary DeWys, R.N., both of whom are certified by Seattle’s NCATS project. Following the training, the three raters conducted five pilot tests with families not involved in the investigation, and achieved a .91 interrater reliability coefficient for the NCATS and a .99 interrater reliability coefficient for the HOME. The raters were not aware of the group assignment of the families they observed. The investigator conducted qualitative interviews of families and staff at the end of the program. Five families were interviewed, three from the nurturing group, and two from the comparison group. The FGC director and four other staff members were also interviewed regarding the selected families. 115 3.6. DATA ANALYSES Three methods of analysis were utilized in addressing the research questions: Idescriptive/narrative, comparative statistics, and associative techniques. 3.6.1. QescriptivelNarrative Descriptive statistics were used to identify and present trends related to the sample, as well as central tendencies and variations on all measures. Qualitative data supple- mented quantified results, summarizing the interpersonal processes and observations that occurred throughout the project. This information was obtained from the research journal, family and staff interviews, and focused on: a. the social context in which the experiment occurred b. observed positive and negative attitudes/feelings c. case studies of family changes The qualitative data allowed more detailed and concrete analysis of the influence of social support, particularly a planned community intervention, in the lives of families. 3.6.2. Comparative Statistics Comparative statistics were used to obtain evidence regarding the extent to which utilization and type of involvement in the Family Growth Center influenced parenting beliefs and parental competence. Using a quasi-experimental design, a series of analyses of covariance (ANCOVA) were conducted in order to compare groups of participants relative to the dependent variables, while controlling for pretest scores. Pretest scores were utilized as covariates because it was 116 assumed that the confounding effects of parent/family traits would be reflected in pretest as well as posttest scores. By controlling for differences at pretest, these factors would also be controlled. Using level of utilization as the independent variable, the least involved participants were compared to the most involved participants on the dependent measures, while con- trolling for the independent effects of pretest scores. This procedure was also used to compare the nurturing group to the comparison group on outcome measures, with the same covariate. 3.6.3. Associative Techniques Associative techniques, including correlational and regression analyses, were utilized in order to determine relationships among the observed variables, as well as their relative influence in accounting for variation in parenting beliefs and parental competence. Several of the stated hypotheses were addressed using these techniques, including the key research question: what is the relative influence of a community prevention program on parenting beliefs and parental competence? As Dunst suggests (1986) , the extent to which an early intervention program accounts for a signi- ficant amount of variance in outcome measures will document the unique contributions of the program. The theoretical model presented in chapter one provided the framework for a series of hierarchical multiple regression analyses. The 117 total variance (R2) accounted for was determined in a sequential manner; with the sets of independent variables entered cumulatively in a specified hierarchy dictated by Figure 3 . 1. Two series of regression analyses were conducted. Since one of the prevention strategies of the Family Growth Center is to influence parenting beliefs and parental competence, the relative effect of planned community support on AAPI, HOME, and NCATS scales was of interest. In the first series of regression analyses, parent/family traits and FGC support were the independent factors with parenting beliefs (AAPI scales) and parental competence (NCATS, HOME) the outcome measures. The parent/family trait variables were divided into two sets, parent history and present situation, as were the social support factors, which included FGC level of utilization and type of service. The model posits that parent history will have the most influence on outcomes, followed by current family situation, FGC level, and, finally, type of service: Y = X1 + X2 + X3 + X4 + errOr where Y refers to parenting outcome, X1 to parent history, X2 to current family situation, X3 to FGC utilization, and X4.- to FGC type of service. Preliminary analyses were conducted to clarify the intercorrelations among the parent/ family traits, and to identify those factors which are the most powerful influences on parenting outcomes. The three most powerful parent history and current family situation 118 variables were then included in the final regression equation. The purpose of this first set of analyses was to determine the extent to which parent history, current family situation, and FGC utilization and service type explained parenting outcomes. Of particular interest was that portion of variance in outcomes which is explained by the FGC program, after the influence of parent/family traits is partialled out. In the second series of regression analyses, the extent to which parenting beliefs mediate observed relations between parent/family traits and parental competence was studied. The NCATS and HOME scores were used as outcome measures, and the total AAPI score was included in the equation as an independent variable before parent/family traits. This analysis revealed the proportion of variation in parental competence that was explained by parenting beliefs, and illuminated the extent to which parenting beliefs mediated parent/family traits in determining parental competence. If parenting beliefs in fact mediate parent and family traits in determining parental competence, then one would expect that these traits would explain little of the variance in the NCATS and HOME scores, beyond that accounted for by the AAPI. Educational and support services aimed at changing parenting beliefs could therefore be considered viable methods for influencing observed parental competence. 119 IV. RESULTS The primary goal of this research was to determine the relative influence of the Family Growth Center in explaining both parenting beliefs and parental competence. In addition, other questions of interest were derived from the empirical model for the investigation (Figure 1.3). The presentation of results is organized by these questions, as follows: 1) In what way are parent/family traits related to parenting beliefs and parental competence? 2) To what degree is utilization of FGC social support associated with parenting beliefs and with parental competence? 3) Areparent/family traits correlated with utilization of FGC social support? 4) When confounding variables are controlled, is there a difference in parenting outcomes between high/low FGC utilization groups and between the nurturing program group and a comparison group? 5) What is the relative influence of a community prevention program (the Family Growth Center) on parenting beliefs and parental competence? 6) To what extent do parenting beliefs mediate the relationship between parent/family traits and parental competence? Following the presentation of statistical analyses which address these questions, insights gained from case studies of individual families participating in the investigation conclude the chapter. 120 4.1. IflE ASSOCIATION §§TW§EN EARENTZEAMILY TRAIIS AND PARENTINQ OUTQOMES In this section, specific parent and family traits which have been associated with child maltreatment are outlined and their relations to parenting beliefs and parental competence are examined. Two categories of parent/family traits were operationalized: parent history and family situation. Parent history included the level of abuse experienced in childhood (Abuse), maternal education (Educ), and maternal age at the birth of the first child (Age). Family situation variables included marital status (a positive score on marital denotes mother is married), family income (Income), and. number of children (Number). Table 4.1 summarizes zero-order correlations found between each of these variables and parenting belief scales, including the total AAPI score, and its four domains: expectations (exp), empathy (emp), corporal punishment (corp), and role reversal (role). Positive scores on the AAPI denote appropriate expectations, high levels of empathy, low dependence on corporal punishment, and low levels of role reversal with the child. Relations were analyzed over two testing periods, at the time of the pretest, and twelve weeks later at post testing. A different version of the AAPI measure was used at pre and post intervals, to avoid bias due to testing. The observed associations over both testing periods reveals both the strength and the reliability of these relationships. 121 TABLE 4.1 gotteiatiogg: Pategtzzgmily Traits and AAE; Scores Note: Relations were examined using both post test (first row) and pretest (second row) scores to clarify the reliabi- lity of observed associations. n=31 Abuse Age Educ Marital Income Number AAPI -.429** .377* .558*** .413** .453** -.358* -.454** .387* .340* .433** .585*** -.391* Exp -.106 .302* .368* .241+ .302* -.245+ -.251+ .254+ .019 .284+ .544*** -.254+ Emp -.437** .512** .583*** .429** .443** -.558*** -.405** .299* .343* .418** .555*** -.255+ Corp -.392** .257+ .461** .292+ .406** -.134 -.146 .162 .031 .182 .373* -.296+ Role -.551*** .324* .577*** .454** .340* -.361* .624*** .511** .612*** .504** .480** -.450** + >.1O probability * >.05 probability ** >.01 probability ***>.001 probability All relationships are in the predicted direction. Childhood abuse is strongly' associated. with parenting' beliefs, the higher the level of abuse experienced by parents in child- hood, the lower the scores on non-abusive parenting beliefs. Of the six parent/family traits, childhood abuse shows the strongest relationships to role reversal and empathy. Age at primiparity and maternal education are also associated, in the positive direction, with AAPI scores: once more, these relationships are strongest in the empathy and role reversal 122 subscales. The data suggest, however, that parent history factors are only weakly and inconsistently associated with expectations and corporal punishment. Family situation variables are also highly correlated with AAPI scores, with family income demonstrating the most consistent and strongest relationships to all of the sub- scales. This is the only parent/family trait that is reliably and significantly associated with expectations and corporal punishment. The higher the family income, the more appropriate the expectations and the less reliance on corporal punishment. Marital status and number of children are moderately associated with empathy and role reversal; married women evidenced more empathy and less role reversal, as did women with fewer children. Turning' to jparental competence, Table 4.2 summarizes relations between behaviors observed at the two test periods using the NCATS and HOME scales and the above parent/family traits. In addition to the total score (NCATS), four of the NCATS six subscales were used in this study: sensitivity to cues (sens), response: to child. distress (distr), social- emotional growth fostering (social), and cognitive growth fostering (cogn). The HOME includes a total score (HOME), as well as six domains: emotional and verbal responsivity (respon), avoidance of restriction and punishment (punish), organization of physical and temporal environment (environ), provision of appropriate play materials (play), maternal 123 TABLE 4.2 Corteiations; Fgmiiy Bigk Eggtgts with NCATS, HOME Scotgs Note: Relations were examined using both post test (first row) and pretest (second row) scores to clarify the reliabi- lity of observed associations. n=31 Abuse Age Educ Marital Income Number NCATS -.159 .339* .682*** .276+ .270+ -.170 Sens -.279+ .293+ .609*** .251+ .134 -.226 -.345* .484** .337* .389* .196 -.264+ Distr -.289+ .320* .348* .344* .290+ -.350* -.054 .267+ .409** .142 .031 .000 Social .007 .266+ .472** .133 .081 -.032 -.055 .133 .226 .264+ .240 .031 Cogn -.368* .482** .723*** .277+ .130 -.459** -.313* .008 .311* .091 .001 -.343* HOME -.598*** .474** .661*** .394** .138 -.495** -.580*** .390* .609*** .421** .263+ -.448** Respon -.309* .366* .708*** .309* .097 -.239+ -.264+ .269+ .450** .274+ .418** -.230 Punish -.297+ .260+ .338* .161 .196 -.166 -.250+ .171 .334+ .082 .016 -.058 Environ -.309* .205 .462** .221 -.186 -.277+ -.381* .042 .139 .153 -.029 -.418** Play -.403** .392* .446** .329* .353* -.592*** -.282+ .271+ .463** .386* .346* -.292+ Involve -.686*** .360* .568*** .231 .008 -.638*** -.533** .282+ .585*** .333* .186 -.502** Variety -.619*** .462** .297* .457** .041 -.241+ -.510** .392* .275+ .347* -.007 -.244+ + >.10 probability * >.05 probability ** >.Ol probability ***>.001 probability 124 involvement with the child (involve), and opportunities for variety in daily stimulation (variety). Once more, relationships are in the predicted direction on both scales, though the HOME scores are more consistently and strongly associated with parent/family traits. Parent history factors are consistently associated with all of the HOME subscales, with maternal education and childhood abuse showing the most reliable relationships to home environ- ment. Childhood abuse is strongly related, in the negative direction, to the total HOME score and to maternal involve- ment and variety in daily stimulation subscales. Education is strongly correlated with maternal responsivity, as well as maternal involvement and total HOME scores, while age at primiparity shows consistent association to variety in stimu- lation. Maternal education is the one factor which is consistently related to the NCATS scales, showing moderate to strong relationships at both test periods to the total score and to sensitivity to cues, response to distress, and to cognitive growth fostering. Childhood abuse and number of children are negatively correlated with cognitive growth fostering at both intervals, while maternal age at primiparity is mildly associated with sensitivity to cues and response to distress. Married women and women with fewer children are more likely to score highest on the HOME. Marital status is most power- fully associated with appropriate play materials and variety 125 in stimulation, while number of children is associated most strongly, in the negative direction, with maternal involve- ment. Family income is not consistently associated with any of the NCATS scales, and is weakly associated with only one of the HOME subscales, appropriate play materials. These correlational analyses support Proposition 1 of this investigation, providing evidence that high risk parent and family traits are associated with abusive parenting beliefs and with low levels of parental competence. Hypothesis 1.1 was supported. Parental childhood abuse was negatively correlated with scores on the AAPI, NCATS, and HOME inventories. Hypothesis 1.2 was supported. Maternal age at primiparity was positively correlated with scores on the AAPI, NCATS, and HOME inventories. Hypothesis 1.3 was supported. Maternal education was positively correlated with scores on the AAPI, NCATS, and HOME inventories. Hypothesis 1.4 was partially supported. Family income was positively correlated with scores on the AAPI and the HOME inventories. Hypothesis 1.5 was supported. Number of children was negatively correlated with scores 'on the AAPI, NCATS, and HOME inventories. 126 4-2-IEEA—JIAIl—SSO ONWw—UII ONQEEQESQEEBIAEQ mmmmw Of particular interest in this study is the influence of a planned community intervention program (the Family Growth Center) on parenting beliefs and parental competence. It was hypothesized that mothers who were more highly involved in the FGC program would have higher scores on the parenting outcome scales. Pearson correlations were examined using both pre and post test scores, thereby providing evidence of the reliability of identified associations. Table 4.3 summarizes the results of these analyses. All relationships are in the predicted direction, although relatively few of the correlations consistently reached statistical significance. The strongest associations are between FGC level of utilization and the HOME scales, including total HOME score, as well as three of its six subscales: avoidance of restriction and punishment, appropriate play materials, and maternal involvement. The higher the FGC utilization, the higher the scores on these factors. The data also suggest that FGC utilization is positively associated with total AAPI and its corporal punishment subscale, as well as cognitive stimulation, as measured by the NCATS scale. 127 TABLE4.3 mmre tOS' mmiato 6003mm Note: Relations were examined using both pretest and post test scores to clarify the reliability of observed associations. Parenting FGC Level of Utilization Outcome ‘ (n=31) Pretest r Post test r AAPI .265+ .342* Expectations .142 .153 Empathy .201 .312* Corporal Punishment .365* .263+ Role reversal .182 .418** NCATS .523** .167 Sensitivity to cues .203 .067 Response to distress .063 .061 Social-emotional .038 .025 Cognitive stimulation .495** .249+ HOME .489** .437** Maternal Responsivity .132 .303* Avoid Restriction/Punishment .544*** .464** Organization of Environment .267+ .216 Appropriate Play .375* .449** Maternal Involvement .421** .409* Variety in stimulation .193 .085 + >.10 probability * >.05 probability ** >.01 probability- ***>.001 probability The above analyses support Proposition 2 of this investiga- tion, providing evidence that utilization of FGC social support is positively related to non-abusive parenting beliefs and parental competence. level of utilization, the higher the scores on the AAPI, NCATS, and HOME scales. 128 Generally, the higher the Hypothesis 2.1 was supported. Level of FGC utilization was positively correlated with scores on the Adult-Adolescent Parenting Inventory. Hypothesis 2.2 was partially supported. Level of FGC utilization was positively correlated with scores on the HOME inventory, and relations were in the predicted direction on the NCATS measures. 4.3. TEE ASSOCIATION BETWEEN PARENTZFAMIL! IBALTS gut QIILL: EAIIQH 9E EQQ §Q£EQBI It was hypothesized that high risk parent/family traits would be associated with low levels of FGC utilization. This hypothesis was tested by examining the correlations between FGC utilization and specified parent history and family situation variables. Results are presented in Table 4.4. TABLE 4.4 Correlations LflglgfiLQQiLlLLi—U° 'zton_i_bwt WM Factor (n=31) Level of Utilization Parental childhood abuse -.235 Maternal age at primiparity -.137 Maternal education .063 Marital status .071 Family income -.232 Number of children in family -.211 129 Several of the relations are in the expected direction, though none reached statistical significance. Parental childhood abuse came close to statistical significance (f=.11) and it appears that this factor is at least weakly related to FGC utilization. Two of the relations are in unexpected directions: family income and age at primiparity. The higher the family income, the lower the utilization, and the lower the age at primiparity, the higher the utilization. Again, these relationships, however, were not statistically significant. The data did not support proposition 3: high risk parent/ family traits were not significantly related to low utiliza- tion of social support. Hypothesis 3.1 was not supported. There were not significant negative correlations between parental childhood abuse nor number of children and FGC utilization, although the observed relations were in the predicted direction. Hypothesis 3.2 was not supported. There were not significant positive correlations between FGC utilization and maternal age at primiparity, or maternal education, or family income. 4.4. TEE DlEEEBEEQEfi IE EABEHIIH§,QHIQQME§ BEIEEEE HIQELLQE HILLIZAIIQR BED BETKEEE HQBIHBIH§.AHQ QOMPARISON GROUES The associations found between FGC utilization and parenting outcomes suggest that the community intervention program may influence parenting beliefs and parental competence. However, given the strong influence of parent/family traits 130 on these same outcomes, it is likely that factors such as childhood abuse and maternal education confound the observed relations between FGC utilization and scores on the AAPI, NCATS, and HOME scales. To discern the unique contribution of FGC support on parenting outcomes, it is necessary that the influences of these confounding variables be partialled out so that the unique contribution of the FGC factors can be determined. In this section, the results of two one-way analyses of covariance (ANCOVA) are reviewed. The ANCOVAs clarify the contributions of FGC utilization and type of service on parenting beliefs and parental outcomes. In the first analysis, two groups of parents were compared, those who fell below the median in level of utilization (low, n=17) and those who scored above the median (high, n=14). In the second analysis, parents participating in the thirteen-week nurturing program (n=15) were compared with a group of parents participating in other FGC services only (n=16). In both ANCOVA’s, pretest scores on all the measures, obtained twelve weeks prior to post tests, were used as covariates. It was assumed that the confounding effects of childhood abuse, :maternal education, age at primiparity, and other selection biases would be reflected in pretest as well as post test scores. Differences in parenting outcomes between high/low utilization groups and between nurturing and comparison groups, after three months of FGC service, would thus be clarified by these analyses. 131 4.4.1. Qiftetencgs getween lggn and Low Utilizatign groups Table 4.5 summarizes mean scores and standard deviations for high and low utilization groups over the two test periods. The high utilization group had higher mean scores than the low utilization group on all three of the measures, AAPI, NCATS, and HOME inventories. Surprisingly, however, raw mean scores on most of the measures show a slight decrease from pretest to post test for both groups, rather than the expected increase. Only the HOME score for the high involve- ment group increased slightly during the twelve week interval. Possible reasons for these findings are discussed in chapter five. Despite the unexpected post test mean scores, it is still relevant ' to determine whether there is a difference in parenting outcomes between the two groups of parents, after the influence of pretest scores is partialled out. Before moving to the results of the ANCOVA, it is important to note, however, that there is a significant heterogeneity of variance between the two groups on 12 of 17 post test measures, as well as four of the pretest scales. This lack of homogeneity diminishes the power of the ANCOVA to pick up differences between the two groups. 132 . TABLE 4.5 AHQQEA 11 MEL! §QQBE§.A§Q §IAEDABD.2E¥1ATLQH§ Low/High Level of FGC Utilization with Parenting Beliefs and Parental Competence Covariate (X) = Pretest Scores on Dependent Measures Level of FGC L- Low (n=17) H- High (n=14) Low: Post Low: Pre High: Post High:Pre Outcome L-Yl L-Sy L-Xl L-Sx H-YZ H-Sy H422 H-Sx AAPI 123.9 20.2* 132.0 19.6* 136.6 11.8* 137.7 11.0* Exp 22.2 4.4 24.8 4.0 23.9 3.3 25.4 3.6 Emp 33.8 5.5 34.0 5.9 36.9 3.7 35.3 3.9 Corp 36.8 6.9* 40.8 5.7 39.9 3.2* 43.1 3.5 Role 30.3 7.6* 32.4 6.5* 35.9 3.8* 33.9 3.7* NCATS 60.8 7.4 60.2 4.4 62.6 4.7 64.1 2.9 Sens 9.5 1.6 9.9 1.0 9.9 1.2 10.2 .9 Distr 10.1 2.0* 9.9 2.0 10.7 1.1* 10.4 1.3 Social 8.8 1.7* 9.4 1.1 9.3 1.0* 9.3 1.1 Cogn 12.5 2.9* 13.0 1.8 13.6 1.3* 14.5 1.2 HOME 35.8 7.5* 36.0 6.5* 41.0 2.0* 40.2 2.6* Respon 9.3 1.7* 9.5 1.2 10.0 1.0* 9.6 1.4 Punish 6.0 1.7* 6.1 1.5 7.1 .9* 7.1 1.1 Environ 4.9 1.1* 5.3 1.0* — 5.6 .6* 5.8 .6* Play 7.3 1.7* 7.4 1.5 8.5 .7* 8.2 1.1 Involve 4.6 2.0* 4.4 1.9 5.6 .7* 5.5 .8 Variety 3.6 1.3 3.1 1.3 4.1 .9 3.9 1.1 * Cochran Homogeneity of Variance Test: Significant (p <.05) differences in variances between two groups. Glass and Hopkins (1984) report that several researchers have studied the empirical consequences of violating the assump- tion of homogeneity of variance and conclude that both the t- test and ANOVA are robust with respect to violation of this assumption when n’s are equal. When n's are not equal, as in this study, the tetest and ANOVA are conservative with respect to committing Type I errors if the larger sample is 133 associated with the larger variance. When the larger n and larger variance are paired, the true probability of a Type I error is always less than the nominal probability. This is true in all cases summarized in Table 4.5; the larger variances occur in group 1, the low utilization group, which is also the larger sample. Therefore, when critical F-values are set at .05, the probability of rejecting a true null hypothesis is actually less than .05. Glass and Hopkins state that there is no need to be concerned about violating the homogeneity of variance assumption under these condi- tions, although the power of the test is obviously diminished. To increase the power in this test, reducing risk of a Type II error, an alpha of .10 was used as the critical value for rejecting the null hypothesis (Ho: there is no difference between the low and high utilization groups). Results of the Analysis of Covariance for high and low FGC utilization are presented in Table 4.6. - The analysis indicates that level of FGC utilization has a significant influence on parenting beliefs when pretest scores are controlled. Two of the AAPI subscales were significantly affected by FGC level: empathy and role reversal. The only competence measure significantly affected by utilization was provision of appropriate play materials. Although the high utilization group had higher mean scores on all the measures, these differences did not reach statistical significance on any of the other parental competence scales when pretest scores were controlled. 134 TABLE 4.6 SUMMARY OF ANALYSIS OF COVARIANCE 1 Low/High Level of FGC Utilization with Parenting Beliefs and Parental Competence Covariate = Pretest n=31 Mean Squares F-Statistic Outcome Pretest Level Within Pretest Level AAPI 4513.41 524.44 136.56 33.05*** 3.84+ Exp 130.97 16.13 11.52 11.37** 1.40 Emp 414.00 34.95 8.55 48.42*** 4.09* Corp 132.30 29.95 27.52 4.81* 1.09 Role 631.86 137.67 17.03 37.09*** 8.08** NCATS 364.25 19.64 28.84 12.63*** .68 Sens 11.53 .37 1.70 6.78** .22 Distr 8.14 2.10 2.56 3.l8+ .82 Social 8.76 2.61 1.76 4.99* 1.49 Cogn 45.26 .11 3.79 11.93** .03 HOME 710.84 6.84 8.88 80.08*** .77 Respon 10.99 3.02 1.67 6.56** 1.81 Punish 29.11 1.28 1.02 28.49*** 1.25 Environ 7.45 1.38 .67 11.16** 2.07 Play 37.05 1.72 .50 74.19*** 3.45+ Involve 50.65 .11 .81 62.53*** .13 Variety 25.16 .27 .42 60.50*** .65 + p<.10 * p<.05 ** p<.01 *** p<.001 The results of the first analysis of covariance provides evidence supporting' Proposition ‘4. When confounding variables. are controlled, positive parenting outcomes increase as a function of the level of FGC utilization. Over a twelve week interval, mothers who were highly involved in the Family Growth Center were significantly different than 135 low utilization mothers in level of empathy, role reversal, and provision of appropriate play materials in the home. Hypothesis 4.1. was supported. When pretest scores were controlled, parents with a high level of FGC utilization had significantly higher scores on the AAPI than parents in the low utilization group. Hypothesis 4.2. was partially supported. Parents in the high utilization group scored higher on a subscale of the HOME inventory than parents in the low utilization group. 4-4-2. Differencesfiwtwe memo 5 The second analysis of covariance aimed to clarify the differential. influences of services available at the Family Growth Center. Of particular interest was the effect of the "nurturing program" (Bavolek and Comstock, 1985), a structured program designed to change dysfunctional and abusive parenting beliefs, including the four AAPI con- structs which are associated with child maltreatment: expectations, empathy, corporal punishment, and role reversal. Each week for thirteen weeks (1) , parents and their children participated in structured group activities which addressed topics such as behavior management, personal power, limitations of hitting, child development, stress management, communication, and personal safety. Children ‘were assigned to age-appropriate developmental groups, which also focused on the themes being addressed by their parents. 136 For example, while adults were learning about behavior management techniques such as "time out", children were learning about consequences for behaviors and what time out means . The nurturing group was compared to a group of parents who were involved in other educational and support activities available at the Center, but had not yet participated in the nurturing group. other services at the center during the twelve week period included: 0 'Drop-in child care 0 Summer Fun Play Group - a special 6 week workshop of activities for parents and children to enjoy together 0 Annual FGC picnic 0 Clothing exchange 0 Parent’s.place - an informal time for parents and their infants to meet together, hear speakers, and share refreshments; child care was provided for older children. 0 STEP program - (Systematic Training for Effective Parent- ing) a six week course for parents in discipline and communication co-sponsored by the Region 13 Substance Abuse Prevention Education program 0 HOPE (Helping Ourselves Parent Effectively) - a support group for parents in crisis as well as for parents who are dealing with daily stresses of raising children 0 Adult Children of Alcoholic Parents - a 3 week workshop co-sponsored by Region 13 Substance Abuse Prevention Education program. All services were available to anyone who wished to partici- pate. l) The Bavolek and Comstock program is actually 15 weeks in duration but due to the Labor Day holiday and school start- up, the summer program was condensed into 13 weeks. 137 As in the first analysis, pretest scores obtained at the beginning of the investigation were used as the covariate in the analysis to identify differences between the nurturing and comparison groups on the outcome measures. Table 4.7 summarizes mean scores and standard deviations for the two groups at pre and post test intervals. TABLE 4.7 539923 21 Mfléflfi AND filAEDABD DEYIAIlgflfi Comparison/Nurturing Groups with Parenting Beliefs and Parental Competence Groups: C=Comparison (n=16) N=Nurturing (n=15) Comp Post Comp Pre Nurt Post Nurt Pre Outcome c481 C-Sy c4i1 C-Sx N472 N-Sy N-X2 N-Sx AAPI 133.4 12.3* 138.0 2.6 125.6 22.1* 130.9 9.3 Exp 23.7 3.6 25.2 3.8 22.2 4.4 24.9 3.8 Emp 36.8 3.4* 36.0 3.8 33.5 5.8* 33.1 5.8 Corp 38.6 4.5 42.1 4.8 37.7 6.9 41.6 5.1 Role 34.2 4.6* 34.7 4.4 31.5 8.3* 31.3 6.0 NCATS 61.8 6.2 61.5 4.6 61.5 6.6 62.5 3.9 Sens 10.0 1.1 10.1 1.1 9.4 1.6 10.1 .9 Distr 10.5 1.5 9.9 2.0 10.2 1.8 10.4 1.3 Social 8.9 1.5 9.2 1.3* 9.1 1.3 9.5 .7* Cogn 13.2 2.2 13.6 1.7 12.8 2.4 13.8 1.8 HOME 39.9 2.9* 38.9 3.0* 36.3 8.1* 36.8 7.2* Respon 9.8 .8* 9.4 1.4 9.5 1.9* 9.7 1.3 Punish 6.7 1.2 6.6 1.3 6.3 1.8 6.7 1.6 Environ 5.6 .8 5.7 .8 4.9 1.1 5.3 .9 Play 8.0 1.1* 7.9 1.1 7.7 1.8* 7.7 1.6 Involve 5.6 .6* 5.4 .8* 4.5 2.2* 4.3 2.1* Variety 4.2 .7* 3.8 1.0 3.5 1.4* 3.1 1.4 * Cochran Homogeneity of Variance Test: differences in variances between two groups. 138 Significant (p <.05) Participants in the nurturing group generally scored lower than the comparison group on both pre and post test scales. This was expected due to differences in parent/family traits between the two samples (see chapter 3). Again, scores on the post tests are, for the most part, slightly lower than scores on the pretests. Explanations for these findings are discussed in chapter five. Once more, the two groups have significant differences in variance on several of the outcome measures and pretest scores. Inn this instance, however, the smaller sample (nurturing group) had the larger variance. When the larger sample has the smaller variance, the true alpha is greater than the nominal probability of a Type-I error (Glass and Hopkins, 1984). Because the difference in sample sizes is so slight (nurturing group n= 15, comparison group n=16) , the effect of heterogeneity of variance on alpha is minimal in this case, with actual probability still falling below .10 when the nominal alpha is .05. However, to avoid a Type I error, the critical value for testing the null hypothesis is set at p<.01 (Ho: There is no difference between the nurturing and comparison groups). Results of the second analysis of covariance are summarized in Table 4.8. 139 TABLE 4.8 SQMMABY QE AEAL1§1§ 9E QQYABIAEQE 2 Type of FGC Involvement with Parenting Beliefs and Parental Competence Nurturing (n=15) and Comparison (n=16) Groups Covariate = Pretest Mean Squares F-Statistic Outcome Pretest Type Within Pretest Group AAPI 4799.97 36.44 153.98 31.17*** .24 Exp 135.23 13.95 11.59 11.66** 1.20 Emp 378.69 8.87 9.48 39.94*** .94 Punish 173.24 2.79 28.49 6.08* .10 Role 676.89 1.35 21.90 30.90*** .06 NCATS 380.44 11.34 29.13 13.06*** .39 Sens 12.41 2.81 1.61 7.68** 1.74 Distr 10.40 1.76 2.57 4.05* .69 Social 8.12 .01 1.85 4.39* .01 Cogn 55.76 2.67 3.70 15.06*** .72 HOME 820.46 17.17 8.51 96.44*** 2.02 Respon 11.85 1.60 1.73 6.86** .93 Punish 38.36 1.47 1.01 37.80*** 1.45 Environ 7.79 1.00 .68 11.43** 1.47 Play 45.79 .16 .56 82.49*** .29 Involve 49.05 .36 .80 61.23*** .45 Variety 22.65 .39 .41 55.04*** .96 * p<.05 ** p<.01 *** p<.001 Proposition 5 was not supported by the data. When pretest scores were controlled, parenting beliefs and parental competence did not increase as a function of Family Growth Center type of service. Hypothesis 5.1 was not supported. When pretest scores were controlled, parents who participated in the nurturing program 140 did not score higher on the AAPI than parents who received other services only. Hypothesis 5.2 was not supported. Parents participating in the nurturing program did not score higher on the NCATS and HOME inventories than parent who received other services only. 4.5%» Bfilfifluflfli Ifllléflflflflfi Diliflflfl IZDQLEX Sflflflflflfl QEQEEEB SEE EABDEIIDD DELLEED AND BABDEIAL QDMEDIDEDE A hierarchical multiple regression analysis was performed to determine the relative influence of the Family Growth Center program on parenting beliefs and parental competence. Based on the conceptual model presented in chapter one, it was hypothesized that parent history and family situation are fundamental determinants of both parenting beliefs and parental competence. Beyond these primary explanations, it was hypothesized that a community support program contributes significantly to parenting outcomes. Parent history factors were entered first, followed by immediate family situation variables, then FGC level of utilization, and finally FGC type of service. At each step, the preceding variables were statistically controlled, so that the independent contribu- tion of the variable in question was known. This section of chapter four describes the results of these analyses. 141 4.5.1. intgtgopteiatipns Anpng the independent Variapies To determine the relative influence of each independent variable on parenting outcomes, intercorrelations among the selected factors were examined first. Table 4.9 summarizes the relations among parental childhood abuse, maternal education, maternal age at primiparity, marital status, number of children, family income, level of FGC utilization, and FGC type of service. TABLE 4.9 Intercorrelations: Independent Variabies Abuse Educ Age Marital Number Income Level Abu -- Edu -.392* -- Age -.341* .602*** -- Mar -.486*** .384* .579*** -- Num .499** -.351* -.494** -.271+ -- Inc -.200 .274+ .516** .638*** -.087 -- Lev -.235 .063 -.137 .071 -.211 -.232 -- Type .340* -.215 -.433** -.105 .204 .104 .071 + <.10 * <.05 ** <.01 *** <.001 In this sample, there are high intercorrelations among the parent/family traits. All associations are in the expected direction. The more abuse suffered by the mother in child- hood, the younger she is at the birth of her first child, the higher the number of children she bears, and the less likely she is to be married. In addition, the younger the mother is at the birth of her first child, the lower her education; 142 and, the lower her education, the higher the number of children and the less likely she is to be married. Participation in the nurturing program is positively associated with parental abuse and negatively correlated with maternal education. Mothers who joined the nurturing program were more likely to have suffered childhood abuse and were less educated than their counterparts in the comparison group. This finding is not surprising since several of the mothers who attended the nurturing program were referred by public and private agencies, including the Probate Court. 4.5.2. Selection pf Variables pp; Regression Analysis In regression analysis, extreme multicollinearity, where independent variables are highly associated with each other, creates serious estimation problems because it produces large variances for the slope estimates and, consequently, large standard errors (Lewis-Beck, 1986). The best unbiased estimates for the regression equation are obtained with variables that are highly correlated to dependent measures, but only moderately or weakly associated with each other. In addition, the more variables included in the regression equation, the lower the power available to identify signifi- cant factors. It is therefore important to carefully select indicator variables and eliminate those factors which do not significantly contribute to the outcome, when other variables are controlled. 143 Two preliminary regression analyses were conducted with the goal of clarifying the relations among the independent factors, as well as their independent contributions to AAPI, NCATS, and HOME scores. The first set of analyses involved a series of hierarchical regression equations in which each of the parent and family variables were regressed on each other. This procedure revealed the degree to which each of the independent factors explained the others and clarified the interrelationships among the parent/family traits. First, education was predicted using parental childhood abuse and age at primi- parity. Next, marital status was explained using parental childhood abuse, maternal education, and age at primiparity as independent variables. Then, number of children was the dependent variable, with parental childhood abuse, maternal education, age at primiparity, and marital status as pre- dictors. Finally, income was predicted using maternal educa- tion, age at primiparity, marital status, and number of children as independent factors. See Table 4.10. It should be noted that adjusted R-square statistics instead of multiple R-square values are used in this and all remaining regression analyses in this study. The multiple R squared is the proportion of variance in the dependent variable associated with variance in the independent variables; the adjusted R2 is corrected for the number of cases. A small number of cases relative to the number of 144 variables in the multiple regression, can bias upwards the estimate of R2 (Nie, et al., 1975). Given the small sample size in this study, the more conservative estimate of explained variance was used. TABLE 4.10 w ___1___8tatist'cs f2: Examining Weiss: Enables n=31 Criterion/ Step 1 Step 2 Step 3 Step 4 Predictor Beta Beta Beta Beta Maternal Education: Childhood Abuse -.317+ -.139 Age at Primiparity .593*** Adj. R2 .068+ .378*** Marital Status: Childhood Abuse -.456** -.360* -.305+ Maternal Education .301+ -.002 Age at Primiparity .505** Adj. R2 .179** .237** .374** Number of Children: Childhood Abuse .468** .384* .341* .400* Maternal Education -.265 -.028 -.027 Age at Primiparity -.395+ -.493* Marital Status .194 Adj. R2 .191** .229** .302** .293** Family Income: Maternal Education .334+ -.062 -.084 -.061 Age at Primiparity .594* .314 .407+ Marital Status .502** .488* Number of Children .160 Adj. R2 .080+ .256** .412*** .406** + <.10 * <.05 ** <.01 *** <.001 145 The data suggest that age at primiparity strongly predicts maternal education, mediating the influence contributed by parental childhood abuse. Together, these two factors explain nearly 40% of the variance in maternal education. See Figure 4.1. Figure 4.1 Exnlaining Maternal Educatign - . 139 Abuse 3 Educ -.29;:\\\3 ////////6593*** Age R2= .378*** Age at primiparity also appears to be a powerful predictor of marital status, mediating its relationship with both maternal education, and to a lesser degree, childhood abuse. In fact, when age at primiparity is added to the regression equation, the influence of maternal education on marital status disap- pears. In this sample, whether or not a woman is married depends primarily on the age at which she had her first child, though childhood abuse continues to influence this variable after age at primiparity is added to the equation. See Figure 4.2. 146 Figure 4 . 2 Explaining Marital Statns -.298+ .593*** Abuse fi‘ Age 7 Educ -.305* .505** -.002 Marital R2=.374** Similarly, the number of children in the family is also predicted primarily by parental childhood abuse and age at primiparity. When these two factors are controlled, neither maternal education nor marital status are significant predictors of family size. Figure 4.3 adds these findings to the results already discussed. Figure 4.3 Manning N___rumbe 2f §___nhildre -.298+ .593*** Abuse 7* Age > Educ -.305* -.493 .400* .505** Number 6 Marital .194 R2=.298** 147 Finally, family income is explained. primarily' by ‘marital status and age at primiparity. The influence of maternal education is once more mediated when these two factors are added to the equation. The number of children does not add significantly to the predictive value of the equation. Figure 4.4 combines all the findings of the first set of hierarchical regression analyses, illustrating the observed relations among parent history and immediate family situation variables. Figure 4 . 4 The Interrelations Among Independent Variables -.298+ .593*** Abuse ; Age ¢Educ .305* -.493 .407+ R2=.378*** .400* .505** .L Number Marital $ Income R2=.298** R2=.374** R2=.406*** The data suggest that parental childhood abuse directly influences age at primiparity, number of children, and marital status, and indirectly contributes to level of maternal education and income. Age at primiparity directly influences education, number of children, marital status, and family income. Marital status directly influences family income. The amount of common association among all of these 148 variables make multicollinearity probable if all are included in the regression analyses. To determine which of the six independent factors are best eliminated in order to produce the best fit equation, a second set of preliminary analyses used all six in predicting total scores for the AAPI, NCATS, and HOME measures. Parent history factors were entered first (parental childhood abuse, age at primiparity, and maternal education), followed by immediate family situation (family income, number of children, marital status). Table 4.11 summarizes results of this analysis. Looking first at parenting beliefs, parental childhood abuse is a significant explaining factor until education and family income are added to the equation, at which time its influence diminishes. It appears that education and income mediate the influence of childhood abuse on parenting beliefs. Together, the first three variables explain 38% of the variance in parenting beliefs. Neither number of children or marital status improve the equation, after these factors are controlled. Maternal education is the only one of the parent/ family traits that significantly predicts parenting behaviors measured by the NCATS observation. Neither parental child- hood abuse nor age at primiparity influence these scales, and family income, family size, and marital status do not contri- bute when education is controlled. 149 TABLE 4.11 Wilma essoitatistieezergerentiuggytgemee n=31 Criterion/ Step 1 Step 2 Step 3 Step 4 Step 5 Step 6 Predictor Beta Beta Beta Beta Beta Beta AAPI Abuse -.429* -.340+ -.246 -.230 -.143 -.147 Age .261 .025 -.175 -.290 -.287 Educ .446* .472* .482* .481* Income .368* .422* .426* Number -.224 -.223 Marital -.009 Adj. R2: .154* .186* .287** .381** .381** .353** NCATS Abuse -0159 -0049 0116 .124 .143 .168 Age .322 -.092 -.187 -.213 -.234 Educ .783*** .796*** .798*** .799*** Income .174 .186 .155 Number -.050 -.056 Marital .072 Adj. R2: -.011 .049 .422*** .424*** .377** .377** HOME Abuse -.598*** -.494** -.393** -.399** -.345* -.312+ Age .305+ .053 .127 .057 .029 Educ .474*** .465** .471** .473** Income -.135 -.102 -.144 Number -.l38 -.l45 Marital .096 Adj. R2: .333*** .396*** .523*** .519*** .495*** .494*** + <.10 * <.05 ** <.01 *** <.001 Parental childhood abuse is a powerful factor in explaining the HOME scores, added to the equation. even when all the successive variables are Age at primiparity adds to the explaining power of childhood abuse, until education is con- 150 trolled, then, its influence. disappears. Income, family size, and marital status do not contribute significantly to the family environment, as measured by the HOME scale, after childhood abuse and maternal education are controlled. Together, these two variables explain 52% of the variance in HOME scores. The variables which appear to be most significantly influen- tial in predicting parenting outcomes, as measured by the three scales, include parental childhood abuse, maternal education, and family income. When these three variables are controlled, age. at jprimiparity, family size, and. marital status do not contribute to the explanatory power of the regression model. The latter three variables were therefore eliminated from the final regression analysis, which examined the relative influence of the Family Growth Center program in explaining parenting outcomes. 4.5.3. B_