PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DATE DUE 5/08 K:IPrq’/Acc&Pres/CIRCIDatoDue.indd EAR EARLY SOCIOEMOTIONAL DEVELOPMENT IN THE CONTEXT OF FAMILY BOUNDARY AMBIGUITY By Erika London Bocknek A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSPHY Family and Child Ecology 2009 13A dcxcl impor circur ps5ch bdflS? 0H gc: emtllli b6t\'\'c “lib F SUDpo “fist: DredjC ABSTRACT EARLY SOCIOEMOTIONAL DEVELOPMENT IN THE CONTEXT OF FAMILY BOUNDARY AMBIGUITY By Erika London Bocknek Early socioemotional development, in general, and self—regulation, in particular, develops in the context of early parenting relationships. This suggests there are important implications for young children’s development when the boundaries circumscribing the parent-child relationship become ambiguous through parental psychological or physical absence. Boundary ambiguity, a theory which describes the basis for individual dysfunction in the context of diffuse family boundaries and is based on general systems theory, may be especially applicable to the study of early toddler emotion regulation. The two primary types of boundary ambiguity are defined as a) parental psychological absence with physical presence or b) parental psychological presence with physical absence. However, boundary ambiguity has not been applied, to date, in studies focusing on the relationships between very young children and their parents. The purposes of this study were to: 1) define the construct of psychological absence with physical presence in the early parent-child relationships, 2) explore the impact of psychological absence on the development of children’s early self-regulation between 14 and 36 months of age, and 3) explore the impact of psychological presence with physical absence on toddlers’ (14-36 months of age) regulatory outcomes. Results supported a proximal method for measuring psychological absence using a latent construct and further demonstrated that mothers’ psychological absence significantly predicted children’s emotion regulation related behaviors across toddlerhood; these fl Er) It rIi r. 0f: results were consistent across racial/ethnic groups and did not differ for boys as compared to girls. Additionally, results demonstrated that fathers’ consistent physical presence significantly predicted toddlers’ emotion regulation related behaviors across toddlerhood; these results are discussed within the framework of boundary ambiguity, suggesting that consistency in presence is a proxy for organization of the fathering role. Taken together, study findings illustrate the ways in which young children make meaning of abstract family attributes, such as role boundaries, reflecting family organization in self-regulatory behaviors. ACKNOWLEDGEMENTS I would like to sincerely thank my committee chairperson, Dr. Holly E. Brophy- Herb, who has led by example and whose mentoring has guided my development as a researcher. I would also like to thank my committee members, Dr. Kathleen B. J ager, Dr. Marsha Carolan, and Dr. Hiram E. Fitzgerald, each of whom is an integral part of my professional development and has generously shared his/her time and wisdom with me. I would further like to thank Dr. Lorraine McKelvey, Dr. Laurie Van Egeren, Dr. Alexander von Eye, Dr. Rachel Schiffman, and Dr. Thomas Luster for their support. I am grateful to the children and families who participated in the Early Head Start Research and Evaluation Project; they are the most important partners in the pursuit to improve outcomes for vulnerable children and their families. Finally, I would like to thank my husband, Andrew D. Bocknek, for his love, encouragement, and listening ear. TABLE OF CONTENTS LIST OF TABLES ........................................................................................................... viii LIST OF FIGURES ............................................................................................................ ix CHAPTER 1 INTRODUCTION ............................................................................................................... 1 Statement of the problem ............................................................................................... 2 Development of Emotion Regulation in the Family System ................................... 2 Significance of the Problem .......................................................................................... 3 Operational Definitions of Boundary Ambiguity .................................................... 3 Conceptual Model and Hypotheses ............................................................................... 4 Measurement of Boundary Ambiguity .................................................................... 5 Hypotheses .................................................................................................................. 11 CHAPTER 2 REVIEW OF THE LITERATURE ................................................................................... 13 Boundary Ambiguity and Families ............................................................................. 15 Holistic Attributes of Families .............................................................................. 15 Meaning-Making Processes and Boundary Ambiguity ......................................... 16 Development of Boundary Ambiguity as a Construct .......................................... 17 Known Outcomes for Children as a Result of Boundary Ambiguity .................... 18 Boundaries Circumscn'bing the Early Parent-Child Relationship ............................... 21 Role-Specific Attributes of Mothering and Fathering ........................................... 22 Measurement of the Father-Child Relationship .................................................... 23 Emotion and Meaning-Making in the Parent-Child Relationship ............................... 26 Meaning in the Mother-Child Relationship ........................................................... 27 Meaning in the Father-Child Relationship ............................................................ 29 Meaning-Making and Toddlers’ Development ..................................................... 31 Summary ...................................................................................................................... 37 CHAPTER 3 METHOD .......................................................................................................................... 38 Procedures ................................................................................................................... 38 Main Study Procedures .......................................................................................... 38 Fathers Sub-study Procedures ............................................................................... 39 Study Measures ........................................................................................................... 39 Psychological Absence ........................................................................................ 400 Parent-Child Interaction ........................................................................................ 42 Depressive Symptoms ........................................................................................... 42 Child Outcome Measures ...................................................................................... 43 Covariate Measures ............................................................................................... 44 Fathers’ Continuous Physical Presence/Absence .................................................. 46 Missing Data ................................................................................................................ 46 V . v "3’3 Effect Sizes .................................................................................................................. 48 General Plan for Analyses ........................................................................................... 49 Power and Sample Size ............................................................................................... 50 CHAPTER 4 PLAN FOR PRESENTATION OF STUDY FINDINGS ................................................. 52 General Summary of Chapters .................................................................................... 52 Between-Group Comparisons ..................................................................................... 53 Comparisons Based on Racial/Ethnic Groups ....................................................... 53 Comparisons Based on Child Gender .................................................................... 53 CHAPTER 5 MATERNAL PSYCHOLOGICAL ABSENCE AND TODDLERS’ SOCIOEMOTIONAL OUTCOMES .......................................................................... 55 Sample Description ..................................................................................................... 55 Child Characteristics ............................................................................................. 56 Mother Characteristics ........................................................................................... 56 Results ......................................................................................................................... 57 Hypothesis 1 (Mothers) ......................................................................................... 57 Hypothesis 2 .......................................................................................................... 63 Discussion .................................................................................................................... 68 Psychological Absence and Toddlers’ Socioemotional Development .................. 68 Relationship Between Maternal Detachment and Child Emotion Regulation Over Time .............................................................................................................. 70 Relationship Between Risk and Psychological Absence ....................................... 72 Child Contributions to Relational Boundaries ...................................................... 72 Comparisons Based on Child Gender .................................................................... 74 Limitations to the Current Study ......................................................................... 744 Implications for Research and Practice ................................................................. 75 CHAPTER 6 PATERNAL PSYCHOLOGICAL ABSENCE AND TODDLERS’ SOCIOEMOTIONAL DEVELOPMENT ................................................................... 77 Sample Description ..................................................................................................... 78 Father Characteristics ............................................................................................ 78 Child Characteristics ............................................................................................. 79 Results ......................................................................................................................... 79 Hypothesis 1 (Fathers) ........................................................................................... 79 Hypothesis 3 .......................................................................................................... 82 Hypothesis 5 .......................................................................................................... 83 Discussion .................................................................................................................... 83 Psychological Absence Among Fathers ................................................................ 83 Fathers’ Lack of Empathy and Parenting and Child Outcomes ............................ 85 Paternal Depressive Symptoms and Parenting and Child Outcomes .................... 86 Limitations to the Current Study ........................................................................... 86 vi CHAPTER 7 PATERNAL PSYCHOLOGICAL PRESENCE AND TODDLERS’ LONGITUDINAL REGULATORY DEVELOPMENT ........................................................................... 88 Sample Description ..................................................................................................... 88 Father and Mother Characteristics ......................................................................... 88 Child Characteristics ............................................................................................. 89 Results ......................................................................................................................... 89 Hypothesis 4 .......................................................................................................... 89 Discussion .................................................................................................................... 93 Fathers’ Continuous Presence as a Proxy for Organized Relational Boundaries.. 93 Father Absence and Family Meaning Making ...................................................... 95 Comparisons Based on Child Gender .................................................................... 97 Comparisons Based on Racial/Ethnic Groups ....................................................... 98 Limitations to the Current Study ......................................................................... 100 CHAPTER 8 FINAL DISCUSSION AND CONCLUSIONS .............................................................. 101 Comparisons Based on Child Gender ........................................................................ 102 Comparisons Based on Racial/Ethnic Groups ........................................................... 103 Children’s Meaning-Making Processes and Holistic Family Attributes ................... 103 Future Implications for Research and Practice .......................................................... 104 APPENDIX B ATTRITION ANALYSES .............................................................................................. 119 Mother Parent Interview ............................................................................................ 1 19 Mother-Child Three-Bag Task .................................................................................. 1 19 Child Bayley Assessment .......................................................................................... 120 Comparisons Based on Missingness Among Mothers and Children ........................ 120 Fathers’ Data ............................................................................................................. 126 REFERENCES ................................................................................................................ 128 vii Tdbil Tabh Tdbit Table Table LIST OF TABLES Table 1. Conceptual and operational definitions of key constructs ................................... 12 Table 2. Sample demographic characteristics of mothers in complete sample ............... 1 13 Table 3. Sample demographic characteristics of fathers’ directly observed subsample. 114 Table 4. Descriptive statistics for variables in cross-sectional maternal psychological absence and boundary ambiguity measurement model and path model. ........... 115 Table 5. Descriptive statistics for variables composing emotion regulation and maternal detachment latent grth curves ........................................................................ 116 Table 6. Description of children’s contact with non-residential biological fathers across toddlerhood. ....................................................................................................... 1 17 Table 7. Mean scores for toddlers’ emotion regulation at each timepoint within groups based on consistency of father presence. ........................................................... 118 viii Fleurc Figure Figure Figurei Figure ' Figure 5 LIST OF FIGURES Figure 1. Conceptual model defining psychological absence. ........................................ 105 Figure 2. Conceptual model demonstrating the longitudinal development of parental detachment and toddlers’ emotion regulation based on linear latent grth curve measurement models .............................................................................. 106 Figure 3. Conceptual model demonstrating consistency of father presence predicting longitudinal development of toddlers’ emotion regulation. ............................. 107 Figure 4. Measurement of boundary ambiguity in the mother-child relationship ........... 108 Figure 5. Relationship between maternal detachment and emotion regulation across toddlerhood based on latent growth curve models. .......................................... 109 Figure 6. Relationship between consistency of father presence and emotion regulation across toddlerhood in a single-domain latent grth curve model. ................. 110 Figure 7. Relationship between consistency of father presence and emotion regulation across toddlerhood in a single-domain latent grth curve model among White families. ................................................................................................. 111 Figure 8. Relationship between consistency of father presence and emotion regulation across toddlerhood in a single-domain latent growth curve model among Black families. ............................................................................................................ 112 CHAPTER 1: INTRODUCTION Research demonstrates that early self-regulation develops in the context of early parenting relationships (Calkins, 2004; Calkins & Johnson, 1998; Denham, 1998; Sroufe, 1996). A view of this principle through the lens of general systems theories (von Bertalanffy, 1968; Minuchin, 1974) would suggest that the early parent-child relationship is a subsystem of the broader family system, and, further, that processes within this subsystem reflect family characteristics which are holistic and systemic, such as relational boundaries. Within families, each subsystem is defined by semi-permeable boundaries which are determined by relational processes. The purpose of family structure and boundaries is to meet the developmental needs of individual family members. When a subsystem breaks down, the family system is unable to meet the developmental needs of individual family members (Hanna & Brown, 2004). Boundary ambiguity theory identifies risks presented by the disorganization of family roles when such a subsystem breaks down. Specifically, boundaries are ambiguous when it is unclear who is in or out of the family. There are likely important implications for boundary ambiguity in the early parent-child relationship though there is a paucity of research demonstrating it. Fitzgerald, Barnes, and Almerigi (2007) suggest that the scientific study of toddler development must occur within theoretical models grounded in systems theory. Based on this perspective, the theory of boundary ambiguity provides a useful fi'amework with which to discuss children’s self-regulation in the context of the parent-child relationships; boundary ambiguity theory considers the family a developmental system of which the parent-child relationships are subsystems. w ‘1'. v1.0 r' demo earl} c 10th} 1' C: use emu Eisenbcr critical ir Socioemot RaVer. 31. l( Parenting , Denham. 1 reg‘dlaIOQ-z disengage? (Sam (fit) I}; Come“ ma‘ Statement of the problem The purpose of the current study was to explore family boundary ambiguity as demonstrated within mother-child and father-child relationships and the related impact on early emotion regulation (measured via emotion regulatory related behaviors) among toddlers in a low-income population. Development of Emotion Regulation in the Family System Emotion regulation refers to the ability to manage one’s arousal and the ability to use emotional information as a regulator of behavior (Cole, Martin, & Dennis, 2004; Eisenberg & Spinrad, 2004). Emotions provide children, as well as adults later on, with critical information about themselves and the social environment, and regulatory skills represent a core competency in predicting later socially adaptive behaviors (Denham, 1998, Thompson & Goodvin, 2007). Children living in poverty are exposed to an accumulation of multiple risk factors and are more likely to experience adverse socioemotional outcomes, including poor regulatory skills, in particular (for review, see Raver, 2004). The development of children’s emotion regulation occurs in the contexts of parenting relationships and family systems (Calkins, 2004; Calkins & Johnson, 1998; Denham, 1998; Sroufe, 1996). Furthermore, research demonstrates that young children’s regulatory-related behaviors become disorganized in the presence of caregiver disengagement (Tronick, Ricks, & Cohn, 1982) as well as in disorganized family systems (Sameroff, 1989). Therefore, a better understanding of the pathways through which context may impact the development of young children’s regulatory competencies needs to be explored. ii; l'CU‘u’ldllUl' is the mo Lame-.1. 1 Ti Ll'tl'OUgh p associatct psycholoé BOih £1er Tt likely to e sample of Slimli‘lt‘an shows [ha be less lik. are at sign LYOHSRUI Fur blOngjCaI l dispropomk Significance of the Problem As described, context significantly impacts young children’s development of self- regulation. Furthermore, the family, and more specifically, the parent-child relationship, is the most significant context for young children’s development (Zeanah, Boris, & Larrieu, 1997). Operational Definitions of Boundary Ambiguity The current study operationalized boundary ambiguity via psychological absence through parental detachment, lack of empathy, and parenting stress, constructs often associated with caregiver depressive symptoms, and boundary ambiguity via psychological presence through inconsistent physical presence of non-residential fathers. Both areas represent significant social issues, particularly among low-income families. To begin, mothers of young children living in poverty are disproportionately likely to experience depressive symptoms, with one recent study finding that 75% of their sample of low-income single mothers of toddlers and preschoolers demonstrated significant depressive symptoms (Peden, Rayens, Hall, & Grant, 2004). Research clearly shows that children of depressed mothers, particularly in low-income families who may be less likely to have access to social support and resources for mental health services, are at significant risk for adverse outcomes (Cummings & Davies, 1994; Field, 1992; Lyons-Ruth, Connell, & Grunebaum, 1990). Furthermore, young children in poverty are highly likely to grow up without their biological fathers living in their homes, often due to structural forces that disproportionately impact low-income families, such as higher incarceration rates and disproportionate minority confinement and multi-partner fertility and low marriage rates prograi‘.‘ father pr Ei‘aluatzt demonstr fathers ar (8.1g. Bull Bronte—Ti 2000i Ox absence “’1 effects of ti Concretizin, Bou; Was explUTL‘i and psycholi ambiguity. F relationship a (Tamis-LeMonda, Shannon, Cabrera, & Lamb, 2004; Black, Dubowitz, & Starr, 1999). Research demonstrates that father absence is related to numerous worrisome outcomes for children (Amato & Rivera, 1991; Seltzer & Bianchi, 1988) and, further, that father presence is a protective factor for children living in at-risk circumstances (J eynes, 2003; Martinez, DeGarmo, & Eddy, 2004). The families enrolled in Early Head Start programming may represent a resilient subpopulation of low-income families in regard to father presence. Research based on data from the Early Head Start Research and Evaluation Project (EHSRE; Love, Kisker, Ross, Raikes, Constantine et al., 2005) demonstrates that this group of children are highly likely to experience contact with their fathers and further that such contact relates to better outcomes for the youngest children (e. g., Boiler, Bradley, Cabrera, Raikes, Pan, Shears, & Roggrnan, 2006; Cabrera, Moore, Bronte-Tinkew, Halle, West et al., 2004; Vogel, Bradley, Raikes, Boller, & Shears, 2006). Overall, families experiencing caregiver depressive symptoms and/or father absence will benefit from targeted interventions informed by theory that describes the effects of these significant social issues on family structure and processes, thus concretizing the pathways between these noted social stressors and children’s outcomes. Conceptual Model and Hypotheses Boundary ambiguity, which served as the theoretical framework for this study, was explored in separate models testing psychological absence with physical presence and psychological presence with physical absence, two different typologies of boundary ambiguity. Further, models will test psychological absence in the mother-child relationship as well as the father-child relationship. Psychological presence will only be EARL-ht“ and inter descr Ol-Cl'l preset tested in the father-child relationship because data are not available for this sample describing children in families in which mothers are physically absent. See Figures 1-3 for conceptual models. Figure 1 presents a conceptual model in which the construct of boundary ambiguity is tested as a latent construct indicated by three manifest variables and embedded in a path model including parental depressive symptoms, parent-child interaction, and child socioemotional outcomes. Figure 2 presents a conceptual model describing the relationship between psychological absence and child emotion regulation over time. Figure 3 presents a conceptual model describing the consistency of father presence over toddlerhood predicting child emotion regulation over time. Measurement of Boundary Ambiguity Boundary ambiguity has historically been studied with self-report measures completed by the family member who remains/ survives following the absence or ambiguous loss. The measure most often used is the Boundary Ambiguity Scale (Boss, 1977). This measure is not appropriate for very young children, though a revision has been published specifically for use with older children whose parents have divorced (Boss, Greenberg, & Pearce-McCall, 1990). Boundary ambiguity is thus measured via the subjective interpretation of the remaining/surviving family member. It is this issue of perception that most inhibits the study of boundary ambiguity in relationships parents have with infants and toddlers. As research respondents, infants and toddlers are not able to communicate their perceptions about an ambiguous loss of a parent in specific terms, and therefore measurement of potential resulting boundary ambiguity is challenging. A few authors have used proximal measures of boundary ambiguity when the aforementioned self-report measure administered to a family member was not appropriate or most useful (e.g., Buehler & Pasley, 2000; Lee & Whiting, 2007; Madden-Derdich, Leonard, & Christopher, 1999; Serovich, Price, Chapman, & Wright, 1992). Though the measures used in these other studies also present self-report data provided by the survivor of an ambiguous loss, the alternative methodologies provide support for using proximal methods to define boundary ambiguity, as is the case in the current study. Carroll, Olson, and Buckmiller (2007) published a 30-year review of the body of knowledge of boundary ambiguity research and determined that the field would benefit from additional methods to measure boundary ambiguity as well as additional analytical protocols. The present study seeks to respond to this research implication. Structural equation modeling (SEM) is a useful approach to define boundary ambiguity in the parent-child relationship by determining the validity of the construct of psychological absence, indicated by proximal observed variables which support the dimensions of the original boundary ambiguity scales, in the overall model and its predictive power of children’s outcomes. The boundary ambiguity scales measure a participant’s perspective on the loss of a family member, combine cognitive and affective dimensions of meaning making to measure boundary ambiguity (Bagarozzi & Berry, 1991). Of great significance to the theory, and therefore to the measurement of the construct, ambiguity is defined by family members who have survived the loss, and the definition depends on their perception of the loss and its impact on the greater family system (Boss, 2007; Boss, 2002). It is this issue of perception that presents the greatest challenge to measuring boundary ambiguity in the relationship between a parent and his/her young child as there are likely challenges to discerning a young child’s perceptions about the incongruence in the physical and psychological states of his/her primary caregiver, particularly if the absence has persisted over time. Study hypotheses reflect a presumption that perception for young children will be reflected behaviorally through less optimal socioemotional outcomes. Measurement of Psychological Absence The method utilized in the current study to measure psychological absence retains the cognitive and affective components of the Boundary Ambiguity Scale. As family- level perception is so central to the measurement of this theory, the parent’s perception of self in the context of his/her role as parent will be included. Though this is different from the perception of the survivor of an ambiguous loss, in this case, the child, the perception of the family via the primary caregiver may be retained. Also, an observer’s rating of the behavioral manifestations of psychological absence will be included to support the parent’s perception. Next, the relationship between boundary ambiguity and child-level outcomes is hypothesized to be mediated by parent-child interaction. This latent factor includes observer ratings of dyadic engagement, providing a behaviorally-based measure of the parent’s and child’s perceptions of their relationship. Therefore, this structural model is hypothesized to be an effective way to measure boundary ambiguity in the parent-child relationship. As Figure 1 demonstrates, parental depressive symptoms will be included in the model to hypothesize depressive symptoms as a correlate with, and distinct construct from, psychological absence. Measurement of Psychological Presence with Physical Absence Psychological presence with physical absence, in this study, is defined as the father’s physical absence with ambiguity related to his psychological presence. The inconsistencies that often exist when fathers are physically absent from their children’s it lives. espe. ambiguitit Parent-Ci Tl. and child DC“ c‘ltlp 20in ibi J, (hr 0 u uh lives, especially early on when families are developing parenting roles, may predict ambiguities in family boundaries. Parent-Child Interaction as Mediator The main hypotheses postulate that the relationship between boundary ambiguity and child outcomes will be mediated by interactions between parents and children. Developmental theory (Sameroff, 1975; Sameroff & Chandler, 1975; Sameroff & F iese, 2000b) indicates that parents and children mutually contribute to children’s development through these interactions. Therefore, a latent construct will be utilized to capture parent and child positive contributions to interactions, measured by observer ratings on play tasks of parental supportive behaviors and child engagement behaviors. Impact of Boundary Ambiguity on Toddlers ’ Socioemotional Development Emotion regulation. The main outcome variable of interest related to the current models is toddlers’ emotion regulation. As described, this construct relates to a child’s ability to self-organize, and self-organization is likely to be associated with family-level organization. The field of study related to emotion regulation is relatively new and a bit varied in terms of operationalizing and measuring this construct. Included by many researchers in the definition of emotion regulation is the control of internal affective states (Cole, Martin & Dennis, 2004; Eisenberg & Spinrad, 2004; Thompson, 1994). However, measuring young children’s internal states is a challenge, particularly in the absence of biophysical markers (Fitzgerald, Barnes, & Almerigi, 2007). Observation rating scales, such as the Emotion Regulation subscale of the Bayley Behavior Rating Scales (BBRS; Bayley, 1993), assess children’s regulation-related behaviors, such as the child’s ability to change tasks during the test and handle fi'ustration. Eisenberg, (Thanti proces “huh tonm; Headf Rain: toinxt behax~ “iflitl measu- the C0 01‘ the pran QUICOI lnterx: earl}'s Champion, and Ma (2004) note that a particular strength of measuring regulation-related processes is that the act of regulation is thus considered separately from emotionality, which varies among children based on temperament. This scale has been used effectively to measure toddler regulatory abilities in other studies also utilizing the National Early Head Start Research and Evaluation Project dataset (e. g., Raikes, Robinson, Bradley, & Raikes, 2004). Orientation and engagement. In the hypothesized cross-sectional model seeking to initially validate a psychological absence construct (see Figure 1), a second Bayley behavior rating scale, the Orientation and Engagement subscale, will also be included with the Emotion Regulation subscale on a single factor. Including more than one measure of socioemotional development in the measurement model will better support the complete model; the purpose of the cross-sectional model, primarily, is to test the fit of the variable factor loadings on the psychological absence factor in the broader context of parental depressive symptoms, parent-child interaction, and child socioemotional outcomes. Orientation and engagement measures a child’s cooperation with the interviewer, positive affect, and interest in the test materials. These behaviors relate to early social skills and, as described, early relationship skills have their foundation in the interactions between parents and young children. Furthermore, this measure has been used effectively in other studies measuring the impact of parent-child relationships on children’s outcomes (e.g., Vogel, Bradley, Raikes, Boller, & Shears, 2006). Contextual Covariates Four additional variables will be included in all models to control for other known significant influences on early development, particularly within the parent-child and family contexts. Parental household composition. First, the parent’s living arrangements will be included. Research (e. g., Aronson & Huston, 2004) demonstrates that young children experience differential socioemotional outcomes based on their parent’s marital status and the presence of alternate caregivers and further that married parents may be more likely to demonstrate a sense of parental satisfaction (Rogers & White, 1998). Child temperament. Additionally, children’s temperament will be included because of the known associations between temperamental characteristics and regulatory skills (Rothbart & Derryberry, 1981). Early Head Start intervention. Children in this sample were assigned to either the program group, receiving Early Head Start intervention, or the comparison group and were free to receive other services in their community. Early Head Start is known to impact parenting and child outcomes (Love, Kisker, Ross, Raikes, Constantine et al., 2005), and, therefore, one might expect these two groups of families to demonstrate differences in the current model. Cumulative risk. Finally, cumulative risk will be entered into the model. All of the families in this sample are considered “at-risk” based on their eligibility status for Early Head Start programming, the main inclusion criterion. However, among the included families, the amount of risk varies, and the accumulation of demographic risk factors, such as income and employment, is known to be a better predictor of outcomes than any 10 unuk LeBo more behA\ Predi fihor lune. and ti p COUIn [odd]. single risk factor on its own (Sameroff & F iese, 2000a). Contemporary research (Puma, LeBeouf, Carta, Rodriguez, Spellman, & Watt, unpublished manuscript) suggests using more specified cumulative measures of risk based on indicators such as parenting behaviors rather than a measure of cumulative demographic risk, dependent on the hypothesized model outcomes. However, guidance for building a risk factor related to children’s socioemotional outcomes is not well-available, and the demographic risk index has been utilized effectively in other research (e. g., Raikes, Pan, Luze, Tamis-LeMonda, Brooks-Gunn et al., 2006). Therefore, it will be included in the current study with this caveat noted. Hypotheses There were four primary objectives in the current study. The first objective was to test a latent factor structural model of psychological absence in early mother-child relationships and father-child relationships. The second objective was to examine the predictive value of psychological absence for parent-child interactions and toddler emotion regulation in both mother-child relationships and father-child relationships over time. The third objective was to examine the relationship between psychological absence and the development of toddlers’ emotion regulation among mothers and their children as compared to fathers and their children. The fourth objective was to examine the effects of continuity of father presence, conceptualized as psychological presence, on the growth of toddler self regulation over time. In support of these objectives, the following hypotheses were proposed: 1) The sample data will fit the hypothesized latent factor measurement model describing psychological absence within the broader path model including 11 2) 3) 4) 5) Table 1 parent-child interaction, parental depressive symptoms, psychological absence, and toddlers’ socioemotional outcomes (self-regulation and orientation and engagement). Mothers’ psychological absence over toddlerhood (14-36 months) will significantly predict change over time in children’s emotion regulation. Fathers’ psychological absence will significantly predict change over toddlerhood (14-36 months) in children’s emotion regulation. Continuity of the physical presence in the early father-child relationship will significantly predict change over toddlerhood (14-36 months) in children’s emotion regulation. There will be a greater effect size in the relationship between psychological absence and toddlers’ regulatory outcomes in the mother-child relationship as compared to the father-child relationship due to mothers’ greater responsibility for primary care early on. Conceptual and Operational Definitions of Key Constructs Construct Conceptual Definition Operational Definition Psychological Psychological absence is Psychological absence is Absence conceptualized as parent’s operationalized by a latent factor physical presence in the indicated by Parental Detachment, parenting role but emotional coded by an independent observer unavailability to his/her child during semi-structured play task (Boss, 1999, 2004, 2006). (N ICHD Early Child Care Research 12 Table 1 Continued Network, 1997); the Lack of Empathy subscale of the Adult- Adolescent Parenting Inventory (Bavolek, 1984); and the Parenting Distress subscale of the Parenting Stress Index (Abidin, 1995). physical absence from a child’s life, partially or wholly, while remaining psychologically present through a lack of family re-organization in which the meaning of his absence is Depressive Depressive Symptoms is Depressive Symptoms is Symptoms conceptualized as a greater operationalized in this study with presence of symptoms the Center for Epidemiological consistent with depressive Study Depressive Symptoms Scale symptoms, such as disinterest short form (CESD; Radloff, 1977; in daily living, low appetite, Ross, Mirowsky, & Huber, 1983). and sadness (Radloff, 1977). Psychological Psychological presence is Psychological presence is Presence conceptualized as a father’s operationalized through the mother’s report of the father’s involvement at the child’s 14- month, 24-month, and 36-month birthday-related assessments; a variable collapses these reports into a dichotomous score of the father’s consistent presence across 13 Table 1 Continued coherently understood (Boss, 1999, 2004, 2006). toddlerhood or lack thereof. Parent-child Parent-child interaction is Parent-child interaction is behavioral skills typically significant in early childhood: self—regulation and orientation and engagement. interaction conceptualized as reflecting operationalized by a latent factor the quality of the parent-child indicated by the Child Engagement relationship. subscale and the Parental Supportiveness subscale, both measured through independent observer ratings during a semi- structured play task (NICHD Early Child Care Research Network, 1997) Toddlers’ Socioemotional outcomes are Socioemotional outcomes are socioemotional conceptualized through operationalized through a latent outcomes toddlers’ development of factor indicated by the Emotion Regulation subscale and the Orientation and Engagement subscale of the Bayley Behavior Rating Scales (BBRS; Bayley, 1993). 14 CHAPTER 2: REVIEW OF THE LITERATURE Boundary Ambiguity and Families The construct of boundary ambiguity is defined as “a state in which family members are uncertain in their perception about who is in or out of the family and who is performing what roles and tasks within the family system” (Boss & Greenberg, 1984, p. 536). Such roles and tasks, indicators of family boundaries, include the physical and psychological processes relevant to particular family roles. Boundary ambiguity theory is most concerned with the family-level perception of boundaries, as opposed to the structure itself, based on a symbolic-interactionist perspective (Boss, 2007; Boss, Greenberg, & Pearce-McCall, 1990). Holistic Attributes of Families Healthy families have clear interpersonal boundaries, contributing to the non- summative, organized whole (Minuchin, 1974; Minuchin, Rosman, & Baker, 1978). Holistic family features include patterns, processes, and roles unique to the family and the ways in which it is organized (McHale & Sullivan, 2008; Minuchin, 1985). Many researchers emphasize that change, or stress, alone does not produce negative outcomes, but, rather, coping and meaning-making are the keys to determining the impact stress will have on a system and its components (Boss, 1982; Boss, 1980; Hobfoll & Spielberger, 1992; Kazak, 1992). At the level of the family, shared meanings reduce ambiguity and uncertainty and support the coordination of responses among family members (Patterson & Garwick, 1994). Therefore, interactions between family members are critical in defining family health. Literature related to family systems emphasizes the importance of consistency, 15 org. (Brt BFUI Cl‘ia: C C) L: t finnf beha‘ parnt devcl proce Unakes cmnm chili-1rC childre compel Leerkeé them 1a. (Lie-berT clarity, and predictability regarding family interactions. Sameroff (1989) defines the organized system of interactions within the family, informed by the bioecological (Bronfenbrenner, 1979; Bronfenbrenner, 1989; Bronfenbrenner & Ceci, 1994; Bronfenbrenner & Morris, 1998) and transactional (Sameroff, 1975; Sameroff & Chandler, 1975; Sameroff & Fiese, 2000b) models of development, as a network of codes, interacting and regulating behavior at multiple levels. Through family codes and daily interactions, children learn behavior that is acceptable, according to their unique family rules, imparted by their parents’ behaviors. Further, children’s responsive behaviors impact the plastic development of family codes through their continuing participation in interactions. The idea of family relationships as a context for development is inherent in the structural nature of families and represents an active process of regulation across development. Meaning-Making Processes and Boundary Ambiguity Thus, meaning-making is central to the study of boundary ambiguity. This issue makes it both challenging to study with very young children with a limited capacity to communicate their subjective experiences and significant to study with very young children, given the wealth of scholarship demonstrating the important ways in which children’s early experiences in relationships informs developing socioemotional competencies via internal representations (Bowlby, 1969/1973/1980; Crockenberg & Leerkes, 2000). Children’s early relationships with significant primary caregivers give them lasting expectations about who they are and what they can expect from others (Lieberman, 1997). Furthermore, role theory posits that a role is a set of expectations 16 ‘ LG? I...“ I ind: Stmc remar about the combined attributes of “actor-other” identities, meaning that roles are socially constructed by two people in a given relationship (Merton, 1957). In the current study, young children were hypothesized to actively participate in family relationships and indeed derive meaning from the ambiguities in their relationships with parents who are either psychologically absent though physically present, due to risk factors such as depressive symptoms, or psychologically present though physically absent, as is hypothetically related to biological father absence. Further, toddlers were hypothesized to reflect this meaning-making process through regulatory behaviors. As described, theory related to boundary ambiguity emphasizes the ways in which individuals perceive dissonance in the physical and psychological presence of a family member, promoting family-level disorganization, and contributing to individual dysfunction. Central to this theory are a symbolic-interactionist perspective and foundational knowledge from family systems theories (Minuchin, 1974; Minuchin, Rosman, & Baker, 1978; von Bertalanffy, 1975) and family stress theory (McCubbin & Patterson, 1982, 1983), that emphasize the pathways among family meaning-making, family patterns, and likely outcomes. Further, clinical (e. g., Minuchin, 1974) and empirical literature (e. g., Lewis, Beavers, Gossett, & Phillips, 1976) demonstrate that individual psychological and development health reside in coherent and stable family structures. Development of Boundary Ambiguity as a Construct Boundary ambiguity theory has been developed over the last 30 years, though it remains a small body of knowledge. It first appeared as a construct in a 1977 article 17 ‘— mil psv psw fam written by Pauline Boss. Boss (197 7) identified two types of ambiguous loss: psychological presence with physical absence (Type I) and physical presence with psychological absence (Type II). Psychological presence (Type I) is defined by surviving family members’ maintenance of a lost family member’s role without that family member being physically present. Psychological presence may occur in relation to life stressors such as divorce, illness, or military service. Psychological presence may be the result of normative life transitions and mark a period of reorganization in a family that is only temporary, particularly when a family defines the stressor as normative and maintains resources to effectively cope (Boss, 1980). In other circumstances, when the meaning the family makes of the stressful life event is negative, when the meanings the members of the family make of the event are incongruent, when the family is unable to make a coherent meaning of the stressor, and/or when the family does not maintain adequate resources to cope, the ambiguity in the family may cause dysfunction (Boss, 1980). Psychological absence (Type 11) is defined by a family member’s physical maintenance of his/her family role while behaving as absent at the level of psychological and emotional processes. Examples of predictors of psychological absence include depressive symptoms, mental illness, substance abuse, and dementia. Known Outcomes for Children as a Result of Boundary Ambiguity The research on boundary ambiguity has mostly focused on the risks present for adult members of families amidst family-level stressors such as depressive symptoms, mental illness, substance abuse, and dementia (Carroll, Olson, Buckmiller, 2007). A limited number of studies have presented empirical findings demonstrating the impact of ambiguous loss on older children (e. g., Bocknek, Sanderson, & Britner, 2009; Buehler & 18 Pasley, 2000; Madden-Derdich & Arditti, 1999; Rosenberg & Guttmann, 2001; Taanila, Laitinen, Moilanen, & Jarvelin, 2002). These studies demonstrate that family boundary ambiguities negatively impact children’s attachment security (e.g., Madden-Derdich & Arditti, 1999), adjustment processes (e.g., Buehler & Pasley, 2000; Rosenberg & Guttmann, 2001), and mental health outcomes (e. g., Bocknek, Sanderson, & Britner, 2009; Taanila, Laitinen, Moilanen, & J arvelin, 2002). Furthermore, these studies demonstrate that inconsistencies between psychological presence and physical presence relate significantly to children’s less optimal functioning in the context of family processes. One qualitative study (Lee & Whiting, 2007) examined the effects of ambiguous loss generally, compounding Type I and Type 11 loss, on children as young as 2 years old. This study raised important considerations about typical symptomology that might arise in children experiencing ambiguous loss, including blocked coping processes and helplessness leading to depressive symptoms and anxiety. A modest scholarship describes the impact on older children’s socioemotional development as a result of violations in the emotional, psychological, and behavioral rules governing interactions within the parent-child family subsystem, particularly in relation to parentification, triangulation, and enmeshment, a very different area of inquiry in terms of boundary disturbances (e. g., Byng-Hall, 2002; Jacobvitz, Hazen, Curran, & Hitchens, 2002; Kerig, 1995; Kerig, 2005). Hazen, J acobvitz, and McFarland (2006) describe the need to better explore the effects of boundary disturbances in early childhood, noting that documented effects on older children have their origins in early caregiving relationships. 19 3550:: III Psychological absence. Psychological absence is known to have a significant association with emotional and psychological pathology for the family member who experiences the loss, such as might be the case for the partner or adult child of an Alzheimer’s patient (e.g., Boss, Caron, Horbal, & Mortimer, 1990; Thomas, Clement, Hazif-Thomas, & Leger, 2001). However, the construct of psychological absence has not been empirically studied in regard to the relationship between parents and their infants and toddlers. Psychological absence is often associated with depressive symptoms, and there is extensive research on outcomes for children of depressed mothers (e.g., Cummings & Davies, 1994; Field, 1992; Lyons-Ruth, Connell, & Grunebaum, 1990). Psychological presence. As is true for the construct of psychological absence (Type II), no literature exists to describe the effects of psychological presence (Type I) on young children. However, the literature relating to psychological presence often focuses on fathers whose roles have become ambiguous in the system due to family structural changes. This issue has particular salience to the study of young children; father presence and/or involvement with their young children is a topic of much contemporary interest (see, for example, Cabrera, Ryan, Mitchell, Shannon, & Tamis-LeMonda, 2008). Pauline Boss (1977) writes that it is possible for family members to perceive and act upon a particular family member’s presence even when that person is not physically present. “If members of the family are not reconciled to the fact that he will never again return to the family, they will do nothing to shift role tasks and reorganize either by taking someone else into the system in his place or by shifting roles within the system to cover his former tasks. Such a father, though physically absent, is 20 to adapt 'i- dexehrpr ndesgox relation t in terms . Hnehens describe Childhoo. Cafeg‘lVin and Iheir (Beam); 1979; BR, 199313n( Chandicr. m the Inf “MIG mt] (CPOCkenl psychologically present and causes ambiguity of boundary for that family system” (p. 141). Boss goes on to write that “a system needs to be sure of its components” (p. 142) in order to adapt to stress. A modest scholarship describes the impact on older children’s socioemotional development as a result of violations in the emotional, psychological, and behavioral rules governing interactions within the parent-child family subsystem, particularly in relation to parentification, triangulation, and enmeshment, a very different area of inquiry in terms of boundary disturbances (e.g., Byng-Hall, 2002; Jacobvitz, Hazen, Curran, & Hitchens, 2002; Kerig, 1995; Kerig, 2005). Hazen, J acobvitz, and McFarland (2006) describe the need to better explore the effects of boundary disturbances in early childhood, noting that documented effects on older children have their origins in early caregiving relationships. Boundaries Circumscribing the Early Parent-Child Relationship There is a significant body of literature describing interactions between children and their parents, drawing on multiple salient theories, including attachment theory (Bowlby, 1969/1973/1980), the bioecological model of development (Bronfenbrenner, 1979; Bronfenbrenner, 1989; Bronfenbrenner & Ceci, 1994; Bronfenbrenner & Morris, 1998) and the transactional model of development (Sameroff, 1975; Sameroff & Chandler, 1975; Sameroff & Fiese, 2000b). Mothers and fathers are similarly significant in the lives of young children yet their interactions with their children are often distinct. While mothers may be more likely to be the primary caregivers to young children (Crockenberg & Leerkes, 2000), fathers are known to make important contributions to 21 «1 1.3011113; c. i Parke. 1 \Rd} s. E predictin need-s. pr predictin predictin role in m but OHC‘H likely to regulate - interact“ Christin; R019 Der \ that a Cl‘. relation, Conslder dCVejOp] 2094:. Z.- young children’s socioemotional development, as well, (MacDonald & Parke, 1984; Parke, 1981; Roberts, 1998; Youngblade, Park, & Belsky, 1993), though in more diverse ways. Role-Specific Attributes of Mothering and Fathering During early childhood in particular, mothers provide bonding behaviors, predicting children’s security-exploration; organized structure and responsiveness to needs, predicting children’s physiological regulation and affective regulation; protection predicting vigilance; teaching predicting learning; play predicting play; and discipline predicting self-control (Sameroff, 1989). During early childhood, fathers also play a key role in the development of regulatory skills through rich interactions with their children, but often in ways that are different from mothers’ normative processes. Fathers are more likely to provide opportunities for children to experience high emotional arousal and regulate those emotions through play and through more active forms of parent-child interaction (Roggman, Boyce, & Cook, 2001; Roberts, 1998; Roggrnan, Boyce, Cook, Christiansen, & Jones, 2004). Role Development of Biological Mothers Winnicott (1965) famously wrote, “there is no such thing as a baby,” inferring that a child could not be understood in isolation from his/her primary caregiving relationship. The early relationship between a child and his/her primary caregiver is considered to be the most significant context for children’s early socioemotional development (see for example Berlin & Cassidy, 2000; Bomstein & Tamis-LeMonda, 2004; Zeanah, Boris, & Larrieu, 1997). Mothers are most likely to fulfill the role of primary caregiver to children, especially during early childhood (Crockenberg & Leerkes, 22 . < -. 4 ”Will 3. “- l 1. sanhhc. Johnson. R015 D51 involxcg weH.AJ fadiers. 2006,_[ firm i l i e) definun 3002b , panlcul 2000). Mothers may be most likely to provide supportive behaviors like sensitive scaffolding and emotion language to help children learn regulatory strategies (Calkins & Johnson, 1998; Denham, 1998; Sroufe, 1996). Role Development of Biological Fathers Hofferth, Pleck, Stueve, Bianchi, and Sayer (2002) indicate that biological fathers, compared to social father figures or step-fathers, are the most likely to be involved with their children, and residential status plays a role in level of involvement as well. Also, there is wide variation in the amount of contact children have with their fathers, with the majority experiencing some contact, particularly among Early Head Start children in the current sample (Boller et al., 2006; Cabrera et al., 2006; Vogel et al., 2006). Differences in level of contact are often dependent on structural differences in families, such as the mother’s marital status (Hofferth et al., 2002). Because the definition of fatherhood, for non-residential fathers in particular, varies widely (Lamb, 2002b) there is a high potential for ambiguities and inconsistencies to occur in this role, particularly early in childhood when the rules and patterns related to family roles in young families may still be evolving. However, research (e. g., Vogel et al., 2006) shows that young children demonstrate more adaptive regulatory competencies as a function of father connectedness and that children with unstable or transient relationships with their fathers are less likely to demonstrate optimal outcomes. Measurement of the Father-Child Relationship As noted, researchers indicate that there is likely more diversity in the fathering role than there is in the mothering role (Lamb, 2002b). Due to this diversity, dernographers suggest that capturing a nationally-representative sample of fathers based 23 on children with their e fathered or well Rese minority r identi iea Shannon det ning social. 2: lParlte~ desenb also m (mach: eXpIQ; K370 11 on children is difficult, as compared to identifying mothers, given that not all fathers live with their children, and, further, that some fathers may not even be aware they have fathered children (Hofferth, 2002). These issues translate to community-based research as well. Researchers demonstrate that societal structural influences, such as disproportionate minority confinement and multi-partner fertility, undermine clear and consistent father identification and involvement, particularly in low-income populations (Tamis-LeMonda, Shannon, Cabrera, & Lamb, 2004; Black, Dubowitz, & Starr, 1999). Furthermore, defining who the father is (i.e., who to study) across families varies based on individual, social, and cultural definitions that are more diverse than the definition of mothers (Parke, 2000; Roggrnan, Fitzgerald, Bradley, & Raikes, 2002). Lamb (1975, p. 245) described fathers as “the forgotten contributors to child development.” Cowan (1997) also made this criticism of research, calling for a family systems perspective of attachment in which the relationships among mothers, fathers, and children are equally explored. Known Associations Between Fathering and Child Development In their review of father-child attachment, van lJzendoorn and De Woolf (1997) describe several studies (e. g., Easterbrooks & Goldberg, 1984; Grossman & Grossman, 1992; Schneider, Rosen, & Rothbaum, 1993) in which (based on non-significant findings) there are no associations between paternal behaviors, such as supportiveness and quality of assistance, and infant/toddler attachment security. The authors also cite a few studies in which there are significant associations demonstrated between fathers’ parenting behaviors and infants’ attachment security (e. g., Cox, Owen, Henderson, & Margand, 1992; Goossens & van IJzendoom, 1990), pointing to a mixed set of findings 24 1-. oxerall i construe offazher current :' signifies and chili the} u er mediate ' Vogel et with ch11. Black fa: relations} .llé’dia l0 ,- Rt Children‘s family CO} cOmplexit 0’1 Child O and SOCial 2007)-Var. Cll'CUmSCn, TElallQnshl-l overall in regard to the particular construct of attachment. In regard to a broader set of constructs, researchers have similarly found mixed results relative to the predictive power of father engagement on children’s outcomes (Pleck, 1997). In the dataset utilized in the current study in particular, Cabrera, Shannon, & Tamis—LeMonda (2007) found significant relationships between paternal behaviors (supportiveness and intrusiveness) and children’s socioemotional outcomes when children were 2 years old but not when they were 3 years old, suggesting that additional variables, including perhaps temporality, mediate the relationships between fathering and children’s socioemotional development. Vogel et a1. (2006) demonstrated that father involvement was most clearly associated with child outcomes among White families though not as consistently associated among Black families, suggesting that cultural meaning-making processes may mediate the relationship. Mediators of the F ather—Child Relationship Researchers explain the inconsistencies in the research on father involvement and children’s outcomes by suggesting that a complex interplay of factors within the broader family context predicts relationships between fathers and their children; a level of complexity that is perhaps unique to the father-child relationship. The effects of fathering on child outcomes may be direct or indirect, based on multiple variables in the personal and social contexts of the family (Cabrera, Fitzgerald, Bradley, & Roggrnan, 2007).Variables describing the structural patterns of the family, particularly those circumscribing the mother, predict the level of involvement fathers may have in their relationships with children including the quality of their relationships with the children’s 25 Der C0114 ax ai Emt "ads ”liar mothers as well as the marital statuses of both mothers and fathers (Hofferth, Pleck, Stueve, Bianchi, & Sayer, 2002). Emotion and Meaning-Making in the Parent-Child Relationship Researchers describe affect as an organizing feature of experience (Emde, 1989; Denham, 1998; Sroufe, 1996). Emde (1989, p. 44) writes, “Affect provides a sense of coherence over time and a continual orientation about what is familiar.” In the early primary caregiving relationship, the child cues the parent and depends on the consistent availability of his/her parent to respond in order to develop healthy functioning over time. Emde and Easterbrooks (1985, p. 86) refer to a parent’s emotional availability as an “adaptively patterned system.” Morris, Silk, Steinberg, Myers, and Robinson (2007) model the processes within the family environment that impact the meanings children make of their relationships, specifying pathways in the model among the child observation of family processes, parenting practices, and emotional climate of the family, all of which are informed by parent and child individual-level characteristics. Family interactions communicate abstract concepts like stability and coherence, informing children’s internal representations of relationships, and contributing to overall well-being (Cummings, Schermerhom, & Keller, 2008; Reiss, 1989). Stern (1989, p. 54) writes, “. . .relational patterns at the outset reside outside of any one individual and are constituted only during the interaction between two or more individuals. They are mutually derived even though subjectively constructed. They are both objective events and subjective experiences.” In this way, the development of the infant’s experience of relationships, and self in relationship, is intersubj ective, facilitated 26 by the t rerr. er: Liebem mepnn processr mhnom inhnnm Fuhhdt hlutual} suchrntv sisnima Parents 3 aCCUratel by the parent’s availability and modified by the child’s own internal traits, such as temperament and negativity (Martin, Clements, & Crnic, 2002; Zeanah, Boris, Bakshi, & Lieberman, 2000). The deve10pment of intersubj ectivity underscores the ways in which the primary parent and infant construct meaning in mutual regulatory and empathic processes (Tronick & Weinberg, 1999) and is a central function of the early parent-child relationship (Bomstein & Tamis-LeMonda, 2004). Meaning in the Mother-Child Relationship Conceptual models attending to mother-child interactions provide useful information to ground the current study within scientific theory, such as the Goodness-of- Fit Model (Thomas & Chess, 1977), the Affect Attunement Model (Stern, 1985), and the Mutual Regulation Model (Tronick, Cohen, & Shea, 1986). The common factor among such models is the emotional interdependence of primary caregiver and child and the significance of reciprocity (Osofsky & Thompson, 2000). The process through which parents and babies match mental states and interpret and respond to each others’ cues accurately in order to promote the child’s socioemotional development may be referred to as affect attunement, sensitive scaffolding, reciprocity, maternal sensitivity, or contingent interactions (Crockenberg & Leerkes, 2000; Osofsky & Thompson, 2000; Fonagy & Target, 2005; Stern, 1985). Disturbances in these interactions predict children’s decreased ability to self-regulate and increased proportion of negative affective behaviors (Seifer & Dickstein, 2000), inferring a process in which children internalize their family relationship patterns. Bomstein and Tamis-LeMonda (2004) describe among the primary functions of early mother-child interaction, the task of helping children both experience diverse 27 emotions a: da elmeCI systems is I". directed ant derelopmer sxrnbonc p to the curren labeling emr- interaction5_ Olk'n emotion FainsilbepK‘, socialization letiphy-Hcfl SPEL‘ll‘ICaHy‘ r Kupanoflj & l ”at? r "a! Dt’p The Dr Children OCCur touch (“'61.an- emotions and then learn to manage and regulate those emotions. This occurs through the development of dyadic processes inherent in the attachment and caregiving behavioral systems which are rooted in ethological theories suggesting that these processes are goal- directed and organized to promote development. Mothers scaffold their children’s development in multiple domains. For example, Fiese (1990) demonstrated that toddlers’ symbolic play was more complex during interactions with their mothers than when playing alone and that reciprocal forms of interaction, as opposed to intrusive or disengaged parenting behaviors, predicted the level of complexity in which toddlers engaged. Mothers’ socialization of socioemotional competencies is of particular relevance to the current study. Research demonstrates that parents may use direct strategies, such as labeling emotions, and indirect strategies, such as warmth and supportiveness during interactions, to support children’s developing understanding and management of their own emotional processes (Gottman, Fainsilber-Katz, & Hooven, 1996; Gottrnan, Fainsilber-Katz, & Hooven, 1997; Smith & Walden, 2001). Mothers’ emotion socialization strategies in specific predict socioemotional competencies in toddlerhood (Brophy-Herb, Schiffrnan, Bocknek, Dupuis, Fitzgerald et al., under review), and, more specifically, regulatory competencies (Cummings & Davies, 1996; F abes, Leonard, Kupanoff, & Martin, 2001). Maternal Depressive Symptoms The processes through which mothers’ depressive symptomology affect young children occur in each of the communicative domains—facial expressions, voice, and touch (Weinberg & Tronick, 1998). Parental depressive symptoms may impact mutual 28 in” ‘ ' T5331 ‘— Germ I \"‘ mt Li. li‘rdit T011". '1 .3 C44 *3 depre intera need: regulatory processes in parents and children (Weinberg, Olson, Beeghly, & Tronick, 2006). Also, maternal sensitivity may moderate the impact of depressive symptoms on early child developmental processes (Campbell, Brownell, Hungerford, Spieker, Mohan et al., 2004). In regard to interactive quality, depressed parents may be behaviorally withdrawn or intrusive or be able to interact positively with their babies (Weinberg & Tronick, 1998). Carter, Garrity-Rokous, Chazan-Cohen, Little, & Briggs-Gowan (2001) demonstrate associations between maternal depressive symptoms and less optimal mother-child interactions, but they also point to the importance of assessing additional individual and environmental risks in determining outcomes. Rogosch, Cicchetti, and Toth (2004) describe the multiple levels of system influences, including both maternal and paternal contributions, predicting the negative emotional environment surrounding depressed mothers and their toddlers. The heterogeneity of outcomes in parent-child interaction when parents are depressed suggests that additional theoretical models are needed to explain these differential outcomes and their impact on children. Meaning in the Father-Child Relationship As compared to mother-child interactions, researchers report greater heterogeneity in the levels of sensitivity fathers demonstrate in interactions with their young children, dependent on individual differences including intergenerational patterns, amount of responsibility for care, and indicators related to proximity (for review, see Lamb, 2002a). For example, Cowan, Cohn, Cowan, and Pearson (1996) demonstrated that fathers who had loving and secure relationships with their own parents were more sensitive and involved with their children than fathers who had less optimal experiences. Brophy-Herb, Gibbons, Omar, & Schiffinan (1999) demonstrated that low-income fathers 29 who lit ed their infra bet“ esp, f, its predict alone doe.- relationsh leh. 21'. Fa sum as "l‘ Cllllt'lren'S Research ' reflex-am.c behaVlorg Similar a5: (e.g., M 51c Father‘s p Compared and regUla Christians. fathers as t In 1 particular 1 model (Laz- who lived with their children were more likely to show sensitivity in interactions with their infants than were non-residential fathers. Proximal indicators of the relationship between father and child are likely particularly important in defining the relationship and its predictive power (Hofferth et al., 2002), but researchers also indicate that proximity alone does not consistently predict positive outcomes; the quality and continuity of the relationship between father and child has important implications for child development (Lamb, 2002b). Fathers may be characterized within historically and socially constructed roles such as “breadwinner” and “disciplinarian,” and they may be integral in socializing children’s perceptions of gender roles (Fitzgerald, Mann, & Barratt, 1999; Lamb, 1987). Research demonstrates patterns of interaction specific to fathering that have particular relevance to early socioemotional development. As much as mothers’ nurturing and behaviors are clearly linked to early socioemotional development, researchers make similar associations between fathers’ play behaviors and children’s early deve10pment (e. g., MacDonald & Parke, 1986; Paquette, 2004; Roggrnan, Boyce, & Cook, 2001). Father’s play is likely especially emotionally evocative through more physical play, compared to mothers’ typically verbal and didactic play, predicting patterns of arousal and regulatory competencies in young children (Roberts, 1998; Roggrnan, Boyce, Cook, Christiansen, & Jones, 2004). To this end, Roggrnan, Boyce, and Cook (2001) refer to fathers as the primary playmate in contrast to mothers as the primary caregiver. In better understanding the pathways between fathering and child outcomes, particular theoretical models have been advanced. An oft-cited example is a tripartite model (Lamb, Pleck, Charnov, & Levine, 1985) which suggests that father involvement 30 includes pater paternal are e resportstbn’rrt Roggman (2U derelopme .1: relationships ; “I Q mnifl 111\ 0l\ C father-child 0 nature of chi h The ct Edwards and IllOml‘is Of 39‘ ”16th Chili Thoug Olim‘ams und preschOOlers E includes paternal engagement (e. g., interaction with the child such as play or teaching); paternal accessibility (e.g., temporal and proximal indicators); and paternal responsibility (e. g., making arrangements for care). Cabrera, Fitzgerald, Bradley, and Roggrnan (2007) posited a heuristic model, consistent with theories based on developmental systems perspectives, which incorporates mediated and moderated relationships among predictors of father involvement as well as the relationships between father involvement and child outcomes. Overall, researchers describe a need to explore father—child outcomes based on theories which emphasize the systemic and transactional nature of children’s development (Roggrnan, Fitzgerald, Bradley, & Raikes, 2002). Meaning-Making and Toddlers ’ Development The current study adopts the definition used by Brownell and KOpp (2007) and Edwards and Liu (2002), defining toddlerhood flexibly as the period between 12 and 36 months of age. The sample in the current study is comprised of parents and their children when the children are 14 to 36 months of age. Though there is a wealth of scholarship describing socioemotional development of infants under 1 year old and their families (e. g., Moore & Calkins, 2004) and preschoolers above age 3 years and their families (e.g., Warren & Stifter, 2008), there is a dearth of research describing socioemotional development of toddlers and their families with such an approach, with notable exceptions (e.g., Adarnson, Bakeman, & Deckner, 2004; Easterbrooks & Goldberg, 1984; Weinberg, Beeghly, Olson, & Tronick, under review). Toddlerhood is a unique developmental period, as described, and requires its own focused inquiry. 31 carsgi 21h)" )- others i 1 Edit at becom, engage l Adam Supper " I I- Brownell and Kopp (2007, p. 2) describe the period from 1 to 3 years of age as a “watershed in development” based on the many developmental transitions children make during this brief period of life. The period of toddlerhood is characterized by increased mastery of the physical and social world, the growth of conscious self-hood, greater understanding of self-other relations, increased participation in (and understanding of) system-level rules and symbolic practices, and more diverse strategies to self-regulate in accordance with this developing awareness of the social context (Brownell & Kopp, 2007). Among the developmental tasks of this period is the need to become connected to others and become a member of social groups, including the broader family system (Edwards & Liu, 2002). Through these developmental processes, toddlers increasingly become active participants in family systems with better mastery of symbol-infused engagement and mental representations of relationships based on repeated experiences (Adarnson, Bakeman, & Deckner, 2004; Crockenberg & Leerkes, 2000), skills that likely support the symbolic processes inherent in family codes, patterns, and transactions. Toddlers ’ Development Within the Parent-Child Relationship This time period is known to be particularly complex in regard to children’s emotional development within the parent-child relationship and likely requires increasingly complex versions of intimacy defined less and less by the physical care provided in infancy (Lieberman & Slade, 2000). In the development of the attachment as well as caregiving behavioral systems, toddlers and parents are increasingly forming reciprocal relationships, and children are increasingly participating in attenuating the balance of proximity-seeking and exploratory behaviors (Crockenberg & Leerkes, 2000). Toddlerhood marks a period of life when there are major transitions in several other areas 32 Oi‘deVC’lOPT are lc‘lmmE bah seek F the tin-Pom selthood. C‘ in c« Bottle} '5 >< Much resea. Robertson :1 significant i separation ti cognitive. pl Hinshaw-Fu studies relat: institutionali ai‘ailability t (Sroufe 8; \\ pSYCllOlCrgiCa of development which support this rapid development in socioemotional skills: children are learning to walk and talk and are able to use new skills of symbol-use and mobility to both seek proximity as well as greater independence. F raiberg (1959) described among the important parenting tasks during toddlerhood as guiding children in acquiring selfliood, conscience, and self-control. In considering the effects of separation between primary caregiver and child, John Bowlby’s seminal work (1969/ 1973/ l 980) on the attachment relationship is most salient. Much research in addition to Bowlby’s work, such as research by Spitz (1945; 1946) and Robertson and Robertson (Robertson & Robertson, 1971; 1989), has documented the significant impact on the development of young children who experience prolonged separation from their primary caregivers, including the presence of delays in the cognitive, physical, and socioemotional domains of development (for review, see Hinshaw-Fuselier, Heller, Patton, Robinson, & Boris, 2007). The majority of these studies relate to the physical separation of parent and child, such as in the context of child institutionalization. However, research also demonstrates that a mother’s psychological availability to her child may be even more ameliorative than her physical presence, (Sroufe & Waters, 1977; van den Boom, 2004) suggesting significant implications of psychological absence, even if physically present. Because of the aforementioned developmental trajectories for parents and their toddlers, separation or ambiguity in the parenting role likely has particular implications for children at this age. Weinberg, Beeghly, Olson, and Tronick (under review) tested an adaptation of the still-face paradigm with toddlers. Their study findings demonstrate that toddlers, as compared to infants, have greater representational capacities and more 33 41f. lfillDC‘l COEIlll 5mm; nCrrT ‘rfl‘ 1 :1 Chi: complex regulatory skills that determine their behavior when faced with a situation that violates their assumptions about their relationships with their mothers. Furthermore, toddlers are likely to search for reasons for the confusing experience, reflecting deeper meaning-making skills that support coping when stressed. Finally, the authors found that toddlers were more likely than infants to maintain an organized state when mothers temporarily withdrew regulatory support, suggesting the importance of history and continuity in the relationship overall in predicting toddlers’ regulatory competencies. Emotion Regulation as a Reflection of Family Boundaries Of specific interest in the current study were the ways in which ambiguities in family boundaries may predict the development of emotion regulation. Thompson (1994) defined emotion regulation as individuals’ attempts to monitor, evaluate, and modify their emotional reactions, as part of a larger process around developing goal-directed behaviors. Crockenberg and Leerkes (2000, p. 63) write, “. . .reactivity is an internal characteristic of infants that together with exogenous factors contributes to differences in emotion regulation.” As infants develop into social toddlers, they become increasingly able to communicate their emotional states to parents, eliciting responsiveness from parents, and developing an understanding of the parenting role in modulating affect, learning that arousal need not lead to disorganization (Sroufe, 1996). This is the process in normative dyads. When relationships are disrupted, or the parent is not available for normative scaffolding of regulatory competencies, children’s developmental processes will likely be similarly disrupted. As described, the early parenting relationships are important contexts for children’s self-regulatory development, and, further, there may be different role-typical 34 to :he robe sn;:e- mthe Chess 16211;): from dano parenting attributes for mothers as compared to fathers which contribute to children’s early development. Though the processes in the mother-child relationships as compared to the father-child relationship may be different in some ways, they both require a parent to be continuously engaged with his/her child to develop a history and repertoire of shared behaviors. Furthermore child characteristics, temperament in particular, contribute to the aforementioned potential for goodness-of-fit between parent and child (Thomas & Chess, 1977). Calkins, Dedmon, Gill, Lomax and Johnson (2002) demonstrated that temperamental characteristics predicted the kinds of regulatory strategies infants used from a cross-sectional perspective, and Hill-Soderlund and Braungart-Rieker (2008) demonstrated that temperamental characteristics measured in infancy predicted effortful control strategies five years later. The ability to regulate emotions helps individuals adapt to their environments and experience competence in social interactions (Thompson & Goodvin, 2007). Emotions, even as early as infancy, provide individuals with critical information about the world around them, and are informed by and inform, or contribute to the construction of meaning, in social interactions (Denham, 1998; Witherington, Campos, & Hertenstein, 2004). Thus, this relates to the importance of meaning-making that is central to boundary ambiguity theory through the fundamental principle of symbolic-interactionism. From a symbolic-interactionist perspective, human beings respond to stimuli based on the meanings they assign to those stimuli and this process generally occurs in the context of social interaction. The idea that young children develop behaviors based on their perceptions of their social interactions (and the meanings they assign to those perceptions) is well-studied in 35 Oil”: CT £;l)r .\l;r.'. incrcd .Ad1er ll] IhlS behax: other areas of research, such as in the related scholarships of attachment (e.g., Ackerrnan & Dozier, 2005; Kochanska, 1995; Kochanska 2001) and infant mental health (e.g., Murray, 1991; Zeanah & Anders, 1987). Generally, infant mental health theorists, referencing attachment theory (Bowlby, 1969/1973/1980), describe the ways in which parents’ own internal representations impact caregiving. Children build internal working models of relationships based on early caregiving relationships which impacts firture behaviors in relationships (Crockenberg & Leerkes, 2000). Research utilizing strange situation assessments (for review, see Calkins, 2004) demonstrates that the development of young children’s regulatory strategies depends on their expectations, based on historical caregiving experiences, of the availability of their parents to provide and scaffold external regulatory strategies. This process is particularly relevant during toddlerhood: social referencing develops by the end of the first year, supporting reciprocity in this part of the caregiving relationship, and toddlers are increasing making meaning out of continuous relational experiences (Moore, 2007). Adverse psychological outcomes are often associated with a lack of parental availability in this regard. While structural family therapists emphasize the importance of consistent behavioral manifestations of an organized family system, child development scholars emphasize the importance of regular patterns of interaction over time in supporting a young child’s healthy socioemotional development (Sroufe, 1989). During these early social exchanges, particularly within the cumulative repetition of these exchanges, young children develop eflectance, the sense of what can and cannot be accomplished. In the absence of parental availability, or inconsistency in exchanges, children may become helpless (Tronick, Ricks, & Cohn, 1982). 36 Ct'fi‘t 6C \Vnile outcor pro\ic léidilt la - thk‘hlbt Summary In summary, children’s regulatory competencies are impacted by the meaning derived from social interactions, particularly within the early parenting relationships. While it is known that children’s regulation is impact by parental depressive symptoms as well as parental absence, new methodologies are needed to increasingly understand the direct pathways between the inconsistencies in these relationships and children’s outcomes and inform well-designed interventions. Boundary ambiguity theory likely provides a useful theoretical framework to study outcomes for children when these relationships are inconsistent, adopting a systemic approach and emphasizing the role of holistic, organizing features of families. 37 631'“ :10” e» ‘ I! > "”r CHAPTER 3: METHOD Procedures The final samples reflected 2,632 mother-child dyads for whom live observation data and self- report measures were utilized; 501 father-child dyads for whom live observation data and self-report measures were employed; and 1,477 father-child dyads for whom only mother-reported involvement was available. Further information about sample characteristics are presented with study results, and synopses of sample demographic characteristics are provided in Tables 2 and 3. Main Study Procedures The current study reflects a secondary analysis of data from the National Early Head Start Research and Evaluation Study (EHSRE; Love, Kisker, Ross, Raikes, Constantine et al., 2005). In the EHSRE, data were collected from 3,001 focus children and their primary caregivers, the majority of whom (99%) were mothers. Data were collected near the children’s 14-month, 24—month, and 36-month birthdays. Key study instruments included a parent interview, observations of the parent and child in play, and the Bayley Mental Development Index (Bayley, 1993) with behavioral rating scales. Families were assessed at later time points as well for the main evaluation study, but only the data collected during toddlerhood will be used in the current study. Children and their families were enrolled in the main evaluation study based on their eligibility for Early Head Start services; families (excluding 24 families who chose to leave the study prior to random assignment) were randomly assigned to the program (receiving Early Head Start programming; n=l,503) or comparison groups (n= 1,474). Families in the comparison group were free to access other services in their communities. 38 Fathers Sub—Study Procedures Of the 17 research sites in the EHSRE, 12 sites also participated in a study of fathers (Boiler, Bradley, Cabrera, Raikes, Pan, Shears, & Roggman, 2006). Fathers were identified by the children's mothers, who were participants in the EHSRE. Mothers were asked during the EHS interviews about the involvement of the child’s biological father and. If the biological father did not live with the mother and child, mothers were asked to indicate if there was a man who was “like a father” to the child and, if so, what his level of involvement was with the child. Mothers were informed about the fathers’ sub-study, and their assistance was requested to contact these men. Fathers were contacted, and participating fathers of focus children were interviewed at the children’s 24-month and 36-month birthday-related assessments at the 12 sites. The interviews included many of the same questions asked of mothers in the main study, and the same observational procedures were repeated, as well, to assess father-child interaction. Study Measures The main study included a parent interview with self-report data and an observation of mothers and children playing together in semi-structured tasks. The same parent interview and observation of parent-child interaction was repeated with fathers and children. Also, the Bayley Mental Development Index (Bayley, 1993) was administered to children by trained data collectors, and the children’s regulatory behavior during the Bayley screening was coded. Measures relevant to this study are described below with reliability information and mean scores for the current sample of mothers (N=2,960) as well as for the current sample of fathers (N=501). 39 Cfllt‘ili 1 ‘t. 130x 0 meua dimen "You. Psychological Absence Psychological absence, conceptualized as a parent’s physical presence with emotional unavailability, was operationalized as parental distress in the parenting role, lack of empathy, and parental detachment. These measures were administered to fathers and mothers. Parenting Distress Parenting distress is measured by total scores on the Parenting Distress subscale of the Parenting Stress Index-Short F orm (Abidin, 1995). This subscale measures the level of distress a parent feels in his/her role as parent. Cronbach’s alpha for internal consistency is reported to range from .81 to .90. This measure was included to represent the ways in which a parent might be psychologically absent within an affective dimension. Items include, “You feel trapped by your responsibilities as a parent,” and “You often have the feeling that you cannot handle things very well.” In the current sample, mothers had a mean score of 25.21 (SD=8.85,' min=12.00; max=58.00) and fathers had a mean score of 19.1 1 (SD=5.49; min=12.00; max=43.00). Parental Lack of Empathy Parental lack of empathy was measured by the total scores on the Lack of Empathy subscale of the Adult-Adolescent Parenting Inventory (AAPI) (Bavolek, 1984). Bavolek (1984, p. 6) defines empathy as, “awareness of a child's needs [that] entails the ability of the parent to understand the condition or state of mind of the child without actually experiencing the feelings of the child.” Therefore, this measure was included to represent the ways in which a parent might be psychologically absent from his/her role as a parent from a cognitive perspective. This measure has a test-rest reliability alpha 40 dual-c- llldlcg «51%., coefficient of .76 (Bavolek, 1984). In the current sample, mothers had a mean score of 1.98 (SD=.84; min=.92; max=5.00) and fathers had a mean score of 3.06 (SD=.67; min=2.00; max=5 .00). Parental Detachment Parental detachment was coded from a free play task (N ICHD Early Child Care Research Network, 1997). During the ten-minute play task, the parent was provided three cloth bags of toys and invited to play with his/her child however he/she wished as long as he/she used the bags in a specified order. The vague directions of this task have the intention of eliciting naturally occurring behaviors (Love et al., 2005). The instructions given to parents for the play task were as follows: “Our last activity will take about 10 minutes. We would like you and (CHILD) to spend this time with the toys in these three bags. During this activity, you may play with (CHILD) if you like. Just to remind you, please face front and try to stay on the mat. Please start with Bag #1, move on to Bag #2, and finish with Bag #3. Do you have any questions?” Parental detachment includes behaviors such as inattention to the child and a seeming lack of engagement and awareness; mothers were rated on a 7-point scale, ranging from 1 to 7, with higher scores indicating greater detachment. In the current sample, mothers had a mean score of 1.24 (SD=.49; min=1.00; max=6.00) and fathers had a mean score of 1.14 (SD=.23; min=1.00; max=3.00). This measure was included to demonstrate the ways in which a parent might be psychologically absent from his/her role as a parent as far as behavioral manifestations of parental emotional disconnection. It is important to note that detachment, suggesting the behavioral implication of psychological absence, is differentiated from negativity through which a parent might be present, though in a negative way. 41 Flirt”? CCSCTl 3 meet Parent-Child Interaction Parental Supportiveness Parental supportiveness was coded from the same semi-structured play task described above. Parental supportiveness is a composite measure of three highly correlated subscales: parental sensitivity, cognitive stimulation, and positive regard. Sample participants were rated on a 7-point scale on each subscale, and an average score across these subscales was computed for a final score. In the current sample, mothers had a mean score on the supportiveness composite of 3.90 (SD=.78; min=1.00; max=6.33) and fathers had a mean score of 4.1 l (SD=.55; min=2.00; max=6.33). Child Engagement Child engagement, again, was coded from the same play task described above; children were coded in play with their mothers and separately with their fathers. The subscale of child engagement measures the degree to which a child shows, initiates, and/or maintains contact with his/her parent during the play task. Children were rated on relevant behaviors from a scale of l to 7, with higher scores indicating greater engagement. In the current sample of mother-child dyads, children had a mean score of 4.70 (SD=.82; min=1.00; max=7.00). In the current sample of father-child dyads, children had a mean score of 5.01 (SD=.57; min=3.00; max=7.00). Depressive Symptoms The Center for Epidemiological Studies-Depressive Symptoms (CES-D) short form measures symptoms of depressive symptoms and includes 12 items taken from the full, 20—item CES-D scale (Radloff, 1977; Ross, Mirowsky, & Huber, 1983). The scale does not diagnosis depressive symptoms, but it does discriminate between depressed 42 with it Chlldre Emmic and 16st respondents and non-depressed respondents. Items include restless sleep, lack of energy, and poor appetite. Cronbach’s alpha for the short form is .88. Scores on the scale range from 0 to 36. In the current sample, mothers had a mean score of 7.77 (SD=6.36; min=.00; max=36.00) and fathers had a mean score of 4.95 (SD=5. 1 9; min=.00; max=34.00). Child Outcome Measures The scales utilized in the current study are defined by the behavior rating scales of the Bayley Scales of Infant Development, 2nd ed. (Bayley, 1993). These scales measure a child’s behavior during the Bayley Mental Development Index (BMDI) assessment. The BMDI evaluates sensory/perceptual acuities, discriminations, and response; acquisition of object constancy; memory learning and problem solving; vocalization and beginning of verbal communication; basis of abstract thinking; habituation; mental mapping; complex language; and mathematical concept formation. The test administrator also coded the child’s behavior during test administration, and these results yielded two summary subscales: orientation and engagement and emotion regulation. For both scales, higher scores indicate greater competence. Orientation and Engagement The orientation and engagement rating subscale measures a child’s cooperation with the interviewer, positive affect, and interest in the test materials. In the current study, children had a mean score on this scale of 3.79 (SD=.61; min=1.20; max=5.00). Emotion Regulation The emotion regulation rating subscale measures a child’s ability to change tasks and test materials, negative affect, and frustration with tasks during the assessment. Both 43 rela’. from ln t}: I'CCOi grant pictui 199% ram, inelud C until. risk int SO'ClOer Panama scales have a minimum of 1.00 and maximum of 5.00. In the total sample (N=2,632), children had a mean score on this scale of 3.92 (SD=.63; min=1 .00; max=5.00). C ovariate Measures Household Composition Mothers reported who lived in their homes with them. The research question in the current study focused on children’s socioemotional competence within primary relationships. Therefore, it might be expected that the presence of other adults in the home could impact these outcomes, especially when a parent is psychologically absent. In the context of maternal psychological absence, this variable could be meaningfully recoded so that married two-parent households=1, other adults in the home=2, and single parents=3; this recoding scheme is replicated from Chazan-Cohen, Ayoub, Pan, Roggrnan, Raikes et al., 2007. In the current study, 674 women (26%) reported they were married, 1,038 women (3 9%) reported they lived with other adults (including children’s grandparents), and 915 women (35%) reported they lived with no other adults. The picture looked quite different among fathers. The majority of fathers in the current study (99%) reported the presence of a spouse or partner in their homes, indicating little to no variance in this construct among participating fathers. Therefore, this variable is only included in models pertaining to mothers. Cumulative Risk The cumulative risk index utilized in the current study is a composite variable of risk indicators shown to be significant in preliminary analyses in predicting children’s socioemotional competence. These indicators of risk, for mothers and fathers, were parental education (no high school diploma), age at birth of first child (teen parenthood), 44 lend‘ Inclu unemployment, single parenthood, and receipt of welfare. These variables were dummy coded into dichotomous variables and summed. Therefore, this variable is an index of the number of risks present. Cumulative risk has been shown to be a powerful predictor of outcomes in general (Sameroff & Fiese, 2000a) and within this dataset in particular (e. g., Raikes, Pan, Luze, Tamis-LeMonda, Brooks-Gunn et al., 2006). In the current study, mothers had a mean of 2.65 risks (SD=1.15; min=.00; max=5.00), with more than half (55%) of the sample indicating the presence of 3 or more risks. Fathers overall demonstrate less risk on this index with a mean of 1.09 risks (SD=.98; min=.00; max=4.00), and 91% of the sample reporting 2 risks or fewer. Program Status The program status variable indicates whether or not a family received the Early Head Start intervention. In the EHSRE, Early Head Start services have been shown to positively impact child- and parent-level outcomes (Love et al., 2005). Fifty-one percent of respondents were assigned to the comparison group while 49% of respondents were assigned to the comparison group among mother-child dyads as well as father-child dyads. Child Emotionality Children’s emotionality, a temperamental characteristic, was assessed at the 14 month birth-related assessment in the EHSRE study, using the Emotionality subscale of the Emotionality, Activity, Sociability, and Impulsivity (EASI) Temperament Survey (Buss & Plomin, 1984). The complete EASI is a parent-report measure about her child’s tendency towards these four dimensions. Sample items on the emotionality subscale include, “He/She often fusses and cries,” and “He/she cries easily.” In the current sample 45 0161": min: overall (N=2,632), children had a mean score on this subscale of 2.97 (SD=.94; min=1.00; max=5.00). Fathers’ Continuous Physical Presence/Absence This variable is created from the information mothers provided at each time point (14-, 24-, and 36-month birthday-related assessments) describing fathers’ involvement with their children. At each of the three timepoints, the biological mother respondent was asked: 1) whether or not the biological father lived with the child (yes or no) and 2) how much contact the biological had with the child (every day or almost every day, a few times a week, a few times a month, about once a month, less often that than or never). See Table 6 for a description of the sample. Based on these two variables, a binary code was created for a new variable specifying whether or the biological father was involved with the child at each timepoint A biological father was considered involved if he lived with the child or spent at least a few times a month with the child. Then, a dichotomous variable was created describing whether or not the child’s biological father was involved at all timepoints (yes or no), representing continuous presence throughout toddlerhood (14-36 months). Missing Data In all longitudinal, community-based samples, attrition is a potential concern, and Acock (2005) suggests that it is an issue particularly relevant to family research, with as much as 50% of respondents often dropping from longitudinal studies. Other studies (e. g., Chazan-Cohen, Ayoub, Pan, Roggman, Raikes et al., 2007) have shown that, in the EHSRE dataset, participants who remained in the study demonstrated fewer risk factors than those who dropped out. Therefore, missing data were carefully and thoroughly 46 was reviewed to explore the mechanisms related to missingness (see Appendix B). Data may be missing completely at random (MCAR), missing at random (MAR), or non-ignorable (N I). In research related to children and families, data are not likely to be MCAR but rather MAR or NI (Acock, 2005). Data may be considered missing at random if the pattern of missingness is explicable in some way but is not related to a participant’s true status on a given variable for which data is missing (Acock, 2005; Musil, Warner, Yobas, & Jones, 2002). Such conditional randomness may be based on common mechanism variables like demographics, including those contained in the cumulative risk index in this study (Acock, 2005). Based on attrition analyses (see Appendix B), the missingness was determined to be MAR, and analyses were based on imputed data. Missing data were imputed prior to analyses in SPSS 17.0 using the Expectation Maximization (EM) algorithm (Dempster, Laird, & Rubin, 1977). Predictor variables were imputed separately from outcome variables, and data were used from respondents’ scores at other time points and on related measures to inform imputation. The EM method implements a maximum likelihood (ML) approach to iteratively impute missing values by using two steps in each iteration: an Expectation-step and a Maximization-step. ML approaches are ideal for preparing data for latent variable models and are appropriate when data is determined to be MAR (Musil, Warner, Yobas, & Jones, 2002; Schafer & Graham, 2002). Another potential approach to impute missing data would be to use Full Information Maximum Likelihood (FIML) procedures concurrent with SEM analyses. However, scholars agree that F IML and EM should yield similar results and have largely trivial differences in bias and efficiency (Enders, 2006; Schafer & Graham, 2002). 47 O'- ‘4‘. iw‘ ti. .‘r‘ Eulk ‘ “.l ‘v uni—.1 .: l 5nUk IC- D: A ' N '5».- MRI: bent-9,. SlZe. Wt 130.“) Side“ Effect Sizes Effect sizes are reported via the standardized path coefficients (see Figure 1) as well as the squared multiple correlations of the latent factors. Effect sizes communicate the strength of measured relationships in a study and report findings in a way that can be interpreted broadly (Hedges, 2008). Standardized solution effect sizes are important to guide the interpretation of statistical relationships and superior to significance testing alone (Abelson, 1995). It is a complex matter to determine appropriate effect size guidelines as different endogenous variables in different areas of research are known to be more or less difficult to predict (Kline, 2005). Cohen (1988) offers the following guidelines for use in the social sciences to interpret the effect sizes of standardized path coefficients: values equaling .10 or less may indicate a “small” effect, values around .30 may indicate a “moderate” effect, and values equaling .50 or more may indicate a “large” effect. These guidelines are broad, and researchers have recently suggested using more specific empirical benchmarks relevant to the particular measures, target population, and interventions addressed (Hill, Bloom, Black, & Lipsey, 2008). Other researchers exploring Early Head start program impacts on a range of outcomes found effect sizes ranging from .10 to .20 on measures of child behavior and parenting (Love, Kisker, Ross, Raikes, Constantine et al., 2005). A meta-analysis of 33 studies exploring the relationship between maternal depressive symptoms and child behavior problems found a mean effect size, weighted based on sample sizes, of .29 (Beck, 1999). These data should be considered in interpreting the effect sizes presented for the current study. 48 ‘h 71, :‘h\al l .lif Were General Plan for Analyses The main modality for all analyses in this dissertation was structural equation modeling (SEM): SEM will be utilized in cross—sectional models as well as to specify grth curves with LISREL 8.80. SEM is particularly useful to test the current hypotheses because it allows a researcher to measure unobserved variables with one or more manifest indicators. This attribute partially responds to the issue discussed earlier: it is difficult to directly observe (e. g., interview young children regarding) boundary ambiguity in the early parent-child relationship. Furthermore, SEM accounts for error and construct-irrelevant variance (Boller, 1989), which is valuable in most community-based family research in which a given outcome is likely influenced by a multitude of nested variables that cannot be parsimoniously accounted for in a single model. As a note, in all studies, the paths of factors indicated by a single manifest variable was constrained to 1.00 and its error variance constrained to .00 (Kelloway, 1998). Furthermore, covariates were freed to predict all endogenous latent factors throughout each model. Fit indices are reported to interpret the results in the current study. Raykov, Tomer, and Nesselroade (1991) indicate that determining good fit in SEM is not a simple quest and should not be based on single fit index. Rather, the authors suggest that researchers report the following: chi-square index, degrees of freedom, and corresponding p-value; the goodness-of-fit index (GFI) if using LISREL; and the standardized root mean square residual (SRMR). This set of indices are reported and considered together, therefore, based on these guidelines; however, when reporting the two-group models, the GFI will be replaced with the Comparative Fit Index (CFI), as the GFI is not provided for the global fit of a multi-group model in LISREL. Furthermore, 49 . v., the root mean square error of approximation is a useful index of fit because it is independent of sample size (Bollen, 1989), and this statistic will be presented as well. Hu and Bentler (1999) indicate, as rules of thumb, that the GFI should be .90 or above, the SRMR should be.08 or below, and the RMSEA should be .06 or below. While these cut- off criteria do not vary widely in research for the GF I and SRMR, there is some variation in cut-off criteria for the RMSEA, and Hu and Bentler’s .06 cut-off is among the more conservative guidelines. Other researchers (e. g., Cummings, Keller, & Davies, 2005; Michalik, Eisenberg, Spinrad, Ladd, Thompson et al., 2007) cite Browne and Cudeck (1993) who advise a cut-off of .08 to determine reasonable fit. Because the current studies are exploratory in nature, and the results and accompanying discussion points are framed in this way, the more liberal cut-off of .08 will be applied if the other fit indices represent good fit. Power and Sample Size Power is the probability that one may reject a null hypothesis when it is false and therefore should be rejected (Cohen, 1988). Ascertaining power in structural equation modeling (SEM) is a complicated matter given that one is testing a complete model with multiple parameters describing regression coefficients as well as error variance, and guidelines which require single-parameter testing may prove unwieldy. As a result, general rules of thumb have been advanced suggesting ad hoc guidelines such as 5 to 10 cases per parameter or sample sizes with a lower bound of 100-200 overall (see for example Boomsma, 1983; Gebring and Anderson, 1985). However, such ad hoc approaches are often criticized as non-specific, and researchers suggest that the necessary number of observations should be determined by model-specific criteria (Tanaka, 1987 ). 50 . .L‘, Jle.Z: haven . L .t‘- tes ‘3'“): .s1. McCallum, Brown, and Sugawara (1996) suggest calculating the lower bound of sample size parameters with the following criteria: the target RMSEA fit indices for the alternative hypothesis and null hypothesis, the significance (or) level, the model’s degrees of freedom, and the target power statistic. Power is an estimate of probability and ranges from 0 to 1, with .80 generally accepted as an acceptable cut-off; as such, this was utilized as the guideline in the current study. Also, the RMSEA for the alternative hypothesis was set at .05 while the RMSEA for the null hypothesis was set at .08 in order to test the likelihood that the alternative hypothesis represented a close fit with the data while the null hypothesis represented a less ideal fit. Degrees of freedom specific to each model were tested. NIESEM (Dudgeon, 2003) was the software utilized to run these calculations based on the given parameters, presenting the necessary lower bound of observations for each model based on a minimum power statistic of .80. Actual power is implied through several fit indices presented for each model, as described earlier. 51 Dene E; i’ a. Cu .«l - \.056 R601 \ ¥ C0n: CHAPTER 4: PLAN FOR PRESENTATION OF STUDY FINDINGS The current presentation of analyses and results is organized into three chapters: Maternal Psychological Absence and Toddlers’ Socioemotional Outcomes (Chapter 5), Paternal Psychological Absence and Toddlers’ Socioemotional Outcomes (Chapter 6), and Paternal Psychological Presence and Toddlers’ Longitudinal Regulatory Development (Chapter 7). Each chapter relates to a different subsample from the EHSRE, and detailed sample descriptions will be included in each chapter. A brief discussion will close each chapter, and then, in Chapter 8, broader themes related to the complete set of analyses will be discussed. General Summary of Chapters Two sets of analyses are included in Chapter 5, Maternal Psychological Absence and Toddlers’ Socioemotional Outcomes: one set of analyses will test the construct of psychological absence within the sample (N=2,632) of mother-child dyads and then test the relationship over time between psychological absence and todders’ regulatory development. In Chapter 6, Paternal Psychological Absence and Toddlers’ Socioemotional Outcomes, analyses are presented regarding the relationships between psychological absence and toddlers’ socioemotional outcomes among the sample of father-child dyads (N=501) from whom data were collected by direct observation and self-report interviews by fathers. In Chapter 7, Paternal Psychological Presence and Toddlers’ Longitudinal Regulatory Development, analyses are presented testing the relationship between fathers’ continuous presence across toddlerhood and children’s longitudinal regulatory 52 (11515.10 Rifldi Oncul 9LR01 Elm-bf; 1 1'16 m development, based on the sample of families (N=l,477) from whom data were collected regarding fathers’ continuous involvement through mothers’ report. In each chapter, analyses pertain to the complete reported sample for that chapter. Some analyses did not support the hypotheses, which will be described. Furthermore, post-hoe analyses involving model fit around child gender and racial/ethnic group provided a basis for subgroup comparison. Therefore, each set of analyses were repeated in multigroup SEM models to compare the results between subgroups of each sample specific to each chapter based on racial/ethnic groups and child gender. Though these between-group differences are not central to the main hypotheses, the body of relevant scholarship suggests there may be important differences which should be addressed. Between-Group Comparisons Comparisons Based on Racial/Ethnic Groups Previous research demonstrates that the construction of family roles differ based on cultural norms (e.g. Barbarin, 1983; Campos, 2008; McAdoo, 1993; Raikes, Summers, & Roggman, 2005) though there is not specific guidance related to culture and boundary ambiguity at present. Therefore, the current hypotheses are tested to compare the fit of the models between racial/ethnic groups. Comparisons Based on Child Gender Further, research demonstrates that boys and girls often differ in regard to early regulatory development, with girls typically presenting as more adaptive than boys (e.g., Kochanska, Coy, & Murray, 2001; Kochanska, Murray, Jacques, Koenig, & Vandegeest, 1996), and further that parenting processes may play a role in these differential outcomes (e.g., Calkins & Johnson, 1998). However, other research indicates that there may be 53 some heterogeneity in outcomes in the body of research related to gender differences; boys are not always less adaptive, or gender differences are not apparent, particularly over time, contrasting the findings that girls are typically more adaptive (e. g., Mayes & Carter, 1990; Stifter & Spinrad, 2002). Such heterogeneity indicates that a two-group model would allow a detailed comparison of the current hypotheses as they may differ based on child gender. In addition, research indicates that father presence and resulting outcomes may differ based on child gender. Fathers may be more likely to be present with older sons (Amato & Keith, 1991; Wood & Repetti, 2004), and research demonstrates that boys at varying ages may be particularly vulnerable when fathers are not present (F eldman, 2003; Pederson, Rubenstein, & Yarrow, 1979). However, this area of research is less well- studied among very young children, and thus the current analyses will contribute to the literature by exploring differences based on child gender. 54 CHAPTER 5: MATERNAL PSYCHOLOGICAL ABSENCE AND TODDLERS’ SOCIOEMOTIONAL OUTCOMES Participants were excluded from the current study based on a few criteria. To begin, 20 children died after enrolling in the study], and these families were not included in the analyses. Another four families left the study before random assignment to the program or comparison groups, and these families were excluded from the current study. After these exclusions, 99% (n=2960) of the primary caregiver respondents identified themselves as the biological mothers to focus children while less than 1% (n=17) identified themselves as biological fathers (n=9) and grandparents and alternate caregivers (n=8). Because the current study focuses on family organization, homogeneity in the primary caregiver role clarifies the interpretation of results, and therefore, only families in which the primary caregiver respondents were biological mothers were included. Finally, there were no data reported for 328 families at any of the three time points on any of the key instruments used in the current study, and these families were also excluded from the current analyses. The final sample size, then, is 2,632 families. Sample Description The sample in this study included 2,632 mother-toddler dyads. F ifty-one percent (n=1,333) of these families were randomly assigned to receive Early Head Start services while 49% (n=1,298) were assigned to the comparison group and free to access other services in their communities. l . . . . . These 20 cases included mrscarrrages in cases when mothers entered the study while pregnant. 55 Child Characteristics F ifty-one percent of these children were boys (n=1 ,3 34) and 49% were girls (n=1,298). Children were, on average, 15.02 months old (SD=1.48; min=11.10, max=22.24) at the time of the 14-month birthday-related assessment; 25.13 months old (SD=1.56; min=20.70, max=32.65) at the time of the 24-month birthday-related assessment; and 37.48 months old (SD=1.96; min=33.24, max=54.10) at the time of the 36-month birthday-related assessment. Mother Characteristics The main study inclusion criterion was eligibility for Early Head Start enrollment, suggesting sociodemographic risk. Based on the cumulative risk index created for the current study which summed the number of risks mothers reported at study enrollment, more than half the sample (57%; n=1,344) reported the presence of 3 or more risks. Mothers in the current sample (N=2,632) were, on average, 22.65 years old (SD=5.64; min=12, max=46) at the time of random assignment in the study. Thirty-nine percent (n=1,006) of the sample reported they were less than 20 years old at the birth of their children. Seventy-six percent (n=1 ,913) of mothers had a high school education/GED or less; nearly half the sample (n=1,186; 47%) did not complete high school at the time of random assignment. Also at the time of random assignment, 54% (n=1 ,3 83) of mothers reported they were neither employed nor in school, and 34% (n=729) reported receipt of welfare at this time. At the time of the child’s 14-month birthday-related assessment, 26% (n=674) of mothers reported living with husbands, 35% (n=915) reported living alone, and 40% (n=1 ,03 8) reported living with other adults. 56 Families in the EHSRE were over-sampled based on racial/ethnic minority status. In the current sample (N=2,632), 968 (38%) families were described as White; 894 (35%) families were described as Black; 602 (23%) families were described as Hispanic; and 116 (5%) families were described as “other” which included Asian-American, Arab- American, and Native American families. The small numbers of each of these racial/ethnic minority groups that were categorized in a single group as “other” did not allow for separate within-group analyses. Results Hypothesis 1 (Mothers) This section describes the analyses and results, among the sample of mothers (N=2,632) pertaining to Hypothesis 1 (alternative): The sample data will fit the hypothesized latent factor measurement model describing psychological absence within the broader path model including parent-child interaction, maternal depressive symptoms, psychological absence, and toddlers’ socioemotional outcomes (emotion regulation and orientation and engagement). Power Analyses To begin, power analyses demonstrate (based on the criteria described in Chapter 3) that the necessary lower bound of observations necessary for this model was 336 cases; hence, analyses proceeded with the sample of 2,632 cases. The data were analyzed with a cross-sectional structural equation model (SEM) describing psychological absence (see Figure 1). The primary purpose of testing this model was to validate the factor structure for measuring psychological absence in the early mother-toddler relationship. This SEM model included four latent factors comprised of eight observed variables. 57 Additionally, cumulative maternal risk, child emotionality, mother’s living arrangements, and participation in Early Head Start programming were included as covariates. Complete Model Fit The model fits the data fairly well (x2=301.98, df=28, p<.05; RMSEA=.O6; GFI=.96; SRMR=.03) (see Figure 4). Measurement model. The first latent factor describing maternal depressive symptoms was indicated by mothers’ scores on the CES-D at the child’s 36-month birthday-related assessment. This latent factor was freed to covary with a second latent factor describing maternal psychological absence, and these variables were significantly associated. The latent factor describing psychological absence included three indicators: mothers’ scores on the parenting distress subscale of the PSI, the Lack of Empathy subscale of the AAPI, and the Parental Detachment subscale of the three-bag play task. As noted, this factor’s validity was of particular interest in the current study, and the factor loadings demonstrated a reasonable fit. The third latent factor, parent-child interaction was indicated by two observed variables, both measured during the 3-bag play task: child engagement and parent support. The fourth latent factor, child socioemotional outcomes, was indicated by two observed variables: the Bayley Behavior Rating Scales measuring emotion regulation and orientation and engagement. The factor loadings of variables indicated these final two factors demonstrated good fit within the model. See Figure 4 for the complete set of parameter coefficients. Path model. As hypothesized, psychological absence significantly negatively predicted the quality of mother-toddler interaction as well as toddlers’ socioemotional outcomes. Further, while maternal depressive symptoms also significantly negatively 58 predicted the quality of mother-toddler interaction and toddlers’ socioemotional outcomes, the effect size was smaller than that of the regression paths between psychological absence and the outcome factors. Finally, as hypothesized, the quality of mother-child interaction significantly predicted child socioemotional outcomes. The model’s effect sizes in predicting both mother-child interaction (R2=.33) and child socioemotional outcomes (R2=.32) were fairly robust. Parent-child interaction as mediator. In this model, parent-child interaction was hypothesized to partially mediate, or explain, the relationship between maternal depressive symptoms and child socioemotional outcomes, as well as the relationship between maternal psychological absence and child socioemotional outcomes. Partial mediation refers to the relationship among three variables when “X,” in this case maternal depressive symptoms and maternal psychological absence (in separate mediation models), predicts “M,” parent-child interaction, “M” predicts “Y,” child socioemotional outcomes, and “X” continues to predict “Y” in the context of these relationships (Baron & Kenny, 1986). To determine if partial mediation occurs among these variables, Holmbeck’s steps for testing mediation with SEM were followed. Those steps are: 1) good model fit for the complete model; 2) good model fit when constraining to zero the path between X and M and the path between M and Y, only estimating the direct effect; and 3) no improvement in model fit for the complete model when only the direct effect is constrained to zero. As described, the full model demonstrated acceptable fit. When only the direct effect was estimated, the model, again, demonstrated reasonable fit (x2=531.14, df=30, p<.05; RMSEA=.08; GFI=.91; SRMR=.O7). In the iteration in which the direct effect was 59 constrained to zero and the indirect paths were estimated, the fit was again reasonable (x’=376.7o, dfi29, p<.05; RMSEA=.O7; GFI=.94; SRMR=.O4). Based on the chi-square distribution, the difference between this chi-square statistic and that of the full model is significant, indicating that the complete model does indeed represent an improvement in fit above the iteration in which the direct effect is constrained to zero, violating Holmbeck’s third assumption. However, the stated hypothesis was that mother-child interaction would partially mediate the relationship between psychological absence and child socioemotional outcomes; if the model fit did not worsen when the direct effect was constrained to zero, this would have represented a full mediation relationship. Therefore, mother-child interaction indeed partially mediated the relationship between psychological absence and child socioemotional outcomes based on these results. Predictive validity of covariates. The covariates were freed to predict all four latent factors: maternal depressive symptoms, psychological absence, mother-child interaction, and child socioemotional outcomes. Household composition had a significant effect on both psychological absence and mother-child interaction, perhaps indicating that less residential social support (e. g., single mothers) predicted greater psychological absence and lower quality mother-child interaction. Additionally, cumulative maternal risk significantly and positively predicted both maternal depressive symptoms and psychological absence. Experimental status (EHS program group or comparison group) positively predicted the quality of mother-child interaction, suggesting that families receiving Early Head Start intervention fared better than their peers in the comparison group on this outcome. 60 Interestingly, child emotionality significantly, positively predicted all four outcomes. These findings suggest that children who demonstrate greater emotionality early on (14 months) are more likely to have mothers who are depressed and/or psychologically absent later (36 months). However, in the context of this model, children with greater emotionality early on were also more likely to have higher quality mother- child interaction and socioemotional outcomes later on. Comparisons Based on Racial/Ethnic Groups Among the 2,632 mother-child dyads included in these analyses, 968 identified themselves as White, 602 identified themselves as Hispanic, and 894 identified themselves as Black. Each of these subgroups exceeds the lower bound of cases determined to be necessary via power analyses (n=336). Furthermore, a second analysis was run to determine the necessary lower bound of observations for the complete sample in a single three-group model: 1,009 cases. Therefore, single and multi-group analyses proceeded based on racial/ethnic groups. The same structural equation model was tested within each group. Results demonstrated that the model fit well among White families (xz=84.29, djE28, p<.05; RMSEA=.05; GFI=.97; SRMR=.03), Hispanic families (78:58.52, df=28, p<.05; RMSEA=.O4; GFI=.97; SRMR=.03), and Black families (x2=108.15, dfi28, p<.05; RMSEA=.O6; GFI=.94; SRMR=.04). A multi-group model was tested with the path between the psychological absence factor and the parent-child interaction factor and the path between the boundary ambiguity factor and the child outcomes factor constrained in independent iterations. There was no significant change in the chi-square statistic between these iterations and an initial iteration in which all paths were freely estimated 61 across groups. These results suggest that there are no major differences based on racial/ethnic groups in the relationships between the boundary ambiguity factor and the parent-child interaction and child outcomes. Comparisons Based on Child Gender As described, 51% (n=1,334) of the children in the current sample were boys, and 49% (n=1,298) were girls. These numbers exceed that which is necessary for a two- group model with 52 degrees of freedom: 844. This model was tested in a two-group SEM model comparing the fit for girls versus boys. The fit of the model in which paths were freely estimated demonstrated a good fit with the data (x2=320.02, dfi52, p<.05; RMSEA=.O6; CFI=1.00; SRMR=.04). The fit of the model significantly worsened, based on the chi-square distribution for 5 degrees of freedom at p<.05, when the paths among maternal psychological absence, maternal depressive symptoms, mother-child interaction, and child socioemotional outcomes were constrained to be invariant for boys as compared to girls (x2=375.33, dfi57, p<.05; RMSEA=.O7; CFl=1.00; SRMR=.04). This suggests there is a significant difference in the fit of the model for boys as compared to girls. In the model in which all parameters are freely estimated, the relationship between maternal depressive symptoms and parent-child interaction was significant for boys (B=-.02, t>1.96), though with a very small effect size, but it was not significant for girls (B=-.00, t<1.96). Furthermore, there was a difference in the amount of variance in child outcomes explained by parent-child interaction between genders; though this path was significant in both models, the effect size was larger among boys (B=.52, t>1.96) than among girls (B=.21, t>1.96). The same was true about the relationship between 62 maternal depressive symptoms and child outcomes: the relationship was significant for both boys and girls, but the effect size was larger for boys (B=-.13, t>1.96) than it was for girls (B=-.O6, t>l .96). However, the paths among psychological absence, parent-child interaction, and child outcomes were not significantly different for boys as compared to girls. Overall, these findings suggest that maternal depressive symptoms may have a greater effect on boys’ socioemotional outcomes via the parent-child relationship, but that these differential effects are not well-explained by psychological absence, which does not appear to differ in its impact based on child gender. Hypothesis 2 This section describes the analyses and results, among the sample of mothers (N=2,632) pertaining to Hypothesis 2 (alternative): Mothers’ psychological absence will significantly predict change over time in children’s emotion regulation. Power analyses For the hypothesized model (see Figure 2), power analyses demonstrated that the lower bound of necessary observations was 289 cases. The current sample size exceeds this number, and therefore analyses proceeded. Cross-Domain Latent Growth Curve Model Development Cross-domain latent growth curve modeling was utilized to test this hypothesis. Latent growth curve modeling is a statistical technique that supports the assumptions of nested data (Duncan, Duncan, & Hops, 1998). This approach has several advantages over other statistical approaches to analyzing longitudinal data including the ability to model error among predictors (Burchinal, Nelson, & Poe, 2006). These are both particularly advantageous in the current study: the complexity of early development within the 63 parent-child relationship is such that no single model could account for all of the variance in a given set of constructs, so the ability to model error will support better model fit. Latent growth curve measurement models. Two growth curves were tested: one describing the development of parental detachment across toddlerhood, and one describing emotion regulation. It was expected that the level and the slope of parental detachment would predict the level and slope, respectively, of emotion regulation, based on the referenced literature describing the development of the young child as embedded in the development of the parent-child relationship. Analyses demonstrated there were significant differences between scores at the 14-month time point as compared to the 36- month time point in parental detachment (t=12.92, p<.001), indicating the likelihood of change over time in this construct such that parents demonstrated less detachment at each successive timepoint. Conceptually the growth curve describing parental detachment is used to represent psychological absence. Just one variable from the three-variable factor tested within Hypothesis 1 was used in this set of analyses in order to create the most parsimonious model given the amount of error variance demonstrated in the model in which a 3-variable factor was used. Parental detachment was chosen as the best single indicator since it is an independent observation of the parent behaving in an emotionally unavailable way with her with her child. The measurement model for each growth curve was hypothesized to best fit a linear trajectory because there are only three timepoints analyzed (Reddy, Rhodes, & Mulhall, 2003). The fit of each growth curve to a no—growth, linear, and quadratic (u- Shaped) trajectory was analyzed separately to test this hypothesis. The difference among the three, from a measurement perspective, is in how the slope loadings are constrained 64 when the intercept is constrained to 1 and the 24-month timepoint set as the main intercept. In a no-growth model, the slope loadings are constrained to zero. In a linear model, the slope loadings are constrained to O, 1, 2, increasing at regular intervals. In a quadratic model, the slope loadings are constrained to O, 3, and 0. Analyses demonstrated that the parental detachment growth curve, as well as the child emotion regulation growth curve, best fit a linear trajectory. T ime-invariant predictors. Maternal depressive symptoms, measured at the child’s 14-month birthday-related assessment, were also included as a time-invariant predictor. Covariates include cumulative maternal risk, child emotionality, mother’s living arrangements, and program status, all measured at the 14-month time point as well. Complete Model Fit Once the measurement models were determined, the complete model was analyzed, testing the interrelationships among the constructs. This model added a new dimension to the previous set of cross-sectional analyses by focusing on emotion regulation and determining the impact of the main predictors on the rate of change in emotion regulation and maternal detachment, as well as the relationships between intercepts and slopes. Furthermore, a latent growth curve approach is an ipsative approach which inherently embs the complexity of development, supporting the assumption that individual children will develop in unique ways based on unique individual contexts. In the complete model (see Figure 5), the intercept of maternal detachment was freed to predict the intercept of emotion regulation, while the slope of maternal detachment was freed to predict the slope of emotion regulation. The covariates, program 65 status, child emotionality, mother’s living arrangements, maternal cumulative risk, maternal depressive symptomology, all measured at the child’s 14-month birthday-related assessment, were freed to predict each growth factor in both growth curves. The model demonstrated good fit with the data (x2=201.88, dfi—34, p<.05; RMSEA=.04; GFI=.98; SRMR=.04). In modeling the growth curves, the intercept was set at 14 months of age. As hypothesized, maternal detachment measured at the child’s 14-month birthday-related assessment significantly negatively predicted child emotion regulation at 14 months of age. However, the rate of change, or the slope, of maternal detachment during toddlerhood did not significantly predict the rate of change in children’s emotion regulation. Mothers in the program group demonstrated less detachment than did mothers in the comparison group at the child’s 14-month birthday-related assessment, but mothers in the program group demonstrated a slightly more rapid increase in detachment during toddlerhood. Children with greater emotionality demonstrated less adaptive regulatory competencies at the 14-month birthday-related assessment but also demonstrated more rapid rates of grth in regulatory competencies across toddlerhood. Greater cumulative risk among mothers predicted greater maternal detachment at the child’s l4-month birthday-related assessment, but less cumulative risk at this timepoint predicted greater increases in detachment during toddlerhood. Finally, greater maternal depressive symptoms at the child’s 14-month birthday-related assessment predicted greater maternal detachment and less adaptive regulatory competencies at the child’s l4-month birthday- related assessments, though depressive symptoms had no significant relationship to the 66 rates of change in these constructs. There were no significant effects of mothers’ living arrangements in the current model. Comparisons Based on Racial/Ethnic Groups As noted in Hypothesis 1, among the 2,632 families included in these analyses, 968 identified themselves as White, 602 families identified themselves as Hispanic, and 894 families identified themselves as Black. Power analyses demonstrated that at least 396 cases per group were necessary to analyze the three-group model described below. Therefore, analyses proceeded. The complete model was run in a three-group model comparing the hypothesized relationships among White, Hispanic, and Black families. The initial model in which all paths were freely estimated demonstrated a good fit with the data (x2=249.26, df=102, p<.05; RMSEA=.04; CF I=.95; SRMR=.04). In the next iteration, the paths between the growth curves within Black families and Hispanic families were constrained to be invariant from that of White families. This model continued to demonstrate a good fit with the data (x2=253.16, df—‘106, p<.05; RMSEA=.04; CF I=.95; SRMR=.04). There was no significant difference between the chi-square values based on four degrees of freedom at p<.05, suggesting that there are no major differences based on racial/ethnic groups in the relationships between maternal detachment and child emotion regulation during toddlerhood in the context of the current model. Comparisons Based on Child Gender The current sample includes 1,334 boys and 1,298 girls, as noted. Power analyses demonstrated that 348 cases per group were necessary to analyze the two-group model described below. Therefore, analyses proceeded. 67 The model was run in a two-group model comparing these relationships between boys and girls. The initial model in which all paths were freely estimated demonstrated a good fit with the data (x2=243.88, dfi68, p<.05; RMSEA=.04; CF I=.95; SRMR=.04). In the next iteration, the paths between the grth curves were constrained to be invariant between groups. This model continued to demonstrate a good fit with the data (x2=246.95, dfi-70, p<.05; RMSEA=.04; CFI=.95; SRMR=.04). There was no significant difference between the chi-square values based on two degrees of freedom at p<.05, suggesting that there are no major differences based on child gender in the relationships between maternal detachment and child emotion regulation during toddlerhood in the context of the current model. Discussion These findings support the initial hypotheses related to the proposed measurement model and path model. Proximal measures of affective, cognitive, and behavioral components of psychological absence significantly predict a single latent factor. Further, that factor significantly predicted parent-child interaction and child outcomes and, when correlated with depressive symptoms, predicted more variance in the endogenous factors than depressive symptoms alone. As a complete model, the results show that the model is a good fit with the data, and the validity of measuring the construct of psychological absence in this way is bolstered as the model fit did not differ across racial/ethnic groups. Psychological Absence and Toddlers’ Socioemotional Development While the hypotheses in the current study cannot be linked to a direct scholarship on boundary ambiguity in the early parent-child relationship, they can be discussed in the context of known outcomes for children of parents who are emotionally withdrawn. The 68 current study seeks to build on this scholarship by providing a theoretical foundation grounded in systems theory to explain the impact of these known behavioral outcomes on children when those behaviors converge into psychological absence resulting in boundary ambiguity, a family-level risk indicator and relational characteristic. The caregiving behavioral system, defined as the primary caregiver’s internal working system that motivates behavior to provide care to one’s child, is a subjective and internal system of experiences and is reciprocal to the attachment system of the child (Solomon & George, 2000). With this consideration in mind, the parent-child relationship is a unique subsystem of the family and as such requires unique methods of measurement accounting for the primary caregiver’s internal experiences. The measurement model of the current study supports this notion. Research shows that the level of boundary ambiguity in a family system, as a result of a particular stressor, is moderated by the primary caregiver’s ability to contribute to the family’s coping abilities (Buehler & Pasley, 2000). Therefore, children experiencing boundary ambiguity in their relationship with their primary caregivers might experience the primary effects of the psychological absence and secondary effects of limited family-level coping strategies because the psychologically absent member of the family is the primary caregiver. Emotion regulation and positive engagement are likely related to effective proactive coping strategies (Brenner & Salovey, 1997; Eisenberg & Spinrad, 2004). Emotion regulation is a primary task of early socioemotional development, and support from the primary caregiver supports the development of regulatory capabilities (Cole, Martin, & Dennis, 2004; Crockenberg & Leerkes, 2000). Diener and Mangelsdorf (2000) demonstrated that young children’s 69 coping skills improve with the presence of a familiar caregiver. The dyadic relationship between a child and his/her primary caregiver likely has the greatest impact on young children’s socioemotional development (Crockenberg & Leerkes, 2000). The role of the parent-child relationship in early emotional development is well- known within infant mental health scholarship. In particular, the process of reciprocity is based in the shared caregiver-child emotional life and is a defining characteristic of the early parent-child relationship, promoting optimal behavioral and affective development (Osofsky & Thompson, 2000). Bowlby (1969, 1973, 1980) posited that separation from one’s primary caregiver predicted less optimal psychosocial development. The current study builds on this scholarship, defining similar sensitive outcomes for children when these developmental processes are unavailable, though the caregiver incongruently physically maintains the role of primary caregiver. Relationship Between Maternal Detachment and Child Emotion Regulation Over Time The second model which measured parental detachment and emotion regulation over time provides further information about this concept. As described, maternal detachment was chosen to represent psychological absence mostly for reasons of parsimony. The factor loadings for the psychological absence latent factor in the first model, discussed above, were reasonable, but a good amount of error variance remained. This construct needs firrther exploration in this or a similar sample. The findings support Hypothesis 2 with a small to moderate effect size. Mothers who appeared more detached in play, based on a context of risk (maternal depressive symptoms, cumulative demographic risk, children’s emotionality, and single parenthood) had children who demonstrated less adaptive regulatory competencies, controlling for Early Head Start 70 program intervention. Children’s regulatory competencies fit a linear trajectory, supporting previous research defining this as a developmental construct (Raikes, Robinson, Bradley, Raikes, & Ayoub, 2007). The construct of parental detachment also fit a linear trajectory, indicating that there is change across time in the current sample within this construct. Interestingly, the relationship with some of the covariates demonstrates that an upward linear trajectory (e.g., increasing detachment over time) may be normative given that families who demonstrated less detachment at 14 months (families in the program group and families with fewer cumulative risk indicators) demonstrated more rapid rates of growth in detachment across toddlerhood. However, the effect sizes for these relationships are very small and, therefore, are presented with great caution. Further, while children whose mothers were detached were less likely to demonstrate adaptive regulatory strategies, the rate of grth of maternal detachment did not predict the rate of grth in children’s regulation. Future inquiry might further explore this issue, seeking to understand the conditions under which mothers become increasingly detached over time to determine if it is indeed normative and perhaps related to children’s increasing independence-seeking behaviors, or non-normative and perhaps related to increases in social stressors low-income families may experience as children get older. In the current model, depressive symptoms predicted greater detachment and less adaptive emotion regulation early on, as expected, though depressive symptoms did not predict the rate of growth of either construct. Further, there were no significant between-group differences in the relationships between maternal detachment and regulatory development over time based on child gender or . 71 Relationship Between Risk and Psychological Absence Of note, the covariates, EHS program participation, parental living arrangements, and cumulative risk, were significant predictors of some of the major outcome factors. It is reasonable to expect program participation to predict some of the variance in outcomes in parent-child interaction, as has been in the case in previous research (Love et al., 2005; Chazan-Cohen et al., 2007). The effect size was somewhat moderate, and, in the context of the model accounting for mothers’ psychological absence and depressive symptoms, it is important to consider the ways an intervention might improve outcomes for parents and children. Furthermore, results show that living arrangements had an impact on the tested outcomes, suggesting that residential social support may contribute to better interactions between mothers and young children. Research supports this finding in so far as mothers who receive support from other adults who are able to provide care for children, including romantic partners and family members, may experience less depressive symptoms and greater parenting competence (Erel & Burman, 1995; Morris, Silk, Steinberg, Myers, & Robinson, 2007; Silver, Heneghan, Bauman, & Stein, 2006). Child Contributions to Relational Boundaries Interestingly, child emotionality significantly, positively, predicted all four outcomes such that children who demonstrated more emotionality early on had mothers who were both more depressed and more psychologically absent later on but also better parent-child interaction and socioemotional outcomes later on. In the context of a model accounting for mothers being withdrawn from their children, it appears that children who were fussier may have been more likely to get their needs met and achieve more optimal outcomes later on. While some research defines difficult infant temperament as a stressor 72 (e.g., Cutrona & Troutman, 1986), other research indicates that mothers may be more, and not less, responsive to children who are fussier, dependent upon child characteristics such as gender (e. g., Crockenberg & Smith, 2002), or social stressors (van den Boom & Hoeksma, 1994). Crockenberg and Leerkes (2003), in a review of the relevant literature, suggest there may be two areas of risk explaining the contrasting findings: 1) competing demands for mothers’ time, such as is likely the case for multi-stressed mothers in low-income populations, predict that mothers may only attend to babies when they are fussy, and use the time when babies are calm to accomplish other tasks; or 2) mothers with internal traits which predispose them to irritability, including depressive symptoms, may respond slowly or inappropriately when their infants demonstrate distress. In support of the first area of risk, van den Boom and Hoeksma (1994) indicate that a sample of mothers defined as high-risk were more likely to demonstrate sensitivity to their crying infants and less likely to demonstrate sensitivity to their children when they were in a quiet alert state. Similarly, the findings in the current study may indicate that infants who demonstrate greater emotionality are more likely to receive sensitive care from their mothers who may otherwise be preoccupied. However, the second area of risk pertains to mothers who are depressed, and there may be implications for this to occur in the current sample as well, though the findings did not support it. This may be explained within a framework of boundary ambiguity: children who are fussier, in fact, may be more consistently initiating parent-child interactions which include sensitivity and thus contributing to a shared repertoire of predictable interactions, the basis for clear relational boundaries. 73 Comparisons Based on Child Gender Model comparisons based on racial/ethnic groups and gender indicate that this model did not differ between subgroups in relation to the measurement or predictive validity of psychological absence. However, there were differences in the way that maternal depressive symptoms, in the context of this model, predicted outcomes for young children, such that boys appear to be at greater risk than do girls. This finding is consistent with previous research that demonstrates that infant and toddler boys may be at greater risk than girls for adverse outcomes with their mothers are depressed (e. g., Carter, Garrity—Rokous, Chazan-Cohen, Little, & Briggs-Gown, 2001; Early & Jung, 1987; Shaw & Vondra, 1995; Murray, F iori-Cowley, Hooper, & Cooper, 1996). Carter et al. (2001) describe this differential outcome as a biological vulnerability, suggesting that boys have fewer internal strengths helping them to attenuate discontinuities that occur as a result of mothers’ emotional unavailability. The authors further demonstrate that, while boys of depressed mothers may be at greater risk in some regards, girls appear to be more likely to experience adverse outcomes as a result of low- quality mother-child interactions. Therefore, the current study may demonstrate that, although boys are at greater risk inherently when mothers are depressed, gender differences are washed out when accounting for the behavioral manifestations of psychological absence, as this relational construct actually may put girls at greater risk. Limitations to the Current Study The results of this study are generally robust statistically and theoretically, but there are a few limitations to the interpretations of the findings. The measurement of perception of boundary ambiguity requires further discussion. The research precedent for 74 studying boundary ambiguity, particularly psychological absence, is through the self- report of the family’s primary caregiver who may also be classified as a survivor of the ambiguous loss and is thus best able to report family-level perception of ambiguity (e. g., Boss, Greenberg, & Pearce-McCall, 1990; Kaplan & Boss, 1999). When studying the impact of parental psychological absence, or the situation when the family’s primary caregiver is also the psychologically absent person, accommodations must be made in this respect. The use of the mothers’ perception in this study may be a limitation based on research precedent, but the delimitation, as noted earlier, is that the parent-child relationship is a unique subsystem and measurement may actually benefit from the parent’s perception of self. Therefore, this complicated issue would likely gain from additional scholarly dialogue. This study seeks to bridge separate areas of research, and the measurement model is tenuously grounded in theory without previous scholarship to directly inform its conceptualization. Future research should seek to replicate this or similar models and support or refute its findings. The factor loadings of the psychological absence latent factor in the current study suggest that there may be other and potentially superior ways to measure this construct in this population, though this study represents a reasonable starting point. Implications for Research and Practice This study also suggests several implications for future research. As noted, the measurement of boundary ambiguity in the early parent-child relationship would benefit from further scrutiny. This study used proximal indicators of measurement; additional studies could similarly discern if proximal methodology is generally sufficient or if this 75 area of measurement should eventually converge into a specified process of measurement. Finally, the long-terrn consequences of boundary ambiguity and its impact on additional dimensions of children’s early development would benefit from empirical research. It is well-known that early socioemotional development in general is known to be an important predictor of later social competence and psychological functioning (Eisenberg, Cumberland, & Spinrad, 1998; Salovey & Mayer, 1990). This study also suggests implications for intervention and practice. As demonstrated, the impact of the Early Head Start intervention on children’s outcomes was significant, particularly in the context of a parent’s psychological absence. Study results indicate that psychological absence may be an important construct for interventions to address, and the measurement of a proximal construct provides specific guidance for the ways in which psychological absence might manifest in the early parent- child relationship. In specific, interventions that target meaning-making processes in families may be most effective, helping at-risk parents create consistent definitions of family roles and relational boundaries, contributing to an increase in predictable, sensitive interactions between parents and children. 76 CHAPTER 6: PATERNAL PSYCHOLOGICAL ABSENCE AND TODDLERS’ SOCIOEMOTIONAL DEVELOPMENT Among the 3,001 children in the EHSRE, 1,747 children participated in research sites that were involved in the father sub-study. In those sites, participant mothers were asked to voluntarily provide contact information for the children’s biological and/or social fathers. Among the fathers whose contact information was provided, approximately 1,000 fathers consented to participate in the overall Early Head Start fathers’ study which collected data at the children’s 24-month and 36-month (but not 14- month) birthday-related assessments from fathers directly (self-report interview) and through observation of the fathers and children engaged in the aforementioned three-bag play task (Cabrera, Tamis-LeMonda, Lamb, & Boller, 1999). Of those 1,000 fathers, data were available for 759 fathers (among the 2,632 mother-toddler dyads described in Chapter 4) that specified the father’s relationship to the focus child as biological or social, residential or other. The other 240 fathers were missing this information because the child’s mother did not provide it. Because this variable defining the child’s relationship to the man who provided data is important to the interpretation of findings in the current study, the presence of data for this variable was the main inclusion criteria. Based on analyses exploring issues of attrition, a final sample of 501 fathers were included in the current study. This subsample (n=501) represents the men who provided at least partial data (at least the self-report interview) at the 36-month timepoint which is the main timepoint of interest for these analyses. A smaller subsample (39%) of these dyads had complete data for the three-bag 77 task, and data were imputed for these variables for the remainder of the 501 father-child dyads. See Appendix B for detailed description. Sample Description Father Characteristics At children’s 24-month birthday-related assessments, the majority of the current sample of fathers (N=501) were residential, biological fathers (n=3 50; 70%). The remaining participants were non-residential, biological fathers (n=68; 14%), residential, non-biological father figures (n=65; 13%), and non-residential, non-biological father figures (n=18; 4%). There were not enough men in these groups to employ analyses comparing the hypothesized models between biological fathers (n=418), non-biological fathers (n=83), or to compare residential fathers (n=415) with non-residential fathers (n=86). Therefore, for simplicity, the term "father" was used to describe any type of paternal relationship with the child. At the child’s 36-month birthday-related assessment, 68% (n=340) of participants were residential biological fathers, while 15% (n=73) fathers were non-residential biological fathers. Among non-biological fathers, 14% (n=72) were residential fathers while 3% (n=16) were non-residential. Fathers in this sample (N=501) were, on average, 30.21 years old (SD=7.96; min=18, max=79) at the time of the child’s 36-month birthday-related assessment. At this time, fathers reported an average monthly income of $1,738.31 (SD=1111.47; min=0, max=7200). Fathers demonstrated fewer demographic risks than did participating mothers; the vast majority of the sample (91%) reported 2 or fewer risks on the cumulative risk index. Furthermore, 99% of the men who reported information about living arrangements at the child’s 36-month birthday-related assessment reported living 78 with a spouse or partner, while less than 1% reported living with other adults. Furthermore, 58% of those reporting information about living arrangements indicated they were married to the focus child’s mother. Child Characteristics The children in the current sample, reflecting fathers and their children, were demographically similar to the children in the complete sample (described in Chapter 4). The current sample included 244 girls (49%) and 257 boys (51%). Furthermore, 255 children (51%) were in the Early Head Start program group, and 246 children (49%) were in the comparison group. At the 36-month birthday-related assessment, children were, on average, 36.97 (SD=1.63, min=33.27, max=48.65) months old. Rwflfi Hypothesis 1 (Fathers) This section describes the analyses and results, among the sample of fathers (N=501) pertaining to Hypothesis l (alternative): The sample data will fit the hypothesized latent factor measurement model describing psychological absence within the broader path model including parent-child interaction, paternal depressive symptoms, psychological absence, and toddlers’ socioemotional outcomes (orientation and emotion regulation). Power Analyses Power analyses demonstrated that 368 cases were necessary as a lower bound for the model described below. Therefore, analyses proceeded. The analyses describing the factor structure of psychological absence at the 36-month birthday-related assessment (see Chapter 4) were repeated with the father sample to test its within-group validity. The 79 same set of procedures followed with the data collected from mother-child dyads was followed with the data collected from father-child dyads (see Figure 1). Complete Model Fit Paternal detachment, lack of empathy, and parental distress were loaded on a single factor describing psychological absence, correlated with paternal depressive symptoms, and parent-child interaction was hypothesized to partially mediate the relationships between paternal depressive symptoms and child outcomes, as well as between psychological absence and child outcomes. The complete model was a mediocre to poor fit with the fathers’ data (x2=128.28, dfi25, p<.05; RMSEA=.O9; GFI=.96; SRMR=.O6). The squared multiple correlation coefficients suggest that this model explains only small to moderate amounts of variance in father-child interaction (R2=.05) and child outcomes (R2=.18). Measurement model. Fathers' supportiveness towards children (8:159) and children’s engagement towards their fathers (B=.59) did not load well onto a single latent factor. Also, the variables hypothesized to fit a single latent factor describing psychological absence did not load well onto a single factor: lack of empathy ([3=.94), parent distress ([3=.15), and paternal detachment (B=.02). Based on the factor loadings, this factor is largely represented by fathers’ lack of empathy, and the remainder of the results will be discussed in this context. As noted previously, lack of empathy is conceptualized as the cognitive manifestation of psychological absence. Path model. Fathers’ lack of empathy significantly negatively predicted child outcomes (B=-.20, t>1.96), as well as parent-child interaction (B=-.21, t>1.96), both with moderate effect sizes. Fathers’ depressive symptoms also significantly, negatively 80 predicted parent-child interaction (B=—.O8, t>l.96), with a somewhat small effect size, but not child outcomes (B=-.O6, t<1.96). Father-child interaction significantly, positively predicted child outcomes (B=.19, t>1 .96) with a moderate effect size. Predictive validity of covariates. As with mothers, the following covariates were included: Early Head Start program participation, child emotionality at 14 months, and (paternal) cumulative risk. Fathers’ living arrangements were not included as a covariate because there was little to no variance demonstrated in this construct among fathers. The covariates entered into the model as part of a context of risk appear to provide little explanation of the variance in outcomes. Fathers’ cumulative risk significantly, negatively predicted child outcomes (B=-.10, t>1 .96) and significantly, positively predicted depressive symptoms (B=.21, t>1 .96), both with moderate effect sizes. The other covariates, program status and child emotionality, did not demonstrate significant associations with fathers’ depressive symptoms, fathers’ psychological absence, father-child interaction, or child outcomes. Comparisons Based on Child Gender Power analyses demonstrate that 434 cases were needed for a two-group model, and therefore, analyses proceeded with between-group comparisons based on racial/ethnic groups and child gender. A two-group model was employed to determine if there were significant differences in the model for boys as compared to girls. First, a model was analyzed with all parameters freely estimated; it was not a good fit with the data. Second, the model was analyzed with the relationships among psychological absence, depressive symptoms, parent-child interaction, and child outcomes constrained to be invariant between groups. 81 The model fit did not change. Based on this information, the model is not more robust among either girls or boys. Comparisons Based on Racial/Ethnic Status Given the relatively small sample sizes of Hispanic fathers and Black fathers in this sample, these groups were collapsed (N on-White) in order to compare the fit of the model between White and non-white fathers (including 20 fathers who identified themselves as racial/ethnic/ethnic minorities other than Black and Hispanic). There was no significant difference in the chi-square statistic between the model in which all parameters were freely estimated and the model in which relationships among psychological absence, depressive symptoms, parent-child interaction, and child outcomes were constrained to be invariant between groups. The results suggest there is little to no variation in the relationships among psychological absence, depressive symptoms, parent-child interaction, and child outcomes between White fathers and non- White fathers in the current sample. However, this finding is presented cautiously given the great diversity represented by the “non-White” group. Hypothesis 3 This section pertains to Hypothesis 3 (alternative): Fathers’ psychological absence will significantly predict change over time in child emotion regulation competencies. Because data were not available for the father-child play task at more than two time points, it was not possible to measure paternal detachment over time among the current sample of fathers (N=501). Data were available for fathers at children’s 24-month birthday-related assessments, and one plausible strategy would be to measure the predictive validity of paternal detachment at 24 months of age as a time-invariant 82 predictor of a latent growth curve measuring children’s emotion regulation over time. However, the fathers included in the current study (N=501) demonstrated little to no variation in detachment during play with their children: 92% of the sample scored “1 ” or “Very Low” on this measure. Therefore, analyses seeking to describe the relationship between fathers’ psychological absence and children’s regulatory competencies did not proceed. On its own, this finding may provide important information about this sample and the themes addressed in this study, and therefore, this result is further explored in the discussion section below. Hypothesis 5 Hypothesis 5 (alternative) proposes: There will be a greater effect size in the relationship between psychological absence and toddlers’ regulatory outcomes in the mother-child relationship as compared to the father-child relationship due to mothers’ greater responsibility for primary care early on. However, because the model described above was such a poor fit with the fathers’ data, analyses comparing this model with the model fitted to the mothers’ data (described in Chapter 5, Hypothesis 1) did not proceed. Discussion Psychological Absence Among Fathers The current study suggests that the psychological absence measurement model, as it is currently designed, does not fit well among the current sample of low-income fathers. Further, this appears to be the case regardless of family or child gender. There are several potential explanations for this finding. 83 Sample Bias First, fathers who could be located and recruited were more likely to be a high- functioning group who were well-involved with their children. This is evidenced by the final sample composition, demonstrating that the largest subgroup of fathers were residential biological fathers, as well as by the variable means demonstrating very low detachment. Other researchers (e.g., Cabrera, Tamis-LeMonda, Lamb, & Boller, 1999) have noted that selection bias in this dataset is a challenge, based on mothers’ gatekeeping role determining fathers’ participation in this study as well as potentially in children’s lives. For these reasons, the construct of psychological absence may have been less likely to apply to the fathers in the current sample. Implications of Fathers ’ Typical Residential Patterns for the Measurement of Psychological Absence Though the data does not fit the current measurement model, these findings still raise important questions about the general applicability of the construct of psychological absence to fathers, in general, and low-income fathers in particular. Scholarship and policy related to father involvement often describe a pervasive problem of father absence and under-involvement in low-income populations (Cabrera & Peters, 2000). However, research with low-income fathers who maintain proximity to their children illustrates that these men appear to demonstrate warmth and sensitivity in interactions with their infants and toddlers (Brophy-Herb, Gibbons, Omar, & Schiffrnan, 1999) and, when physically available to children, make important contributions through sensitive interactions to their children’s development (Cabrera, Shannon, & Tamis-LeMonda, 2007; Kelley, Smith, 84 Green, Bemdt, & Rogers, 1998; Shannon, Tamis-LeMonda, London, & Cabrera, 2002; Tamis-LeMonda, Shannon, Cabrera, & Lamb, 2004). In this context, the current study findings suggest that physically present fathers may be unlikely to be psychologically absent in the parenting role. While mothers may maintain their roles as caregivers physically if psychologically absent, likely based on compulsory role attributes and expectations, fathers may be less likely to maintain proximity if experiencing risk factors for psychological absence. This may be particularly true in low-income populations where father absence is prolific and, therefore, increasingly normative, impacting social role expectations about fatherhood. While there is limited research assessing the mental health characteristics directly of absent fathers, some research demonstrates that physically absent fathers may be more likely to demonstrate mental health risk factors such as depressive symptoms and antisociality than more involved fathers (J affee, Caspie, Moffitt, Taylor, & Dickson, 2001; Pfiffirer, McBumett, & Rathouz, 2001). Therefore, lack of fit of this model with this sample of generally involved fathers may have broader implications for the relationship between physical and psychological presence of low-income fathers. Fathers ’ Lack of Empathy and Parenting and Child Outcomes The explanations above describe the poor fit of the psychological absence latent factor measurement model. Therefore, the findings regarding the path model should be interpreted cautiously. The psychological absence factor, based on the factor loadings, is disproportionately indicated by lack of empathy scores. The significant finding between this variable and child outcomes, as well as between lack of empathy and parent-child interaction, demonstrated a moderate effect size. Previous research illustrates that fathers’ 85 attitudes and cognitive beliefs about the fathering role in abstraction plays a critical role in their quality of engagement with their children (Freeman, Newland, & Coyl, 2008; Gaertner, Spinrad, Eisenberg, & Greving, 2007). This finding is consistent with this research, demonstrating the ways in which fathers’ attitudes impact children’s socioemotional outcomes directly and indirectly via father-child interaction, or their own supportiveness and children’s reciprocal engagement. Paternal Depressive Symptoms and Parenting and Child Outcomes Fathers’ depressive symptoms significantly, negatively predicted parent-child interaction, with a somewhat small effect size, but not child outcomes. However, fathers demonstrated very low rates of depressive symptoms in this sample (M=4.94; min=.00, max=34.00), suggesting that this finding is difficult to interpret. Research related to paternal depressive symptoms is more limited than that of maternal depressive symptoms, but some empirical associations have been made. For example, Rarnchandani, Stein, Evans, and O’Connor (2005) demonstrated a significant relationship between paternal depressive symptoms postnatally and later behavioral problems when children were preschoolers, even after controlling for maternal depressive symptoms. Therefore, the weak relationship between paternal depressive symptoms and parenting and child outcomes is likely a function of the lack of variation in this variable in the current sample. Limitations to the Current Study In general, the model as a whole does not appear to fit this sample of fathers well nor explain the relationships between fathers’ emotional availability to their children (or lack thereof) when physically present and child’s socioemotional outcomes. The weak 86 relationships between the covariates and the outcomes further underscore the poor fit of this model. Therefore, future inquiry should explore ways to measure psychological absence, and among what groups of fathers or in what contexts this construct might be relevant, and further explore the potential for discontinuities in physical and psychological presence to exist among fathers and impact children’s outcomes. As described in the section on attrition analyses (see Appendix B), a large portion of the data were imputed based on scores provided by sample participants at an earlier timepoint. Attrition analyses demonstrate that children whose fathers were available at both the 24-month and 36-month timepoints to participate indeed had better socioemotional competencies at both timepoints than their peers who dropped out of the study. This is a significant limitation to the current study. However, these findings in and of themselves provide further information about the ways low-income families may organize around the role of the father, signifying the importance of physical presence of fathers. 87 CHAPTER 7: PATERNAL PSYCHOLOGICAL PRESENCE AND TODDLERS’ LONGITUDINAL REGULATORY DEVELOPMENT The analyses in this section pertain to the total sample (N=2,63 2) of families. The main inclusion criterion for these analyses was the presence of mother-report data indicating whether or not biological fathers were present across toddlerhood. Fathers did not need to be physically present in children’s lives in order for families to be included in the current sample. Biological fathers are the focus in this section because hypothesis 4 and its accompanying plan for analysis seeks to test whether or not continuity in the fathering role predicts children’s regulatory development, implying that continuity in this role is expected to reflect organization in the family. The biological father’s role is more likely to carry normative social expectations contributing to coherence in family patterns. Fifty-six percent of the sample (n=1,477) had complete information for the variable described in Chapter 3 (Methods) that ascertains whether or not mothers report biological fathers to be continuously present throughout toddlerhood or not. Among those children, sixty-eight percent of children (n=1 ,004) had biological fathers who were continuously present while 32% (n=473) had biological fathers who were not continuously present. Children who were missing this information were excluded from analyses, resulting in a sample size of 1,477 for this set of analyses on the effects of continuous father presence on child outcomes. Sample Description Father and Mother Characteristics In the current sample, mothers reported that biological fathers were, on average, 25.51 years old (SD=6.69, min=14, max=52) at the time of the child’s birth. Fifty-one 88 vi (.1 percent of mothers reported, at 36 months, that their children (n=754) usually lived with their biological fathers. Fifty percent of mothers (n=731) reported they were married to or cohabiting with the focus child’s biological father, and 6% of mothers (n=87) reported they were in a non-cohabiting romantic relationship with the focus child’s biological father at 36 months. Child Characteristics At the time of the 36-month, birthday-related assessment, children in this sample were, on average, 37.35 months old (SD=1.89, min=33.24, max=54.10). Fifty percent (n=732) were girls, and 50% were boys (n=745). Also, 52% (n=770) were receiving Early Head Start intervention while 48% (n=707) were in the comparison group. Results Hypothesis 4 The analyses and results presented in this section pertain to Hypothesis 4 (alternative): Continuity of the physical presence in the early father-child relationship will positively predict change over time in children’s emotion regulation. Analyses were conducted to examine the relationship between mothers’ report of fathers’ physical presence over time and child emotion regulation over time (see Figure 3). Power Analyses Power analyses indicated that a lower bound of 727 observations were necessary for the current model; therefore, analyses proceeded. Cross-Domain Latent Growth Curve Model Development As described, the variable indicating fathers’ continuous presence is by maternal report, indicating that a child’s father was either consistently present or not (and therefore 89 either inconsistently present or consistently physically absent) across toddlerhood. This variable will be included as a time-invariant predictor of a single-domain growth curve, measuring child emotion regulation across toddlerhood, from 14 to 36 months (see Figure 3). The procedures for modeling latent growth curves described in the subsection on the development within the mother-child relationship (see Chapter 4) were similarly followed. The growth curve describing child emotion regulation was tested separately from the complete model to determine, among the smaller subsample of focus in the father-child analyses (N=l,477), if a linear trajectory continued to best fit the data. Indeed, a linear model demonstrated the best fit with the data. Complete Model Fit The final model demonstrated an acceptable fit with the data (x2=106.13, dfil 1, p<.05; RMSEA=.08; GFI=.98; SRMR=.05) (see Figure 6). Fathers’ consistent presence in their children’s lives demonstrated a small, significant effect on children’s emotion regulation at 36 months (intercept), but there was no significant relationship between fathers’ presence and the rate of growth (slope) over toddlerhood. Children who received Early Head Start programming demonstrated a more rapid rate of growth (s10pe) in regulatory competencies across toddlerhood as did children with greater emotionality at 14 months. Children whose mothers experienced greater cumulative risk early on, at 14 months, demonstrated poorer self regulatory competencies at 36 months. Comparisons Based on Gender Power analyses demonstrated that a two-group model comparison required at least 814 cases in the total sample. The current sample, out of 1,477 total cases, included 744 boys and 732 girls. Descriptive data demonstrated that, in this sample, toddler boys were 90 not more likely than toddler girls to have their biological fathers consistently present across toddlerhood (x2=.03, p>.10). Out of 744 boys in the current sample, 508 (68%) had biological fathers consistently present; the same was true for 496 (68%) out of 732 total toddler girls. The odds ratio of boys to girls is 1.02, demonstrating that boys and girls had almost the exact same odds of having fathers consistently present. A two-group iteration of the latent growth curve model was estimated to compare the impact of continuous father presence on children’s emotion regulation over toddlerhood between boys and girls. The two group model in which all paths were fieely estimated demonstrated an acceptable fit with the data (x2=99.02, df=22, p<.05; RMSEA=.O7; CFI=.85). When the paths between the predictor variables and the growth curve latent factors was constrained to be equal between groups (x2=103.58, dfi30, p<.05; RMSEA=.06; CF I=.86), the model fit did not significantly change based on the chi-square distribution for 8 degrees of freedom at p<.05. These results indicate that the relationship between father presence and emotion regulation over time is not significantly different for boys as compared to girls. Comparisons Based on Racial/Ethnic Groups This model was estimated in a three-group iteration to determine potential significant differences between White, Black, and Hispanic families. In White families, 70% of mothers (n=424) in the current sample reported biological fathers were consistently present across toddlerhood. In Black families, 54% of mothers (n=248) reported biological fathers were consistently present. In Hispanic families, 82% of mothers (n=259) reported biological fathers were consistently present. There were statistically significant differences among these numbers 08:78.36, p<.001). The odds 91 ratio demonstrates that White children were 1.99 times more likely to have their fathers consistently present across toddlerhood than were Black children. Hispanic children were 3.97 times more likely than Black children and 1.99 times more likely than White children to have their fathers consistently present. Power analyses demonstrated that a three-group model could proceed with at least 888 cases, and therefore, analyses were carried out. The fit of the model with all parameters freely estimated had similar fit statistics as the single-group model (x2=1 10.21, df=33, p<.05; RMSEA=.O7; CFI=.84). When paths were constrained in the relationships between the predictor variables and the child outcome factors, the chi- square increased (x2=129.84, df=49, p<.05; RMSEA=.06; CFI=.84) but not significantly so based on the chi-square distribution for 16 degrees of freedom at p<.05. However, the regression coefficients for the relationship between father presence and child outcomes appeared very different in the model describing White families (see Figure 7) as compared to the model describing Black families (see Figure 8). In White families, father presence significantly predicted children’s emotion regulation at 36 months (intercept) with a moderate effect size (6:118). In Black families, father presence did not significantly predict either latent growth factor. Therefore, a second two-group iteration was estimated to determine if these apparent differences were statistically significant. The iteration in which all parameters were freely estimated between White and Black families demonstrated an acceptable fit with the data (x2=88.53, df=22, p<.05; RMSEA=.08; CFI=.84). In a separation iteration in which the paths between the predictor variables and the growth factors were constrained to be invariant, the fit worsened based on the chi-square statistic (x’=102.71, df=30, p<.05; RMSEA=.07; CFI=.83). The chi- 92 square difference represents a significant difference, based on the chi-square distribution for 8 degrees of freedom, indicating statistical support for the finding that continuous father presence predicts children’s emotion regulation at 36 months within White families but not Black families. Other iterations determined there were no significant differences between Hispanic families and White families, nor between Black and Hispanic families Discussion a» The findings in this study demonstrate that, overall, toddlers whose biological fathers were consistently present across toddlerhood were more likely to demonstrate more optimal regulatory competencies than were toddlers whose biological fathers were inconsistently present or not present at all during this developmental period, particularly among White families. Fathers ’ Continuous Presence as a Proxy for Organized Relational Boundaries As described, children’s regulatory development largely depends on the history of, and continuity experienced in, parenting processes (Sroufe, 1989). An assumption of the currently tested models is that absent fathers are likely psychologically present in families in some way, particularly early on when families are still organizing and re- organizing in relation to parenting roles, given that early childhood normatively represents a period of transition in roles as new mothers and fathers become parents (Duvall & Hill, 1948; Hill, 1986; Mederer & Hill, 1983). This psychological presence may be associated with a temporary period of disorganization as the primary caregiver or caregivers determine distribution of parenting role attributes, defining role boundaries. However, psychological presence may contribute to ambiguity of boundaries over time if the father’s potential to be present is unclear or such re-organization does not occur for 93 it”? \ \\ other reasons, including socially constructed norms about gendered parenting roles which make it difficult for other family members to take over what is assumed to be the father’s responsibility. Family life cycle theory indicates that age 3 is approximately the time when the next major family life cycle transition occurs, as the child enters preschool (Duvall & Hill, 1948; Hill, 1986; Mederer & Hill, 1983). Therefore, the longitudinal perspective of this model across toddlerhood, ending at the child’s 36-month birthday- related assessment, accounts for a period of time in which new parenting roles should nonnatively be consolidated. Development of Role Boundaries in the Father—Child Relationship Roles theorists posit that relationship-based roles are negotiated by those participating in a given relationship (Merton, 1957) and that well-being for those in the relationship may be determined by the ability to satisfactorily fulfill role expectations (Thoits, 1991). Transitions in roles are likely to be a challenge to one’s sense of efficacy in fulfilling new role expectations (Thoits, 1991). There may be particular risk during the early parenting years when rules and patterns related to family roles may still be evolving. Varying definitions of fatherhood (Lamb, 2002b) contribute to a high potential for ambiguities and inconsistencies to occur in this role, and scholars further suggest there is a lack of clarity as to what normatively should be expected of fathers (Doherty, Kouneski, & Erickson, 1998). When such ambiguities exist, families may struggle to adapt, and children’s perception of this disorganization is likely to be reflected behaviorally as dysregulation. However, when fathers are consistently present, the fathering role is more likely to follow a normative trajectory of development that father and child co-construct, and, therefore, children may experience a greater sense of external 94 v , a" organization, contributing to more optimal internal organization. Social father figures may not be equally psychologically present in the family, compared to biological fathers, as the social father’s role is not likely to carry the same set of normative expectations as the biological father’s role. Development of the Fathering Role and the Family Life Cycle The question of how the biological father will be involved may be a lingering I? question in the early parenting years (as opposed to when children are older and families have had years to establish family patterns) (Bell, Goulet, Tribble, Paul, Boisclair, & Tronick, 2007). As described, when children are born, families are normatively moving from one state of equilibrium to another, demarcating temporary disorganization (Bell et al., 2007); any departure of the biological father early in a child’s life may intensify the patterns of disorganization/re-organization. F amilies who report biological fathers as consistently present may have clarity on role/involvement; families who report inconsistent or even no contact, may still be clarifying the processes inherent to the fathering role. Father Absence and Family Meaning Making The current study did not directly measure family’s perceptions of organization and meaning derived from presence or loss of fathers, but, rather, used fathers’ consistency of presence across time as a proxy indicator. Therefore, this study is presented from an exploratory perspective of this construct, suggesting areas for future inquiry. Future scholarship might address mediators of the relationship between consistent father presence and children’s regulatory outcomes, including measures of 95 family perceptions of role development and qualitative indicators about the coherence in meaning derived from fathers’ presence or absence. Overall, nearly four of every ten children in the United States are growing up in homes without their biological fathers consistently present (Horn, 2006). However, these findings may eclipse the non-residential but consistently present fathers, and other research with this EHSRE dataset in particular demonstrates that low-income children may experience significant contact with their fathers. For example, 80% of 2-year-olds in the EHSRE had some degree of access to their biological fathers; by 36 months, 72% of these children had contact with their biological fathers, and researchers report that father presence was generally consistent for these children across toddlerhood (Boiler et al., 2006; Cabrera, Moore, Bronte-Tinkew, Halle, West et al., 2004). Boundary ambiguity is “a state in which family members are uncertain in their perception about who is in or out of the family and who is performing what roles and tasks within the family system” (Boss & Greenberg, 1984, p. 536). The meanings that family members assign to the loss predict the potential for ambiguities in boundaries to occur and place individuals and systems at risk. In response to a loss (psychological or physical), families’ whose individual and collective meaning-making processes are less consistent, less coherent, or unresolved are less likely to reorganize effectively. Children participate in and experience their families’ meaning-making processes. Therefore, mothers’ reports of the degree of consistency in fathers’ presence in the family may be an important indicator of such meaning-making processes, and, further, children’s regulatory competencies may demonstrate the ways in which toddlers reflect their understanding of family-level meanings and organization. 96 Comparisons Based on Child Gender These findings did not significantly differ based on child gender. Boys and girls did not differ in regard to whether or not fathers were consistently present, nor did they differ in regard to whether or not consistent presence of their fathers predicted regulatory competencies over time. This is surprising given the research that demonstrates that fathers may be more likely to be involved with their sons, including direct father involvement such as engaging in caregiving and play and indirect father involvement such as attending school meetings, especially when they are school-age and older (Amato & Keith, 1991; Wood & Repetti, 2004), and that boys may be more likely to benefit from fathers’ presence at varying ages (Feldman, 2003; Pederson, Rubenstein, & Yarrow, 1979). However, the body of literature related to fathering and child gender is still small (e. g., Crouter, Perry-Jenkins, Huston, & McHale, 1987; Diener, Mangelsdorf, McHale, & Frosch, 1998; Snow, J acklin, & Maccoby, 1983), and there remains gaps in the body of knowledge about these relationships. The findings herein may be interpreted in a few ways. First, this study may suggest that the ways in which families organize in regard to parenting roles may not differ based on child gender. Further, it might be that child gender differences predicting father involvement may not be as present during early childhood as when children are older. It is also likely that children in Early Head Start programming benefitted from targeted initiatives to help keep fathers engaged. Wood and Repetti (2004) suggest that father differences in involvement based on child gender are likely to reflect the attitudes and behaviors of mothers around fathering role expectations and child gender. Because the current study measured father presence based on maternal report, it is a possible explanation of the findings that mothers may be 97 :- a1.J“u"'r-‘. J. JL "5‘" less likely to differentiate their expectations about fathering based on child gender when children are very young. On the other hand, this study did not account for the gender of other children living in the home, and, if parenting role development does indeed relate to child gender, variation in sibling gender of participant families would likely matter. This finding, therefore, is presented cautiously, noting the limitations of important excluded variables such as the gender of other children in the home and information about the quality of interactions between fathers and children, and simply suggests that this is an area that likely would benefit from further, future inquiry. Comparisons Based on Racial/Ethnic Groups Tun-arm '1 in» ._._._ The findings, however, did differ based on racial/ethnic groups such that the relationship between consistent father presence and toddlers’ regulatory competencies were more robust among White families than among Black families, echoing previous research with Early Head Start children (Vogel et al, 2006). New families may be reflecting social and cultural norms to a higher degree than more well-established families as they search for and develop their own rules, roles, and patterns. Children in Black families are likely to be raised in female-headed households, with the majority of Black children, approximately 60%, living within this kind of family composition, suggesting this is increasingly the norm in this community (McAdoo & McAdoo, 2002). Black families are more likely to emphasize broader kinship networks in defining patterns of caregiving while White families are more likely to emphasize the nuclear, two-parent family, and men and women in Black families are more likely to parent in egalitarian ways than are men and women in White families (Barbarin, 1983; McAdoo, 1993). Further, the fathering role may reflect social issues and community norms. 98 Fatherhood is often defined by the provider role, across cultures, and disproportionate rates of incarceration, unemployment, and educational truancy among young Black men may contribute to the changing expectations of the fathering role in low-income Black families. These cultural and social norms that inform parenting role development fiirther reflect the ways in which the measurement of fathers’ presence is a proxy for family organizational patterns and the meaning families derive from such presence. White children were more likely to have fathers consistently present across toddlerhood as compared to Black children, and a lack of consistent father presence disproportionately impacted White children in regard to poorer related outcomes. Because consistent father presence may be more normative among White families, the meaning about father presence or absence may be particularly salient to family organization and thus predictive of children’s self-regulation. This is not to say, however, that fathers are not important contributors to child development in Black families, and research consistently demonstrates that father presence is a protective factor, particularly for children in racial/ethnic minority groups (Jeynes, 2003; Martinez, DeGarmo, & Eddy, 2004). This study does not measure the multitude of ways fathers interact with, protect, and care for their children and, therefore, these findings cannot be extended to the whole of what is fathering in White families, Black families, or Hispanic families. Instead, these findings may be interpreted within a framework suggesting the importance of organization in the fathering role and creating a match between what is and what is expected. 99 Limitations to the Current Study There are important limitations to the interpretation of the current study to be addressed. To begin, the fit of the model overall, with an RMSEA statistic of .08 is acceptable in the context of the other fit indices presented, but it does not represent a very close fit of the data to the model. Furthermore, the effect size of consistent father presence on the development of children’s emotion regulation is small. Considering this r“ information overall suggests that, while consistent father presence is likely important in regulatory development, there are additional variables not accounted for in the current , model which play an important role in children’s regulatory development. Most notable . among these is the measurement of the children’s relationships with their mothers, the R primary caregivers in the current study. Furthermore, the implicit assumption of this study is that family’s sense of coherence and organization mediates the relationship between consistent father presence and children’s regulatory development. Future inquiry might include such a measure to determine if model fit increases with its presence. 100 CHAPTER 8: FINAL DISCUSSION AND CONCLUSIONS Overall, this collection of studies explores the ways in which boundary ambiguity may manifest among young children and their parents in low-income populations. In specific, results suggest that children demonstrate less adaptive regulatory competencies when mothers are psychologically absent while still physically present and, fiirther, that children demonstrate less adaptive regulatory competencies when fathers are persistently, or inconsistently, physically absent. In the latter case, current family systems models, particularly boundary ambiguity, suggest that in these cases fathers may still be psychologically present via incomplete family role organization, which can create ambiguous definitions of relationships. These themes may be particularly salient among low-income families for whom social stressors accumulate and impact family functioning (Evans, Boxhill, & Pinkava, 2008; McCubbin & Patterson, 1983). The core assumption of boundary ambiguity theory is that a psychological family exists, describing one’s perceptions about who makes up the family, and that the psychological family can differ from the physical family (Boss, 2007). The findings in this study demonstrate that the consistent presence of primary caregivers in children’s lives is important, and, further, imply that inconsistencies may contribute to less optimal regulatory development for young children. The picture for mother-toddler dyads differs from father-toddler dyads, suggesting further that these roles have specific circumscribed boundaries. Such boundaries may be challenging to reorganize for very young families, who are likely still establishing role patterns, when parents are psychologically or physically unable to fulfill those roles. 101 A major difference between the maternal and paternal samples was present in regard to living arrangements. Many of the women in the sample were single while others were married and still others lived with other adults. However, the majority of the men in the study lived with romantic partners/spouses rather than alone or with other adults. This was likely related to the recruitment procedures in that many of these men were located because they were married to the focus children’s mothers, while the mothers in the study comprised a larger sample with greater diversity in living arrangements. Even the men not married to the mothers of the focus children were likely to be living with other partners and spouses, and this may still have reflected a greater degree of stability in living arrangements, contributing to their participation in the study. Still, these very b- different kinds of family compositions are likely important contributors to the ways in which family boundaries are created and maintained. Indeed, results demonstrated that single mothers demonstrated greater psychological absence as compared to mothers who reported the presence of alternate caregivers. Comparisons Based on Child Gender Interestingly, there were mixed findings based on comparisons between boys and girls. Within the set of analyses related to maternal psychological absence and children’s regulatory outcomes, boys’ outcomes differed from girls in relationship to maternal depressive symptoms, but psychological absence appeared to equally impact boys and girls. Furthermore, there were no gender differences based on father’s physical absence. Taken together, considering previous research demonstrating child gender differences as a result of differential parent-child interaction, and within the framework guiding this 102 study, these findings may suggest that ambiguities in family boundaries impact boys and girls similarly. Comparisons Based on Racial/Ethnic Groups There were also mixed findings related to between-group comparisons based on racial/ethnic groups. Though there were no significant differences based on racial/ethnic groups in the maternal models of psychological absence and detachment, there were 1,. significant differences based on racial/ethnic groups in the model exploring the relationship between continuous biological father presence and young children’s regulatory competencies. These findings taken together suggest there are likely differences in how cultural groups define fatherhood, but the definition of motherhood V may be more consistent across motherhood. Children’s Meaning-Making Processes and Holistic Family Attributes These studies raise important questions about how young children’s participation in family structural processes is both conceptualized and measured. Utilizing statistical methodologies which support the assumptions of latent constructs and nested data, the findings presented here indicate that children reflect meaning derived from the degree of consistency in relationships with their primary caregivers. A mismatch between perceptions regarding caregivers’ psychological and physical presence may result in less adaptive regulatory strategies for young children. Further, the implication of this study is that such a mismatch occurs within caregivers’ ability to organize and regulate family codes and patterns effectively, which may be jeOpardized when mothers are depressed, or fathers are absent. Neither scenario includes a ritualized way to bring closure but often rather requires a span of time for family members to create a sense of meaning, adapt, 103 parenting efficacy, as well as social support, and these may be lacking for many new parents in low-income populations (McCubbin & Patterson, 1983; Raikes & Thompson, 2005) Future Implications for Research and Practice The study of boundary ambiguity is likely important to continue to explore during this particular part of the lifespan given that the toddlerhood years are a period of great change in development overall and self-regulation skills in particular. As described, previous research demonstrates that toddlers are increasingly developing the skills necessary to participate in the psychological family including participating in system- level rules and symbolic practices (Adamson, Bakeman, & Deckner, 2004, Brownell & Kopp, 2007), developing mental representations of relationships based on repeated experiences (Crockenberg & Leerkes, 2000), seeking to develop social relationships and attain group membership (Edwards & Liu, 2002), and engaging in interactions in more emotionally complex ways (Fiese, 1990; Lieberman & Slade, 2000). Therefore, the clinical implications of this study are that interventions should target parents’ awareness of role attributes and family organization patterns in the families in which they were raised, seeking to ascertain expectations, as well as the families they are creating, including their understanding of their children’s developing competencies. Further, interventions may be most effective, when primary caregivers are psychologically unavailable or physically unavailable, if they help family members develop more organized patterns of caregiving so that children have the opportunity to attain developmental competencies within predictable and consistent relationships. 104 Appendix A Figures and Tables Figure 1. Conceptual model defining psychological absence. CES-D PSI AAPI Parental Depression Parenting Lack of Detachment Distress Empathy Parental Parental Depression Psychological Absence Cumulative / \ v / \ EH3 Risk < \ l v Program ‘ ... ( Status EASI ’ . ‘ Child A ,“ ‘ Household Emotionality \ [’p ‘ r Composrtron Child Par ent- Socio- Child Emotional Interaction Outcomes Child Parental BBRS. BER," Onentatlon Emotlon Engagement Support and . E Regulation ngagement 105 Figure 2. Conceptual model demonstrating the longitudinal development of parental detachment and toddlers’ emotion regulation based on linear latent growth curve measurement models. 14m 24m 36m Emotion Emotion Emotion Regulation Regulation Regulation 1“ Emotion Regulation Emotion Regulation Intercept Household Composition ‘v/I \Vv EASI Ch'ld )9 V Emotiolnality Cumulative ‘ Risk \ ‘ ‘ \ EHS CESD- ‘ ‘ ; Program Parental , \ ,’ \ Status Depression Parental Detachment Parental Detachment Intercept 14m 24m 36m Parental Parental Parental Detachment Detachment Detachment 106 Figure 3. Conceptual model demonstrating consistency of father presence predicting longitudinal development of toddlers’ emotion regulation. COHSiSlenC)’ Cumulative EHS EASI Of Father Risk Program Chlld . Presence Status Emotionality Emotion Regulation Emotion Regulation Intercept 14m . 24m 36m Emotion Emotion Emotion 1 Regulation Regulation Regulation 107 Figure 4. Measurement of boundary ambiguity in the mother-child relationship. CES-D PSI AAPI Parental Depression Parenting Lack of Detachment Distress Empathy 41* .63* .32* Parental Parental Depression Psychological Absence . .l 1* ' Cumulative I n EHS Risk v w Program V . v Status .08* .0 .17* EASI é Q Child A ,A b Household Emotionality /’ Q _'04 Composition Parent- Child Child Socio- Interaction Emotional Outcomes '79]. 34* .73: .51" BBRS BBRS Engagement . Support and Regulation ‘ 1 Engagement L *=t>l.96 X2=301-98, df=28, p<.05; RMSEA=.06; CFI=.95; NFI=.95; SRMR=.O3 108 n" 'F Figure 5. Relationship between maternal detachment and emotion regulation across toddlerhood based on latent grth curve models. 14m 24m 36m Emotion Emotion Emotion Regulation Regulation Regulation 1W Emotion Emotion f Regulation Regulation Intercept Household - \. v Composition W v EASI ‘ .0 ' \ -.15* b.08 Cl‘uld Emotionality Cumulative Risk .3. a a! ., CESD- Program . 1* Parental A Status Depression .0 r A Parental Parental Detachment Detachment Intercept r V] 2 14m 24m 36m Parental Parental Parental Detachment Detachment Detachment *=t>l .96 x’=201.88, dfi28, p<.05; RMSEA=.04; CFI=.95; NFI=.94; SRMR=.04 109 Figure 6. Relationship between consistency of father presence and emotion regulation across toddlerhood in a single-domain latent growth curve model. Consistency Cumulative EHS EASI of Father Risk Program Child . Presence Status Emotionality Emotion Regulation Emotion Regulation Intercept 14m 24m 36m Emotion Emotion Emotion Regulation Regulation Regulation *=t>1 .96 x2=106.13, df=11, p-value=.00; RMSEA=.08; GFI=.98; SRMR=.05 110 Figure 7. Relationship between consistency of father presence and emotion regulation across toddlerhood in a single-domain latent grth curve model among White families. Consistency Cumulative EHS EASI of Father Risk Program Child . Presence Status Emotionality 67 -.26* .36*-.O4 '92 18* M 69* Emotion Emotion Regulation Regulation Intercept 14m 24m 36m Emotion Emotion Emotion I Regulation I Regulation Regulation *=t>l.96 x2=53.30, df=11, p-value=.00; RMSEA=.08; GFI=.98; SRMR=.05 111 Figure 8. Relationship between consistency of father presence and emotion regulation across toddlerhood in a single-domain latent growth curve model among Black families. Consistency Cumulative EHS EASI Of Father Risk Program Clllld . Presence Status Emotionality Emotion Regulation Emotion Regulation Intercept 14m I 24m . 36m Emotion Emotion Emotion Regulation Regulation Regulation *=t>1 .96 x2=35.24, df=11, p-value=.04; RMSEA=.O7; GFI=.98; SRMR=.05 112 Table 2 Sample Demographic Characteristics of Mothers in Complete Sample (N =2, 632) Mother Characteristics % N Household Composition Single 35 915 Married 26 674 Living with other adults 39 1,038 Employment Unemployed 54 l ,3 83 School/T raining 22 555 Employed 24 613 Education Less than high school 47 1,186 High school diploma or GED 29 727 More than high school 24 622 113 "l Table 3 Sample Demographic Characteristics of Fathers ’ Directly Observed Subsample (N=501) Father Characteristics % N Residential Status Lives with Child 82 412 Does Not Live with Child 18 89 Employment Unemployed 7 28 Employed 93 391 Education Less than high school 10 48 High school diploma or GED 57 280 More than high school 33 161 114 Table 4 Descriptive Statistics for Variables in Cross-Sectional Maternal Psychological Absence and Boundary Ambiguity Measurement Model and Path Model Variable Mean (SD) Minimum Maximum Skewness Kurtosis Parental Depressive 4.95 (5.19) .00 34.00 1.82 4.68 Symptoms Psychological Absence Parenting Distress 19.12 (5.49) 12.00 43.00 .85 .66 lack of Empathy 3.06 (.67) 2.00 5.00 .59 -.06 Parental Detachment 1.14 (.23) 1.00 3.00 4.27 21.88 Parent-Child Interaction Parental Supportiveness 4.11 (.55) 2.00 6.33 .15 3.26 Child Engagement 5.01 (.57) 3.00 7.00 -.77 3.50 Child Socioemotional Outcomes Emotion Regulation 3.97 (.71) 1.43 5.00 -.81 .67 Orientation and 3.85 (.73) 1.60 5.00 -.50 -.32 Engagement 115 Table 5 Descriptive Statistics for Variables Composing Emotion Regulation and Maternal Detachment Latent Growth Curves Mean (SD) Minimum Maximum Skewness Kurtosis Emotion Regulation 14 Months 3.69 (.61) 1.14 5.00 ~.79 24 Months 3.63 (.69) 1.00 5.00 -.74 36 Months 3.92 (.63) 1.00 5.00 -.92 . Parental I Detachment 14 Months 1.62 (.88) 1.00 7.00 2.20 5.97 24 Months 1.45 (.74) 1.00 7.00 2.80 10.65 36 Months 1.24 (.49) 1.00 6.00 3.81 19.98 116 Table 6 Description of Children ’5 Contact with Non-Residential Biological Fathers Across Toddlerhood 14 Months 24 Months 36 Months Every day or almost every day 152 (27%) 143 (26%) 1 16 (21%) A few times a week 121 (21%) 106 (19%) 110 (20%) A few times a month 106 (19%) 127 (23%) 141 (25%) About once a month 47 (8%) 50 (9%) 41 (7%) Less often than that 55 (10%) 67 (12%) 86 (15%) Never 89 (16%) 61 (11%) 63 (11%) 117 Table 7 Mean Scores for Toddlers ' Emotion Regulation at Each T imepoint Within Groups Based on Consistency of Father Presence Father Consistently Father Not Present Across Consistently Present Toddlerhood Across Toddlerhood Emotion Regulation at 14 Months 3.73 (.69) 3.67 (.67) Emotion Regulation at 24 Months 3.67 (.82) 3.56 (.79) Emotion Regulation at 36 Months 3.97 (.77) 3.86 (.76) 118 Appendix B Attrition Analyses Analyses described below attend to potential differences among participants based on missing data. Mother Parent Interview Among the sample of 2,632 mothers, 1,690 mothers (64%) had complete data for the parent interview at the three timepoints; 264 mothers (10%) had complete data for the parent interview at the l4-month and 24-month birthday-related assessments but not for ; the 36-month assessment; 203 mothers (8%) had complete data for the parent interview at E only the 14-month birthday-related assessment; 172 mothers (7%) had complete data at the 14-month and 36-month birthday-related assessments but not the 24-month birthday- related assessment; 141 mothers (5%) had complete data at the 24-month and 36-month birthday-related assessments but not the l4-month assessment; 95 mothers (4%) had complete data at the 36-month birthday-related assessment but not at the other two timepoints; 56 mothers (2%) had complete data at only the 24-month birthday-related assessment but not at the other two timepoints; and 11 mothers (<1%) had no data for the parent interview across timepoints. Mother-Child Three-Bag Task Among the sample of 2,632 mothers, 1,153 mothers (44%) had complete data for the three-bag task at the three timepoints; 325 mothers (12%) had complete data for the three-bag task at the l4-month and 24-month birthday-related assessments but not for the 36-month assessment; 290 mothers (11%)had complete data for the three-bag task at only the l4-month birthday-related assessment; 178 mothers (7%) had complete data at 119 the 14-month and 36-month birthday-related assessments but not the 24-month birthday- related assessment; 204 mothers (8%) had complete data at the 24-month and 36-month birthday-related assessments but not the l4-month assessment; 117 mothers (4%) had complete data at the 36-month birthday-related assessment but not at the other two timepoints; 104 mothers (4%) had complete data at only the 24-month birthday-related assessment but not at the other two timepoints; and 261 mothers (10%) had no data for the three-bag task across timepoints. Child Bayley Assessment Among the sample of 2,632 children, 1,321 children (50%) had complete data for the Bayley at the three timepoints; 320 children (12%) had complete data for the Bayley at the 14-month and 24-month birthday-related assessments but not for the 36-month assessment; 269 children (10%) had complete data for the Bayley at only the l4-month birthday-related assessment; 153 children (6%) had complete data at the 14-month and 36-month birthday-related assessments but not the 24-month birthday-related assessment; 199 children (8%) had complete data at the 24-month and 36-month birthday-related assessments but not the 14-month assessment; 107 children (4%) had complete data at the 36-month birthday-related assessment but not at the other two timepoints; 97 children (4%) had complete data at only the 24-month birthday-related assessment but not at the other two timepoints; and 166 children (6%) had no data for the Bayley across timepoints. Comparisons Based on Missingness Among Mothers and Children A series of one-way analyses of variance (ANOVA) were run to determine potential group differences on the study measures and continuous demographic 120 indicators. The Scheffe test was chosen as the post-hoe test for the one-way ANOVA’s because it is considered the most conservative and best test to compensate for potentially spurious results when there are multiple comparisons (Tabachnick & Fidell, 1996). To test differences in the categorical data, cross-tabulations were run and the Pearson Chi- square statistic is reported for significant categorical differences. All demographics reported in sample characteristics were tested for differences across groups as well as the indicators on the cumulative risk index as well as the index itself. Among participants who had data for the parent interview at the l4-month birthday-related assessment (n=2,329), four groups were compared. There were no significant differences on the study measures. While there was a significant ANOVA on the parent supportiveness measure (F =3.71, p<.05), post-hoc Scheffe tests revealed no significant differences between groups (p>.10), suggesting the initial ANOVA finding is spurious. There were several significant differences on demographic indicators between these groups. Participants who had complete data at all three timepoints had significantly less cumulative risk than participants who had data missing at any timepoint, but there were no significant differences in risk between the groups who had missing data (F =8.57, p<.001). Mothers who had complete data at all three timepoints were significantly older on average (m=23.01) than mothers who were missing data at the 24-month assessment only (m=21.59), but there were no other significant between-group comparisons (F =5.74, p<.01). One-way ANOVA initially demonstrated a significant difference between groups in gross income at study enrollment (F =3.09, p<.05), but post-hoc Scheffe tests revealed no significant differences between groups (p>.10), suggesting the initial ANOVA finding is spurious. Participants who had complete data at all timepoints were more likely to be in 121 the Early Head Start program group (x2=7.91, p<.05), were more likely to be White (x2=33.12, p<.001), were more likely to have completed high school or greater (x2=20.40, p<.01), and were less likely to receive welfare (78:13.39, p<.01). Among participants who had data at the 14emonth timepoint for the 3-bag task (n=1,946), four groups were compared. There were significant between-group one-way ANOVA’s for the following study measures at the l4-month assessment: Parent Supportiveness (F=3.53, p<.05) and the BBRS Emotion Regulation subscale (F =2.80, p<.05). Scheffe post-hoe analyses demonstrate no significant between-group differences on these measures, indicating the ANOVA findings are likely spurious. Participants who had complete data all timepoints had significantly lower cumulative risk than participants who did not have data at the 24-month assessment; there were no other significant between-group differences in terms of cumulative risk (F =7.32, p<.01). Participants who had complete data at all timepoints were more likely to be White (38:37.80, p<.001), were more likely to be employed (76:15.41, p<.05), and were less likely to receive welfare (x2=9.12, p<.05). Among participants who had data at the 14-month timepoint for the Bayley (n=2,063), four groups were compared. There were no significant differences between groups on study measures. Participants who had complete data all timepoints had significantly lower cumulative risk than participants who did not have data at the 24— month assessment; there were no other significant between-group differences in terms of cumulative risk (F =7.16, p<.001). Participants who had complete data at all timepoints were more likely to be White (x2=29.17, p<.01), were more likely to be employed (x2=16.25, p<.05), and were less likely to receive welfare (78:14.81, p<.01). 122 Among participants who had data at the 24-month timepoint for the parent interview (n=2, 1 51), four groups were compared. There were no significant differences between groups on study measures. Participants who had complete data at all three timepoints had significantly lower risk than participants missing data at the 36-month timepoint only and participants missing data at the l4-month timepoint only, but there were no other significant differences between groups (F =6.32, p<.001). One-way ANOVA comparisons demonstrated a significant difference between groups based on mother’s age (F =3.93, p<.01) and income (F =3.40, p<.05), but post-hoe tests do not demonstrate significant comparisons, suggesting these relationships are likely spurious. Participants with complete data were more likely to be in the program group (x2=12.14, p<.01), were more likely to be White (38:19.00, p<.05), and were more likely to have complete high school or greater (x2=17 .71, p<.01). Among participants who had data at the 24-month timepoint for the three-bag task (n=1,786), four groups were compared. There were no significant differences between groups on study measures. Participants who had complete data had significantly less cumulative risk than participants who completed only the latter two timepoints as well as participants who completed only the 24-month timepoints (F =7.93, p<.001). One-way ANOVA comparisons demonstrated a significant difference between groups based on income (F =2.76, p<.05), but post-hoe tests do not demonstrate significant comparisons, suggesting these relationships are likely spurious. Participants with complete data were more likely to be White (78:37.58, p<.001), were more likely to have complete high school or greater (38:15.68, p<.05), and were more likely to be employed (78:20.22, p<.01). 123 Among participants who had data at the 24-month timepoint for the Bayley (n=1,93 7), four groups were compared. Participants who had complete data , as compared just to the group of participants who only had data at the latter two timepoints, had significantly less cumulative risk (F=8.l7, p<.001) and significantly higher income (F =6.88, p<.001). Participants with complete data were more likely to be in the program group (x2=9.88, p<.05), White (x2=33/05, p<.001), were more likely to have completed high school or greater 08:19.75, p<.01), were more likely to be employed (78:14.14, 3 . p<.05), and were less likely to receive welfare (x2=8.56, p<.05). Among participants who had data at the 36-month timepoint for the parent 3...; interview (n=2,098), four groups were compared. Participants who had complete data had i significantly higher scores on Child Engagement at the 36-month assessment than participants who only had data for the parent interview at the 36-month assessment, but there were no other differences between subgroups (F =3.01 , p<.05). Participants who had complete data across timepoints had significantly less cumulative risk as compared to all the other groups when data was missing at any timepoint (F =9.44, p<.001). Mothers who had complete data across timepoints were significantly older (F=5. 10, p<.01) and had higher income (F =4.48, p<.01) than mothers who had did not have data at the 24-month assessment but completed the first and last assessment. Participants who had complete data across timepoints were more likely to have a high school education or greater (x2=23.29, p<.01), were more likely to be employed (78:18.30, p<.01), and were less likely to receive welfare (x2=11.49, p<.01). Among participants who had data at the 36-month timepoint for the three-bag task (n=1 ,652), four groups were compared. There were no significant differences between 124 groups on the study measures. Participants with complete data across timepoints had significantly lower risk than participants missing data at any timepoints (F =10.73, p<.001). One—way ANOVA comparisons demonstrated a significant difference between groups based on income (F =2.96, p<.05), but post-hoe tests do not demonstrate significant comparisons, suggesting these relationships are likely spurious. Participants with complete data were more likely to be White (x2=31.56,p<.001), were less likely to have been a teen parent when the focus child was born (XL—“8.79, p<.05), were more likely to have completed high school or greater (38:17.99, p<.01), and were more likely to be employed (x2=25.71, p<.001). Among participants who had data at the 36-month timepoint for the Bayley (n=1 ,780), four groups were compared. One-way ANOVA comparisons demonstrated a significant difference between groups based on Parental Distress at the 36-month assessment (F=2.72, p<.05) as well as in regard to maternal age (F =2.73, p<.05), but post-hoe tests do not demonstrate significant comparisons, suggesting these relationships are likely spurious. Participants who had complete data across timepoints had significantly less cumulative risk as compared to all the other groups when data was missing at any timepoint (F =13. 1 8, p<.001). Participants who had complete data across timepoints had significantly higher income than participants who had complete data at only the 24-month and 36-month assessments (F =7.59, p<.001). Participants with complete data were more likely to be White (38:25.79, p<.01), were more likely to be married 08:16.82, p<.05), were more likely to have completed high school or greater (x2=25.36, p<.001), were more likely to be employed (38:25.50, p<.001), and were less likely to receive welfare (78:14.98, p<.01). 12S Overall, there appear to be significant differences relative to missingness based on demographic indicators, some of which relate to identifying risk among families, but there do not appear to be major associations between missingness and the main study variables. The single exception to this is the significant finding that participants who only completed the 36—month parent interview have significantly lower scores on the Child Engagement measure than participants who completed all three timepoints. However, this finding was not replicated when comparing groups based on missingness on the play task data, nor were there any other significant differences between groups on study measures based on missingness. Therefore, overall, the missingness is likely MAR as opposed to N1, and I will proceed with imputation of the mother and child data. Fathers’ Data Direct measurement of fathers via fathers’ self-report and observations of fathers and their children in a play task are only used at the 36-month timepoint in the current study, so missingness will only be reported at this timepoint. As described, data are complete at the 24-month timepoint based on the inclusion criteria for the study, so the data at this timepoint will be used to inform later imputation. Among the complete study sample of 759 fathers, 501 men participated in the 36- month father interview, and 196 men participated in the three-bag task. Biological fathers were more likely than non-biological fathers to have complete data on the three-bag task, and residential fathers were more likely than non-residential fathers to have complete data on the three-bag task (38:10.33, p<.05). There were no other demographic differences between fathers who had complete data on the father interview or the three- bag task and those who did not. 126 Fathers’ scores on the study measures did not differ at the 24-month assessment if they dropped out subsequently. However, the children of the men who did not complete the parent interview had significantly worse scores on the BBRS Emotion Regulation scale at 36 months (r=-3.13, p<.01). The children of the men who did not complete the three-bag play task had significantly worse scores at 24 months (Emotion Regulation: t=- 2.01, p<.05; Orientation and Engagement: r=-l .96, p<.05) and 36 months (Emotion 1. Regulation: t=-4.56, p<.001; Orientation and Engagement: t=-2.31, p<.05) (but not 14 months) on both of the Bayley Behavior Rating Scales. Among the 501 men who did complete the father interview at the 36-month assessment, the children of the men who did not also complete the play task had significantly worse scores on the Bayley Behavior Rating Scales at 36 months (Emotion Regulation: t=-3.67, p<.001; Orientation and Engagement: t=-2.24, p<.05) but not at the earlier timepoints. Within that group of 501, a series of t-tests revealed that this result held in various subgroups, including only the program group, only residential fathers, and only children without established biomedical and environmental risks. While it appears that the children of the men who dropped out were significantly different, there are few differences among the men themselves. Reducing the sample size to 196 would not only reduce power in the main study analyses, but doing so would also throw out, perhaps unnecessarily, the variability in the data from the father interview (e. g., CES-D scores, Lack of Empathy scores, and Parental Distress scores) and create a different bias in the study findings. Therefore, the sample of 501 fathers will be utilized, data will be imputed for fathers with missing information in the three-bag task, and this issue will be noted among study limitations. 127 REFERENCES Abelson, R. P. (1995). Statistics as principled argument. Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Abidin, R. R. (1995). Parenting Stress Index (3rd ed.). Odessa, FL: Psychological Assessment Resources. Ackerman, J. P. & Dozier, M. (2005). The influence of foster parent investment on . children's representations of self and attachment figures. Journal of Applied I Developmental Psychology, 26(5), 507-520. Acock, A. C. (2005). Working with missing values. Journal of Marriage and Family, . 67(4), 1012-1028. L- Adarnson, L. B., Bakeman, R., & Deckner, D. F. (2004). The development of symbol- infused joint engagement. Child Development, 75(4), 1171-1187. Amato, P. R. & Keith, B. (1991). Parental divorce and the well-being of children: A meta-analysis. Psychological Bulletin, 110, 26-46. Amato, P. R. & Rivera, F. (1991). Paternal involvement and children’s behavior problems. Journal of Marriage and the Family, 61, 375-384. Antonovsky, A. & Sourani, T. (1988). Family sense of coherence and family adaptation. Journal of Marriage and the Family, 50, 79-92. Aronson, S. R. & Huston, A. C. (2004). The mother-infant relationship in single, cohabiting, and married families: A case for marriage? Journal of Family Psychology, 18(1), 5-18. Bagarozzi, D. A. & Berry, J. T. (1991). A review of the boundary ambiguity scales. American Journal of Family Therapy, 18(4), 393—397. Bavolek, S. (1984). Handbook for the Adult-Adolescent Parenting Inventory. Park City, Utah: Family Development Resources, Inc. Bayley, N. (1993). Bayley Scales of Infant Development: Manual (2nd ed. ). New York: Harcourt B & Company. Berlin, L. J. & Cassidy, J. (2000). Understanding parenting: Contributions of attachment theory and research. In J. D. Osofsky & H. E. Fitzgerald (Eds), WAIMH handbook of infant mental health (pp. 133-170). New York: Wiley & Sons. 128 Bertalanffy, L. von. (1968). General systems theory. New York: George Braziller. Bertalanffy, L. von (1975). Perspectives on General Systems Theory: Scientific- philosophical studies. New York: George Braziller. Black, M. M., Dubowitz, H., & Starr, R. H. (1999). Afiican American fathers in low- income, urban families: Development, behavior, and home environment of their three-year-old children. Child Development, 70(4), 967-97 8. Bocknek, E. L., Sanderson, J ., Britner, P. A. (2009). Ambiguous loss and post-traumatic stress in a sample of school-age children of prisoners. Journal of Child and Family Studies, 18(3), 323-333. Bollen, K. A. (1989). Structural equations with latent variables. New York: John Wiley & Sons. Boller, K., Bradley, R., Cabrera, N., Raikes, H., Pan, 3., Shears, J., & Roggman, L. (2006). The Early Head Start father studies: Design, data collection, and summary of father presence in the lives of infants and toddlers. Parenting, 6, 117 — 143. Boomsma, A. (2000). Reporting analyses of covariance structures. Structural Equation Modeling, 7(3), 461-483. Bomstein, M. H. (2002). Parenting infants. In M. H. Bomstein (Ed.), Handbook of parenting, Vol. 1: Children and parenting (2nd ed., pp. 3-43). Mahwah, NJ: Lawrence Erlbaum Associates. Boss, P. (2007). Ambiguous loss theory: Challenges for scholars and practitioners. Family Relations: Special issue an ambiguous loss, 56, 105-111. Boss, P. (2002). Family stress management: A contextual approach. Thousand Oaks: CA, Sage. Boss, P. (1977). A clarification of the concept of psychological father presence in families experiencing ambiguity of boundary. Journal of Marriage and the Family, 39, 141—151. Boss, P., Caron, W., Horbal, J ., & Mortimer, .I. (1990). Predictors of depressive symptoms in caregivers of dementia patients: Boundary ambiguity and mastery. Family Process, 29, 245—254. Boss, P., & Greenberg, J. (1984). Family boundary ambiguity: A new variable in family stress theory. Family Process, 23, 535—546. 129 Boss, P., Greenberg, J ., & Pearce-McCall, D. (1990). Measurement of boundary ambiguity in families. Station Bulletin (5 93—1 990). Minneapolis: University of Minnesota, Agricultural Experiment Station. Boss, P. G. (1980). Normative family stress: Family boundary changes across the lifespan. Family Relations, 29, 445-450. Boss, P. G. (1992). Primacy of perception in family stress theory and measurement. Journal of Family Psychology, 6, 1 13-1 19. Bowlby, J. (1969, 1973, 1980). Attachment and loss: Vol. 1-3. New York: Basic Books. Bronfenbrenner, U. (1979). The ecology of human development: Experiments by nature and design. Cambridge, MA: Harvard University Press. Bronfenbrenner, U. (1989). Ecological systems theory. In R. Vasta (Ed.), Six theories of child development: Revised formulations and current issues (pp. 185-246). Greenwich, CT: JAI Press. Bronfenbrenner, U. & Ceci, S. J. ((1994). Nature-nurture reconceptualized in developmental perspective: A bioecological model. Psychological Review, 101, 568-586. Bronfenbrenner, U. & Morris, P. A. (1998). The ecology of developmental process. In W. Damon (Series Ed.) & R. M. Lerner (Vol. ed.), Handbook of child psychology: Vol. 1. Theoretical models of human development (5th ed., pp. 993-1028). New York: Wiley. Brophy-Herb, H. E., Gibbons, C., Omar, M. A., & Schiffrnan, R. F. (1999). Low-income fathers and their infants: Interactions during teaching episodes. Infant Mental Health Journal: Special issue on fathers and infants, 20(3), 305-321. Brophy-Herb, H. E., Schiffrnan, R. F., Bocknek, E. L., Dupuis, S. B., Fitzgerald, H. E., et al. (under review). Toddlers' social-emotional competence in the contexts of parental emotion beliefs and contingent responsiveness. Browne, M. W. & Cudeck, R. (1993). Alternative ways of assessing model fit. In K. A. Bollen, & J. S. Long (Eds), Testing Structural Equation Models (pp. 136—162). Beverly Hills, CA: Sage. Buehler, C. & Pasley, K. (2000). Family boundary ambiguity, marital status, and child adjustment. Journal of Early Adolescence, 20(3), 281-308 . Burchinal, M., Nelson, L., & Poe, M. (2006). Growth curve analysis: An introduction to various methods for analyzing longitudinal data. Monographs of the Society for Research in Child Development, 71(3), 65-87. 130 Buss, A. H., & Plomin, R. (1984). Temperament: Early developing personality traits. Hillsdale, NJ: Erlbaum. Byng-Hall, J. (2002). Relieving parentified children’s burdens in families with insecure attachment patterns. Family Process, 41, 375-388. Cabrera, N., Fitzgerald, H. E., Bradley, R. H., & Roggman, L. (2007). Modeling the dynamics of paternal influences on children over the life course. Applied Developmental Science, 11 (4), 185-189.Cabrera, N. & Peters, H. E. (2000). Public policies and father involvement. Marriage and Family Review, 29(4), 295- 314. Cabrera, N. J ., Ryan, R. M., Mitchell, S. J ., Shannon, J. D., & Tamis-LeMonda, C. S. (2008). Low-income, nonresident father involvement with their toddlers: Variation by fathers' and ethnicity. Journal of Family Psychology, 22(4), 643- 647. Cabrera, N. J ., Shannon, J. D., & Tamis-LeMonda, C. (2007). Fathers’ influence on their children’s cognitive and emotional development: From toddlers to pre-k. Applied Developmental Science, 11 (4), 208-213. Cabrera, N. J ., Tamis-LeMonda, C. 8., Lamb, M. E., & Boller, K. (1999). Measuring father involvement in the Early Head Start evaluation: A multidimensional conceptualization. Paper presented at the National Conference on Health Statistics, Washington, DC. Calkins, S. (1994). Origins and outcomes of individual differences in emotion regulation. Monographs of the Society for Research in Child Development, 59, (No 2/3), 53- 72. Calkins, S. D., Dedmon, S. E, Gill, K. L., Lomax, L. E., & Johnson, L. M. (2002).Frustration in infancy: Implications for emotion regulation, physiological processes, and temperament. Infancy, 3 (2), 175-197. Calkins, S.D. & Johnson, M.C. (1998). Toddler regulation of distress to frustrating events: Temperamental and maternal correlates. Infant Behavior and Development, 21, 379-395. Campbell, S. B., Brownell, C. A., Hungerford, A. H., Spieker, S. J., Mohan, R., & Blessing, J. S. (2004). The course of maternal depressive symptoms and matemal sensitivity as predictors of attachment security at 36 months. Development and Psychopathology, 16, 231-252. Carroll, J. S., Olson, C. D., & Buckmiller, N. (2007). Family boundary ambiguity: A 30- year review of theory, research, and measurement. Family Relations: Special issue on ambiguous loss, 56, 210-2302. 131 Carter, A. S., Garrity-Rokous, E., Chazan-Cohen, R., Little, C., & Briggs-Gowan, M. J. (2001 ). Maternal depressive symptoms and comorbidity: Predicting early parenting, attachment security, and toddler social-emotional problems and competencies. Journal of the American Academy of Child & Adolescent Psychiatry, 40(1), 18-26. Chazan-Cohen, R., Ayoub, C., Pan, B. A., Roggman, L., Raikes, H., McKelvey, L., Whiteside-Mansell, L., & Hart, A. (2007). It takes time: Impacts of Early Head Start that lead to reductions in maternal depressive symptoms two years later. Infant Mental Health Journal: Special issue on Early Head Start, 28(2), 151-170. Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). New York: Academic Press. Cole, P. M., Martin, S. E., & Dennis, T. A. (2004). Emotion regulation as a scientific construct: Methodological challenges and directions for child development research. Child Development, 75(2), 317— 333. Cottrell, D. & Boston, P. (2002). Practitioner Review: The effectiveness of systemic family therapy for children and adolescents. Journal of Child Psycholog and Psychiatry, 43(5), 573-586. Cowan, P. A. (1997). Beyond meta-analysis: A plea for a family systems view of attachment. Child Development, 68(4), 601-603. Cowan, P. A., Cohn, D. A., Cowan, C. P., & Pearson, J. L. (1996). Parents’ attachment histories and children’s externalizing and internalizing behaviors: Exploring family systems models of linkage. Journal of Consulting and Clinical Psychology, 64, 53-63. Crockenberg, S. B. & Leerkes, E. (2000). Infant social and emotional development in family context. In C. H. Zeanah (Ed.), Handbook of infant mental health (2nd ed., pp. 60—90). New York: Guilford Press. Crockenberg, S. B. & Leerkes, E. (2003). Infant negative emotionality, caregiving, and family relationships. In A. C. Crouter & A. Booth (Eds), Children ’s influence on family dynamics (pp. 57-78). Crockenberg, S. B. & Smith, P. (2002). Antecedents of mother—infant interaction and infant initability in the first 3 months of life. Infant Behavior and Development, 25(1), 2-15. Crouter, A. C., Perry-Jenkins, M., Huston, T. L., & McHale, S. M. (1987). Processes underlying father involvement in dual-eamer and single-earner families. Developmental Psychology, 23, 431-440. 132 Cummings, E. M. & Davies, P. T. (1994). Maternal depressive symptoms and child development. Journal of child psychology and psychiatry, 35(1), 73-122. Cummings, E. M. & Davies, P. T. (1996). Emotional security as a regulatory process in normal development and the development of psychopathology. Development and Psychopathology, 7, 123-139. Cummings, E. M., Keller, P. S., & Davies, P. T. (2005). Towards a family process model of maternal and paternal depressive symptoms: Exploring multiple relations with child and family functioning. Journal of Child Psychology and Psychiatry, 46(5), 479-489. Cummings, E. M., Schermerhom, A. C., Keller, P. S., & Davies, P. T. (2008). Parental depressive symptoms, children’s representations of family relationships, and child adjustment. Social Development, 1 7(2), 278-305. Cutrona, C. E. & Troutman, B. R. (1986). Social support, infant temperament, and parenting self-efficacy: A meditational model of postpartum depressive symptoms. Child Development, 57, 1507-1518. Dempster, A. P., Laird, N. M., & Rubin, D. B. (1977). Maximum likelihood estimation from incomplete data via the EM algorithm. Journal of the Royal Statistical Society, Series B, 39, 1—38. Diener, M. L. & Mangelsdorf, S. C. (2000). Behavioral strategies for emotion regulation and toddlers: Associations with maternal involvement and emotional expressions. Infant Behavior & Development, 22, 569—583. Doherty, W. J ., Boss, P. G., LaRossa, R., Schumm, W. R., & Steinmetz, S. K. (1993). F arnily theories and methods: A contextual approach. In P. G. Boss, W. J. Doherty, R. LaRossa, W. R. Schumm, & S. K. Steinmetz (Eds.), Sourcebook of family theories and methods (pp. 3-30). New York: Springer. Dudgeon, P. (2003). NIESEM: A computer program for calculating noncentral interval estimates (and power analysis) for structural equation modeling. Melbourne, Australia: Department of Psychology, University of Melbourne. Duncan, T. E. & Duncan, S. C. (2004). An introduction to latent growth curve modeling. Behavior Therapy, 35, 333-363. Duncan, T. E., Duncan, S. C., Hops, & HOps, H. (1998). Latent variable modeling of longitudinal and multilevel alcohol use data. Journal of Studies on Alcohol, 59(4), 399-408. Duvall, E. M., & Hill, R. L. (1948). Report of the committee on the dynamics of family interaction. Washington, DC: National Conference on Family Life. 133 MEI-F13 .ml #415. r ' . a Edwards, C. P. & Liu, W. (2002). Parenting toddlers. In M. H. Bomstein (Ed.), Handbook of parenting, Vol. 1: Children and parenting (2nd ed., pp. 45-71). Mahwah, NJ: Lawrence Erlbaum Associates. Eisenberg, N., Cumberland, A., Spinrad, TL. (1998). Parental socialization of emotion. Psychological Inquiry, 9, 241—273. Eisenberg, N. & Spinrad, T. L. (2004). Emotion related regulation: Sharpening the definition. Child Development, 75, 334-339. Eisenberg, N., Zhou, Q., Spinrad, T. L., Valiente, C., Fabes, R. A., & Liew, J. C. (2005). Relations among positive parenting, children’s effortful control, and externalizing problems: A three-wave longitudinal study. Child Development, 76(5), 1055- 1071. Emde, R. N. (1989). The infant’s relationship experience: Developmental and affective aspects. In A. J. Sameroff & R. N. Emde (Eds), Relationship disturbances in early childhood (pp. 33-51). New York: Basic Books. Enders, C. K. (2006). Analyzing structural equation models with missing data. In G. R. Hancock & R. O. Mueller (Eds), Structural equation modeling: A second course (pp. 313-342). Greenwich, CT: Information Age Publishing. Erel, O. & Burman, B. (1995). Interrelatedness of marital relations and parent-child relations: A meta-analytic review. Psychological Bulletin, 118, 108-132. Evans, G. W., Boxhill, L., & Pinkava, M. (2008). Poverty and maternal responsiveness: The role of maternal stress and social resources. International Journal of Behavioral Development, 32(3), 232-237. Evans, V. J. (2004). Foreword. In R. D. Day & M. E. Lamb (Eds), Conceptualizing and measuring father involvement (pp. ix-xiii). Mahwah, NJ: Lawrence Erlbaum Associates. F abes, R. A., Leonard, S. A., Kupanoff, K, & Martin, C. L. (2001). Parental coping with children’s negative emotions: Relations with children’s emotional and social responding. Child Development, 72 (3), 907-920. Field, T. (1992). Infants of depressed mothers. Developmental Psychopathology, 4, 49- 66. F iese, B. H. (1990). Playful relationships: A contextual analysis of mother-toddler interaction and symbolic play. Child Development, 61(5), 1648-1656. Fitzgerald, H. E., Barnes, J. & Almerigi, J. (2007). A systemic approach to assessment of normative and atypical socioemotional function in toddlers. In C.A. Brownell & 134 C. B. Kopp (Eds), Socioemotional development in the toddler years: Transitions and transformations (pp. 451-482). New York: The Guilford Press. Fitzgerald, H. E., Mann, T., & Barratt, M. (1999). Fathers and infants. Infant Mental Health Journal: Special issue on fathers and infants, 20(3), 213-221. Fonagy, P. & Target, M. (2005). Bridging the transmission gap: An end to an important mystery of attachment research? Attachment and Human Development, 7(3), 333- 343. F raiberg, S. H. (1959). The magic years: Understanding and handling the problems of early childhood. New York: Scribner’s. Garber, J., & Dodge, K. A. (Eds) (1991). The development of emotional regulation and dysregulation. Cambridge: Cambridge University Press. Garcia-Coll, C., Surrey, J. L., & Weingarten, K. (1998). Mothering against the odds: Diverse voices of contemporary mothers. New York: Guilford Press. Garey, A. I. & Arendell, T. (1999, April). Children, work, and family: Some thoughts on “mother blame”: Working paper no. 4. Berkeley, CA: Center for Working Families. ' Gottman, J., Fainsilber-Katz, L., Hooven, C. (1996). Parental meta-emotion philosophy and the emotional life of families: Theoretical models and preliminary data. Journal of Family Psychology, 10, 243-268. Gottman, J ., Fainsilber-Katz, L., Hooven, C. (1997). Meta-emotion: How families communicate emotionally. Mahwah. NJ: Lawrence Erlbaum Associates. Hanna, S. M. & Brown, J. H. (2004). The practice of family therapy: Key elements across models. Belmont, CA: Thomson, Brooks/Cole. Hazen, N., Jacobvitz, D., & McFarland, L. (2005). Antecedents of boundary disturbances in families with young children: Intergenerational transmission and parent-infant caregiving patterns. In P. K. Kerig (Ed.), Implications of parent-child boundary dissolution for developmental psychopathology: "Who is the parent and who is the child? " (pp. 85-110). New York: The Haworth Press. Hill, R. (1958). Generic features of families under stress. Social Casework, 49, 139-150. Hill, R. (1986). Life Cycle Stages for Types of Single Parent Families: Of Family Development Theory. Family Relations, 35(1), 19-29. Hinshaw-Fuselier, S., Heller, S. S., Parton, V. T., Robinson, L., & Boris, N. W. (2007). Trauma and attachment: The case for disrupted attachment disorder. In J. D. 135 Osofsky (Ed.), Young children and trauma: Intervention and treatment (paperback edition, pp. 47-68). New York: The Guilford Press. Hobfoll, S. E. & Spielberger, C. D. (1992). Family stress: Integrating theory and measurement. Journal of Family Psychology, 6, 99-1 12. Hofferth, S. (2002). The demography of fathers. In C. S. Tamis-LeMonda & N. Cabrera (Eds), Handbook of father involvement: Multidisciplinary perspectives (pp. 31- 32). Mahwah, NJ: Lawrence Erlbaum Associates. Hofferth, S., Pleck, J., Stueve, J. L., Bianchi, S., & Sayer, L. (2002). The demography of fathers: What fathers do. In C. S. Tamis-LeMonda & N. Cabrera (Eds), Handbook of father involvement: Multidisciplinary perspectives (pp. 63-90). Mahwah, NJ: Lawrence Erlbaum Associates. Hu, L., & Bentler, PM. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6(1), 1-55. Jackson, D. & Mannix, J. (2004). Giving voice to the burden of blame: A feminist study of mothers’ experiences of mother blaming. International Journal of Nursing Practice, 10, 150-158. Jacobvitz, D., Hazen, N., Curran, M., & Hitchens, K. (2004). Observations of early triadic family interactions: Boundary disturbances in the family predict depressive and anxious symptoms in middle childhood. Development and Psychopathology, 16(3), 577-592. Kaplan, L. & Boss, P. (1999). Depressive symptoms among spousal caregivers of institutionalized mates with Alzheimer’s: Boundary ambiguity and mastery as predictors. Family Process, 38(1), 85—103. Kazak, A. E. (1992). Stress, change, and families: Theoretical and methodological considerations. Journal of Family Psychology, 6, 120-124. Kerig, P. K. (1995). Triangles in the family circle: Effects of family structure on marriage, parenting, and child adjustment. Journal of Family Psychology, 9, 28- 43. Keri g, P. K. (2005). Revisiting the construct of boundary dissolution: A multidimensional perspective. Journal of Emotional Abuse, 5(2/3), 5-42. Klein, D. M. & Jurich, J. A. (1993). Metatheory and family studies. In P. G. Boss, W. J. Doherty, R. LaRossa, W. R. Schumm, & S. K. Steinmetz (Eds), Sourcebook of family theories and methods (pp. 31-67). New York: Springer. 136 Kline, R. B. (2005). Principles and practice of structural equation modeling (2nd ed.). New York: The Guilford Press. Kochanska, G. (1995). Children’s temperament, mothers’ discipline, and security of attachment: Multiple pathways to emerging internalization. Child Development, 66, 597-615. Kochanska, G. (2001). Emotional development in children with different attachment histories: The first three years. Child Development, 72(2), 474-490. Kochanska, G., & Coy, K. C., & Murray, K. T. (2001). The development of self regulation in the first four years of life. Child Development, 72(4), 1091-1111. Lamb, M. E. (1975). Fathers: Forgotten contributors to child development. Human Development, 18, 245-266. Lamb, M. E. (2002a). Infant-father attachments and their impact on child development. In C. S. Tamis-LeMonda & N. Cabrera (Eds), Handbook of father involvement: Multidisciplinary perspectives (pp. 93-118). Mahwah, NJ: Lawrence Erlbaum Associates. Lamb, M.E. (2002b). Nonresidential fathers and their children. In C. S. Tamis-LeMonda & N. Cabrera (Eds), Handbook of father involvement: Multidisciplinary perspectives (pp. 169-184). Mahwah, NJ: Lawrence Erlbaum Associates. LaRossa, R. & Reitzes, D. C. (1993). Symbolic-interactionism and family studies. In P. G. Boss, W. J. Doherty, R. LaRossa, W. R. Schumm, & S. K. Steinmetz (Eds), Sourcebook of family theories and methods (pp. 135-162). New York: Springer. Lavee, Y., McCubbin, H. I., & Patterson, J. M. (1985). The Double ABCX Model of family stress and adaptation: An empirical test by analysis of structural equations with latent variables. Journal of Marriage and the Family, 4 7(4), 81 1-825. Lee, R E. & Whiting, J. B. (2007). Foster children's expressions of ambiguous loss. The American Journal of Family Therapy, 35(5), 417 - 428. Lewis, J. M., Beavers, R. W., Gossett, J. T., & Phillips, V. A. (1976). No single thread: Psychological health in family systems. New York: Brunner/Mazel. Lewis, M. D., Lamey, A. V., & Douglas, L. (1999). A new dynamic systems method for the analysis of early socioemotional development. Developmental Science, 2, 457- 475. Lieberman, A. F. & Slade, A. (2000). Parenting toddlers: Developmental and clinical considerations. In J. D. Osofsky & H. E. Fitzgerald (Eds), WAIMH handbook of infant mental health (pp. 25—56). New York: Wiley & Sons. 137 Love, J. M., Kisker, E. E., Ross, C., Raikes, H., Constantine, J ., Boller, K., Brooks-Gunn, J., Chazan-Cohen, R., Tarullo, L. B., Brady-Smith, C., Fuligni, A. S., Schochet, P. 2., Paulsell, D., & Vogel, C. (2005). The Effectiveness of Early Head Start for 3- year-old children and their parents: Lessons for policy and programs. Developmental Psychology: Special section on the effectiveness of Early Head Start, 41(6), 885—901. Lyons-Ruth, K., Connell, D.B., & Grunebaum, H.U. (1990). Infants at social risk: Maternal depressive symptoms and family support services as mediators of infant development and security of attachment. Child Development, 61, 85—98. MacDonald, K. & Parke, R. D. (1984). Bridging the gap: Parent-child play interaction and peer interactive competence. Child Development, 55, 1265-127 7. Madden-Derdich, D. A., & Arditti, J. A. (1999). The ties that bind: Attachment between former spouses. Family Relations, 48, 243—249. Madden-Derdich, D. A., Leonard, S. A., & Christopher, F. S. (1999). Boundary ambiguity and co-parental conflict after divorce: An empirical test of a family systems model of the divorce process. Journal of Marriage and Family, 61 , 588— 598. Marvin, R. S. & Stewart, R. B. (1993). A family systems framework for the study of attachment. In M. T. Greenberg, D. Cicchetti, & E. M. Cummings (Eds), Attachment in the preschool years: Theory, research, and intervention (pp. 51- 86). Chicago: University of Chicago Press. Martin, S. E., Clements, M. L., & Crnic, K. A. (2002). Maternal emotions during mother- toddler interaction: Parenting in affective context. Parenting: Science and Practice, 2(2), 105-126. McCallum, R. C., Browne, M. W., Sugawara, H. M. (1996). Power analysis and determination of sample size for covariance structure modeling. Psychological Methods, 1(2), 130-149. McCubbin, H. I. & Patterson, J. M. (1982). Family adaptation to crisis. In H. I. McCubbin, A. Cauble, & J. M. Patterson (Eds), Family stress, coping, and social support (pp. 26-47). Springfield, IL: Charles C. Thomas. McCubbin, H. I. & Patterson, J. M. (1983). The family stress process: The Double ABCX model of adjustment and adaptation. In H. I. McCubbin, M. B. Sussman, & J. M. Patterson (Eds), Social stress and the family: Advances and developments in family stress theory and research (pp. 7-37). New York: The Haworth Press. McDonald, R. P. & Ho, M. H. R. (2002). Principles and practice in reporting structural equation analyses. Psychological Methods, 7(1), 64-82. 138 McHale, J. P & Sullivan, M. J. (2008). Family systems. In A. M. Gross & M. Hersen (Eds), Handbook of clinical psychology, vol. 2 (pp. 192-226). New York: Wiley. Mederer, H. & Hill, R. (1983). Critical transitions over the family life span: Theory and research. Marriage and Family Review, 6, 39-60. Michalik, N. M., Eisenberg, N., Spinrad, T. L., Ladd, R, Thompson, M., & Valiente, C. (2007). Longitudinal relations among parental emotional expressivity and sympathy and prosocial behavior in adolescence. Social Development, 16(2), 286- 309. Minuchin, S. (1974). Families and family therapy. Cambridge, MA: Harvard University Press. Minuchin, P. (1985). Families and individual development: Provocations from the field of family therapy. Child Development, 56, 289-302. Minuchin, S., Rosman, B.L., & Baker, L. (1978). Psychosomatic families. Cambridge, MA: Harvard University Press. Moore, G. A. & Calkins, S. D. (2004). lnfants' vagal regulation in the still-face paradigm is related to dyadic coordination of mother-infant interaction. Developmental Psychology, 40(6), 1068-1080. Morris, A. S., Silk, J. S., Steinberg, L., Myers, S. S., & Robinson, L. R. (2007). The role of the family context in the development of children’s emotion regulation. Social Development, 1 6, 361—388. Mulsow, M., Caldera, Y. M., Pursely, M., Reifinan, A., & Huston, A. C. (2002). Multilevel factors influencing maternal stress during the first three years. Journal of Marriage and Family, 64(4), 944-956. Murray, L. (1991). lntersubjectivity, object relations theory, and empirical evidence from mother-infant interactions. Infant Mental Health Journal, 12(3), 219-232. Murray, L., Fiori-Cowley, A., Hooper, R., & Cooper, P. (1996). The impact of postnatal depressive symptoms and associated adversity on early mother-infant interactions and later infant outcome. Child Development, 67, 2512-2526. Musil, C. M., Warner, C. B., Yobas, P. K., & Jones, S. L. (2002). A comparison of imputation techniques for handling missing data. Western J oumal of Nursing Research, 24, 815—829. NICHD Early Child Care Research Network. (1997). The effects of infant child care on infant-mother attachment security: Results of the NICHD study of early child care. Child Deve10pment, 68, 860—879. 139 Osmond, M. W. 8c Thorne, B. (1993). Feminist theories: The social construction of gender in families. In P. G. Boss, W. J. Doherty, R. LaRossa, W. R. Schumm, & S. K. Steinmetz (Eds), Sourcebook of family theories and methods (pp. 591-622). New York: Springer. Osofsky, J. D., & Thompson, M. D. (2000). Adaptive and maladaptive parenting: Perspectives on risk and protective factors. In J. P. Shonkoff & S. J. Meisels (Eds), Handbook of early childhood intervention (2nd ed., pp. 54—75). New York: Cambridge University Press. Palkovitz, R. (2002). Involved fathering and child development: Advancing our understanding of good fathering. In C. S. Tamis-LeMonda & N. Cabrera (Eds), Handbook of father involvement: Multidisciplinary perspectives (pp. 1 19-140). Mahwah, NJ: Lawrence Erlbaum Associates. Parke, R. D. (1981). Fathers. Cambridge: Cambridge University. Parke, R. D. (2000). Father involvement: A developmental perspective. Marriage and Family Review, 29, 43-58. Paquette, D. (2004). Theorizing the father—child relationship: Mechanisms and developmental outcomes. Human Development, 47, 193—219. Pleck, J. (1997). Paternal involvement: Levels, sources and consequences. In M. Lamb (Ed.), The role of father in child development (pp. 66—103). New York: John Wiley & Sons. Pleck, J. (2007). Why could father involvement benefit children? Theoretical perspectives. Applied Developmental Science, 11(4), 196-202. Puma, J ., LeBeouf, W., Carta, J ., Rodriguez, E., Spellman, M., & Watt, N. (unpublished manuscript). Optimizing the power of prediction of a cumulative risk index. Radloff, LS. (1977). The CES-D scale: A self-report depressive symptoms scale for research in the general population. Applied Psychological Measurement, 1, 385— 401. Raikes, H., Pan, B.A., Luze, G., Tamis-LeMonda, C., Brooks-Gunn, J ., Constantine, J., Tarullo, L. B., Raikes, H. A., & Rodriguez, E. T. (2006). Mother-child bookreading in low-income families: Correlates and outcomes during the first three years of life. Child Development, 77(4), 924—953. Raikes, H. A. & Thompson, R. A. (2005). Efficacy and social support as predictors of parenting stress among families in poverty. Journal of Infant Mental Health, 26(3), 177-190. 140 Rane, T. R. & McBride, B. A. (2000). Identity theory as a guide to understanding fathers’ involvement with their children. Journal of Family Issues, 21(3), 347-366. Raykov, T., Tomer, A., Nesselroade, J. R. (1991). Reporting structural equation modeling results in Psychology and Aging: Some proposed guidelines. Psychology and Aging, 6(4), 499-503. Reddy, R., Rhodes, J. E., & Mulhall, P. (2003). The influence of teacher support on student adjustment in the middle school years: A latent growth curve study. Development and psychopathology, 15, 1 19-138. Reiss, D. (1989). The represented and practicing family: Contrasting visions of family continuity. In A. J. Sameroff & R. N. Emde (Eds), Relationship disturbances in early childhood (pp. 191-220). New York: Basic Books. Roberts, P. (1998). Fathers’ time. In E. M. Junn & C. J. Boyatzis (Eds), Child growth and development, annual editions, 98/99 (pp. 146-152). Guilford, CT: Dushkin. Patterson, J. M. & Garwick, A. W. (1994). Levels of meaning in family stress theory. Family Process, 30, 287-304. Robertson, J. & Robertson, J. (1971). Young children in brief separation: A fresh look. Psychoanalytic Study of the Child, 26, 264-315. Robertson, J. & Robertson, J. (1989). Separation and the very young child. London: Free Association Books. Rogers, S. J. & White, L. K. (1998). Satisfaction with parenting: The role of marital happiness, family structure, and parents’ gender. Journal of Marriage and the Family, 60(2), 293-308. Roggman, L. A., Boyce, L. K., & Cook, J. (2001). Widening the lens: Viewing fathers in infants’ lives. In H. E. Fitzgerald, K. H. Karraker, & T. Luster (Eds), Infant development: Ecological perspectives (pp. 193-220). New York: Routledge Falmer. Roggman, L. A., Fitzgerald, H. E., Bradley, R. H., & Raikes, H. (2002). Methodological, measurement, and design issues in studying fathers: An interdisciplinary perspective. In C. S. Tamis-LeMonda & N. Cabrera (Eds), Handbook of father involvement: Multidisciplinary perspectives (pp. 1-30). Mahwah, NJ: Lawrence Erlbaum Associates. Roggman, L. A., Boyce, L. K., Cook, J ., Christiansen, K., & Jones, D. (2004). Playing with daddy: Social toy play, Early Head Start, and developmental outcomes. Fathering, 2(1), 83-108. 141 Rogosch, F. A., Cicchetti, D., & Toth, S. L. (2004). Expressed emotion in multiple subsystems of the families of toddlers with depressed mothers. Development and Psychopathology, 16, 689-709. Rosenberg, M., & Guttmann, J. (2001). Structural boundaries of single-parent families and children’s adjustment. Journal of Divorce & Remarriage, 3 6(1—2), 83—98. Ross, C.E., Mirowsky, J ., & Huber, J. (1983). Dividing work, sharing work, and in- between: Marriage patterns and depressive symptoms. American Sociological Review, 48, 809—823. Sameroff, A. J. (1975). Early influences on development: Fact or fancy? Merrill—Palmer Quarterly, 21, 267-294. Sameroff, A. J. (1989). Principles of development and psychopathology. In A. J. Sameroff & R. N. Emde (Eds), Relationship disturbances in early childhood (pp. 17-32). New York: Basic Books. Sameroff, A. J. (1994). Developmental systems and family functioning. In R. D. Parke & S. Kellam (Eds), Exploring family relationships with other social contexts. Lawrence Erlbaum, Hillsdale, NJ. Sameroff, A. J. (2000). Ecological perspectives on developmental risk. In J. D. Osofsky & H. E. Fitzgerald (Eds), WAIMH handbook of infant mental health: Vol. 4. Infant mental health in groups at high risk (pp. 1-29). New York: Wiley. Sameroff, A. J ., & Chandler, M. J. (1975). Reproductive risk and the continuum of caretaking casualty. In F. D. Horowitz, M. Hetherington, S. Scarr-Salapatek, & G. Siegal (Eds), Review of child development research: Vol. 4 (pp. 187—244). Chicago: University of Chicago Press. Sameroff, A. J. & Emde, R. N. (1989). Relationship disturbances in context. In A. J. Sameroff & R. N. Emde (Eds), Relationship disturbances in early childhood (pp. 221-238). New York: Basic Books. Sameroff, A. J. & Fiese, B. H. (1992). Family representations of development. In I. E. Sigel, A. V. McGillicuddy-DeLisi. & J. J. Goodnow (Eds), Parental belief systems: The psychological consequences for children (2mI ed., pp. 347-369). Sameroff, A. J. & F iese, B. H. (2000a). Models of development and developmental risk. In c. H. Zeanah (Ed.), Handbook ofinfant mental health (2“d ed., pp. 3-19). New York: The Guilford Press. Sameroff, A. J. & Fiese, B. H. (2000b). Transactional regulation: The developmental ecology of early intervention. In J. P. Shonkoff & S. J. Meisels (Eds), Handbook 142 of early childhood intervention (pp. 135-159). Cambridge, UK: Cambridge University Press. Schafer, J. L. & Graham, J. W. (2002). Missing data: Our view of the state of the art. Psychological Methods, 7(2), 147-177. Seifer, R. & Dickstein S. (2000) Parental mental illness and infant development. In C. H. Zeanah (Ed.), Handbook of infant mental health (2nd ed., pp. 145—160). The Guilford Press, New York. Seltzer, J. A. & Bianchi, S. M. (1988). Children’s contact with absent parents. Journal of Marriage and the Family, 50, 663-677. Serovich, J. M., Price, S. J ., Chapman, S. F., & Wright, D. W. (1992). Attachment between former spouses: Impact on co-parental communication and parental involvement. Journal of Divorce and Remarriage, 17(3—4), 109—1 19. Shaw, D. S. & Vondra, V. I. (1995). Infant attachment security and maternal predictors of early behavior problems: A longitudinal study of low-income families. Journal of Abnormal Child Psychology, 23(3), 335-357. Silver, E., Heneghan, A., Bauman, L., & Stein, R. (2006). The relationship of depressive symptoms to parenting competence and social support in inner-city mothers of young children. Maternal and Child Health Journal, 10(1), 105-1 12. Spitz, R. A. (1945). Hospitalism: An enquiry into the genesis of psychiatric conditions in early childhood. Psychoanalytic Study of the Child, 1, 53-74. Spitz, R. A. (1946). Anaclitic depressive symptoms: An enquiry into the genesis of psychiatric conditions in early childhood. Psychoanalytic Study of the Child, 2, 131-342. Sroufe, L. A. (1989a). Relationships and relationship disturbances. In A. J. Sameroff & R. N. Emde (Eds), Relationship disturbances in early childhood (pp. 97-124). New York: Basic Books. Sroufe, L. A. (1989b). Relationships, self, and individual adaptation. In A. J. Sameroff & R. N. Emde (Eds), Relationship disturbances in early childhood (pp. 70-94). New York: Basic Books. Sroufe, L. A. & Waters, E. (1977). Attachment as an organizational construct. Child Development, 7, 92-126. Stern, D. (1989). The representation of relational patterns: Developmental considerations. In A. J. Sameroff & R. N. Emde (Eds), Relationship disturbances in early childhood (pp. 52-69). New York: Basic Books. 143 Stern, D. (1985). The interpersonal world of the infant: A view from psychoanalysis and developmental psychology. New York: Basic Books. Taanila, A., Laitinen, E., Moilanen, I. & J arvelin, M. J. (2002). Effects of family interaction on the child’s behavior in single parent or reconstructed families. Family Process, 41(4), 693-708. Tabachnik, B. G., & Fidell, L. S. (1996). Using multivariate statistics (3rd ed.). New York: HarperCollins College Publishers. Tamis-LeMonda, C. S., Shannon, J. D., Cabrera, N. J ., & Lamb, M. E. (2004). Fathers and mothers at play with their 2- and 3-year-olds: Contributions to language and cognitive development. Child Development, 75(6), 1806-1820. Tanaka, J. S. (1987). How big is big enough? Sample size and goodness of fit in structural equation models with latent variables. Child Development, 58(1), 134- 146. Thomas, A., & Chess, S. (1977). Temperament and devel0pment. New York: Brunner/Mazel. Thomas, P., Clement, J. P., Hazif-Thomas, C., & Leger, J. M. (2001). Family, Alzheimer’s disease and negative symptoms. International Journal of Geriatric Psychiatry, 16, 192—202. Thompson, R. A., & Goodvin, R. (2007). Taming the tempest in the teapot: Emotion regulation in toddlers. In C. A. Brownell & C. B. Kopp (Eds), Transitions in early socioemotional development: The toddler years (pp. 320-344). New York: Guilford. Tomarkin, A. J. & Waller, N. G. (2005). Structural equation modeling: Strengths, limitations, and misconceptions. Annual Review of Clinical Psychology, 1, 31-65. Tronick, E., Cohen, J. F ., & Shea, E. (1986). The transfer of affect between mothers and infants. In T. B. Brazelton & M. W. Yogman (Eds), Ajfective development in infancy (pp. 11—25). Norwood, NJ: Ablex. Tronick, E., Ricks, M., & Cohn, J. F. (1982). Maternal and infant exchange: Patterns of adaptation. In T. Field & A. Fogel (Eds), Emotion and early interactions (pp. 83- 100). Mahwah, NJ: Lawrence Erlbaum Associates. Tronick, E. & Weinberg, M. K. (1999). Gender differences and their relation to maternal depressive symptoms. In S. L. Johnson, A. M. Hayes, T. M. Field, N. Schneiderrnan, & P. M. McCabe (Eds), Stress, coping, and depressive symptoms (pp. 23-34). Mahwah, NJ: Lawrence Erlbaum Associates. 144 van den Boom, D. C. (2004). First attachments: Theory and research. In G. Bremner & A. F ogel (Eds), Blackwell handbook of infant development (pp. 296-325). Malden, MA: Blackwell Publishing. van den Boom, D. C. & Hoeksma, J. B. (1994). The effect of infant irritability on mother- infant interaction: A growth curve analysis. Developmental Psychology, 30, 581- 590. van IJzendoom, M. H. & De Wolff, M. S. (1997). In search of absent father-meta analyses of infant-father attachment: A rejoinder to our discussants. Child Development, 68(4), 604-609. Warren, H. K. & Stifier, C. A. (2008). Maternal emotion-related socialization and preschoolers' developing emotion self-awareness. Social Development, 1 7(2), 239-258. Weinberg, M. K., Beeghly, M., Olson, K. L., & Tronick, E. (under review). Still-face paradigm for young children: 272 year-aids ' reactions to maternal unavailability during the still- ace. Weinberg, M. K., Olson, K. L., Beeghly, M., & Tronick, E. Z. (2006). Making up is hard to do, especially for mothers with high levels of depressive symptoms and their infant sons. Journal of Child Psychology and Psychiatry, 4 7(7), 670-683. Weinberg, M. K. & Tronick, E. Z. (1998). Emotional characteristics of infants associated with maternal depressive symptoms and anxiety. Pediatrics, 102(5), 1298-1304. Whitchurch, G. G. & Constantine, L. L. (1993). Systems theory. In P. G. Boss, W. J. Doherty, R. LaRossa, W. R. Schumm, & S. K. Steinmetz (Eds), Sourcebook of family theories and methods (pp. 325-3 52). New York: Springer. Winnicott, D. (1965). The maturqtional processes and the facilitating environment. New York: International Universities Press. Witherington, D. C., Campos, J. J ., & and Hertenstein, M. J. (2004). Principles of emotion and its development in infancy. In G. Bremner & A. Fogel (Eds), Blackwell handbook of infant development (pp. 427-464). Malden, MA: Blackwell Publishing. Wood, J. J. & Repetti, R. L. (2004). What gets Dad involved? A longitudinal study of change in parental child caregiving involvement. Journal of Family Psychology, 18(1), 237-249. Youngblade, Y. M., Park, K. A., & Belsky, J. (1993). Measurement of young children’s close fiiendship: A comparison of two independent assessment systems and their 145 associations with attachment security. International Journal of Behavioral Development, 16, 563-587. Zeanah, C. H. & Anders, T. F. (1987). Subjectivity in parent-infant relationships: A discussion of internal working models. Infant Mental Health Journal, 8(3), 237- 250. Zeanah, C. H., Boris, N. W., & Larrieu, J. A. (1997). Infant development and developmental risk: A review of the past 10 years. Journal of the American Academy of Child and Adolescent Psychiatry, 36, 165—178. 146 "11111111le11111IIlIIlIIlIlfllelll“