UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Exploring the stepgap : how parents' ways of coping with daily family stressors impact stepparent-stepchild… Preece, Melady 2000

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2000-566048.pdf [ 7.31MB ]
Metadata
JSON: 831-1.0099539.json
JSON-LD: 831-1.0099539-ld.json
RDF/XML (Pretty): 831-1.0099539-rdf.xml
RDF/JSON: 831-1.0099539-rdf.json
Turtle: 831-1.0099539-turtle.txt
N-Triples: 831-1.0099539-rdf-ntriples.txt
Original Record: 831-1.0099539-source.json
Full Text
831-1.0099539-fulltext.txt
Citation
831-1.0099539.ris

Full Text

EXPLORING THE STEPGAP: How Parents' Ways of Coping with Daily Family Stressors Impact Stepparent-Stepchild Relationship Quality in Stepfamilies By M E L A D Y PREECE B. A . University of Los Angeles, California 1993 M . A . University of British Columbia 1996 A THESIS SUBMITTED I N P A R T I A L F U L F I L L M E N T O F T H E REQUIREMENTS FOR THE D E G R E E O F DOCTOR OF PHILOSOPHY In T H E F A C U L T Y OF G R A D U A T E STUDIES Department of Psychology The University of British Columbia We accept this thesis as conforming to the required standard October 2000 © Melady Preece, 2000 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. The University of British Columbia Vancouver, Canada DE-6 (2/88) 11 ABSTRACT This research focuses on husbands' and wives' perceptions of parent-child relationship quality in stepfamilies. One goal was to examine the links between parents' ways of coping with family stressors and changes in parent-child relationship quality over time. A related goal was to consider the difference in relationship quality for parents' own children and parents' stepchildren. This difference was referred to as the "stepgap". It was expected that characteristics of the family, characteristics of the children, and parents' ways of coping with family stressors would all have an impact on relationship quality. It was also expected that some of these characteristics would alter the "stepgap". Multilevel analyses of family data (Snijders, 1995) were used to replicate consistent findings in the stepfamily literature and extend them by allowing for the drawing of within-family conclusions. For Time 1 relationship quality, characteristics of children and characteristics of the family were modeled on parents' perceptions of relationship quality with individual children. At both levels, the influence of these characteristics on the "stepgap" was also considered. The initial sample interviewed at Time 1 consisted of 154 couples. Of these, 142 couples also participated at Time 2. Husbands initially rated the closeness and tension they perceived in their relationships with 404 children (191 stepchildren, 213 own children). Wives rated the closeness and tension they perceived in their relationships with 407 children (204, stepchildren, 203 own children). Results provided evidence of a "stepgap" in relationship quality for both husbands and wives. However, results also indicated that relationship quality was affected by child age, amount of time spent in the I l l family home, whether there were children from the current union, and the number of years the stepfamily had been in existence. A subsample of these families (81 couples) also provided daily diary data that were used to explore lagged daily relations between parents' reports of affection and tension with children and stepchildren, and parents' ways of coping with family stress. Three ways of coping relevant for interpersonal stressors were examined: compromise, confrontation, and interpersonal withdrawal. Results provided evidence of a direct relationship between parents' ways of coping with family stress and changes in daily relationship quality in terms of affection from children and tension with children. To link the microlevel and the macrolevel, aggregated variables describing parents' typical way of coping with family stressors across a seven-day period were used to explain changes in relationship quality two years later. Results of these analyses indicated that husbands' and wives' coping predicted change not only in the quality of their relationship with children in the stepfamily, but also affected their spouses' stepgap in relationship quality. iv TABLE OF CONTENTS A B S T R A C T i i T A B L E OF CONTENTS iv T A B L E S vii LIST OF FIGURES ix A C K N O W L E D G M E N T S x INTRODUCTION 1 P R O B L E M S A N D C H A L L E N G E S OF STEPFAMILY L IFE 2 APPROACHES TO STEPFAMILY R E S E A R C H 3 CONSIDERING T H E STEPGAP 5 P A R E N T - C H I L D RELATIONSHIPS IN T H E STEPFAMILY 7 The Stepfather-Stepchild Relationship 7 The Stepmother-Stepchild Relationship 10 The Remarried Parent-Child Relationship 14 C H I L D R E N ' S CHARACTERISTICS 15 Children's Age or Developmental Stage 15 Children's Gender 16 Time Spent in the Stepfamily Home 17 F A M I L Y CHARACTERISTICS 18 Years Together as a Stepfamily 18 Presence of Children from the Current Union 19 Average Age of Children in the Family 20 STRESS A N D INTERPERSONAL CONFLICT 20 Approaches to the Study of Conflict in Interpersonal Relationships 21 Conflict in Parent-Child Relationships 22 Conflict Management Behaviors 23 Using Interaction Sequences to Study Interpersonal Conflict 25 STRESS A N D COPING IN T H E STEPFAMILY 26 Ways of Coping with Family Stress 27 Compromise 28 Confrontation 30 Withdrawal 31 The Importance of Context ^ 32 M E T H O D O L O G I C A L ISSUES 33 G O A L S OF T H E PRESENT R E S E A R C H 35 V STUDY HYPOTHESES 37 SET 1 37 SET 2 , 39 SET 3 40 METHODS 41 S A M P L E 41 Data Available for Analysis 42 Sample Characteristics 44 Characteristics of Diary Sample 47 M E A S U R E S 47 Macrolevel variables 48 Parent-child relationship quality 48 Children's characteristics 49 Microlevel variables 50 Stepparents' coping strategies 50 Daily parent-child relationship quality 51 STATISTICAL A N A L Y S E S 52 Hierarchical linear modeling 52 Indicator variables 52 Application of multilevel modeling to family data 53 RESULTS 54 A G G R E G A T E D RELATIONSHIP Q U A L I T Y FOR O W N C H I L D R E N A N D STEPCHILDREN 54 Differences between Husbands and Wives 54 Differences Between Own Children and Stepchildren 59 Differences Between Time 1 and Time 2 61 The Average Stepgap 63 Summary of Analyses of Aggregated Relationship Quality Variables 64 M U L T I L E V E L A N A L Y S E S OF T H E E F F E C T OF F A M I L Y A N D C H I L D CHARACTERISTICS O N P A R E N T - C H I L D RELATIONSHIP Q U A L I T Y 66 Preliminary Analyses 66 Child-level Correlations between Children's Characteristics and Relationship Quality 66 Family-level Intercorrelations Among Family Characteristics and Aggregated Relationship Quality Variables 68 Multilevel Analyses of Child and Family Characteristics 71 Husbands' Reports Of Time 1 Relationship Quality 72 Wives' reports of Time 1 relationship quality 83 Multilevel Analyses of Child Characteristics and Family Characteristics as Predictors of Time 2 Relationship Quality 86 Husbands' reports of Time 2 relationship quality 86 Wives' reports of Time 2 relationship quality 89 vi Further investigation of the relations between children's age and change over time 92 D A I L Y F L U C T U A T I O N S IN C O P I N G A N D RELATIONSHIP Q U A L I T Y 95 Preliminary Analyses 95 Intercorrelations between Diary Variables 95 Intercorrelations among aggregated diary variables 98 Contextual Model of Coping and Next-day Affection 100 Contextual Model of Coping and Next-Day Tension 103 Contextual Model of Relationship Quality and Next-Day Coping 104 EFFECTS OF P A R E N T S ' COPING O N T I M E 2 RELATIONSHIP Q U A L I T Y 106 Husbands' Time 2 Relationship Quality 106 Husbands' Time 2 Closeness 108 Husbands' Time 2 tension 110 Wives' Time 2 Relationship Quality 110 Wives' Time 2 Closeness 112 Wives' Time 2 tension 115 DISCUSSION 117 S U M M A R Y OF RESULTS 117 Exploring the Stepgap 117 Differences between husands' and wives' stepgap 118 Changes in the Stepgap over Time 119 Predictors of Parent-Child and Stepparent-Stepchild Relationship Quality 120 The Influence of Children's Characteristics 120 The Influence of Family Characteristics 122 The Effect of Daily Coping 123 Parents' Typical Ways of Coping with Family Stressors 127 IMPLICATIONS OF THIS R E S E A R C H 131 Theoretical Implications 131 Methodological Implications 134 Limitations of the Research 134 C L O S I N G T H E STEPGAP 136 REFERENCES 140 APPENDIX A 156 C O P I N G ITEMS FOR D I A R Y D A T A 156 APPENDIX B 157 STATISTICAL PRINCIPLES BEHIND M U L T I L E V E L M O D E L I N G 157 Model Assumptions 158 V l l TABLES Table 1. Data Available for Analysis 43 Table 2. Stepfamily Characteristics 45 Table 3. Custody Arrangements and Type of Time Spent in Stepfamily Home by Children From Previous Unions 46 Table 4. Aggregated Reports of Relationship Quality: Differences between Husbands and Wives 55 Table 5. Aggregated Reports of Relationship Quality: Differences Between Own Children and Stepchildren 59 Table 6. Aggregated Reports of Relationship Quality: Differences between Time 1 and Time 2 62 Table 7. Stepgap Scores: Differences between Husbands and Wives 64 Table 8. Stepgap Scores: Differences Between Time 1 and Time 2 65 Table 9. Child-level Correlations Among Husbands' and Wives' Time 1 and Time 2 Relationship Quality, Child Age, Child Gender, Time Spent in household, and Stepparent Gender 67 Table 10. Family-level Correlations among Family Characteristics and Relationship Quality Aggregates for Husbands' and Wives' Stepchildren and Own Children 69 Table 11. Child Characteristics and Family Characteristics as Predictors of Time 1 Relationship Quality between Husbands and Children in Stepfamilies and Effect of Family Characteristics on the Stepgap 75 Table 12. Child Characteristics and Family Characteristics as Predictors of Time 1 Relationship Quality Between Wives and Children in Stepfamilies and Effect of Family Characteristics On the Stepgap 85 Table 13. Child Characteristics and Family Characteristics as Predictors of Time 2 Relationship Quality Between Husbands and Children in Stepfamilies, and Effects of Family Characteristics on the Time 2 Stepgap 87 Table 14. Child Characteristics and Family Characteristics as Predictors of Time 2 Relationship Quality Between Wives and Children in Stepfamilies, and Effects of Family Characteristics on the Stepgap 90 vm Table 15. Intercorrelations among Diary Variables 96 Table 16. Intercorrelations, Means, and Standard Deviations of Aggregated Diary Variables 99 Table 17. Three-level Contextual Model of Coping and Next-Day Affection 101 Table 18. Three-level Contextual Model of Coping and Next-Day Tension 103 Table 19. Three-level Contextual Model of Relationship Quality and Next-Day Coping 105 Table 20. Husbands' and Wives' Average Coping as Predictors of Time 2 Relationship quality between Husbands and Children in Stepfamilies 107 Table 21. Husband's and Wives' Average Coping as Predictors of Time 2 Relationship Quality between Wives and Children in Stepfamilies I l l ix LIST OF FIGURES Figure 1. Children from Current Union x Stepgap interaction predicting husbands' Closeness to children in stepfamily at Time 1 78 Figure 2. Mean age of children in family x Stepgap interaction predicting husbands' closeness to children in stepfamily at Time 1 80 Figure 3. Mean age of children in family x Stepgap interaction predicting husbands' tension with children in stepfamily at Time 1 81 Figure 4. Children from current union x Stepgap interaction predicting husbands' tension with children in stepfamily at Time 2 88 Figure 5. Years living together x Stepgap interaction predicting wives' closeness to children in stepfamily at Time 2 91 Figure 6. Husbands' Closeness and Tension with Children in Stepfamily at Time 1 and Time 2 as a Function of Age Category 93 Figure 7. Wives' Closeness and Tension with Children in Stepfamily at Time 1 and Time 2 as a Function of Age Category 94 Figure 8. Wives' Confrontation x Stepgap interaction predicting husbands' closeness to children in Stepfamily at Time 2 108 Figure 10. Wives' Interpersonal Withdrawal x Stepgap interaction predicting wives' closeness to children in stepfamily at Time 2 114 Figure 11. Husbands' Interpersonal Withdrawal x Stepgap interaction predicting wives' tension with children in stepfamily at Time 2 115 X ACKNOWLEDGMENTS I wish to express my gratitude to all the parents in stepfamilies who participated in this study. They gave generously of their time so that others might understand the challenges they face. I have also learned from my friends and family who have found themselves in step-relationships. They have worked uncomplainingly to be the best stepparents and/or stepchildren they could be, and have shown me that complicated family relationships can still be satisfying and rewarding. I would also like to acknowledge all the people who helped me in my efforts to learn multilevel modeling: Doug Willms, Sarah Newth, Lynda Grant, all the members of the multilevel list, Mathilde at Scientific Software International, and especially Tom Snijders. Most of all, I thank my advisor, Anita DeLongis, who took a chance on me, and gave me the room I needed to grow. I have benefited tremendously from her openness and largesse. 1 INTRODUCTION Remarriage... "the triumph of hope over experience." Samuel Johnson, (cf. Boswell, 1770, p. 122) The rate of divorce has skyrocketed in recent years, currently approaching 50% of all marriages. The vast majority of divorced adults will either remarry or establish common-law relationships. Many of these adults will bring children from a previous union into their new situation, creating a stepfamily. It has been estimated that between 25% and 40% of North American children will live in a stepfamily for a time (Ahrons & Rogers, 1987; Glick, 1989). Most of them wil l be children of divorce, already at increased risk for adjustment problems (Saucier & Ambert, 1986). Children from divorced and remarried families generally evidence more problem behaviors and lower levels of psychological well-being than those from never-divorced families (Amato & Keith, 1991a; Emery & Forehand, 1994; Hemerington, 1991b; McLanahan & Sandefur, 1994). For many parents and their children, forming a new family doesn't work out as well as they may have hoped. Research has shown that the presence of stepchildren is a major contributor to the somewhat greater rate of divorce among couples with stepchildren relative to those without stepchildren (Booth & Edwards, 1992; McCarthy, , 1978; Teachman, 1986; White & Booth, 1985). Stepfamilies are often less cohesive and more stressful than "nuclear" families and parent-child relationships are generally more detached, negative, and conflicted (Bray & Berger, 1993). 2 Problems and Challenges of Stepfamily Life Remarriage has the potential to positively influence the overall functioning of a family previously headed by a single parent. In the best possible scenario, the addition of a new stepparent can provide economic advantages, as well as emotional and child-rearing support, to the biological parent (Zill, Morrison, & Coiro, 1993). However, as often as not, the promise of improved family well-being goes unrealized. This is because members of a new stepfamily must cope with a number of stressors unique to the remarried family structure that can be so problematic as to outweigh any advantages. For example, establishing and maintaining a strong marital bond is more challenging when the stepparent must simultaneously work to construct a functional stepparent-stepchild relationship. Similarly, while working to develop a close marital relationship, remarried parents also need to sustain a close relationship with their children from a previous union, and resolve the loyalty conflicts that are likely to emerge (Hetherington &Jodl , 1994). The quality of the stepparent-stepchild relationship is believed by many to be the most important relationship in predicting overall stepfamily happiness (Crosbie-Burnett, 1984; Visher & Visher, 1988). However, it is also the most problematic and stressful stepfamily relationship (Mills, 1984). It is well documented that stepfamilies experience stress (Crosbie-Burnett, 1989a). However, very little is known about how stepfamilies cope with stress on a daily basis, and how their ways of coping affect the quality of the stepfamily relationships. Because many stepparent-stepchild relationships do not have the solid foundation created by early childhood bonding experiences, the way that parents cope with family stress may have an even greater impact on the stepparent-stepchild relationship than on parent-child relationship that began at birth. This study examined the effects of parents' ways of coping with daily stressors on the stepparent-stepchild relationship, as compared to the effects on their relationships with their own children. Approaches to Stepfamily Research Early research on stepfamilies focused mainly on comparisons between blended and intact families as i f intact families were the "gold standard" against which stepfamilies were to be compared. However, this approach resulted in a paucity of information regarding interpersonal processes in stepfamilies, and researchers have now begun to emphasize the importance of studying stepfamilies in their own right (Coleman & Ganong, 1990; Burrell, 1995). One approach to the study of families is to consider the family as a holistic unit, or system, and each dyadic relationship within the family as a subsystem (Broderick, & Smith, 1979; Broderick, 1993). These relationships form identities that are separate and distinct from the identities of the two individuals. From a family systems perspective, "family members are viewed as part of an interdependent emotional and relational system, the parts of which mutually influence one another in different aspects of the family system" (Bray & Berger, 1993). The research presented here extends common approaches to the study of stepfamilies, by focusing on the quality of the relationships between each parent and each child in the stepfamily unit to draw conclusions that apply both within- and between-families. Such an approach also avoids the common error made when macro-level correlations are used to make inferences about micro-level correlations. This type of 4 erroneous interpretation is termed the "ecological fallacy" (Snijders & Bosker, 1999), and occurs whenever relationships between group averages are used to make inferences about relationships between group members. For example, stepmothers consistently report having more problematic relationships with stepchildren than do stepfathers. However, such a result may be affected by differences between men who have custody of their children and men whose children from a previous marriage live with their mother, and visit the stepfamily home only occasionally. Thus, before asserting that this is a true characteristic of stepmothers, we need to compare stepmothers with stepfathers within the same family, controlling for differences in the amount of time each child spends in the stepfamily home, the age of the children, and perhaps also husband's coping behavior. In this study, in order to counter the possibility of erroneous interpretations based on group averages, data are analyzed from the "bottom-up", with information from each parent-child dyad contributing to within-family conclusions that can be drawn about stepfamily relationships. This is achieved by using a statistical technique known as multilevel modeling, or hierarchical linear modeling ( H L M ; Bryk & Raudenbush, 1992). With this type of analysis, children's data are analyzed at the first level of analysis, and nested within families at a higher level of analysis. To the author's knowledge, such an approach to the study of stepfamily data has not been undertaken previously. This study wil l also examine a number of variables that have previously been identified as important predictors of stepfamily relationships, in order to determine whether consistent findings from the literature can be replicated with a within-family analytical approach. Stepfamily relationships can also be viewed from a "microsystem" approach, examining the daily interactions and reinforcements that shape relationships. According 5 to this approach, both actions and interactions in families tend to be regulated by learned response sets. The rigidity of both coercive and affectionate response sets in marriage and family relationships has been well-documented (Patterson, 1982; Weiss, 1978; Gottman, 1979; Fitzpatrick and Ritchie, 1993). It is assumed that..."the family system, like any system, has self-stabilizing properties...families stabilize around habitual patterns of interaction..." (Maccoby, 1984, p 326). This study probes diary data from stepfamilies depicting parents' ways of coping with daily stressors and daily fluctuations in parent-child and stepparent-stepchild relationship quality. The goal is to detect reliable patterns of interaction occurring from one day to the next. The effectiveness of a particular coping strategy can thus be gauged by its' association with relationship quality at a subsequent time point. Such an analytical strategy strengthens causal inference, by ruling out the temporal influence of other factors, such as mood, that may influence self-reports. Parents' daily coping is important because of its expected long-term impact on relationship quality. To verify this, coping outcomes in this study will be considered in two ways. First, coping wil l be used to predict fluctuations in relationship quality from one day to the next. Second, parents' average ways of coping will be used as contextual variables to predict long-term variations in relationship quality. Considering the Stepgap The term "gap" has often been used to describe the degree of disparity between two groups. Probably the most well-known use of the term is in the descriptor, "gender gap" which has been used to describe differences between men and women in terms of 6 income, poverty, voting preferences, occupational status, access to the Internet, etc. Politicians, sociologists, and educators have also used the term "race gap" to refer to differences between people of different racial backgrounds in areas such as income, level of education, and intellectual test scores. For this study, the term "stepgap" was coined as a way of referring to differences between stepchildren and cWlchen-from-birth in terms of parent-child relationship quality. One of the most important determinants of the quality of the relationship between the children and parents in a stepfamily is whether the parent is a parent-from-birth or a stepparent. Stepparents and stepchildren experience each other in a way that is fundamentally different from the way parents-from-birth and their children experience each other (Papernow, 1988). Generally speaking, relational ties between stepparents and stepchildren are weaker than ties between parente-from-birth and their children. Further, research has shown that in homes where prior-marriage children of both parents are present, parents report differential treatment of children (Hetherington & Camara, 1985). Differential treatment of children in stepfamilies may lead to resentment and have long-term implications for the quality of relational ties (Anderson, Hetherington, Reiss, & Howe, 1994). The quality of the stepparent-stepchild bond is important not only while the child is living in the stepfamily household. Families remain important sources of identity and perceived social support long after children have grown up and left home (White, 1994). In weaker relationships, stepparents give less support to stepchildren, and stepchildren are also less likely to support stepparents. It has been demonstrated that perceived supportiveness is more important than the actual exchange of goods and services for 7 health outcomes (House, Umberson, & Landis, 1988). However, stepchildren's perception of available support has been shown to be substantially lower than that of biological children (White, 1994). Less supportive relationships can be characterized as those where there is less closeness and/or greater tension between parent and child. In this study, after demonstrating the existence of the "stepgap", the focus turns to potential moderators of this gap. Specifically it was expected that parents would report greater closeness and less tension with their own children than with their stepchildren. However, it was also expected that characteristics of children and parents would moderate the stepgap. Parent-Child Relationships in the Stepfamily This next section reviews the extant literature on parent-child relationship quality in stepfamilies. Although the review is loosely grouped into issues pertaining to stepfather-stepchild, stepmother-stepchild, and remarried parent-child relationships, factors such as child age and child gender are often inextricably mtertwined. Some research pertaining more specifically to child age and gender is then reviewed. A less often researched influence involves the amount of time children typically spend in the stepfamily home. In this section, the influence of this factor is also discussed. The Stepfather-Stepchild Relationship A positive relationship between stepfathers and stepchildren can lead to positive outcomes for the child and for the family. However, both the age and the gender of the child can have a strong influence on the development of the stepfather-stepchild relationship. Early studies of stepfather families found that the entrance of a stepfather into a father-absent home had a positive effect on boys' cognitive and personality development (Chapman, 1977; Santrock, 1972; Oshman and Manosevitz 1976). In a study of the social behavior of boys and girls between the ages of 6 and 12, it was found that boys in stepfather families tended to be more socially competent than boys in intact families (Santrock, Warshak, & Eliot; 1982). Boys also showed more warmth toward their stepfathers than did girls. Girls in stepfather families, on the other hand, were observed to be more anxious than girls in intact families, and there was a trend for girls in these families to show more anger toward their mothers than boys. In fact, boys in stepfather families have been found to function better than those in either single-parent families or conflict-ridden intact families (Hetherington, 1982). However, i f the remarriage occurred when the boy was older (9 or 10), the relationship with the father was more likely to be negative than in the case of younger children. The best way for a stepfather to handle his integration into the family may differ depending upon the age of the stepchild. Ganong and Coleman (1994) hypothesized a scenario in which a newly married mother's early preoccupation with her spouse may create feelings of jealousy and insecurity in her young children. They may withdraw from her, create conflicts with the stepfather, or misbehave and act out in order to demand their mother's attention. Such behavior may also negatively affect the stepfather-stepchild relationship. It has been suggested that newly remarried mothers may be compensating for the more permissive style of discipline generally engaged in by stepfathers (Bray, Burger, Silverblatt, & Hollier, 1987). 9 The entry of a stepfather into the lives of older children may have a more positive effect on the quality of their relationship with their mothers. Older children may welcome the emotional distance and interpersonal autonomy that comes with their mother's remarriage. This may improve the mother-child relationship and create a positive climate for the stepfather-stepchild relationship to flourish. Indeed, adolescents in stepfather households report more permissive parenting by both parents than those living with stepmothers (Bray et al., 1987). However, permissive parenting may not always benefit the stepparent-stepchild relationship. On the contrary, Hetherington (1993) found that immediate authoritative parenting by the stepfather in remarriages with adolescent children led to more rapid acceptance and more positive outcomes for children. With younger children, however, any kind of controlling behavior tended to lead to greater resistance. With younger children, it appeared that initial support of the mothers' discipline, and the much slower development of a close relationship with the child was most effective. When new stepfathers find themselves rejected in their early attempts to become involved with their stepchildren, they may withdraw from their stepchildren, and relinquish the parental role. As disengaged parenting of stepchildren of all ages is the predominant style of stepfathers (Hemerington, 1993), it may be that such disengagement leads to the common failure of stepchildren to adapt successfully to membership in a remarried family. Little research has examined how mothers' behavior may impact the relationship between stepfathers and stepchildren. From a family systems perspective, "family members are viewed as part of an interdependent emotional and relational system, the parts of which mutually influence one another in different aspects of the family system" 10 (Bray & Berger, 1993, p. 78). Such a perspective would predict that her behavior would be important, although the specifics are unknown. Indeed, the mother's behavior may affect not only stepfather-stepchild relationships but also the father's relationships with his own children. In the past decade, a number of researchers have begun to focus on emotional transmission in families (for a review, see Larson & Almeida, 1999). Some of this work has concentrated on daily patterns of transmission (e.g. Bolger, DeLongis, Kessler, & Wethington, 1989; Repetti, 1989) and some has focused on more microscopic time frames, such as minute to minute interactions (e.g. Repetti & Wood, 1997; Roberts & Krokoff, 1990). Other work has considered transmission between family members using measures spaced over a span of months or years (e.g. Karney & Bradbury, 1995; Patterson & Dishion, 1988). In this study, it is expected that the way one parent in the stepfamily habitually copes with family stressors wil l have an impact on the relationship quality reported by the other parent two years later. In keeping with the theoretical approach of this research, it was also expected that wives' coping would have a differential impact on husbands' relationship quality with their own children and with their stepchildren, either by increasing or decreasing the size of the stepgap. The Stepmother-Stepchild Relationship. Far less research has examined the stepmother-stepchild relationship, probably due to the traditionally lower occurrence of stepmothers residing with their stepchildren. As recently as twenty years ago, 90% of mothers retained sole custody of children in a 11 divorce. However, in recent years, more men are seeking custody of their children, joint custody arrangements have become common, and the number of children residing with stepmothers, at least part of the time, has therefore increased. One contribution of this research is a relaxing of the strict division between stepmothers who have primary care of their stepchildren and stepmothers who co-parent their stepchildren, along with children's own mother. Instead, the amount of time spent in the stepfamily household was considered as a variable having an important influence on stepmother-stepchild relationships. In general, stepmothers tend to experience a great deal more stress, anxiety, depression, and anger about their role in the family than mothers in other family structures do (Santrock & Sitterle, 1987). Hobart (1994) proposed a number of explanations for why stepmothers more often have relationship problems with stepchildren than do stepfathers. First, many husbands may view the caring for and entertaining of children as their wives' responsibility. Both visiting and live-in stepchildren increase the housework load: there is more cooking, cleanings shopping, laundry, etc. In many households, this extra burden is largely on the wife's shoulders. Ambert (1989) found that husbands generally do not increase their level of help in the house when their children are visiting. Second, visiting stepchildren may become involved in conflicts with the wives' children. If a mother intervenes, the stepchildren may perceive her as taking sides with her children against them (Ahrons & Wallisch, 1987). Given the way men and women are socialized in our society, mothers usually care more and are more upset by these conflicts than are fathers (Hobart, 1994). Third, to counteract guilty feelings about the 12 divorce, a father may be more solicitous or extravagant with his children when they visit, and be unwilling to use any form of discipline (Ambert, 1989). The wife may see this as favoritism, and an indication that her husband cares more about his children than about her or her children. Fourth, there are often differences in loyalty feelings children have for their mothers and fathers. For the father's children, their mother is usually the primary parent, and their feelings of loyalty to her are typically very strong. The child may resist closeness with the stepmother to avoid feelings of disloyalty toward his or her own mother. On the other hand, it has been suggested that when children live with their mother and a stepfather, loyalty feelings of children are less likely to interfere, as non-custodial fathers are often less involved with their children than are non-custodial mothers (Brooks, 1985, as cited in Hobart, 1994). In a study of stepmother families (Clingempeel, Brand, & Ievoli, 1984), it was found that stepmothers' relationships with their stepdaughters were generally more negative and detached than stepmothers' relationships with their stepsons. It was suggested that fathers who have custody of their daughters may have a particularly close relationship with them, and stepdaughters may view the stepmother as a threat to the closeness they enjoy with their father. This study also found that for stepfamilies in which the father's children live with the remarried couple, the relationship of the stepmother to her husbands' children was very different from her husband's relationship with those children. When the quality of the relationship between stepmothers and their stepchildren was compared to the quality of the relationship between their husbands and the same children, remarried custodial fathers were found to be closer and more nurturant with their children from a previous 13 marriage than were their new wives. Some differences depended upon the gender of the children. Remarried fathers were more permissive and indulgent with their daughters from a previous marriage than were their new wives, and reported feeling less confident in raising them than did their new wives. The opposite was true for sons, with remarried custodial fathers being more restrictive with their sons than were their new wives, and feeling more confident in raising them, than did their new wives. Stepmothers often make a tremendous effort to competently fulfill their role as a parent to their husbands' children. Remarried fathers and their new wives generally reported equal involvement and responsibility in their roles as parents (Clingempeel et al, 1984). Nonetheless, children in these stepmother families perceived their stepmothers to be less involved with them than were their own fathers. These results suggest a dynamic in stepmother families in which stepmothers take responsibility for a goodly amount of the daily parenting activities, but may not feel that they receive adequate recognition for their efforts, either from their husbands (Ambert, 1989) or from their stepchildren (Clingempeel et al, 1984). In families where husbands are aware of such dynamics, and attempt to cope with their wives' concerns in a positive and proactive manner, wives may find it easier to maintain a positive relationship with their husbands' children. This line of reasoning stimulated the hypothesis for the current research that husbands' ways of coping would have an impact on wives' relationships with their stepchildren. In addition, coming from a family systems approach, it was expected that husbands' ways of coping would have an impact on wives' relationships with their own children as well. Hobart (1988b) has shown that remarried mothers generally report less positive relationships with their stepchildren than with their own children, whereas remarried 14 fathers are less likely to report differences. It was therefore also hypothesized that in this study, the "stepgap " would be more evident for wives than for husbands. It is further expected that husbands' coping may serve to either increase or decrease the stepgap as reported by wives. The Remarried Parent-Child Relationship Some researchers have suggested the quality of the parent-child relationship is negatively impacted by divorce and does not improve with remarriage. However, the effect of divorce on parent-child relationships may be different for sons and daughters, and appears to depend partly upon the gender of the parent. In a prospective study of divorce and remarriage, it was found that even after controlling for marital quality as reported by parents, subsequent divorce was related to less closeness between sons and mothers in divorced families than between sons and mothers in non-divorced families as reported by the children twelve years later (Booth & Amato, 1994). However, no differences in closeness were found for daughters and their divorced mothers as compared to daughters and mothers from intact marriages. Divorce also had a stronger negative impact on daughters' closeness to their fathers twelve years later than it did on sons' closeness to their fathers. A recent meta-analysis found that parental divorce was significantly associated with poorer relationships with both parents, although the mean effect size was stronger for fathers than for mothers (Amato & Keith, 1991). Remarried custodial fathers often report being more actively involved in parenting their children than fathers from intact families, but remarried custodial mothers still report being much more involved in discipline, observation of their children's 15 recreational activities, and provision of comfort and sympathy, than do remarried custodial fathers (Clingempeel et al., 1984). However, adolescents in divorced families and in stepfamilies have also been shown to experience the highest levels of mother-adolescent disagreements and the lowest levels of supervision (Demo & Acock, 1996). No specific hypotheses were formulated for the remarried parent-child relationship. However, whenever applicable, differences in the parent-child relationships for stepchildren and for prior marriage children were explored. Children's Characteristics Children's Age or Developmental Stage Children's developmental stages, as loosely described by their chronological age, have an important influence on parent-child relationships in the stepfamily. It has been suggested that adaptation in stepfamilies is a continuous process with many transitions occurring at different times for different members (Hetherington, 1993). The transition to adolescence is probably the most problematic transition for all children to negotiate successfully. When stepfamily transitions intersect with children's developmental transitions, the potential for stepchildren to experience developmental difficulties increases. It has been consistently found that older children have more difficulty adapting to stepfamily life than do younger children (Bray & Berger, 1993; Hetherington, 1988, 1989, 1993). There is some indication that such difficulties may not resolve over time. For example, one study found that children from all family types displayed more negative behavior as they moved into adolescence. Their relationships with their parents became 16 more positive as they moved into late adolescence, but their relationships with their stepparents tended not to improve (Hetherington & Clingempeel, 1992). Divorced and remarried parents report relationships with their adolescent children that are often difficult and stressful. Data from the Virginia Longitudinal Study of Divorce and Remarriage (Hetherington, 1993) indicated that parents in divorced and remarried families experienced an increase in negative life stresses when their children reached adolescence. Stresses and conflicts between parents and children were also greater i f the remarriage occurred when the children were over 9 years of age. A common result of these stresses and conflicts was the distancing of adolescents from all family relationships. Although some distancing and an increase in autonomy is expected in adolescence, it can sometimes be exacerbated by stepfamily membership, leading to involvement with deviant peer group members and delinquent behavior (Tfemerington, 1993). The important influence of children's age on parent-child relationships in stepfamilies prompted the expectation that in the current research, parents' would report less closeness and greater tension in their relationships with older children than with younger children. Children's Gender There is a consensus in the literature that boys tend to have better relationships with stepfathers than do girls, particularly during adolescence (e.g., Hetherington et al., 1982). One possible explanation is the stepdaughters' physical maturation. Even non-divorced fathers are often disconcerted by their adolescent daughters' developing sexuality and are concerned about expression of physical affection at this time. For 17 stepfathers, this discomfort is likely to be even greater. Stepfathers represent a significant threat to their stepdaughters' well-being. The chances that a girl will be sexually abused by a stepfather are one in six, as compared with a one in forty probability of abuse by her biological father (Finkelhor, 1984). As girls mature, the threat of inappropriate intimacy, whether consciously acknowledged or not, may lead to a decrease in closeness and an increase in tension between stepfathers and stepdaughters. Stepfathers are likely to have more in common with boys, and often develop a close relationship based on engagement in traditionally "masculine" activities that they can share. Several studies have shown that relationships between stepfathers and stepsons tend to become closer over time (e.g., Hetherington, 1993). It was therefore expected that husbands would report greater closeness and less tension with their stepsons than with their stepdaughters.. Time Spent in the Stepfamily Home Another important influence on parent-child relationships in stepfamilies is the amount of time the stepchild spends in the stepfamily home. It has been found that in stepfamilies where the husbands' children from a prior marriage do not live in the home, the wives' children are more attached to both the mother and the stepfather. However, when the husbands' prior marriage children spend more time in the stepfamily household, this attachment between the wives' children and their mother and stepfather weakens. However, stepmothers report more negative relationships with stepchildren who live with their mothers and visit their stepfathers than with stepchildren who live in the stepmothers' homes (Ambert, 1989). This finding led to the prediction that wives would 18 report greater closeness to stepchildren who spend more time living in the stepfamily home. Family Characteristics When studying stepfamilies, it is important to consider differences between stepfamilies that may affect differences in relationship quality between stepchildren and own children, as well as influencing relationship quality overall. Although the statistical approach taken in this study controls for unmeasured differences between families, a few important predictors of relationship quality mentioned in the literature were chosen for closer examination, and for the purposes of replication. These characteristics were also examined for their long-term influences on the stepgap. Years Together as a Stepfamily Many remarried couples believe that after a period of adaptation, conflicts and tension in the stepfamily wil l be resolved, and a new stable family unit wil l emerge. To establish stepfamily integration it is necessary for both parents and children to give up the attitude that nurturance and decision-making reside entirely within the original parent-child relationship, and re-establish them under the umbrella of the new stepfamily relationship. Such a radical transition is bound to be a difficult one for both parents and children. The odds that the members of a stepfamily wil l resolve their differences and become a well-functioning family unit are much poorer than one might hope. Pi l l (1990) found that one or both partners in 41 percent of responding stepfamilies said they entered 19 remarried life expecting that stepfamily members would become as closely bonded as in first-married families. However, 75% reported that their expectations were not fulfilled. In other research, longer remarried couples reported that their families became less cohesive over time and that conflicts about children actually increased (Bray & Berger, 1993; Hetherington, 1993). As this process of adaptation and negotiation offamilial roles is likely to require more skill and resources than are available for many families, it was expected that a greater number of years since the inception of the stepfamily would be related to increased tension not only with stepchildren, but also with children from previous unions. Presence of Children from the Current Union The presence of a half-sibling, a shared child of the remarried parents, has been associated with greater attachment of the mothers' prior marriage children to their stepfather (White & Booth, 1985). Although the mechanisms for such an attachment have not been explored, evolutionary theory would predict such a result. It has been proposed that "homo sapiens is obsessed with kinship" {Tinker, 1997, p. 430). Once we know we are related to other people, we feel a greater measure of solidarity, sympathy, tolerance, and trust towards our relatives. In stepfamilies with children from a current union, perhaps the recognition by the mother's children that they share kinship bonds with their half-sibling promotes feelings of kinship toward the stepfather, and indeed, stronger bonds among all family members. It may also be that stepfathers with a child from the current union take a more active and involved role in the family, positively affecting their relationship with their wives' prior-marriage children. It was expected, 20 therefore, that in stepfamilies where there was at least one child from the current union, husbands would report greater closeness and less tension with their stepchildren.. Average Age of Children in the Family For the same reasons discussed above regarding the importance of the child's age for relationship quality, it was expected that the average age of the children in the family would have a separate effect on relationship quality. In adolescence, children have more expensive needs and wants, and perceptions of differential treatment may be more upsetting. Further, from a family systems view, (Broderick & Smith, 1979; Broderick, 1993), the greater tension reported by remarried parents with adolescent children should affect the relationship quality of all family members. It was therefore expected that when children in the stepfamily were older on average, there would be generally less closeness and more tension with all children. Stress and Interpersonal Conflict From a stress and coping perspective, interpersonal conflict represents a significant source of stress that is likely to be particularly pervasive in stepfamilies. Researchers who study interpersonal conflict have noted that it is difficult, i f not impossible, to separate interpersonal conflict from the other sources of stress we encounter in our daily lives. Conflicts hitchhike on other events. People have conflict over dinner, while on vacation, and when trying to get the kids off to school in the morning. Further, grievances are more likely to be felt when events in the situation are absorbing and stressful—for example, when the car repeatedly stalls or the checkbook doesn't balance—rather than during a lull in the interaction (Sillars & Weisberg, 1987, p. 157). 21 By using a stress and coping approach to study family stressors, it is possible to obtain a much richer sampling of daily interpersonal conflict that could be captured through observational methods. Such an approach argues for an integration of the two streams of research when examining interpersonal stress in context. A consideration of the literature on conflict in interpersonal relationships is therefore pertinent when studying daily family stress in stepfamilies. Approaches to the Study of Conflict in Interpersonal Relationships The potential for conflict always exists when people are interdependent in some way. How people manage conflict tells us a lot about their relationship. Although a number of disciplines have contributed to the literature on interpersonal relations, at their points of convergence they bear a remarkable similarity. There are many definitions of interpersonal conflict, but little consensus. Canary, Cupach, and Messman (1995) described four major categories of definitions used by researchers. They pointed out that conflict can be conceptualized either as (1) a pervasive quality in a relationship, (2) a specific type of episode, (3) a specific type of behavior, or (4) a specific type of behavior within a specific type of episode. For the current research, interpersonal conflict is assumed to have occurred when specific types of behaviors are used in the context of a family stressor. By asking parents to identify a family stressor, and then considering those behaviors that tend to occur only when there is some incompatibility between individuals, a fairly general sampling of family conflicts can be assumed. The perspective taken in this research was nicely articulated by Sprey (1971): "The family process per se is conceived of as a continuous confrontation between participants with conflicting—though not necessarily opposing—interests in their shared fate" (p. 722). Conflict in Parent-Child Relationships The parent-child relationship is an involuntary association. There is an imbalance of power and resources in such a relationship, and an obligation exists for the parent to act as a caregiver under all circumstances. The child, on the other hand, is under a developmental obligation to reach a point where the caregiver is no longer required. These differences in impetus may become a point of contention, leading to increased conflict. Conflict is not necessarily a barrier to good quality relationships between adolescents and their parents. Adolescents often report engaging in negative and dramatic episodes of conflict with their parents. However, adolescents also tend to believe that the parent-child relationship can sustain intense expressions of disagreement (Laursen, 1993). Consistent with this, other research has indicated that conflict between parents and children does not necessarily have long-term negative effects on the parent-child relationship (Paikoff & Brooks-Gun, 1991; Eisenberg, 1992). Whether the stepparent-stepchild relationship may enjoy the same "protected status" that is afforded to parent-child relationships is unknown. It may be that some stepparents do benefit from adolescents' benign attitude toward parent-child disagreements. In a study that asked adolescent stepchildren to describe their relationships with their stepparents, the majority of stepchildren did not perceive their relationships with their stepparents to be particularly conflictual (Ganong 23 & Coleman, 1986). Further, in terms of closeness, 77% of stepchildren in this sample of adolescent and young adults (ages 15 to 22) reported feeling moderately close or close to their stepmothers, and 63% of stepchildren reported feeling moderately close or close to their stepfathers. However, stepparents tend to report a great deal more conflict and tension with their stepchildren than they (the stepparents) are comfortable with, although the causal direction for such conflict is difficult to determine. In a summary of the literature on conflictual interactions between parents and adolescents, it was noted that in more functional parent-adolescent dyads, the probability of reciprocating negative messages was low. However, in dysfunctional dyads, negative messages continue to be reciprocated for longer periods than in functional dyads (e.g., Gottman, 1987; Sillars, 1980b; Ting-Toomey, 1983a). Conflict Management Behaviors Horney (1945) described three basic styles of relating to people: (1) moving towards others, (2) moving against others, or (3) moving away from others. These general strategies correspond to the idea that in times of stress or conflict, people can either work with each other (cooperate, or compromise), against each other (compete, or confront) or move away from each other (reject, withdraw). These strategies have emerged in several empirical studies (e.g., Bell & Blakeney, 1977; Putnam & Wilson, 1982; Ross & DeWine, 1988; Sillars, Colletti, Parry & Rogers, 1982) and run through most discussions of interpersonal conflict (Canary & Cupach, 1988; Canary & Spitzberg, 1989; Sillars, 1980a, 1980b). 24 The various ways that people cope with conflict have also been described along two continua: valence and engagement (Sillars & Wilmot, 1994). From this perspective, conflictual interactions can be classified according to the positivity or negativity of the conflict behaviors, as well as whether the approach is direct or indirect. "Positive and direct" conflict behaviors are represented in the literature by such constructs as integrative tactics (Comstock & Buller, 1991), reasoning (Straus, 1979), and rational discussion (Rosenthal, Demetriou, & Efklides, 1989). "Negative and direct" conflict behaviors are represented by such constructs as verbal aggression, defensive communication (Alexander, 1973) and emotional expression of hostility (Rosenthal et al, 1989). However, when it comes to indirect behaviors, it is usually assumed that the same indirect method can have either a positive or a negative intent. For example, one might withdraw to avoid escalation of a conflict, or one might withdraw to avoid having to deal with the consequences of one's action. Indirect behaviors have been described in the literature by constructs like avoidance (Comstock & Buller, 1991), avoidance of confrontation (Rosenthal et al, 1989), or withdrawal (Whittaker & Bry, 1991). Sprey (1971) has argued that no a priori determination should be made as to whether any particular conflict behavior is positive or negative. However, for most families, "positive and direct" behaviors are likely to be perceived as more effective than either "negative and direct" behaviors or 'indirect" behaviors. Further, "indirect" or avoidant behaviors are generally considered to be preferable to "negative and direct" behaviors (Canary & Cupach, 1988; Canary & Spitzberg, 1990). Indeed, there is something of a consensus among relationship researchers that individuals in quality relationships manage conflict through positive interaction behaviors, which include 25 collaborating with each other, not allowing anger to escalate, and understanding that withdrawal is not a good strategy when maintenance of good relationship quality is the goal (e.g. Gottman, 1994; Sillars & Wilmot, 1994). Using Interaction Sequences to Study Interpersonal Conflict Systems theory suggests that a message is simultaneously a stimulus, a response, and a reinforcer, so that interpersonal communication represents a chain of overlapping and interdependent links (Bateson, 1972). It is assumes that when sequences occur repeatedly, they reflect some implicit, underlying rules of the relationship. Social learning theory has been profitably applied to interaction sequences by researchers such as Patterson (1979, 1982), who examined parent-child interactions and Gottman (1979, 1987), who is best known for his research with married couples. Both showed how negative styles of communication were reinforced in distressed dyads, whereas positive messages went unreciprocated. Although interaction sequences suggest temporal precedence, it is still difficult to establish causality. That is because it is equally plausible for relationships to become dissatisfying because of a pattern of negative interactions, or for patterns of negative interactions to be merely an outward expression of problems within the relationship. Because of this ambiguity, it is generally concluded that a process of "contagion" characterizes communication in poor relationships, because negative expressions elicit reciprocal behavior, and lead to the escalation of conflict (Patterson, 1982). 26 Stress and Coping in the Stepfamily Coping is defined as an individual's efforts to manage those demands appraised as either taxing or exceeding available resources (Lazarus & Folkman, 1984). Two broad functions of coping have generally been emphasized: problem-focused and emotion-focused. Problem-focused coping involves attempts to change the person-environment relation directly whereas emotion-focused coping is geared toward managing negative emotions generated by the stressful situation. When coping with stressors that are primarily interpersonal, however, an additional function emerges. This function of coping, termed relationship-focused coping, describes those modes of coping that are intended to manage, regulate, or preserve relationships during stressful periods (Coyne & Fiske, 1990; DeLongis & O'Brien, 1990; Coyne & Smith, 1991; O'Brien & DeLongis, 1996) . Research in the area of stress and coping has indicated that those daily stressors with the greatest impact on mood and health are usually interpersonal in nature (Bolger, DeLongis, Kessler, & Schilling, 1989; DeLongis & O'Brien, 1990). Further, a number of researchers have concluded that interpersonal factors have a powerful influence on every aspect of the stress and coping processes (for a review, see Taylor, Repetti, & Seeman, 1997) . In the stress and coping literature, however, the effectiveness of various coping strategies has usually been considered in relation to the individual's own outcomes. Interpersonal conflict represents a direct threat to the individual, and is therefore a potent source of stress (Thoits, 1985). Within the family, where the maintenance of good relationships is critical to well-being, the way parents cope with family stress may have 27 direct implications for the quality of family relationships. From this perspective, the most relevant outcomes to consider when examining the effectiveness of coping strategies used with family stressors are their influences on subsequent relationship quality. Some studies have indicated that remarried couples may possess poorer conflict resolution and problem-solving skills than couples in first marriages (Bray et al., 1987; Larson & Allgood, 1987). However, whether this means that remarried couples are typically deficient in their use of coping strategies, or whether stepfamily parents are merely overwhelmed by the degree of stress in their daily lives is unclear. It has been pointed out by a number of researchers that the quality of social relationships may be an important determinant of the coping strategies individuals select (DeLongis & O'Brien, 1990; Dunkel-Schetter, Blasband, Feinstein, & Herbert, 1992; Schreurs & de Ridder, 1997; Pearlin & McCall , 1990). In a stepfamily, strained relationships may negatively affect the employment of effective coping strategies. Ways of Coping with Family Stress Researchers in the communication literature have identified three interpersonally-based options available to individuals faced with a relationship conflict (Fitzpatrick, 1988; Sillars, Colletti, Parry, & Rogers, 1982), which they describe as: (1) avoidance and withdrawal, (2) verbal competition, aggression, and confrontation, or (3) cooperation and compromise. Coping research has also pointed out the importance of withdrawal, confrontation, and compromise as coping strategies used to cope with interpersonal stress (Daylen, 1993; Aldwin & Revenson, 1987; Buss, 1992; McCrae, 1984; Repetti, 1992). 28 The Brief Ways of Coping (BWOC; Preece, DeLongis, O'Brien, & Campbell, 2000) is a coping measure developed for use in diary studies. One of the notable features of this scale is the inclusion of ways of coping that are primarily interpersonal in nature. Three subscales of this measure correspond to the constructs discussed above, and will be examined in this study to determine their influence on relational outcomes. Compromise Compromise can be described as belonging to the class of "positive and direct" behaviors (Sillars & Wilmot, 1994). Although little research has focused on the construct of compromise, perse, there are a number of similar, more often studied, constructs that are relevant. For example, a theory of accommodation processes has been advanced that bears some resemblance to the construct of compromise. Accommodation is defined as an individual's willingness to engage in a constructive reaction given a partner's potentially destructive behavior (Rusbult, Verette, Whitney, Slovik, & Lipkus, 1991). In a series of studies of this construct, it was concluded that the decision to accommodate tends to be associated with features of the relationship. In particular, individuals are more likely to accommodate in relationships where they have a high level of commitment. It was also proposed that there might be a social cost to the decision to accommodate. In a healthy relationship, a fair degree of mutuality in the process of accommodation is expected. However, i f one partner carries most of the accommodative burden in the relationship, they wil l probably experience some personal distress as a consequence. Compromise can also be thought of as a method of cooperating in a social dilemma. A social dilemma can be defined as a situation in which two or more persons 29 in an interdependent relationship are faced with a conflict between maximizing personal (selfish) interests and maximizing collective (family) interests (Komorita, Parks, & Hulbert, 1992). From this perspective, one of the most important determinants of an individual's decision to cooperate is their expectation of reciprocity. The norm of reciprocity predicts that individuals tend to help those who have helped them in the past, and to retaliate against those who have injured them. In other words, individuals are more cooperative when they expect that others will also cooperate. It has been suggested that the idea of reciprocity is at the heart of all stable relationships, and is a basic norm in all social interactions (Thibault & Kelley, 1959). Although normal parent-child negotiation processes most likely include some expectation of reciprocity, for stepparents and stepchildren, the expectation of reciprocity may be less assured. Such problematic processes may parallel those between leaders of two countries who can see no ground for mutual agreement in the settlement of their conflicts. In adversarial situations, opening oneself up to alternative possibilities may seem too threatening, especially when so much is at stake. Each side faces the risk that the adversary wil l interpret a desire for a settlement, or hints of possible concessions, as indicators of weakness. If such will be the interpretation, the adversary may escalate his expectations and increase his pressure. Therefore, the bargainer will face the dilemma of how to yield without appearing weak (Touval, 1975, p. 54). Although this quote is from a case analysis of international negotiation, it is not difficult to replace the roles of the adversary and the bargainer with those of the stepparent and the stepchild, although they both may take turns in either position, depending upon the issue at hand. 30 In a study of husbands and wives coping with daily marital tension (Daylen, 1993), it was found that although wives' use of compromise did not reduce the amount of distress they experienced, it was related to a decrease in the amount of distress reported by their husbands. This suggests that compromise in an intimate relationship may benefit primarily the other, and perhaps the relationship, but carries no immediate benefits to the self. However, i f there is a normed belief in reciprocity, compromise may be thought of by the individual who uses it as an investment that will increase the likelihood of future cooperation. In this study, compromise was conceptualized as an effort to find a solution that is fair to all, involving meeting the other person in the interaction "half-way." In a previous study using the same data set as the current research, the relations between daily appraisals of stress and concurrent coping strategies were examined (Preece, DeLongis, O'Brien, & Campbell, 2000). The use of compromise was associated with stressors appraised as a threat to the coper's well-being. Specifically, such stressors were appraised as involving either the threat of "losing someone's respect or love" or "not getting the support and understanding you want." This suggests that a willingness to compromise may often be motivated by the need to preserve one's place in a relationship. Confrontation Confrontation can be thought of as a "negative and direct" method of coping with conflict (Sillars & Wilmot, 1994). It has been suggested that such competitive, dominating behaviors are most likely to be employed to deal with interpersonal conflict when people are more concerned with themselves than with the other in the situation 31 (Rahim, 1983). Although confronted individuals may concede defeat in the short term, the likelihood of future cooperation will likely be reduced due to the experience of such negative emotions as resentment, anger, hurt, and sadness. Equally problematic, an immediate short-term success provides reinforcement to the confronter, thus encouraging individuals to continue employing similar strategies. Over time, such interaction patterns may lead to the escalation of hostile and aggressive behaviors (Patterson, 1982). The use of conffontive coping has been consistently linked with negative psychological outcomes for the individual (Folkman, Lazarus, Dunkel-Schetter, DeLongis, & Gruen, 1986; Folkman & Lazarus, 1988). When coping with daily marital tension, confrontation has been associated with higher levels of distress for both participants in a dyadic interaction (Daylen, 1993). In this study, confrontation is conceptualized as a hostile and somewhat aggressive manner of achieving the outcomes one desires. It describes a method of coping that includes expressions of anger and a refusal to back down, accompanied by demands that the other concede. Withdrawal Withdrawal can be thought of as an "indirect" conflict management behavior (Sillars & Wilmot, 1994). Some researchers have pointed out that withdrawal can have either a positive or a negative intent. For example, Repetti (1992) conceptualized withdrawal as a method of decreasing arousal after a stressful encounter. In her conceptualization, withdrawal was proposed to be an unconscious behavior, and included such activities as reading the paper or listening to music. In another conceptualization, withdrawal was identified as the conflict management behavior most likely to be taken 32 when concern for the self and concern for the other are both low (Rahim, 1983). In the stress and coping literature, behaviors such as those described above correspond more readily to a coping strategy called Distancing, which involves removing oneself from the problem, not allowing it to "get to you". In this research, the construct named "interpersonal withdrawal" (to distinguish it from "withdrawal") includes items describing a more punitive and negative tone. It is conceptualized as a negative and avoidant response and includes such behaviors as: withdrawing from the other person involved, giving them "the silent treatment", sulking, and efforts to "keep my feelings to myself and to "keep others from knowing about the problem or about my feelings." The Importance of Context In order to cope successfully with interpersonal stressors in stepfamilies, it is likely that a well-developed ability to negotiate settlements between individuals with competing goals would be beneficial. However, contextual influences may interfere with profitable application of such ability. It has been suggested that social conflict can produce rigid thinking and decreasing cognitive complexity, leading to an inability to consider alternate perspectives (Carnevale & Probst, 1998). In more cooperative contexts, however, individuals are more apt to see how trade-offs can lead to mutually beneficial agreements (Carnevale & Pruitt, 1992). Thus, the efficacy of a coping strategy used on a particular day to cope with a specific stressor by one parent will be influenced by the general coping strategies of both parents in the stepfamily. This issue wil l be addressed by considering parents' typical ways of coping as a backdrop for the day-to-33 day interactions between parents and children. To further examine the effects of coping as context, husbands' and wives' typical ways of coping will be used to predict relationship quality two years later. Methodological Issues The study of stepfamilies is fraught with complexities that present methodological problems for researchers. Because of the many variants of stepfamily composition, there is often confusion about what types of stepfamilies are being studied. In many stepfamilies, not all family members necessarily reside full-time in one household. In one stepfamily household there may be several combinations of full- and part-time living arrangements for different children. Some researchers have attempted to develop typologies to describe different types of stepfamilies (Pasley & Ihinger-Tallman, 1982; Clingempeel, Brand, & Segal, 1987; Wald, 1981). These typologies give from 9 to 15 variations of stepfamilies, and do not include the additional complexities that occur when remarried couples have children from the current union. Findings from research on stepfamilies are difficult to generalize, as many studies have focused on only one member of the remarried family, collapsed data across different structural types of stepfamilies, and have failed to control for a number of other variables related to outcomes. Further, for studies that compare means across groups, it is often difficult to obtain the necessary number of participants for each category, let alone try to match across groups for such characteristics as socioeconomic status, age, presence of stepsiblings, or children from the new marriage. 34 Such difficulties have forced researchers to take a more eclectic and flexible view of stepfamilies. Instead of focusing on differences between family structures, it may be more profitable to identify variables that reliably predict differences in important outcomes and use a multivariate approach to understanding stepfamilies. In this research, the characterization of families as complex groups leads to the application of multilevel modeling as a statistical technique for the analysis of data with a hierarchical nesting structure. Variables that describe differences between stepfamilies can then be modeled for a more specific understanding of their influences. In the case of families, children can be thought of as being nested within the parental structure (Snijders, 1995). A l l children in the family have the same parents (or stepparents), so their relationship outcomes are not independent. There are a number of distinct advantages to the use of multilevel models to analyze this type of data. First, simpler methods, such as ordinary least square regression analysis, do not take into account the grouping of data in families, and therefore the models are misspecified and the results reliable. A second advantage of multilevel models is that the observed variance is decomposed into variance due to differences between children and variance due to differences between families so that explanatory variables can be modeled separately on each. A third advantage is that this method of analysis considers variance in the slopes separately from variance at either level. A fourth advantage is that in multilevel analyses, it is possible to compare relationship quality of stepchildren and biological children to the same parent, initially using data from those families who contain both types of children, but "borrowing strength" through an iterative process from those families who are missing data (in this case, children). In 35 other words, for those families in which only the wives' children are described, the data they provide are still considered in the calculation of the final coefficients, even though they cannot provide within-family information on the "stepgap". This type of analysis is also very fruitful for the examination of diary data. Days are considered to be nested within individuals (husbands and wives). In turn, these individuals are nested within families. For the analyses of diary data in this study, a three-level model is used. One of the strong advantages of such a model is that it allows for the generalization of results to the individual, within-family case. When data are aggregated, relations between macro-levels cannot be used to make assertions about micro-level relations. However, with multilevel analyses, we can make statements about micro-level relations, as well as how these micro-level relations may vary depending upon macro-level variables. Goals of the Present Research This research had three main goals. The first goal was to map out the extent of the "gap" in the way parents related to their own children as opposed to their stepchildren. Further, by examining the influences of children's and families' characteristics on relationship quality in stepfamilies, results generally found in the stepfamily literature could be refined by applying them to within-family analyses. The second goal was to explore the day to day relations between parents' coping and daily fluctuations in the quality of parent-child relationships. The third goal was to examine the effect of remarried parents' coping on relationship quality over time, considering 36 ways of coping that may either maximize or minimize the "gap" in the way parents relate to their own children as opposed to their stepchildren. 37 STUDY HYPOTHESES The study hypotheses are listed below. The first set of hypotheses relate to the identification of a "stepgap" in relationship quality, and the replication of findings from the stepfamily literature. The second set of hypotheses concern the diary portion of the study, and focus on day-to-day relationships between coping and relationship quality. The third set of hypotheses considers the relations between parents' coping and relationship quality two years later. Set I Hypothesis I: It was expected that there would be differences in parent-child relationship quality between own children and stepchildren (i.e., the stepgap). HI A: It was expected that the stepgap would be evident for both husbands and wives, but that the stepgap would be greater for wives than for husbands H1B: It was expected that the stepgap would be evident both at Time 1 and at Time 2. H1C: Husbands were expected to report greater closeness with their own children than with their stepchildren. HID: Husbands were expected to report greater tension with their stepchildren than with their own children. HIE: Wives were expected to report greater closeness with their own children than with their stepchildren. 38 H1F: Wives were expected to report greater tension with their stepchildren than with their own children. Hypothesis 2: It was expected that characteristics of individual children would influence parent-child relationship quality in stepfamilies. H2A: Parents were expected to report greater closeness with younger children than with older children. H2B: Parents were expected to report greater tension with older children than with younger children. H2C: Husbands were expected to report greater closeness with male children than with female children. H2D: Husbands were expected to report greater tension with female children than with male children. Hypothesis 3. It was expected that children's characteristics would interact with the stepgap in the prediction of parent-child relationship quality. H3A: Wives were expected to report greater closeness to stepchildren who spent more time in the stepfamily home. Hypothesis 4. It was expected that family characteristics would have an influence on parent-child relationship quality, after controlling for children's' characteristics. H4A: Parents were expected to report less closeness and greater tension with all children in stepfamilies where the children were older on average. H4B: Parents in stepfamilies that had been in existence longer were expected to report less closeness and greater tension with all children than parents in stepfamilies that had been together for a shorter length of time. 39 Hypothesis 5: It was expected that family characteristics would moderate the stepgap. H5A: Husbands in stepfamilies with children from the current union were expected to report greater closeness and less tension with stepchildren than husbands in stepfamilies with no children from the current union. H5B: Wives in stepfamilies with children from the current union were expected to report greater closeness and less tension with own children than wives without children from the current union. In addition to addressing the hypotheses specified above, the long-term effect of family characteristics on the stepgap at Time 2 was also explored. Set 2. The hypotheses for the diary data were more exploratory in nature, and the main goal was to identify patterns of behavior that show a lagged relationship between coping and changes in relationship quality. Hypothesis 6: It was expected that parents' ways of coping with daily family stressors would be related to fluctuations in parent-child relationship quality. The following general hypotheses regarding directionality were also proposed. H6A: It was expected that compromise would be related to improved relationship quality. H6B: It was expected that confrontation would be related to poorer relationship quality. 40 H6C: It was expected that interpersonal withdrawal would be related to poorer relationship quality. Hypothesis 7: It was expected that daily indications of relationship quality would be related to parents' subsequent coping. Set 3 Hypothesis 8: It was expected that parents' average daily coping would be related to relationship quality at Time 2. Two specific hypotheses were proposed: H8A: It was expected that wives' coping (as a set) would moderate husbands' Time 2 stepgap. H8B: It was expected that husbands' coping (as a set) would moderate wives' Time 2 stepgap. In addition, the same general hypotheses regarding directionality for the relations between coping and relationship quality presented in Set 2 were considered for Set 3. 41 METHODS Sample Couples were recruited from the lower mainland of British Columbia, Canada, by means of newspaper and radio advertisements, notices in school newsletters, posters on community bulletin boards, and solicitation at several local stepfamily groups. In our sample, 71 percent reported hearing of the study through the newspaper or radio, and 29 percent were notified through posters, or by a friend. The requirements for participation were that couples be either married or living together and have a child from a previous union living in the home at least 25 percent of the time. Interested couples who qualified were asked to telephone the project office for more information. Upon contacting the project office, couples were sent a letter describing the study's goals and procedures. This letter also noted that all participating couples would be entered in a random $500 drawing. If couples met the qualifications for participation, and agreed to participate in the study, preliminary demographic information was obtained. In-depth telephone interviews were then scheduled separately with each spouse. Following the telephone interviews, respondents were mailed a set of structured diaries to be completed each morning and evening over a period of one week. For this study, only the evening segment of the structured diary was used. Respondents were asked to complete the diary materials and return them in the stamped envelopes provided. 42 In the instructions accompanying the diaries, the importance of each spouse completing the materials independently was emphasized. The instructions read: "We ask that you and your spouse complete all of the study materials separately and that you do not discuss your responses with one another until after the materials have been returned to us." Each spouse was also provided with a number of tabs and instructed to seal each day's diary after completion. It was expected that these instructions would encourage the independent completion of the forms by each participant. Approximately 20 months after the initial interview, couples were re-contacted for a second telephone interview. In this interview, participants again answered questions regarding the quality of all their family relationships. Data Available for Analysis The details of the data available for analysis are presented in Table 1. There were 154 couples who participated in the initial interview. Of these, 142 also participated in the second interview approximately two years later. In addition, 81 couples returned completed diaries for a couple diary response rate of 53 percent. Only the relationship quality with minor children (under the age of 21) were used in this research, because adult children spent little or no time in the stepfamily home, and were thus unlikely to be affected by parents' ways of coping with daily stressors. Children under the age of two were also excluded, as reports of relationship quality with babies were described as uniformly close and lacking in tension. Further, relationship quality for children from the current union was not included, as these children were not anyone's stepchildren, and differed qualitatively from those children who were from previous unions. Further, age was completely confounded with their status, as these children were always the youngest in the stepfamily. Husbands and wives did not always report on the same number of children. There were a number of reasons for this. The primary reason was that the stepparent did not always have a relationship with their partner's child. Sometimes the children lived in another city, were away at school, or did not visit the stepfamily home for some other reason. Therefore, only the parent could describe their relationship with that child, but not the stepparent. In other cases, only one parent participated in the interview. Table 1. Data Available for Analysis Husbands # of children Wives #of children Couples Completed Family information 156 160 154 Completed Interview 1 156 167 154 RQ reports on at least one child at at T1 148 404 154 407 NA RQ reports on at least one own child and one stepchild at T1 89 96 86 Completed Interview 2 142 155 142 RQ reports on at least one child at T1 andT2 128 354 131 346 NA RQ reports on at least one own child atT1 and T2 85 94 NA RQ reports on at least one stepchild atT1andT2 83 89 NA RQ reports on at least one own child and at least one stepchild at T2 81 89 78 Completed Diary Data 81 81 81 Completed diary data and reported on at least one child at T1 and T2 70 200 74 209 NA Contributed at least one day-pair for diary analyses. 35 27 22 Number of day-pairs contributed 90 98 Note: Husbands and Wives did not always report on the same number of children because in some circumstances, although the father maintained a relationship with a child from a previous marriage, his new wife did not know the child, or had only met the child briefly, and therefore did not wish to comment on relationship quality. On a few occasions, a similar circumstance applied to husbands. 44 For the analysis of diary data, only data from parents who reported having contact with both a stepchild and their own child on two consecutive evenings could be used. It would be impossible to interpret lagged results unless this restriction was made. Therefore, only 90 day-pairs from 35 husbands and 98 day-pairs from 27 wives were available for analysis. For these analyses, a three-level model was used, with day-pairs nested within individuals who were then nested within families. This allowed for the analysis of 188 day-pairs, from 62 parents, representing 41 families. Sample Characteristics The characteristics of this sample of stepfamilies are presented in Table 2. The median family income for the families in the study was $68,000 (CDN) per year, indicating a comfortable middle-class standard of living. This is only slightly higher than the average family income reported by Statistics Canada for the area where they lived ($64,778). Husbands had significantly greater personal income than did wives (M= $56,250 and M = $25,550, respectively, t (154) = 6.00, p < .001). The mean education level of individuals participating in the study was 13.86 (ranging from 8 to 17) years, and there was no significant difference in educational level between husbands and wives on a paired f-test. This educational level is about average for their area, according to Statistics Canada (1996). On the Lower Mainland, approximately 75% of individuals have a Grade 12 or higher educational level, and over 50% have some kind of certificate or diploma for training after high school. Husbands were significantly older than their wives were (M= 40.24, and M = 37.17, t (154) = 5.564,/? < .001). This is consistent with previous 45 research showing that remarried couples are more likely to be in age-heterogamous marriages than first-marrieds (Booth & Edwards, 1992). Table 2 Stepfamily Characteristics Husbands Wives Couples Mean income (CDN) $56,150.00 a $25,550 b $68,000.00 Education (years) 13.94 13.78 13.86 Age 40.237 a 37.168 b 38.55 Mean number of children from a previous union 1.65 a 1.37 b Mean age of children from a previous union 12.41 11.66 Mean time spent by children from a previous 6.71 a 10.57 b union in stepfamily home (months per year) Participants reporting at least one previous 93% 93% union Participants previous divorced or separated 88% 91% Participants widowed 5% 2% Mean number of children living in stepfamily 3.14 home: Mean number of years stepfamily living 4.57 together Proportion of stepfamilies with children from 0.31 current union Note: Those means with different subscripts are significantly different from each other, p < .05 The modal number of previous marriages for both husbands and wives was one. In fact, 93% of the participants reported at least one previous union. Of the total number of men and women participating in the study, 88% of husbands and 91% of wives were either divorced or separated from their previous partner. Only 5% of the husbands and 2% of the wives had been widowed. Although the requirement for participation in the study was that one stepchild live in the household for three months per year, data from all children between the ages of 2 and 21, whether living in the stepfamily household or not, 46 were included in the analyses. The mean number of children in each stepfamily was 3.14, ranging from 1 to 8. The average age of the children in each family was 11.19, ranging from 2.4 to 24 years. The average length of time the stepfamily had been in existence was 4.57 years (ranging from 1 to 16). Of all the stepfamilies, 31% had children from the current union. In Table 3, the custody arrangements and amount of time spent by stepchildren in the stepfamily household are presented. In this sample, over 50% of the husbands' children lived with their fathers at least half of the time. This is probably somewhat higher than the national average. According to Statistics Canada, approximately one-third of stepfamilies are what they term "blended" stepfamilies. This term refers to stepfamilies in which both parents have brought children from previous unions to live, at least part-time, in the new stepfamily household. Table 3. Custody Arrangements and Type of Time Spent in Stepfamily Home by Children From Previous Unions Husbands'children Wives'children Custody arrangements: Full custody 35% 74% Ex-partner full custody 25% 1% Joint custody 32% 20% Other 8% 4% Time spent in stepfamily home: None of the time 11% 3% Weekends 3% 0% Vacations 1% 0% Summers and weekends 3% 0% Most of the time 10% 13% all of the time 32% 66% not specified 41% 18% 47 Characteristics of Diary Sample Couples who completed diary data were compared to those who did not on a variety of demographic variables, including education, income, years in the stepfamily, the number of children from the current union, and the average age of children in the stepfamily. The only significant difference between couples who completed diaries and those that did not was the average age of the children. In stepfamilies where couples completed diary data, the children were older on average than in stepfamilies where couples did not complete diary data, (M= 12.02, and M = 9.79, respectively, t (153) = 2.94, p < .01). Two possible explanations for this difference are that either parents with children who were older on average had more time to participate in a daily diary study, or they were more motivated to participate because they were experiencing more problems with their stepchildren. To test this second possibility, couples who completed diaries were compared to those who did not on all aggregated measures of relationship quality. However, neither husbands nor wives in couples who completed diaries reported significantly different levels of tension or closeness than husbands and wives who did not complete diaries. Wives completed more diaries than their husbands did (M= 5.6 and M = 5.0 respectively, paired t (80) = - 3.239,/? < .001). Measures The variables used in this study are described as either macrolevel variables, representing characteristics of the children and the family, or microlevel variables, used in the diary portion of the study. 48 Macrolevel variables Macro-level variables are those variables that were obtained by asking parents to give a general impression of their relationship with their stepchildren and their children from a previous union. These variables were used at the level of the child, as well as being aggregated for some analyses. Aggregates were created separately for own children and stepchildren, and represented the average relationship quality described by parents across each of their own children, and across each of their stepchildren. Parent-child relationship quality At the macrolevel, the parent-child relationship was assessed by parents' ratings of items regarding closeness and tension with each child at Time 1 and Time 2. At the beginning of the interview, the interviewer said, "I would like to start by asking you a few questions about your relationship with (each of) your child(ren) and your step-child(ren). Husbands and wives were asked to rate on a five-point scale how close they felt to each child, and to what extent the relationship with each child was characterized by tension. In each case, "1" meant not at all and "5" meant very. The means and standard deviations of these variables are reported in the results section, along with comparisons between husbands' and wives' mean scores, as well as comparisons between reports for own children and for stepchildren. A significant limitation of this study was the lack of a standardized measure for the assessment of relationship quality. One of the trade-offs for collecting data that describe a number of factors regarding the family milieu is a lack of in-depth reports for every area of interest. Observational data, as well as children's reports, would also 49 provide important insights into the topic of this research. However, for the present study, the focus of interest was on the stepgap, which can be conceptualized at the between-family level of analyses as a latent variable representing the degree of disparity between parents' perceptions of their relationships with their own children, and their perceptions of their relationships with their stepchildren. For this purpose, the broad assessments of relationship quality used in this study provided consistent, face-valid information. A n added advantage was the use of similar constructs to assess relationship quality in the diary study, where longer measures were not feasible. Autocorrelations provided a measure of consistency in ratings from Time 1 to Time 2. For wives', there was a high degree of consistency in reports of closeness with individual children (r = .73). Husbands' reports of closeness were moderately consistent (r = .56). Wives' reports of tension also showed moderate consistency across timepoints (r = .44), whereas husbands' reports of tension, although again somewhat lower, were still moderately consistent (r = .37). Children's characteristics When stepfamilies first contacted the laboratory to indicate their interest in participating in the study, a family information sheet was completed over the telephone. Some of the information obtained pertained to the number of children, their ages, sex, and parentage. Age of the child, sex of the child, and sex of the child's stepparent were used in conjunction with family information and information on relationship quality to test the first set of hypotheses. Other information regarding length of the stepfamily relationship, whether or not there were children from the current union, and the mean age of children 50 in the family were used as family variables expected to impact parent-child relationships in stepfamilies. Microlevel variables These variables were taken from the daily diary, completed twice daily by husbands and wives for seven consecutive days. These data were also aggregated for some of the analyses, for use as contextual variables. Only portions of the diary relevant to the proposed study are described here. Each evening participants were instructed to select a family stressor that had occurred that day. The following instructions were used to guide subjects in their choice: Please describe briefly the most bothersome event or problem you had with someone in your family today. It might have been something as minor as your child's distress over something that happened at school or it might have been a major argument or disagreement. Whatever your most serious family problem was today (no matter how minor or trivial it may seem to you), please describe it here. Stepparents' coping strategies. The use of compromise, confrontation, and interpersonal withdrawal as ways of coping with daily family stressors were measured using the Brief Ways of Coping (BWOC). This scale is the result of psychometric work completed on three previous data sets (Bishop, 1990; DeLongis & Kessler, 1986; Preece, 1994) and a final factor analysis using the current data set (Preece, O'Brien, DeLongis, & Campbell, 2000). Participants were 51 asked to describe their use of each strategy on a three-point scale, indicating either a "1" (not at all), a "2" (a little) or a "3" (a lot). The items and alphas for these scales are listed in Appendix A . Lagged autocorrelations for the diary data indicated a moderate degree of consistency for all three forms of coping. Similar results were obtained for compromise (r = .35), confrontation (r = .32), and interpersonal withdrawal (r = .35). The aggregated coping scores representing the average use of these strategies on all days that a family stressor was reported were used as macrolevel variables describing parents' typical ways of coping with interpersonal stress. Daily parent-child relationship quality. At the daily level, parent-child tension was measured by two items. Parents were asked, "How much tension has there been this evening in your relationship with your child(ren)?" This item was rated on a four-point scale, ranging from "not at all" to "a lot". There was also an option to note "not applicable" i f they had not been with any of their own children that evening. They were also asked, "How much tension has there been this evening in your relationship with your stepchildren)? This item was rated on a similar scale, with an option to choose "not applicable", i f they had not interacted with an stepchild that evening. Closeness was measured by one item on a three-point scale. The item read, "Considering the whole day, to what extent did your child(ren) show affection or support to you?" This question could be answered with a " 1 " (not at all), a "2" (somewhat) or a "3" (a lot). Then the same question was also asked with regard to their stepchildren. Lagged autocorrelations were also calculated for the reports of relationship quality. Affection from stepchildren and affection from own children were both highly 52 consistent across days (r - 56, and r = 50). However, consistency for tension with stepchildren was quite low (r = . 11), although tension with own children was moderately consistent (r = .36). Statistical Analyses Hierarchical linear modeling The hierarchical linear model (HLM) is a variant of the multiple linear regression model for data with a hierarchical nesting structure. For the longitudinal analyses, children (level-1 units) are indicated by / and families by j. The dependent variable is defined at the level of the individual, and is written as For the longitudinal analyses, the dependent variables were parents' reports of either closeness or tension, indicating aspects of the quality of the parent-child relationship as perceived by the parent. For the daily analyses, day-pairs, each corisisting of two consecutive days of diary reports, were the unit of analysis at level 1, parents' gender and average coping were used as control variables at level 2, and at the third level, differences between families were controlled. At the daily level, the dependent variables were (1) reports of affection that day or tension that evening, and (2) coping strategies used. Indicator variables. Indicators, or "dummy" variables, were created to pick out the types of relationships to be compared. One indicator variable, Xjj, was used in the analyses to consider the "stepgap" as a factor predicting relationship quality at the children's level of analyses. This variable was coded "+ 1" for wives' children from a previous marriage 53 and "- 1" for husbands' children from a previous marriage. Thus, when this variable was a significant predictor of relationship quality, it could be interpreted as indicating a significant difference between wives' children and wives' stepchildren (husbands' children) when wives' relationship quality was the dependent variable. Conversely, when husbands' relationship quality was the dependent variable, a significant "stepgap" indicated a significant difference between husbands' stepchildren (wives' children) and husbands own children. A second indicator variable was created to compare male children and female children, with "+ 1" indicating a male child, and " - 1 " indicating a female child. Application of multilevel modeling to family data A detailed description of the statistical analyses used in this research is provided in Appendix B. The discussion therein provides a theoretical background of multilevel modeling. Those aspects of the statistical analysis that are specifically relevant to this study wil l be discussed in more detail along with the pertinent results. 54 RESULTS In this section, I begin by reporting the results of analyses using aggregated relationship quality variables for parents' own children and parents' stepchildren. The main purpose of these analyses was to establish the existence of the stepgap and provide preliminary evidence that it differs for husbands and wives. A n additional goal was to replicate with within-family analyses those effects typically found using between-family analyses. Next, I present multilevel analyses examining the effect of children's characteristics and family structure on parent-child relationship quality in stepfamilies, including the effects of family structure on the stepgap, both in the short-term, and over time (two years later). Then, I examine the day-to-day relations between parents' ways of coping with family stress and reports of tension between parents and children and affection from children towards parents. Finally, I present the results of analyses exarnining the effect of parents' average coping on parent-child relationships two years later as well as an analysis of how the average coping of one parent may moderate the stepgap in relationship quality for the other parent. Aggregated Relationship Quality for Own Children and Stepchildren Differences between Husbands and Wives Table 4 presents the preliminary analyses examining differences in husbands' and wives' reports of relationship quality in stepfamilies. The means and standard deviations of husbands' and wives' reports of closeness and tension with their children, aggregated 55 separately for stepchildren and for own children, are provided here. The purpose of these analyses was to establish the existence of the stepgap and obtain preliminary evidence that it differs for husbands and wives. These results must be interpreted with caution, as they do not control for a number of other influences shown to be relevant to parent-child relationship quality in families. However, these other influences were controlled in the multilevel analyses presented later. Table 4 Aggregated Reports of Relationship Quality: Differences between Husbands and Wives Variable Husbands Wives Paired Statistics M SD M SD R df t Time 1: Own Children Closeness 4.17 0.87 4.63 0.80 0.02 86 - 3.69 *** Tension 2.24 1.00 1.91 1.17 -0.05 86 1.921 Stepchildren Closeness 3.56 0.94 3.01 1.19 0.04 87 3.47*** Tension 2.41 0.93 2.84 1.29 0.07 87 -2.65** Husbands' Children Closeness 4.18 0.87 3.16 1.14 0.27 ** 117 8.99*** Tension 2.22 0.94 2.79 1.19 0.20 * 117 - 4.59 *** Wives' Children Closeness 3.65 0.91 4.64 0.74 0.23 * 120 -10.53 *** Tension 2.34 0.91 1.93 1.11 0.14 120 3.39 *** Time 2 Own Children Closeness 3.95 1.15 4.44 0.77 0.00 78 -3.09** Tension 2.18 1.13 2.16 1.04 0.05 78 0.12 Stepchildren Closeness 3.56 1.11 2.66 1.11 0.00 79 5.16*** Tension 2.39 1.17 2.64 1.20 -0.03 78 -1.31 Husbands' Children Closeness 3.93 1.07 2.77 1.06 0.32 *** 105 9.61 *** Tension 2.21 1.06 2.64 1.15 0.23 * 105 -3.22** Wives' Children Closeness 3.60 1.06 4.39 0.74 0.17 » 104 - 6.84 *** Tension 2.39 1.12 2.26 0.96 0.25 ** 104 1.03 56 Note: Means are slightly different for different analyses because of differ numbers of couples are being compared lp < .10, *p < .05, **p < .01, ***p < .05 The results indicate that at Time 1, husbands' reports of closeness with children from a previous union were significantly lower than their wives' reports of closeness to their children from a previous union, (M= 4.17 and M = 4.63, respectively, paired t (86) = - 3.69,/? < .001). There was also a trend towards husbands reporting greater tension with their own children than their wives reported with their own children, (M= 2.24, and Af = 1.91, respectively, paired t (86) = 1.92,/? < .10). For relationships with stepchildren, however, this pattern was reversed. Husbands reported significantly greater closeness to their stepchildren than their wives did to their stepchildren (M= 3.56 and M= 3.01, paired t (87) = 3.45,/? < .001). Husbands also reported significantly lower average levels of tension with their stepchildren than their wives did with their stepchildren (M= 2.41, M= 2.84, respectively, paired t (87) = - 2.65, p < .01). Taken together, these results indicate that husbands felt closer and experienced less tension with their stepchildren than wives did with their stepchildren, but husbands felt less closeness with their own children than wives did with their own children. Because the t-tests presented above are based on report comparisons with different children (husbands' own children are wives' stepchildren, and vice versa), analyses were also conducted to compare husbands' and wives' reports of relationship quality with the same group of children. The number of couples to be compared is greater for these analyses, and thus the means differ slightly. At Time 1, the results indicated that husbands' reported greater closeness to their own children from a previous union than their current wives did to the same children (M= 4.18 and M= 3.16, respectively, paired t 57 (117) = 8.99,p < .001) as well as less tension with their own children than their wives reported with those children (M= 2.22 and Af = 2.79, paired t (117) = - 4.59, p < .001). Similarly, husbands reported less closeness with their wives' children than their wives did with their own children (M= 3.65 and M= 4.64, respectively, paired t (120) = - 10.53, p < .001) and greater tension with their wives' children than their wives did with their own children (M= 2.34 and M= 1.93, respectively, paired t (120) = 3.39,/? < .001). These results suggest that the quality of the husbands' relationship with his stepchildren was still significantly poorer than the relationship between those children and their own mother. At Time 2, husbands still reported less closeness with their own children than their wives reported with their own children (M= 3.95 and M= 4.44, paired t (78) = -3.09, p < .01). And again, husbands reported being significantly closer to their stepchildren than wives were to their stepchildren (M= 3.56 and M= 2.66, paired t = 5.16, p < .001). However, at Time 2 there was no significant difference between husbands' and wives' reports of tension with their own children, and no significant difference between husbands' and wives' reports of tension with their stepchildren. . These results suggest that differences between husbands' and wives' reports of tension that are related to the new stepfamily structure may dissipate somewhat over time. This also points to the importance of controlling for time in the stepfamily as well as children's age, when looking at changes over time. Because of the possibility that these differences were related to characteristics of the specific children, husbands' and wives' reports of relationship quality with the same children were also compared. Husbands reported being significantly closer to their own 58 children than their wives did to their husbands' children (M= 3.93 and M- 2.77, paired t (105) = 9.61, p < .001) and reported lower levels of tension with them than their wives did (M= 2.21 and M = 2.62, respectively, paired t (105) = - 3.22,/? < .01). For wives' children, on the other hand, husbands reported being significantly less close to them than their wives reported themselves to be (M= 3.60 and M= 4.39, paired t (104) = - 6.84,/? > .001). There were no significant differences between husbands and wives in reports of tension with the wives' children. Because these analyses did not take into account any of the children's characteristics, it is instructive to examine the correlations between husbands' and wives' aggregated reports of relationship quality with the same children. Although a number of children's qualities may affect parent-child relationships, Correlations at Time 1 indicated a significant relationship between husbands' and wives' reports of closeness with both husbands' children (r = .27,/? < .01) and wives' children (r = .23,/? < .05), and a significant relationship between husbands' and wives' reports of tension with the husbands' children (r = .20,/? < .05). However, the relationship between husbands' and wives' reports of tension with the wives' children (r= .14,/? > .10) was not significant. This last result suggests that husbands' reports of tension with wives' children at Time 1 may be based upon concerns not applicable for wives. At Time 2, on the other hand, husbands' and wives' reports of tension corresponded to a greater degree (r = .25, p < .01). 59 Differences Between Own Children and Stepchildren The results reported in Table 5 are taken from analyses that compared aggregated relationship quality with parents' own children to aggregated relationship quality with stepchildren. The significant differences here are evidence of the stepgap in relationship quality that existed in these stepfamilies, and provide support for Hypothesis I. Husbands reported being significantly closer, on average, to their own children than to their stepchildren, both at Time 1 (M= 4.19 and M= 3.59, paired t (89) = 4.68,/? < .001) and at Time 2 (M = 3.99 and M = 3.52, paired t (81) = 2.92,/? < .05). However, there was no significant difference between average tension with own children and stepchildren at Time 1 (M= 2.22 and 2.40, paired t = - 1.33,/? > .10) and only a trend towards a significant difference at Time 2 (M= 2.16 and M= 2.42, paired t = -1.87,/? < .10). Table 5 Aggregated Reports of Relationship Quality: Differences Between Own Children and Stepchildren Own Children Stepchildren Paired Statistics Relationship Quality M SD M SD r Df t Husbands Time 1 Closeness 4.19 0.86 3.59 0.96 0.12 89 4.68*** Time 2 Closeness 3.99 1.15 3.52 1.11 0.17 81 2.92** Time 1 Tension 2.22 0.99 2.40 0.93 0.11 89 -1.33 Time 2 Tension 2.16 1.13 2.42 1.19 0.41 *** 81 -1.871 Wives Time 1 Closeness 4.62 0.78 2.96 1.19 0.33** 96 13.72*** Time 2 Closeness 4.44 0.77 2.63 1.10 -0.02 89 12.69*** Time 1 Tension 1.93 1.15 2.82 1.28 0.44 *** 96 - 6.85 *** Time 2 Tension 2.15 1.03 2.53 1.19 0.12 89 -2.40* tp < .10, * p < .05, ** p < .01, *** p < .05 60 Wives reported being significantly closer on average to their own children than to their stepchildren, both at Time 1 (M= 4.62 and M= 2.96, paired t (96) = 13.72,/? < .001) and at Time 2 (M= 4.44 and M= 2.63, paired t (89) = 12.69,/? < .001). Wives also reported significantly less tension on average with their own children than with their stepchildren, both at Time 1 (M = 1.93 and M = 2.82, paired t (96) = - 6.85, p < .001) and at Time 2 (M = 2.15 and M = 2.53, paired t (89) = - 2.40, p < .04). These results are consistent with Hypothesis IC, which predicted that husbands would be closer to their own children than to their stepchildren, Hypothesis IE, which predicted that wives would report greater closeness to the own children than to their stepchildren, and Hypothesis IF, which predicted that wives would report greater tension with their stepchildren than with their own children. However, Hypothesis ID, which predicted that husbands would report greater tension with their stepchildren than with their own children was not supported by these analyses. These results also suggest that the stepgap continues over the long term, partially supporting Hypothesis IB. However, as these stepfamilies have been in existence of different lengths of time, it was important to also control for the number of years they have lived together, as was done in the multilevel analyses. There were also significant correlations between husbands' reports of tension with own children and with stepchildren at Time 2 (r = .41,/? < .001) and significant correlations between wives' closeness with own children and stepchildren at Time 1 (r = .33,/? < .01) as well as significant correlations between wives' tension with own children and stepchildren at Time 1 (r = .44, p < .001). These correlations suggest that there are influences on the parent-child relationships other than parentage that influence parents' relationships with all children in the stepfamily. One of these influences may be the 61 amount of stressors the family has to deal with, as well as the way that parents are coping with these stressors. This notion was also explored in later analyses. Differences Between Time 1 and Time 2 The next set of results examined whether there were overall differences in relationship quality from Time 1 to Time 2. Before modeling variables on changes in the stepgap over time, it is important to conduct a preliminary examination of such differences. In Table 6 differences between Time 1 and Time 2 relationship quality are reported separately for husbands and wives. Husbands' did not report significantly different levels of closeness and tension with their own children at Time 1 and Time 2. However, husbands' reported greater closeness to stepchildren at Time 1 than at Time 2 (A/= 3.63 and Af= 3.23, paired t (83) = 2.78,/? < .01) and less tension with stepchildren at Time 1 than at Time 2 (Af= 2.23 and Af = 2.60, paired t (83) = - 2.22,/? < .05). These results suggest that there is little change in husbands' relationship quality with their own children from Time 1 to Time 2, but that husbands' generally reported a deterioration in relationship quality with stepchildren from Time 1 to Time 2. However, as the results of these analyses do not consider the effect of children's age or the length of time the stepfamily has been together, they should not be considered definitive until placed in context, as wil l be demonstrated in later analyses. For wives, there was also no difference in reports of closeness with either own children or stepchildren, and no difference in reports of tension with own children from Time 1 to Time 2, but there was a trend towards greater levels of tension with stepchildren at Time 1 than at Time 2 (Af= 2.81 and Af = 2.46, paired t (89) = 1.83, /? < . 10). Taken together, these results suggest 62 that there is more variation across time available for modeling husbands' relationship quality than there is for wives. Table 6 Aggregated Reports of Relationship Quality: Differences between Time 1 and Time 2 Time 1 Time 2 Paired Statistics Variable M SD M SD R df t Husbands Own Children Closeness 4.15 0.89 4.02 0.92 -0.21« 85 0.87 Tension 2.18 1.02 2.25 0.77 -0.09 85 -0.51 Stepchildren Closeness 3.63 0.92 3.23 1.07 0.10 83 2.78 ** Tension 2.23 0.88 2.60 1.06 -0.23* 83 - 2.22 * Wives Own Children Closeness 4.57 0.81 4.37 0.96 0.13 94 1.64 Tension 1.97 1.22 2.21 1.17 0.44 94 -1.40 Stepchildren Closeness 3.08 1.24 3.05 1.09 0.10 89 0.37 Tension 2.81 1.29 2.46 1.22 -0.04 89 1.831 tp<.10,*p<.05,**p<.01,***p<.05 There were few significant correlations between Time 1 and Time 2 reports, suggesting that aggregated reports of relationship quality with children obscure important information that influences fluctuations in relationship quality beyond the broad distinction between own children and stepchildren. For example, the significant negative correlation between husbands' reports of tension with stepchildren at Time 1 and at Time 2 (r = - .23, p < .05) suggests that those husbands reporting higher than average levels of tension with stepchildren at Time 1 were reporting lower than average levels of tension with stepchildren at Time 2 and vice versa. These types of changes indicate that factors 63 related to children that change reliably over time, (e.g., age) may play a significant role in fluctuations in relationship quality. The Average Stepgap The results in Table 7 again examine the stepgap, defined as the differences in relationship quality between stepchildren and own children. However, for these analyses, the stepgap was quantified in another way. For these analyses, average ratings of closeness with stepchildren were subtracted from average ratings of closeness with own children to provide a difference score representing the stepgap in closeness. Similarly, average ratings of tension with stepchildren were subtracted from average ratings of tension with own children to provide a difference score representing the stepgap in tension. A positive closeness stepgap score indicates greater closeness with own children than with stepchildren, and a negative tension stepgap score indicates greater tension with stepchildren than with own children. As the tension stepgap scores are consistently negative, it is necessary to tMnk of their absolute values when mterpreting comparisons for these scores. The results are therefore presented with absolute value signs. At Time 1, husbands' closeness stepgap score was significantly smaller than wives' closeness stepgap score (M= .62 andM= 1.62, paired t (86) = - 4.82,p < .001). Similarly, husbands' tension stepgap score was significantly smaller than wives' tension stepgap score (M= \ -AS \ andM= | - .951, paired t (86) = | - 3.671 ,p <•. 001). At Time 2, husbands' closeness stepgap score was still significantly smaller than wives' closeness stepgap score (Af = .40 and M= 1.81, paired t (78) = - 5.56,p < .001). However, there was no significant difference between husbands' tension stepgap score 64 and wives' tension stepgap score (M= I - .22 | and M= | - .471, paired t (78) = 1.02, p > . 10). These results provide support for Hypothesis I A, which predicted that the stepgap would be more evident for wives than for husbands. However, they also indicate that the stepgap for tension at Time 2 was not significantly different for husbands and wives. Table 7 Stepgap Scores: Differences between Husbands and Wives Husbands Wives Paired Statistics Stepgap M SD JW SD r Df t Time 1 Closeness Stepgap 0.62 1.22 1.62 1.17 -0 .03** 86 - 4.82 *** Tension Stepgap -0.18 1.29 -0.95 1.26 -0.02 86 - 3.67 *** Time 2 Closeness Stepgap 0.40 1.35 1.81 1.43 -0 .03** 78 - 5.56 *** Tension Stepgap -0.22 1.26 -0.47 1.51 -0 .27* 78 -1.02n.s. * p < .05, ** p < .01, *** p < .001 In Table 8, comparisons between the stepgap at Time 1 and Time 2 are presented. The only significant change from Time 1 to Time 2 was for wives' tension stepgap. There was a significant decrease in the tension stepgap from Time 1 to Time 2 (M= \ -.921 andM= | - .431, paired t (85) = | - 3.161 ,p < .01). Summary of Analyses of Aggregated Relationship Quality Variables In summary, the results presented in Tables 1 through 5 provide preliminary evidence that both husbands and wives reported more closeness with their own children than with their stepchildren, and that there were no significant changes either in closeness or in the stepgap over time except for husbands' closeness with stepchildren, which on 65 average, appeared to decrease over time. Wives also reported feeling closer to their own children than husbands did to their own children, whereas husbands reported feeling closer to their stepchildren than wives did to their stepchildren. There appears to be no change in these differences between husbands and wives over time. In terms of tension, husbands did not report significant differences in tension between own children and stepchildren at either Time 1 or Time 2, but wives reported greater tension with stepchildren than with their own children. Table 8 Stepgap Scores: Differences Between Time 1 and Time 2 Time 1 Time 2 Paired Statistics Stepgap M SD M SD r df t Husbands Closeness Stepgap 0.62 1.19 0.47 1.46 0.67*** 81 1.23 Tension Stepgap -0.22 1.26 -0.26 1.27 0.48*** 81 0.24 Wives Closeness Stepgap 1.63 1.17 1.80 1.34 0.56*** 86 -1.33 Tension Stepgap -0.92 1.30 -0.43 1.47 0.47*** 86 -3.16** *p<.05,**p<.01,***p<.001 The stepgap was greater for wives than for husbands for both closeness and tension at Time 1, and for closeness at Time 2. However, wives' stepgap for tension was not significantly different than that for husbands at Time 2, due to a significant reduction in the tension stepgap for wives from Time 1 to Time 2. This reduction appeared to be due to a reduction in wives' reports of tension with stepchildren at Time 2. None of the results reported in this section control for the effects of child age, child gender, or the length of time the stepfamily has been together, or the effect of 66 children from the current union. In the next section, the multilevel analyses are able to provide a more detailed comparison between parent-child relationships with own children and with stepchildren by allowing for the influence of these important child and family characteristics. Multilevel Analyses of the Effect of Family and Child Characteristics on Parent-Child Relationship Quality The next group of results examines the individual and group-level effects of child characteristics and family characteristics on parent-child relationship quality for individual children in stepfamilies. First, I present preliminary results examining the associations between these variables at both levels of analysis. Preliminary Analyses Child-level Correlations between Children's Characteristics and Relationship Quality In Table 9, analyses examining the intercorrelations among children's' characteristics, and correlations between children's' characteristics and relationship quality are presented. For these correlations, children were treated as independent data sources. Therefore, all of these results should be interpreted with caution, and are presented mainly as indicators to aid in the interpretation of later analyses. Further, although significance indicators are given, it should be noted that the large number of tests reported make it unwise to draw conclusions based solely upon any single result. 67 Child age was negatively related to time spent in household. Time in household was also significantly related to stepparent gender, indicating that wives' children spent more time in the stepfamily household than husbands' children did. Table 9 Child-level Correlations Among Husbands' and Wives' Time 1 and Time 2 Relationship Quality, Child Age, Child Gender, Time Spent in household, and Stepparent Gender. Child Age Child Gender Time Spent in Household Stepparent Gender Child Age 0.01 -0.16*** -0.09* Child Gender3 — -0.02 -0.02 Time in Household — 0.44 *** Stepparent gender — Wives Time 1 Closeness -0.14** 0.02 0.45 *** 0.60*** Tension 0.18*** -0.03 -0.14** - 0.30 *** Time 2 Closeness -0.11* 0.01 0.41 *** 0.58 *** Tension -0.05 -0.09* 0.03 -0.10* Husbands Time 1 Closeness - 0.24 *** 0.06 0.091 - 0.21 *** Tension 0.06 -0.05 -0.04 0.07 Time 2 Closeness -0.13* 0.08 0.03 -0.18*** Tension - 0.01 -0.09 0.091 0.08 Note: Sample size ranges from n=346 to n = 407, due to missing data. aChild gender is coded (+1) for males and (-1) for females. bStepparent gender is coded (+'1) when the stepparent is the father and (-1) when the stepparent is the mother. 1 p < .10, * p < .05, ** p < .01, *** p < .001 For both husbands and wives, there was a significant negative relationship between the age of the child and closeness both at both thnepoints. There was a significant positive relationship between child age and tension at Time 1 for wives only. Gender of the child showed no significant effects. Wives' reports of greater closeness were related to more time spent in household at both timepoints. This correlation is 68 confounded by the fact that wives' own children generally spent more time in the stepfamily household than husbands' children. Less time spent in the household was also associated with wives' reports of greater tension. The relations with stepparent gender indicate that both husbands and wives reported being closer to their own children than to their stepchildren. At Time 1, tension was negatively related to stepparent gender, indicating that wives reported less tension with their own children than with their stepchildren at Time 1. Family-level Intercorrelations Among Family Characteristics and Aggregated Relationship Quality Variables Relationship quality in terms of closeness and tension were aggregated separately for own children and stepchildren. These mean relationship quality variables were intercorrelated for husbands and wives, and also correlated with family characteristics. These correlations are presented in Table 10. Again, they are presented for the information of the reader, and should not be interpreted in isolation. In particular, it should be noted that for these correlations, the mean age of the children in the family is confounded with the number of years the stepfamily has been together. The number of years that the stepfamily had been in existence was related to the mean age of the children. The number of years the stepfamily had been in existence was associated with lower levels of closeness to stepchildren as reported by husbands. Being together longer was also associated with wives' reports of less closeness and more tension with their own children and more tension with their stepchildren. R R "3 R CO 12 R a is R a to I bo « -R R o R Q n C s E R 5 O S Q 3 o •« e a - ~ R i ?> Is CD CO i i t - o o CO CO O o o t cn co J eg 1 0 -^ o o o O ) N ^3" CO t- C D o o C D O C D C D T — O CO T — o o r - - <0 CD T— O O M - l O C D O CD CD O CD CM r -O CD T - CM CD CD • V O O O CD co o C D CD ~ t — * OO O T -CD C D CO CO O O ^ J - ^ f OO w *- 1— x— CD T-^ CD —^_ O CD O CD CD CD O O s o o CM T CD O (fi cz o TO 8 a> CO a) CM CD O O O V Q . * CO * E $ -S o £ Y, CD CO. CO * cr O ai v 70 Having children from the current union was associated with a lower mean age of children in the stepfamily, and related to husbands' reports of greater closeness with his stepchildren at Time 1, and less tension with stepchildren at Time 2. When children in the family were older on average, husbands' reported lower levels of closeness with both their own children and their stepchildren at Time 1. The mean age of children in the stepfamily was also negatively related to wives' closeness to their own children at Time 1 and husbands' reports of closeness to their stepchildren at Time 2. Husbands' reports of higher levels of closeness with their own children were associated with less tension with those children at Time 1, and also associated with wives' reports of greater closeness to those children at Time 1 (wives' stepchildren). However, husbands' reports of closeness with their own children at Time 1 were also associated with wives' reports of tension with those children at Time 2. Higher levels of tension with own children at Time 1, as reported by husbands, were associated with wives' reports of higher levels of tension with stepcMldren, as well as lower levels of closeness with their stepchildren and lower levels of closeness with wives' own children. Husbands' closeness to stepchildren at Time 1 was associated with less tension with stepchildren. Further, husbands' reports of closeness with stepchildren at Time 1 were significantly positively related to wives' reports of closeness with those children (wives' own children) at Time 1, and negatively related to wives' reports of tension with those children (wives' own children) at Time 1. Husbands reports of tension with stepchildren at Time 1 were significantly negatively related to tension with stepchildren at Time 2. However, husbands' reports of tension with stepchildren at Time 2 were also significantly 71 negative related to wives' reports of closeness with those children (wives' own children) at Time 2. Wives' reports of greater closeness to their own children at Time 1 were associated with greater closeness and greater tension with their stepchildren at Time 1. Wives' reports of tension with their own children at Time 1 were positively related to their reports of tension with their stepchildren at Time 1. Wives' reports of greater closeness with their stepchildren at Time 1 were associated with lower levels of tension with their stepchildren at Time 1, and positively related to husbands' closeness to those children (husbands' own children) at Time 2 . Husbands' reports of greater closeness to own children were associated with less tension with own children at Time 2, but associated with wives' reports of greater tension with her own children. Further, husbands' reports of higher tension with own children at were positively associated with wives' reports of higher tension with stepchildren. Husbands' higher levels of closeness to stepchildren was associated with lower levels of tension with stepchildren at Time 2. Wives' greater closeness with own children was associated with higher tension with stepchildren at Time 2. Wives' reports of greater tension with own children at Time 2 was associated with wives greater closeness to stepchildren at Time 2. Wives' greater closeness to stepchildren at Time 2 was associated with higher tension with stepchildren at Time 2. Multilevel Analyses of Child and Family Characteristics Next, the multilevel analyses examining the relative effects of child-level characteristics and family characteristics on relationship quality as reported by husbands 72 and wives at Time 1 are presented. Cross-level interactions representing the influence of family characteristics on the "stepgap" were also considered. Deviance statistics are reported, as indicators of whether or not each set of variables improves the fit of the model to the data. Multilevel models are probability models, and therefore some differences between these models and other common statistical methods are worth noting. Sampling theory makes a distinction between design-based inference and model-based inference. With design-based inference, the focus is on the finite population from which the sample was drawn. With model-based inference, on the other hand, a probability model is proposed by the researcher, and inferences may be drawn to the wider population beyond that from which the sample was drawn. Such inference is adequate to the extent that the random variation in the model adequately reflects the effects that are not explicitly included by means of observed variables (Snijders & Bosker, 1999). Husbands' Reports Of Time 1 Relationship Quality Table 11 presents the results of analyses exarnining the simultaneous effects of child characteristics and family characteristics on husbands' reports of closeness and tension with individual children at Time 1. The first line of the first column of table presents the intercept, which represents the mean level of closeness reported within-families. This mean level is a significant predictor of closeness reported for a single child. As the intercept is always a significant predictor in these analyses, it will not be remarked upon again. The B's are non-standardized, representing the average degree of 73 increase or decrease from the mean for each unit of the independent variable. The standard errors and the t-values are also noted. Child-level predictors of husbands' Time 1 relationship quality. After controlling for the various child-level predictors and their interactions with the stepgap, there was no longer a main effect for the stepgap in husbands' reports of closeness. There was a significant effect for child age (B = - .07, t (396) = - 3.62,/? < .01), indicating that husbands reported significantly less closeness with older children than with younger children.1 The significant interaction between the stepgap and child age (B = - .04, t (396) = - 2.33, p < .05) suggests that closeness was even lower for older stepchildren than for older children of the husband from a previous union. This interpretation was confirmed by rurining separate models for husbands' stepchildren and husbands' own children. For husbands' stepchildren, after controlling 1 Note that degrees of freedom for t-tests at the individual level are based on total number of children, o r n - q - l , n = the number of children in the analyses and q = the number of variables used as predictors at the individual level. However, it must be noted that these degrees of freedom are approximations. In fact, the multilevel does not have an exact t-distribution under the null hypothesis. This is because the variance estimate occurring in the denominator does not have a chi-squared distribution, but is a function of several variables with chi-squared distributions (representing the within and between group variances). Generally, the sample sizes wil l be big enough that the ^-distribution is a good approximation and a large number of degrees of freedom may be assumed. For level 1 variables with a total sample size of 30 or more there is no practical problem because the number of degrees of freedom is close to mfinity anyway. For higher level variables, the best approach is to get approximate degrees of freedom by comparing this testing problem with the aggregate OLS approach. For the main effect of a level-2 variable in a 2-level design, then, the degrees of freedom are the same as they would be i f we ran an ordinary regression model, or N - q - 1, where N = number of level-2 units and q number of level 2 explanatory variables. For cross-level interactions, where the level 1 variable has a random slope, the analogous OLS estimates would involve N witliin-group estimates of the slope regressed on the level 2 variable. Again, the degrees of freedom 74 for the other variables in the model, age was a significant negative predictor of closeness (B = -. 11, t (188 ) = - 3.04,p < .01). For husbands' own children, on the other hand, age was not significantly related to closeness (B = - .02, t (210 ) = - .70, p > .10). This result supports Hypothesis H2A, which proposed that parent would report greater closeness with younger children than with older children. However, this appeared to be true only for husbands' stepchildren, not for husbands' own children. Contrary to hypothesis H2C, the sex of the child was not related to husbands' reports of closeness. For husbands' reports of tension at Time 1, the stepgap was a significant negative predictor of tension (B = - .76, t (396) = - 2.42,/? < .05). Child age was a significant positive predictor of tension (B = .07, t (396) = 2.79, p < .01), and there was a significant interaction between the stepgap and time spent in the stepfamily home in the prediction of tension (B = .05, t (396) = 2.72, p < .01). These results suggest that husbands experienced less tension with their stepchildren than with their own children, and more tension with older children than with younger children. The interaction between stepgap and time spent in the family was explored by rurming the model separately for husbands' stepchildren and husbands' own children. There was a trend towards a significant negative effect for time spent in stepfamily home for husbands' own children, (B = - .03, /(210) = - 1.766,/? < .10), suggesting that husbands' reported somewhat lower tension with their own children when they spent more time in the stepfamily home. The slope for husbands' stepchildren was not significant, but was in the opposite direction (B = .04, t would equal N - q - 1, where q is now the total number of cross-level interactions of this slope with level 2 variables (Snijders, personal communication, January 13,2000). 75 (188) = 1.595, p > . 10). The significant interaction indicates that the slopes were significantly different from each other. Table 11. Child Characteristics and Family Characteristics as Predictors of Time J Relationship Quality between Husbands and Children in Stepfamilies and Effect of Family Characteristics on the Stepgap Husbands' Time 1 Relationship Quality Predictors Closeness Tension 6 SE f 8 SE T Intercept 3.98 0.07 60.36 *** 2.12 0.07 30.81 *** Child level predictors Stepgap3 0.30 0.27 1.10 -0.76 0.32 -2.42* Child sexb 0.05 0.05 1.05 -0.03 0.06 -0.44 Child age -0.07 0.02 - 3.62 ** 0.07 0.02 2.79 ** Time spent in stepfamily home 0.03 0.02 1.62 0.03 0.02 1.31 Stepgap x Child sex -0.03 0.05 -0.60 0.06 0.06 1.02 Stepgap x Child age -0.04 0.02 -2.33* 0.03 0.02 1.79* Stepgap x Time spent in stepfamily -0.02 0.01 -1.02 0.05 0.02 2.72** home Family level predictors Yrs. Living tog. -0.02 0.02 -0.78 -0.02 0.02 -1.03 Children current union 0.05 0.19 0.26 0.08 0.19 0.43 Mean age of children -0.04 0.01 -3.15** 0.01 0.01 0.71 Cross-level interactions (Family Characteristics x Stepgap) Yrs. Living tog. 0.00 0.02 -0.25 -0.01 0.02 -0.60 Children current union 0.47 0.17 2.76** -0.26 0.19 -1.41 Mean age of children 0.04 0.02 2.54* -0.04 0.02 -2.30* Deviance Reduction Statistics: Dev. Par. X2 Dev. Par. X2 Null model 1168.70 3 1254.05 3 Child level predictors 1113.43 12 55.27 *** 1239.26 12 14.79* Family level predictors 1101.22 15 12.21 ** 1238.61 15 0.64 Family Characteristics x Stepgap 1090.47 18 10.75* 1231.66 18 6.95* aStepgap compares wives' children (+1) to husbands' children (-1). b Indicator variable compares male children (+1) to female children (-1). Note: N=148, n=404 * p < .10, * p < .05, ** p < .01. *** p < .001, 76 These results support Hypothesis 2B, which predicted that parents would report more tension with older children than with younger children. However, these results do not support Hypothesis HID, which predicted that husbands would experience more tension with their stepchildren than with their own children. Instead, it appears that husbands report less tension with their stepchildren than with their own children, after controlling for other child characteristics. However, this difference is reduced when husbands' own children spend more time in the stepfamily home. Hypotheses H2C and H2D, which predicted that husbands would experience greater closeness and less tension with male children than with female children, were not supported by these analyses. Family-level predictors of husbands' Time 1 relationship quality. At the family-level of analyses, characteristics that differ between families were used to explain variance in the intercepts. Of the three variables, only the mean age of children in the family was a significant predictor of husbands' reports of closeness (B = - .04, t (144) = - 3.15,p < .01). This result suggests that after controlling for individual children's characteristics, the average age of all children in the stepfamily was a significant predictor of the intercept (or mean level of closeness in the stepfamily), and that in families where the children were older on average, husbands reported significantly lower levels of closeness. This result provide partial support for Hypothesis 4A, which proposed that in stepfamilies with older children on average, there would be less closeness and more tension. For husbands' reports of tension, however, family characteristics did not have a significant effect after controlling for child-level characteristics. Thus, Hypothesis 4A, that there would be more tension in general in families as a function of average age of the children, was not supported by these results. 77 Cross-level interactions (family characteristics by stepgap). Cross-level interactions demonstrate whether variables at one level of analysis interact with variables at another level of analysis in the prediction of the dependent variable. In this section, the question is whether family-level variables might moderate the "stepgap" in closeness. In other words, I considered whether characteristics of the family might increase or decrease the difference between fathers' closeness to their own children and fathers' closeness to their stepchildren. A cross-level interaction is examined by modeling the predictor variable at one level on the slope representing the relationship between the lower level predictor and the dependent variable. The slope representing the relationship between "stepgap" and husbands' Time 1 closeness was not significant, but there was significant variability in this effect across husbands.2 When the three variables representing family characteristics were modeled on that slope, the indicator variable for those families who had a child from the current union had a significant effect (JB = .47, t (144) = 2.76, p < .01). This indicates that although fathers were generally closer to their own children than to their stepchildren, in stepfamilies where there was a child from the current union, the gap in closeness was significantly affected. This interaction is portrayed graphically in Figure 1. 2 This indicates that the relationship between these variables differs significantly between husbands. 78 5 Figure 1. Children from Current Union x Stepgap interaction predicting husbands' Closeness to children in stepfamily at Time 1 The slopes describing the interaction were derived by running separate multilevel analyses for husbands' own children and for husbands' stepchildren and plotting the effect of having children from the current union on husbands' reports of closeness for each group.3 In addition to the slopes being significantly different from each other, as indicated by the significant interaction, the slope describing the relations between the families with children from the current union and husbands' closeness to his stepchildren was significantly different from zero in the smaller sample (B = .60, t (112) = 2.23, p < .05). However, the slope describing the relations between the family characteristic of children from the current union and husbands' closeness to his own children was not 79 significantly different from zero (B = .18, t (109) = - 0.81,/? > .10).4 This result provides partial support Hypothesis 5A, which predicted that having children from the current union would moderate the stepgap effect for closeness reported by husbands. Further, this result suggests that as reflected in the stepfamily literature, husbands report a closer relationship with their stepchildren when there are children from the current marital relationship. The mean age of children in the stepfamily also had a significant effect on the stepgap slope (B = .04, t (144) = 2.54, p < .05). This interaction is portrayed graphically in Figure 2, and indicates a reduction in the stepgap for closeness in families with older children on average than in families with younger children on average. Husbands reported being significantly less close to their own children from a previous union when the children in the stepfamily were were older on average, (B = - .06, t (109) = - 3.226, p < .01). Husbands also reported only a slight decrease in closeness with stepchildren in families where children were older on average, but this decrease was not significant, (B = - .026,1(112) = -1.47, p > .10). In this instance, we see that a reduction in the stegap is achieved by a decrease in relationship quality with husbands' own children that is greater than that for husbands' stepchildren. 3 For these post-hoc analyses, all the variables used in the primary analysis were included as control variables. 4 The degrees of freedom for the two analyses do not add up to the number of families involved in the larger analysis, because in many instances, husbands had both children of their own and stepchildren. 80 Figure 2. Mean age of children in family x Stepgap interaction predicting husbands' closeness to children in stepfamily at Time 1. The mean age of children in the stepfamily also had a significant effect on the stepgap slope for tension (B = - .04, t (144) = - 2.30,/? < .05). This interaction is portrayed graphically in Figure 3, and suggests that the stepgap in tension was smaller for families where children are older on average. When the analyses were run separately, the coefficient representing the relationship between mean age and husbands' tension was not significant either for own children (B = .03, t (112) = 1.47,/? > .10) or for husbands' stepchildren (B = - .016, t (109) = - .847,/? > .10). However, the significant interaction indicates that these coefficients are significantly different from each other. 81 Figure 3. Mean age of children in family x Stepgap interaction predicting husbands' tension with children in stepfamily at Time 1. This result provides partial support for Hypothesis 4B, which predicted that in families where children were older, there would be higher levels of tension. However, the interaction indicates that there is again a reduction in the stepgap that is due primarily to an increase in tension with own children and a decrease in tension with stepchildren. In other words, again the reduction in the stepgap appears to be at the expense of relationship quality with own children. These results, taken together, suggest that for husbands, as the average age of children in the family increases, stepfathers report greater tension with own children than with their stepchildren. Deviance reduction statistics The deviance is a measure of model fit that can also be used as a measure of improvement of model fit for subsequent models. Although this 82 statistic is in some ways analogous to R 2 in generalized least squares regression, it differs in that there is no upper bound, so that predictor variables not related to the dependent variable can actually increase the amount of deviance, or variability (Kreft & De Leeuw, 1998). This statistic can be useful when considering whether a set of variables are important predictors, explaining significant variance in the dependent variable, whereas the t statistic only determines the significance of each coefficient independently. In Table 11, deviance in the dependent variable is reported for the "null" model, which is the model of the dependent variable with no predictors. Next to it is the number of parameters estimated in the null model. Underneath the first deviance statistic is the deviance reported after the child level predictors were included in the model. Note that for closeness, this statistic represents a significant reduction in deviance from the null model. This reduction is reported as a chi-square (%2 (9) = 55.27,/? < .001), indicating a significant reduction in deviance. The degrees of freedom for the chi-square is equal to the change in the number of parameters estimated. For tension, the child-level predictors only resulted in a trend towards a significant reduction in deviance (x2(9) = 14.79,/? < .01). These results partially support Hypothesis 2, which predicted that the characteristics of individual children would influence parent-child relationship quality in stepfamilies, but indicate that they are better predictors of closeness than they are of tension. For closeness, family level predictors as a set also resulted in a significant reduction in deviance (x2 (3) = 12.21,/? < .01). This partially supports Hypothesis 4, which predicted that family characteristics would have an influence on parent-child 83 relationship quality, after controlling for children's characteristics. These results indicate that family characteristics are relevant for husbands' reports of closeness, but not for reports of tension. Cross-level interactions also resulted in a significant reduction in deviance for closeness (x2(3) = 10.75,/? < .05) and a trend towards a significant reduction in deviance for tension (x2(3) = 6.95, p < . 10). This result supports Hypothesis 5, which predicted that family characteristics would moderate the stepgap. Wives' reports of Time 1 relationship quality The results of multilevel analyses of wives' reports of closeness and tension with children in stepfamilies are presented in Table 12. Child-level predictors of Time 1 relationship quality The stepgap was a significant predictor of wives' reports of closeness (B = .95, t (399) = 3.72,/? < .001), suggesting that wives were significantly closer to their own children than to their stepchildren. The stepgap was also a significant predictor of tension (B = - .71, t (399) = - 2.25, p < .05), which indicates that wives also reported significantly less tension with own children than with their stepchildren. These results provide strong support for Hypothesis IE and IF, which proposed that wives would report greater closeness and less tension with their own children than with their stepchildren, by showing that these results are evident even after controlling for other child characteristics. The age of the child was a significant negative predictor of wives' reports of closeness (B = - .06, t (399) = - 3.08, p < .01) as well as being a significant positive predictor of wives' reports of tension (B = .07, t (399) = 3.19, p < .01). These results provide strong support for Hypothesis 2A and 2B, which predicted that children's age 84 would be negatively associated with closeness and positively associated with tension. There was a significant interaction between the stepgap and time spent in the stepfamily home (B = - .05, t (399) = - 3.55, p < .01), indicating a reduction in the stepgap for children spending more time in the stepfamily home. When the model was run separately for wives' stepchildren and wives' own children, it became clear that wives reported being closer to their stepchildren when they spent more time in the stepfamily home (B = .09, t (200 ) = 4.92, p < .001). For wives' own children, on the other hand, the relationship was not significant (B = .002, t (199) = .13, p > .10). This result provides support for Hypothesis 3 A , which predicted that wives would report greater closeness to stepchildren who spent more time in the stepfamily home. Family-level predictors of wives' Time 1 relationship quality At the family level of analyses, the number of years living together as a stepfamily was a significant predictor of tension, (B = .07, r (150) = 3.12,/? < .01). There were no significant cross-level interactions between family level-predictors and the stepgap. Deviance reduction statistics For wives, child level-predictors significantly reduced the deviance statistic for both closeness (%2 (9) = 260.91,/? < .001) and tension (%2 (9) = 57.06,/? < .001). Family-level predictors resulted in a significant reduction in deviance for tension only (%2 (3) = 10.46,/? < .05), and there was no reduction in deviance for cross-level interactions. 85 Table 12 Child Characteristics and Family Characteristics as Predictors of Time 1 Relationship Quality Between Wives and Children in Stepfamilies and Effect of Family Characteristics On the Stepgap. Wives' Time 1 Relationship Quality Predictors Closeness Tension B SE T B SE T Intercept 3.93 0.06 62.88 *** 2.35 0.07 32.39 *** Child level predictors Stepgap3 0.95 0.25 3.72 *** -0.71 0.31 - 2.25 * Child sexb 0.02 0.05 0.33 -0.03 0.06 -0.49 Child age -0.06 0.02 - 3.08 ** 0.07 0.02 3.19** Time spent in stepfamily home 0.02 0.02 1.38 0.02 0.02 1.14 Stepgap x Child sex 0.00 0.04 -0.01 0.00 0.06 -0.03 Stepgap x Child age 0.02 0.02 1.13 0.01 0.02 0.57 Stepgap x Time spent in stepfamily home -0.05 0.01 -3.55** 0.02 0.02 1.22 Family level predictors Yrs. Living tog. 0.00 0.02 0.08 0.07 0.02 3.12** Children current union -0.17 0.17 -1.00 -0.10 0.20 -0.51 Mean age of children -0.01 0.01 -0.83 -0.01 0.02 -0.91 Cross-level interactions (Family Char, x Stepgap) Yrs. Living tog. -0.03 0.02 -1.781 0.01 0.02 0.57 Children current union 0.06 0.16 0.36 -0.19 0.19 -1.00 Mean age of children 0.00 0.02 -0.16 -0.01 0.02 -0.79 Deviance Reduction Statistics: Dev. Par. X2 Dev. par. X2 Null model 1351.76 3 1322.42 3 Child level predictors 1090.85 12 260.91 *** 1265.36 12 57.06 *** Family level predictors 1089.05 15 1.80 1254.89 15 10.46 * Family Char. X Stepgap 1085.73 18 3.31 1253.45 18 1.44 aStepgap compares wives' children (+1) to husbands' children (-1). b Indicator variable compares male children (+1) and female children (-1) to average across children (0). Note: N = 154, n = 407 <p<.10,*p<.05,**p<.01,***p<.05 These results provide additional support for Hypothesis 2, which proposed that characteristics of individual children would influence parent-child relationship quality in stepfamilies, and for Hypothesis 4, which predicted that family characteristics would 86 have an influence on parent-child relationship quality, after controlling for children's characteristics. In particular these results indicate that children's characteristics were an important predictor both of wives' reports of closeness and wives' reports of tension, whereas family characteristics were an important predictor of wives' reports of tension only. However, they do not support Hypothesis 5, which predicted that family characteristics would moderate the stepgap. Multilevel Analyses of Child Characteristics and Family Characteristics as Predictors of Time 2 Relationship Quality The next set of analyses explored the long-term effects of child characteristics and family structure on relationship quality at Time 2 after controlling for relationship quality at Time 1. Husbands' reports of Time 2 relationship quality The long-term effects of child characteristics and family characteristics on Time 2 Relationship Quality for husbands are presented in Table 13. From these results, it is evident that no family level predictors of closeness or tension for husbands had any long-term effects after controlling for relationship quality at Time 1. However, for tension, there was a long-term effect of children from the current union on the stepgap at Time 2 (B = - .37, t (124) = - 2.17, p < .05), indicating that for stepfamilies where there is a child from the current union, the stepgap is significantly moderated. The interaction is graphically portrayed in Figure 4. 87 Table 13 Child Characteristics and Family Characteristics as Predictors of Time 2 Relationship Quality Between Husbands and Children in Stepfamilies, and Effects of Family Characteristics on the Time 2 Stepgap Husbands' Time 2 Relationship Quality Predictors Closeness Tension B SE t e SE T Intercept 3.75 0.05 79.09*** 2.25 0.06 39.81 *** Child level predictors Time 1 relationship quality 0.54 0.05 11.54 *** 0.32 0.04 7.20 *** Stepgap3 -0.05 0.05 -0.98 0.03 0.06 0.48 Family level predictors Yrs. Living tog. -0.01 0.02 -0.60 -0.02 0.02 -1.05 Children from current union -0.03 0.15 -0.21 -0.27 0.18 -1.53 Mean age of children 0.01 0.01 0.81 0.00 0.01 -0.27 Cross-level interactions (Family Char, x Stepgap) Yrs. Living tog. 0.02 0.02 0.93 -0.01 0.02 -0.34 Children from current union 0.14 0.16 0.85 -0.37 0.17 -2.17* Mean age of children -0.01 0.01 -1.11 -0.01 0.01 -0.44 Deviance Reduction Statistics: Dev. par. X2 Dev. par. X2 Null model 1034.26 3 1014.33 3 Child level predictors 890.39 7 143.87 *** 955.84 7 58.50 *** Family level predictors 889.56 10 0.83 953.27 10 2.57 Family Char. X Stepgap 886.36 13 3.20 948.17 13 5.10 3Stepgap compares wives' children (+1) to husbands' children (-1). b Indicator variable compares male children (+1) and female children (-1) A/ofe:N = 128,n = 354 t p < .10, * p < .05, ** p < .01, *** p < .05 88 Figure 4. Children from current union x Stepgap interaction predicting husbands' tension with children in stepfamily at Time 2. The regression lines describing the interaction were derived by running separate multilevel analyses for husbands' own children and for husbands' stepchildren and plotting the effect of children from the current union on tension levels for each group of children. The slope for stepchildren was significantly different from zero (B = - .67, t (96) = - 2.33, p < .05), but the slope for husbands' own children was non-significant (B = .32,f(95)= 1.23,/?<.10). This suggests that having children from the current union has long-term effects on the relationship between stepfathers and their stepchildren, resulting in a reduction in the stepgap in tension. This result provides additional support for Hypothesis 4B, which 89 predicted that having children from the current union would moderate the stepgap in relationship quality for husbands. However, family characteristics modeled on the stepgap slope did not result in a significant reduction in deviance. This is because there was not significant variability in this slope between husbands, after controlling for Time 1 relationship quality.5 Wives' reports of Time 2 relationship quality The results of analyses concerning the effect of family characteristics on the stepgap for wives at Time 2 are presented in Table 14. According to the results, even after controlling for closeness at Time 1, there was a significant stepgap in wives' reports of closeness (B = .33, t (343) = 5.22, p < .001). In addition, there was an effect of family characteristics on the stepgap at Time 2. Specifically, the number of years the stepfamily had been in existence was related to a decrease in the stepgap over time (B = - .05, t (342) = - 2.68,/? < .01). The cross-level interactions between family structure variables and the stepgap also resulted in a significant reduction in deviance (x2 (3) = 7.95,/? < .05). The interaction was broken down in the manner described above and plotted. A graphic representation of the interaction is shown in Figure 5. Length of time in the current stepfamily situation was related to wives' reports of significantly lower levels of closeness with their own children (B = - .06, t (103) = - 3.01,/? < .01). On the other hand, length of time in the current stepfamily situation was significantly related to 5 Even when there is no significant variability in the slopes, i f one has a theoretical reason for modeling such variables, it is acceptable to do so. However, conclusions must be 90 wives' reports of higher levels of closeness with stepchildren (B = .06, t (111) = 2.28, p < .05). Table 14 Child Characteristics and Family Characteristics as Predictors of Time 2 Relationship Quality Between Wives and Children in Stepfamilies, and Effects of Family Characteristics on the Stepgap. Wives' Time 2 Relationship Quality Predictors Closeness Tension B SE f B SE t Intercept 3.59 0.05 66.14*** 2.36 0.07 35.20 *** Child level predictors Time 1 relationship quality 0.60 0.05 11.99*** 0.42 0.05 8.22 *** Stepgap3 0.33 0.06 5.22 *** -0.02 0.06 -0.24 Family level predictors Yrs. Living tog. 0.02 0.02 0.86 0.00 0.02 -0.04 Children from current union 0.15 0.16 0.95 -0.16 0.20 -0.81 Mean age of children 0.01 0.01 0.38 -0.04 0.02 -2.38* Cross-level interactions (Family Char, x Stepgap) Yrs. Living tog. -0.05 0.02 -2.68** 0.00 0.02 0.15 Children from current union 0.00 0.15 -0.01 0.31 0.19 1.67 Mean age of children 0.00 0.01 0.17 0.02 0.02 1.21 Deviance Reduction Statistics: Dev. par. X2 Dev. Par. X2 Null model 1180.82 3 1113.66 3 Child level predictors 892.14 7 288.67 *** 1039.61 7 74.05 *** Family level predictors 890.83 10 1.31 1034.89 10 4.72 Family Char. X Stepgap 882.89 13 7.95* 1030.89 13 4.01 aStepgap compares wives' children (+1) to husbands' children (-1). b Indicator variable compares male children (+1) and female children (-1) to average across children (0). A/ofe:N = 131,n = 346 tp < .10, * p < .05, ** p < .01, *** p < .05 drawn with caution, as the lack of resulting reduction in deviance does not support the hypothesis (Snijders, personal communication, April 13,2000). 91 5 n Stepchildren , , , , 2-1 , , , , - 4 - 3 - 2 - 1 0 1 2 3 4 Sample mean (4.57 years) Years living together. Figure 5. Years living together x Stepgap interaction predicting wives' closeness to children in stepfamily at Time 2. These results suggest that for those stepfamilies that have been together longer, wives report a slight, though significant, decrease in the stepgap for closeness. However, it appears that this decrease in the stepgap is achieved partially at the cost of decreased closeness with own children. This result partially supports Hypothesis 4A, which predicted that the number of years the stepfamily had been in existence would be related to less closeness. However, it appears that this decrease in closeness is only for wives' own children. This result also supports Hypothesis 5, which suggested that family characteristics would moderate the stepgap. 92 There was no significant effect of the stepgap on wives' reports of tension at Time 2, after controlling for Time 1 tension. However, there was a significant effect of mean age of children on tension, (JB = - .04, t (127) = - 2.38,/? < .05). This result suggests that after controlling for tension at Time 1, there is a greater reduction in tension for families whose children are older on average than for families whose children are younger on average. Further investigation of the relations between children's age and change over time. Because of the complexities involved in disentangling the effects of age from the effects of time, the effect on children's age on relationship quality was examined from another perspective. Children were assigned to age categories related to their educational level at Time 1; either as pre-school age (under 5), school-age (6 to 11), high-school age (12 to 17), or post high-school (18 to 20). The means of relationship quality for each age group were plotted separately for husbands and wives and are depicted graphically in Figures 6 and 7 below. From these figures, it is clear that both husbands and wives report less closeness and more tension with children on average as a function of their age. However, the figures also suggest a decrease in closeness from Time 1 to Time 2, as well as an increase in tension with younger children and a slight decrease in tension with older children. 93 5 4.5 • p . —I T1 c lose m T2 c lose — y\— . T1 (ansa — ^ — T2 (ansa 2.5 , 5 t 0-5 6-11 12-17 18-20 Figure 6. Husbands' Closeness and Tension with Children in Stepfamily at Time 1 and Time 2 as a Function of Age Category. For the purposes of further exploring the separate effects of children's age and time, a 2(Time) x 4 (Age Category) repeated measures A N O V A was utilized. For husbands' closeness, there was a significant effect for Time, F( l ) = 6.357, p < .05, and a significant effect for Age Category, F(3) = 6.25, p < .001. There was no significant interaction between Time and Age Category, F(3) = .908, p > .10. Follow-up contrasts indicated that for the marginal means, the younger two age groups were significantly different from the older two. This result suggests that husbands reported less closeness with children in the stepfamily overall at Time 2 than at Time 1, and more closeness with children under twelve at the beginning of the study than with children over twelve at the beghining of the study. For husbands' reports of tension, on the other hand, there was no effect of Time, F( l ) = .93, p > .10, or Age Category, F(3) = .97, p > .10. There was a trend towards significance for the interaction of Time and Age Category, F(3) = 2.24, p < .10. These 94 results suggest that for husbands, tension with children in the stepfamily is relatively stable across time and age groups. The repeated measures A N O V A for wives reports of closeness revealed a significant effect of Time, F(l) = 11.99, p < .001, and a trend towards significance for an effect of Age Category, F(3) = 2.26, p < .07. The interaction between Time and Age Category was not significant, F(3) = 89, p > .50. This suggests a general reduction in closeness over time, and a trend towards a difference in closeness due to Age Category. — T 1 c l o s e T 2 c l o » « - - T 1 t * n « « - - T 2 t * n * « 0-5 6-11 12-17 18-20 Figure 7. Wives' Closeness and Tension with Children in Stepfamily at Time 1 and Time 2 as a Function of Age Category. For wives' reports of tension, there was no effect of Time, F( l ) = 1.15, p > .10, or Age Category, F(3) = 2.23, p > .10. However, there was a significant interaction of Time and Age Category, F(3) = 5.16, p < .002. This result suggests that over time, wives reported a reduction in tension with older children and an increase in tension with younger children. 95 Daily Fluctuations in Coping and Relationship Quality This section presents the analyses of daily processes taken from diary data. First, parents' daily coping was used to predict next-day relationship quality for own children and stepchildren. Second, parents' reports of relationship quality with own children and stepchildren were used to predict next-day coping. Note that only data from pairs of days on which both stepchildren and own children interacted with the parent in some fashion were included in these analyses. Preliminary Analyses Intercorrelations between Diary Variables Within-day relationships Daily correlations presented in Table 15 indicate that reporting affection from one's own children was positively related to reports of affection from stepchildren on the same day and negatively related to reports of tension with own children on the same day. Reports of affection from stepchildren were also negatively related to reports of tension with stepchildren on the same day. Affection from stepchildren was positively related to stepparents' use of compromise and negatively related to stepparents' use of interpersonal withdrawal. Tension with own children and tension with stepchildren were related to parents use of confrontation on the same day. However, only tension with stepchildren was related to stepparents' use of interpersonal withdrawal. sD OS 8 O <o •5 tO CO CSI O CSI CO OO o d d CD OO O CM O o d d o csi o o T- O o d co CSI o d d •«r oo CO T - co o o o o d d t- co O T -d d co co o o d d * I CO CO csi csi CO to o o d d co o o d d 1 1 OO CO CO to d d o csi to co co co o T-d d •<- o d d •«- co o o t o -*f T - O d d CJ) r— o o d d cz> o c o csi CO o CO CM o o CO O J •«- o d d co CM oo o CSI T-oo oo co O i— O d d d CSI <z> CD CO o d d Csl csi oo CM O T-d d d co co oo o o o o d d co to o T— CO o d d co -r- co T - co T -l O CM CO CM O f - Oi to O O o d d •*f OO T -T - O O d d d •<— o d d o Tl- CM co T - o o o d d CD o I 2 € C L CU CO CM •o o 32 € C L CD CO _ co cr ® o * .2 § c i O C £ 5? -C := m JC = ,X O C tz «= — CO CD O O *i Q - ° ° — S § - t n CD s 3 O 2 | < O 1 " c f S i t s w ° w co 35 O ^ OJ CU o — CO o '_I I _ o o 8 8^ CO o o p V *°-t p V C L t LO p V 97 Compromise on any day was positively related to the use of confrontation that same day. Further, confrontation on any day was related to interpersonal withdrawal on the same day. However, there was no relationship between compromise and interpersonal withdrawal on the same day. Across-day (lagged) correlations. Affection from own children was positively related to next-day affection from own children. Similarly, affection from stepchildren was positively related to next-day affection from stepchildren. Tension with own children was positively related to next-day tension with own children. However, there was no relationship between tension with stepchildren on one day and tension with stepchildren on the next day. There were some significant cross-day relationships between parents' coping and relationship quality. Parent's use of compromise on one day was positively related to affection the next day, both from own children and from stepchildren. Parent's use of confrontation on one day was related to tension with own children on the next day. There were no significant relationships between interpersonal withdrawal and next-day relationship quality. For indicators of relations between relationship quality and next-day coping, only one correlation was significant. Tension with stepchildren on one day was positively related to stepparents' reports of interpersonal withdrawal the next day. Although these correlations are interesting and suggest a number of potential hypotheses about parents and children, as well as stepparents and stepchildren, they cannot be considered strong evidence because of a number of competing explanations for results. Daily responses are affected by typical response bias, as well as such contextual 98 variables as individual differences in relationship quality, and individual differences in coping behavior. Further, the data are not independent, as some parents contributed more than one day-pair to the analysis. By using multilevel modeling to examine the lagged relations between these variables, and controlling for a number of daily and contextual variables, several alternative explanations for various relationships can be ruled out, so that stronger inferences can be drawn. Intercorrelations among aggregated diary variables The intercorrelations, means, and standard deviations of aggregated diary variables describing relationship quality and coping are presented in Table 16. These mean scores represent the average of all reports by each individual across the seven days of the study. Therefore, although the daily reports are taken only from those day-pairs on which both stepchildren and children interacted with the parent completing the diary on both days, the mean scores include all days for the parent, whether interacting with both stepchildren and children or not. By including these contextual variables in the analysis, it is possible to control for parents' tendency to report using certain coping strategies (response bias), their tendency to use those strategies (typical ways of coping) and the typical quality of their relationship with their stepchildren and/or their own children. By including these variables, inferences about causality can be more confidently proposed. The aggregated diary variables indicated that individuals who typically report receiving more affection from their own children also typically reported receiving more affection from their stepchildren. Interestingly, there was no relationship between affection from own children and tension with own children, although the relationship 99 between affection from stepchildren and tension with stepchildren was significant and negative. At the aggregated level, typical affection from own children and typical affection from stepchildren were both significantly positively related to parents' average reports of compromise. Parents who reported typically high levels of tension with their own children also reported high levels of confrontation. Higher typical use of compromise was significantly positively related to higher typical use of confrontation. However, there were no significant relationships between interpersonal withdrawal and any of the aggregated diary variables. Table 16 Intercorrelations, Means, and Standard Deviations of Aggregated Diary Variables Affection Tension Coping Own Step Own Step Comp Conf Int.With Affection Own Children — Stepchildren 0.61*** -Tension Own Children 0.00 -0.03 — Stepchildren 0.18 -0.28* 0.22 — Coping Compromise 0.34* 0.26* -0.10 0.20 — Confrontation 0.14 0.09 0.28* 0.09 0.26* — Int. Withdraw 0.05 -0.21 -0.05 0.17 0.03 0.08 — Mean 2.33 2.07 1.34 1.37 1.36 1.49 1.30 SD 0.51 0.55 0.39 0.41 0.28 0.32 0.26 Note: N = 64 *p < .05, ** p < .01, *** p < .001 100 Contextual Model of Coping and Next-day Affection In Table 17, the results of a three-level contextual model of coping and next-day affection are presented. Days were the lowest level of analysis, parents were the second level, and at the third level, variance attributable to unmeasured differences between families were controlled. Mean levels of parents' use of compromise, confrontation, and interpersonal withdrawal were calculated using the entire set of diary data. These variables thus provide a marker of the family context, or milieu, in which the relationships between children, stepchildren, parents, and stepparents are played out. By controlling for these variables at the individual level, the generalizability of the day-to-day relationships is increased. Husbands' and wives' data were combined for these analyses. Therefore, gender was also used as an individual-level variable to control for gender differences in daily processes. At the daily level, each coping strategy variable was centered on the mean for that individual, so that significant results indicated a significant increase or decrease from their average levels. Affection on the previous day was not centered, as the purpose of including this control variable was to assess the degree of change as a result of parents' coping on the previous day, in addition to children's reactions on the same day. For all the multilevel analyses of diary data, the original diary sample of 81 families was reduced due to missing data. Only data that had reports of coping for two consecutive days, as well as relationship quality each day for both own children and stepchildren, could be used. This resulted in a sample for these analyses of 188 days of 101 data from 62 parents in 41 families. As the main focus of these analyses is at the daily level, there were sufficient data to draw reliable conclusions at that level. Table 17 Three-level Contextual Model of Coping and Next-Day Affection Next-Day Affection From Own Children From Stepchildren Predictor variables B SE f B SE f Family level Intercept 1.42 0.16 8.78 *** 1.55 0.14 10.79*** Daily Predictors Affection 0.39 0.07 5.86*** 0.24 0.07 3.71 *** Compromise 0.49 0.13 3.75*** 0.18 0.12 1.48 Confrontation -0.06 0.09 -0.67 0.11 0.08 1.36 Interpersonal Withdrawal 0.06 0.13 0.43 0.18 0.13 1.43 Individual Level Predictors Gender -0.14 0.08 -1.76 0.11 0.08 1.42 Mean Compromise 0.22 0.18 1.20 0.26 0.21 1.27 Mean Confrontation 0.08 0.15 0.54 -0.24 0.18 -1.35 Mean Int. Withdrawal -0.07 0.19 -0.36 -0.43 0.23 -1.85 Note: Gender is coded (+ 1) for males and (-1) for females. N (families) = 41, n (parents) = 62, n (day-pairs) = 188 t p < .10, * p < .05, **p < .01, *** p < .001 Compromise was a significant predictor of parents' reports of next-day affection from own children, (B = .49, t (183) = 3.75, p< .001) controlling for affection on the previous day, as well as the use of confrontation and interpersonal withdrawal on the previous day. Affection on the previous day was significantly related to affection the following day (B = .39, t(183) = 5.86, p < .001). 102 At the individual level, gender of the parent and average coping were controlled, by modeling them on the family-level intercept. Thus, several alternative explanations for the relationship between compromise on one day and affection from own children can be ruled out. By controlling for the previous day's affection, response bias and spillover effects can be ruled out as explanations for the results. Further, by including other daily coping variables, it is assured that compromise has a specific relationship with next-day affection from own children, and is not merely due to children's efforts to reconcile after a stressful encounter. By controlling for average use of compromise, confrontation, and interpersonal withdrawal, the possibility that higher levels of compromise lead to higher levels of affection from own children on any day can also be ruled out. For stepchildren, none of the parents' coping variables were predictors of next day affection, although previous day's affection from stepchildren was positively related to affection the following day (B = .24, t(183) = 3.71, p < .001), suggesting that there is some degree of consistency in relationship quality across days. These results, taken together, suggest that when parents use compromise to cope with a family stressor, their own children tend to reward that behavior the next day by an increase in expressions of affection and support. However, stepchildren do not appear to provide such reinforcement. This result supports Hypothesis 6, which proposed that parents' ways of coping with daily family stressors would be related to fluctuations in parent-child relationship quality. However, the results from Table 17 indicate that this is true only for parents' own children. 103 Contextual Model of Coping and Next-Day Tension Results of analyses predicting next-day tension from parents' coping on the previous day are presented in Table 18. Fluctuations in coping on the previous day were not related to next-day tension with parents' own children, although tension on one day was a significant predictor of tension on the next day (B = .30, t (183) = 4.30, p < .001), again indicating a degree of consistency in relationship quality across days. At the individual level, however, average use of confrontation was a significant positive predictor of tension with parents' own children (B = .51, t (57) = 3.16,/? < .01). Table 18 Three-level Contextual Model of Coping and Next-Day Tension Next-day Tension Own Children Step-children Predictor variables B SE t B SE T Family level intercept 0.96 0.11 8.90*** 1.31 0.12 11.33*** Daily Predictors Tension 0.30 0.07 4.30*** 0.05 0.07 0.69 Compromise -0.10 0.16 -0.65 0.00 0.18 0.00 Confrontation -0.05 0.11 -0.44 0.05 0.12 0.38 Int. Withdrawal -0.14 0.17 -0.83 -0.40 0.18 -2.20* Individual Level Predictors Gender 0.09 0.09 0.99 -0.37 0.10 -3.62** Mean Compromise -0.33 0.18 -1.83 0.29 0.20 1.47 Mean Confrontation 0.51 0.16 3.16** 0.05 0.17 0.29 Mean Int. Withdrawal -0.19 0.19 -1.03 -0.03 0.21 -0.15 Note: Gender is coded (+ 1) for males and (-1) for females. N (families) = 41, n (parents) = 62, n (day-pairs) = 188 * p < .10, * p < .05, **p < .01, *** p < .001 For stepchildren, on the other hand, parents' use of interpersonal withdrawal on one day was negatively related to next^day tension, (B = - .40, t(l 83) = - 2.20, p < .05). 104 Unlike the results for parents' own children, tension with stepchildren on one day was not related to tension on the following day. However, at the individual level, there was a significant effect for gender (B = - .37, t(57) = - 3.62), which suggests that stepfathers reported significantly lower levels of tension with stepchildren than stepmothers did. These results, taken together, suggest that stepchildren may also encourage repetition of some of their stepparents' coping behavior. The results indicate that when parents' withdraw as a response to a family stressor, they experience a reduction in tension in their relationship with their stepchildren on the following day. However, interpersonal withdrawal is not related to next-day tension with own children. Contextual Model of Relationship Quality and Next-Day Coping The next set of analyses examined the other half of the process by using parents' reports of daily relationship quality to predict next-day coping strategies. The results of these analyses are presented in Table 19. Average tension and average affection with both own children and stepchildren were used as individual-level predictors, so that both daily effects and contextual effects could be considered. Gender was also controlled. Daily tension and affection with own children and stepchildren were centered on each parent's mean. Further, the use of the same coping strategy on the previous day was controlled. Compromise. A significant predictor of next-day compromise was the use of compromise on the previous day, (B = .32, t (182) = 4.30, p < .001). Further, the receipt of affection from stepchildren was a significant negative predictor of next-day 105 compromise (B = - .17, t (182) = - 2.29, p < .05). Average relationship quality was not related to parents' daily use of compromise. Confrontation There were no significant relationship quality variables that were predictive of next-day confrontation, other than use of confrontation on the previous day (B = .21, t (182) = 2.79, p < .01). Again, average relationship quality was not related to parents' daily confrontation. Table 19 Three-level Contextual Model of Relationship Quality and Next-Day Coping Next-Day Coping Predictor variables Compromise Confrontation Interpersonal Withdrawal 8 SE f 8 SE t 8 SE T Family level Intercept 0.92 0.10 8.83 *** 1.21 0.12 9.73*** 1.28 0.04 36.25 *** Daily predictors Previous day's Coping 0.32 0.07 4.30 *** 0.21 0.08 2.79** 0.14 0.07 1.92* Own Children Tension -0.07 0.06-•1.09 -0.08 0.09 -0.95 0.02 0.05 0.34 Affection 0.07 0.08 0.84 -0.12 0.11 -1.15 0.06 0.07 0.89 Stepchildren Tension -0.04 0.05 -0.73 0.00 0.07 0.07 0.06 0.04 1.39 Affection -0.07 0.08 -2.29* -0.01 0.10 -0.09 -0.01 0.06 -0.12 Individual level predictors Gender 0.11 0.07 1.68 -0.03 0.09 -0.29 -0.04 0.06 -0.66 Mean Tension (Own) -0.12 0.08 -1.53 0.20 0.12 1.66 0.05 0.08 0.60 Mean Affection (Own) 0.05 0.09 0.57 0.05 0.14 0.38 0.11 0.09 1.14 Mean Tension (Step) 0.17 0.10 1.67 0.16 0.14 1.09 -0.06 0.10 -0.57 Mean Affection (Step) 0.11 0.08 1.30 0.02 0.13 0.18 -0.12 0.09 -1.37 Note: N (families) = 41, n (parents) = 62, n (day-pairs) = 188 t p < .10, * p < .05, **p < .01, *** p < .001 Interpersonal Withdrawal. At the daily level, interpersonal withdrawal on the previous day was a significant predictor of parents' use of interpersonal withdrawal on 106 the next day, (B =. 14, t( 182) = 1.92, p < .05. At the individual level, average relationship quality was not related to parents' daily use of interpersonal withdrawal. These results, taken together, indicate that overall, fluctuations in daily relationship quality are not important predictors of either the use of confrontation or interpersonal withdrawal on the following day. This suggests that the use of these two strategies may be influenced by more immediate instigators. However, for compromise, it appears that stepchildren's demonstrations of affection on the previous day may actually reduce the likelihood of parents using compromise as a coping strategy on the following day. Effects of Parents' Coping on Time 2 Relationship Quality In the final section, the long-term effects of coping on relationship quality were considered. These analyses used husbands' and wives' coping to predict closeness and tension in the parent-child relationships at Time 2, controlling for Time 1 relationship quality. Such an analytic strategy provides results that can be interpreted as indications of change over time. Coping scores for husbands and wives are based on their use of compromise, confrontation, and interpersonal withdrawal to cope with daily family stressors approximately 20 months previously. The number of coping reports used ranged from one to seven, depending upon the number of diaries completed. Husbands* Time 2 Relationship Quality The long-term effects of husbands' and wives' coping on husbands' reports of closeness and tension at Time 2 are presented in Table 20. 107 Table 20 Husbands' and Wives' Average Coping as Predictors of Time 2 Relationship quality between Husbands and Children in Stepfamilies Husbands' 12 Relationship Quality Closeness Tension Predictor variables B SE f B SE f Intercept 3.68 0.06 60.28 *** 2.16 0.07 29.35 *** Child level predictors Time 1 Relationship quality 0.54 0.06 9.04 *** 0.21 0.07 2.97 ** Stepgap3 0.04 0.07 0.65 -0.04 0.08 -0.46 Parents' Coping Husbands' Compromise -0.06 0.26 -0.24 0.04 0.31 0.14 Husbands' Confrontation 0.29 0.21 1.35 0.31 0.25 1.21 Husbands' Interpersonal Withdrawal -0.12 0.26 -0.48 0.66 0.31 2.17* Wives compromise -0.36 0.26 -1.41 -0.36 0.31 -1.17 Wives Confrontation 0.11 0.19 0.57 -0.15 0.23 -0.65 Wives Interpersonal Withdrawal 0.27 0.31 0.89 0.08 0.37 0.21 Cross-level interactions (Coping x Stepgap) Husbands' Compromise x Stepgap 0.52 0.29 1.83* -0.02 0.32 -0.07 Husbands' Confrontation x Stepgap 0.16 0.24 0.68 0.17 0.27 0.64 Husbands' Int. Withdrawal x Stepgap -0.28 0.28 -0.99 -0.07 0.33 -0.20 Wives Compromise x Stepgap 0.13 0.28 0.44 -0.09 0.32 -0.28 Wives' Confrontation x Stepgap -0.47 0.21 - 2.23 * -0.12 0.24 -0.52 Wives'lnt. Withdrawal x Stepgap -0.35 0.35 -0.99 0.28 0.39 0.72 Deviance Reduction Statistics: Dev. par. X2 Dev. Par. X2 Null model 577.60 3 566.54 3 Child level predictors 520.43 7 57.17*** 553.64 7 12.90* Husbands'Coping 513.85 10 6.58* 546.39 10 7.25 1 Wives' Coping 512.93 13 0.92 543.68 13 2.71 Husbands' Coping x Stepgap 511.12 16 1.81 543.46 16 0.22 Wives' Coping x Stepgap 502.62 19 8.50* 542.61 19 0.85 aStepgap compares wives' children (+1) to husbands' children (-1). Note: N = 70, n = 200. 1 p < .10, * p < .05, **p < .01, *** p < .001 108 Husbands' Time 2 Closeness. Time 1 closeness was a significant predictor of Time 2 closeness, (B = .54, t (197) = 9.04, p < .001), and child level predictors resulted in a significant reduction in deviance (X2 (4) = 57.17,/? < .001). There was also a trend towards a significant reduction in deviance for husbands' coping as a set (x2(3) = 6.58,/? < .10). However, no individual way of coping had a significant main effect. After controlling for husbands' and wives' average coping, wives' confrontation had a significant effect on the stepgap (B = - .47, t (63) = - 2.23,p < .05). The interaction between wives coping as a set and the stepgap also resulted in a significant reduction in deviance, (%2 (3) = 8.50,/? < .05). This result supports Hypothesis 8A, which proposed that wives' coping as a set would moderate the effect of the stepgap on husbands' relationship quality. Figure 8. Wives' Confrontation x Stepgap interaction predicting husbands' closeness to children in Stepfamily at Time 2. 109 The interaction was broken down and is portrayed graphically in Figure 8. To plot this interaction, the sample was divided into husbands' stepchildren and husbands' own children and the model was run separately for each group. The results indicated that wives' greater typical use of confrontation to cope with family stressors was related to husbands' reports of less closeness with stepchildren. The slope describing the relationship between husbands' closeness to stepchildren and wives' typical use of confrontation was significant, (B = - .74, t (49) = - 2.07, p < .05). However, the slope describing the relationship between husbands' closeness to own children and wives' typical use of confrontation was not significant, (B = .25, t (48) = .73,/? > .10). These results suggest that in families where mothers interact with their children in an aggressive, confrontational manner, stepfathers report a decrease in closeness with their stepchildren over time. Due to stepfathers' lacking a socially defined role, they are likely to feel uncomfortable about mtervening in mother-child conflicts. Additionally, wives may make it clear to their husbands that their participation in mother-child disagreements is not appreciated, particularly i f the husband is taking the child's side. Stepchildren are not likely to develop a close relationship with stepfathers who will not intervene in a negative conflictual interaction, and thus, a decrease in closeness would be the inevitable result. This result provided support for Hyothesis H 6 B , which proposed that confrontation would be related to poorer relationship quality, by showing that confrontation can also have a negative effect relationships between other family members. 110 Husbands' Time 2 tension. Husbands' reports of tension at Time 1 were significant predictors of the level of tension they reported at Time 2, (B = .21, t (197) = 2.97, p < .01) and child level predictors resulted in a significant reduction in deviance (%2 (4) = 12.90, p < .05). In addition, there was a significant main effect of husbands' typical use of interpersonal withdrawal predicting higher levels of tension at Time 2, controlling for Time 1 levels (B - .66, / (63) = 2.17, p < .05). There was also a trend towards a significant reduction in deviance as a result of the inclusion of husbands' coping in the model as a set (x2(3) = 7.25,/? < .10). These results suggest that husbands' coping as a set had an effect on husbands' own reports of tension at Time 2. In particular, husbands who reported higher levels of interpersonal withdrawal reported significantly higher levels of tension at Time 2 than husbands who reported lower levels of interpersonal withdrawal. Wives' coping did not have a significant effect on husbands' reports of tension at Time 2, and neither husbands' nor wives' coping as a set had an effect on the stepgap. These results do not support Hypothesis H8A, which proposed that wives coping as a set would moderate husbands' stepgap. However, they do provide some support for Hypothesis H6C, which proposed that interpersonal withdrawal would be related to poorer relationship quality. Wives' Time 2 Relationship Quality The results of multilevel analyses using parents' coping to predict wives' reports of relationship quality at Time 2, controlling for relationship quality at Time 1, are presented in Table 21. I l l Table 21 Husband's and Wives' Average Coping as Predictors of Time 2 Relationship Quality between Wives and Children in Stepfamilies Wives' T2 Relationship Quality Closeness Tension Variable 8 SE T 8 SE f Intercept 3.57 0.08 42.87 *** 2.30 0.09 24.59 *** Child level predictors Time 1 relationship quality 0.58 0.07 7.96 *** 0.46 0.08 5.71 *** Stepgap3 0.54 0.08 6.60 *** -0.05 0.08 -0.63 Parents' Coping Husbands' Compromise 0.01 0.38 0.01 -0.51 0.41 -1.25 Husbands' Confrontation 0.43 0.28 1.52 0.34 0.32 1.06 Husbands' Interpersonal Withdrawal -0.38 0.33 -1.13 0.53 0.38 1.42 Wives compromise 0.19 0.33 0.56 0.09 0.38 0.23 Wives Confrontation -0.17 0.26 -0.67 -0.11 0.29 -0.38 Wives Interpersonal Withdrawal -0.27 0.43 -0.63 0.39 0.48 0.83 Cross-level interactions (Coping x Stepgap) Husbands' Compromise x Stepgap 0.42 0.32 1.30 -0.13 0.36 -0.35 Husbands' Confrontation x Stepgap -0.29 0.23 -1.26 0.23 0.28 0.83 Husbands' Int. Withdrawal x Stepgap 0.53 0.26 2.01* 0.63 0.31 2.00* Wives Compromise x Stepgap 0.05 0.27 0.19 0.02 0.32 0.06 Wives' Confrontation x Stepgap 0.06 0.21 0.26 0.20 0.25 0.77 Wives'lnt.1 Withdrawal x Stepgap -0.68 0.34 -2.00* -0.35 0.40 -0.88 Deviance Reduction Statistics: Dev. par. X2 Dev. par. X2 Null model 733.45 3 691.24 3 Child level predictors 574.14 7 159.31 *** 656.67 7 34.57*** Husbands' Coping 570.38 10 3.77 647.90 10 8.77* Wives' Coping 568.48 13 1.89 647.39 13 0.51 Husbands' Coping x Stepgap 564.03 16 4.45 642.48 16 4.91 Wives' Coping x Stepgap 560.01 19 4.02 641.37 19 1.12 aStepgap compares wives' children (+1) to husbands' children (-1). b Sex of child is coded (+1) for males and (-1) for females. Note: N = 74, n = 209. * p < .10, * p < .05, **p < .01, *** p < .001 112 Wives' Time 2 Closeness. There was a continued effect of the stepgap on closeness at Time 2 (B = .54, t (206) = 6.60,/? < .001), indicating that wives' reported being significantly closer to their own children than to their stepchildren, even after controlling for differences in closeness at Time 1. Time 1 closeness was also a significant predictor of Time 2 closeness, (B = .58, t (206) = 7.96, p < .001). Child-level predictors, as a set, also resulted in a significant reduction in deviance, (x2 (4) = 159.31, p < .001). After controlling for husbands' and wives' coping, husbands' use of interpersonal withdrawal had a significant effect on the stepgap slope {B = .53, t (67) = 2.01, p < .05). This interaction was broken down and plotted separately as simple slopes for both stepchildren and own children in Figure 9. The figure indicates that husbands' greater typical use of interpersonal withdrawal to cope with family stressors was related to wives' reports of lower levels of closeness with stepchildren at Time 2. This slope was significant in the smaller sample of stepchildren, (B = - .97, t (50) = - 2.04,/? < .05). The effect of husbands' interpersonal withdrawal was not evident, however, for wives' reports of closeness with own children (B = .07, t (52) =. 163, p >. 10). These results provide some support for H6C, which proposed that interpersonal withdrawal would be related to poorer relationship quality, by showing that husbands' interpersonal withdrawal had a negative effect on wives' closeness to stepchildren. 113 ( O in o c CO o to > Own Children Stepchildren 4 3.5 3 2.5 2 1.5 — 1 -y = 0 . 0 7 X + 4.4tf y = -0.97x + 2.65 -1 Husbands' Interpersonal Withdrawal Figure 9. Husbands' Interpersonal Withdrawal x Stepgap interaction predicting wives' closeness to children in stepfamily at Time 2. Husbands' coping did not result in a significant reduction in deviance, as hypothesized. Therefore, Hypothesis H8B, which proposed that husbands' coping as a set, would moderate wives' Time 2 stepgap, was not supported. This was because the variance in the stepgap slopes between wives was not significant. Thus, any variables that were modeled on these slopes could not expect to explain variance that did not exist. This suggests that for most wives, differences in closeness between stepchildren and own children are fairly stable, and therefore subsumed in the Time 1 closeness predictor. 114 Although not hypothesized, there was also a significant moderating effect on wives' closeness stepgap by wives' own use of interpersonal withdrawal (B = - .68, t ( 67) = - 2.00, p < .05). The interaction was broken down and portrayed graphically in Figure 10. Although neither slope was significant in the smaller sample, the significant interaction suggests that the slopes are significantly different from each other. Wives' typical use of interpersonal withdrawal appears to have had a greater influence on reports of closeness with their own children than it did on reports of closeness with their stepchildren. Again, this result suggests that the decrease in the stepgap is at the expense of wives' own children, and not the result of any benefit to stepchildren. Figure 10. Wives' Interpersonal Withdrawal x Stepgap interaction predicting wives' closeness to children in stepfamily at Time 2. 115 Wives' Time 2 tension. Wives reports of tension at Time 1 were significant predictors of tension at Time 2, {B = .46, t (206) = 5.71,/? < .001). Further, child-level predictors resulted in a significant reduction in model deviance, (x2(3) = 34.57, p < .001). Although the inclusion of husbands' coping in the model also resulted in a significant reduction in model deviance, (x2(3) = 8.77, p < .05), no form of coping had a significant effect on its own. However, husbands' greater typical use of interpersonal withdrawal did have a moderating effect on the stepgap (B = .63, t (67) = 2.00, p < .05) for wives' reports of tension with children in the stepfamily. This interaction was broken down and portrayed graphically in Figure 11. 5 , 4.5 -4 -H u s b a n d s ' Interpersonal Wi thdrawa l Figure 11. Husbands' Interpersonal Withdrawal x Stepgap interaction predicting wives' tension with children in stepfamily at Time 2. 116 Figure 11 indicates that husbands' greater typical use of interpersonal withdrawal was significantly related to wives' reports of increased tension with own children, (B = 1.20, t (52) = 2.758, p < .01). However, for stepchildren, the slope was non-significant (B = .34, t (50) = .694,/? > .10). The significant cross-level interaction indicates that these slopes are significantly different from each other, and suggests that husbands' use of interpersonal withdrawal has an important influence on wives' relationship quality with their own children. This result provides support for Hypothesis H8B, which proposed that husbands' coping would moderate wives' Time 2 stepgap. 117 DISCUSSION This research examined the gap between parent-child relationship quality and stepparent-stepchild relationship quality. This gap was referred to as the "stepgap". After establishing the existence of this gap, potential moderators of the size of the gap were explored. In addition, a more detailed examination of daily processes was undertaken to see whether evidence of the "stepgap" was also present in day-to-day interactions. Finally, coping scores from the diary data were used to predict relationship quality two years later. In particular, the effect of one parents' coping on the other parents' reports of relationship quality were considered. Summary of Results Exploring the Stepgap As expected, both husbands and wives reported greater closeness with their own children than with their stepchildren. For husbands, after controlling for gender and time spent in the stepfamily home, as well as stepfamily characteristics, the stepgap in closeness was only evident for older, as opposed to younger children. For wives, even after controlling for children's characteristics and family characteristics, the stepgap for closeness was still evident. This is somewhat consistent with research that has shown remarried mothers to report less positive relationships with their stepchildren than with their own children, whereas remarried fathers were less likely to report such differences 118 (Hobart, 1988). As expected, wives' reported greater tension with their stepchildren than with their own children. Contrary to expectations, however, after controlling for children's characteristics and family characteristics, husbands actually reported greater tension with their own children from a previous union than with their stepchildren. Interactions between the stepgap and time in the stepfamily home suggested that husbands report greater tension with their own children who spend less time in the stepfamily home than with those who spend more time in the stepfamily home. Further, they report greater tension with their older children from a previous union than with their younger children. Perhaps husbands are more likely to experience tension with their own children because of the problems the children and their mother experienced after the divorce, such as economic hardship, reduced accessibility to the father, feelings of abandonment. This tension may offset somewhat their experiences of tension with their stepchildren. Differences between husands' and wives' stepgap When husbands and wives were compared, husbands reported less closeness with their children from a previous marriage than their wives reported experiencing with their own children. Further, husbands reported greater closeness and less tension in their relationships with their wives' children than wives did in their relationships with their husbands' children. Some of these differences may be due to the inequitable distribution of custody, with the majority of wives maintaining physical custody of their children, whereas the majority of husbands' children from a previous marriage lived with their mothers. Husbands are therefore more likely to share households with their stepchildren 119 than wives are to share households with their stepchildren. It has been shown that stepparents are closer to their stepchildren when they spend more time in the stepfamily household. Therefore, this difference between husbands and wives in closeness to stepchildren may be partially due to these custodial differences. However, husbands and wives also differed in their relationships with the same children, which suggests that differences in relationship quality are not solely due to custodial arrangements. Husbands reported being less close to their wives' children than their wives reported being to those children. Further, wives reported more tension in their relationship with their husbands' children than their husbands reported with the same children. Changes in the Stepgap over Time When aggregated reports of relationship quality with own children and stepchildren were used, husbands reported a decrease in closeness and an increase in tension with stepchildren from Time 1 to Time 2. For wives, there was a trend towards a decrease in tension with stepchildren from Time 1 to Time 2. However, when stepgap scores were used, the only decrease in the stepgap was for wives' reports of tension. However, these results did not control for any child or family characteristics. In the multilevel analyses, the process of change over time was revealed to be much more complex, and to depend upon a number of factors which will be discussed in more detail in another section. However, a general picture was given by the analyses examining the effects of Time and Age on closeness and tension. Those results indicated that both husbands and wives reported lower levels of closeness at Time 2 than at Time 1, and 120 husbands also reported less closeness with older children than with younger children at both timepoints. For husbands, there was little change in levels of tension across time. Wives, on the other hand, reported a decrease in tension with older children and an increase in tension with younger children. These results suggest that wives' tension with children is most problematic during early adolescence, but appears to improve in late adolescence. Predictors of Parent-Child and Stepparent-Stepchild Relationship Quality The Influence of Children's Characteristics As expected, children's age had a generally deleterious effect on relationship quality, with both wives and husbands reporting less closeness and more tension with older children than with younger children. Husbands were also less close to their older stepchildren than to their own older children. This result is consistent with the general consensus among stepfamily researchers that adolescents in stepfamilies are at increased risk for all manner of psychosocial, learning, and health-related problems (Bray, 1988, Bray & Hetherington, 1993; Hetherington, Arnett & Hollier, 1988; Vuchinich, Hetherington, Vuchinich, & Clingempeel, 199; Zi l l , 1988). Contrary to expectations, and to results reported in the literature, child gender was not related to relationship quality either for husbands or wives in this study. There are several possible explanations for this. First, it is possible that the effect of gender is a more subtle influence that is washed out by more influential variables, such as age and "stepness". Second, because these analyses are within-family, in order to detect gender 121 differences, a number of parents with both male and female adolescent stepchildren would have to report large differences. Third, the majority of the literature reporting these gender differences is from the early 1980's (e.g., Hetherington, 1982; Finklehor, 1984). Perhaps in the 90's, after a decade of exposure to discussions of safe sex, incest, and date rape, issues regarding sexuality are less secretive, and better understood. It may be that today's young women do not feel as threatened by a stepfather, and vice versa. Fourth, the measure of relationship quality used for this study provided only a very brief snapshot of the relationship from the parent's (or stepparent's) point of view. A more detailed and sensitive measure of relationship quality may be needed to detect differences in relationship quality due to gender. Consistent with previous literature (Ambert, 1989), the amount of time husbands' children from a previous marriage spent in the stepfamily home was an important predictor of relationship quality. When husbands' children spent more time in the stepfamily home, their stepmothers felt closer to them, and their fathers reported lower levels of tension with his children. Some may question whether such an indicator can properly be called a characteristic of the child, as it is quite likely that such arrangements are due more to the parents' attitudes than to the child's wishes. However, as the focus here is on establishing within-family differences, the amount of time spent in the stepfamily household is a variable that differs between children within the stepfamily, and must rightly be considered as a child characteristic. 122 The Influence of Family Characteristics Family characteristics, for the purpose of this study, were conceptualized as those family circumstances that were the same for all children in the stepfamily. The number of years the stepfamily had been in existence, for example, was implicated in parents' reports of relationship quality. Wives in stepfamilies that had been together longer reported more tension with individual children, after controlling for children's age. This could be due to the impact of ongoing stressful interactions over a number of years. However, years in the stepfamily was also related to wives' reports of greater closeness with stepchildren. These results suggest that closeness and tension are not mutually exclusive, and underscore the importance of considering the two indicators of relationship quality separately. In a robust finding that is consistent with findings from the stepfamily literature (e.g., White & Booth, 1985), in stepfamilies where the remarried parents had a child together, fathers' reported greater closeness and less tension with their stepchildren than with their own children from a previous union. In contrast, having a child from the current union had no effect on wives' reports of relationship quality. For this study, the average age of the children in the stepfamily was also conceptualized as a family characteristic. In stepfamilies where the average age of the children was greater, husbands reported less closeness with all children, after controlling for the age of individual children. Family characteristics continued to impact relationship quality two years later, controlling for initial levels. In particular, in stepfamilies with children from the current 123 union, fathers reported a further decrease in tension with stepchildren as compared to own children at Time 2. For wives, however, time did not suggest any great improvement in levels of closeness. Results of longitudinal analyses indicated that the stepgap for closeness was greater at Time 2, controlling for Time 1 closeness. Further, although for stepfamilies that had been together longer, wives' closeness stepgap decreased more than for stepfamilies who had been together for a shorter period, this decrease appeared to be due not only to an increase in closeness with stepchildren, but also to a decrease in closeness with wives' own children. This suggests that the conflictual nature of the stepfamily milieu may have a negative effect on mothers' relationships with their own children from a previous marriage. For wives, the effect of changes in tension over time appeared to be tied to the ages of the children in the stepfamily. Some results indicated that over time wives' reported a decrease in tension with older children and an increase in tension with younger children. This suggests that relationships with adolescent children are particularly difficult for wives, but that some of these difficulties resolve as children mature. The Effect qf Daily Coping It is acknowledged that the connections between coping and relationship quality observed in this study are but a small sampling of the richness of day to day interpersonal interactions in stepfamilies. Certainly there are interactions occurring minute-by-minute and hour-by-hour that are completely absent from the results reported here. It has been found that interpersonal conflict has an immediate effect on mood, but that emotional habituation occurs by the second day for all events except interpersonal conflicts (Bolger, 124 DeLongis, Kessler, & Schilling, 1989). By using a 24-hour lag, the contamination of relationship quality reports by negative mood due to other stressful events could be avoided. The longer lasting effects of interpersonal conflict have been shown to have a physiological component as well. For example, in a study of newly married couples, it was found that negative conflict had immunosuppressive effects that persisted for at least the next day (Kiecolt-Glaser et al, 1993). The results of the exploration of across-day relations between coping and relationship quality indicated that a clear connection could be established, and that fluctuations in daily relationship quality could be considered outcomes of coping. That is, after controlling for average use of compromise, as well as affection from own children on the same day, greater use of compromise on one day was related to an increase in affection and support from parents' own children on the next day. Similarly, use of interpersonal withdrawal on one day, controlling for average use, as well as tension on the same day, was related to a significant decrease in tension with stepchildren on the following day. Across-day associations help to strengthen causal inference, and rule out a number of competing explanations for the findings. The results also emphasized the utility of contextual models as a way of disentangling typical coping responses and average relationship quality from the direct relations between an increase in coping on a specific day and fluctuations in relationship quality on the following day. The results suggest that parents' own children tend to reward their parents for using compromise when dealing with family stressors, whereas stepchildren do not. Of course, we don't know who was favored in the compromising response. Perhaps stepchildren do not reward stepparents' use of compromise because it is less likely to 125 affect them favorably. On the other hand, such lagged responses may be entirely outside of awareness, reflecting deeply ingrained patterns of parent-child interactions. From a Sullivanian perspective, interactions early in childhood lay down memory traces that are activated by current interpersonal situations, and determine behavior in those situations (Carson, 1991). Relations between stepchildren and stepparents do not have that history, and may therefore be more deliberate. A comment by a study participant emphasized the self-consciousness a stepparent may feel when establishing a relationship with stepchildren. This stepmother, who became an instant mother when her husbands' young children moved into their home, said, "I have lost my freedom and my privacy. There are moments of considerable resentment on my part that of course I must bring to my relationship with them. I don't have the natural bond that's felt with a birth-parent and the children. So it's an ongoing effort to extend myself in that way." Stepparents are rewarded, however, for their use of interpersonal withdrawal. The use of interpersonal withdrawal by stepparents was related to a decrease in tension with stepchildren on the following day. This results suggest that parents' use of interpersonal withdrawal as a way of avoiding tension with stepchildren may quickly become a self-sustaining behavior. For tension, the patterns revealed for parents and own children are reminiscent of the coercive-aversive interactions described by Patterson (1983). Parents' reports of tension with their own children on any day were fueled by their greater typical use of confrontation. However, tension with either parents' own children or their stepchildren did not predict parents' coping on the following day. Only parents' use of compromise appeared to be triggered by relationship quality on a previous day, but in the opposite 126 direction to what one might expect. Stepchildren's demonstrations of affection and support predicted a reduction in stepparents' use of compromise the following day, as compared to their average levels. Such a result suggests that stepparents may not think to reward their stepchild's prosocial behavior, but rather interpret such behavior as an opportunity to get the upper hand. Such a response is likely to quickly extinguish any future attempts to get closer to the stepparent. Although no recursive relations were found, the results presented here are not inconsistent with researchers' general observations that relationships and conflict interactions reciprocally reframe each other (Canary, Cupach, & Messman, 1995). Confrontation did not show any lagged relations with relationship quality in either direction. Some researchers have indicated that anger is not nearly as corrosive to relationships as withdrawal (e.g., Gottman & Levenson, 1992). Further, conflict can have a positive role in the development of a relationship (Canary, Cupach & Messman, 1995). It is not always clear empirically whether conflict is a symptom of relationship difficulties, or a sign that active effort is being put forth to develop an alliance. In fact, contrary to expectations, the diary data indicated that neither interpersonal withdrawal nor confrontation have a particularly negative impact on relationship quality at a single timepoint. However, as a habitual form of response to interpersonal conflict, confrontation may become problematic because of the reactions it causes in others. The longitudinal analyses discussed below provide some support for that assumption. Although confronted individuals may concede defeat in the short term, the likelihood of future cooperation wil l likely be reduced due to the resentment they experience. Further, an immediate short-term success provides reinforcement to the confronter, thus 127 encouraging individuals to continue employing similar strategies. Over time, such interaction patterns may lead to the escalation of hostile and aggressive behaviors (Patterson, 1983). Parents' Typical Ways of Coping with Family Stressors. Parental behavior has been shown to be highly stable across time (Holden & Miller, 1999). Therefore, it is reasonable to assume that small effects of parental behavior are likely to become larger effects over time. Further, research on stress and coping has shown that coping behavior, although situation-specific, is also quite stable over time (Carver, Scheier, & Weintraub, 1989; Endler & Parker, 1990). To test the hypothesis that parents' typical ways of coping over a seven-day period would be related to relationship quality two years later, husbands' and wives' aggregated coping scores were used. However, the focus of these analyses was not merely on the relations between coping and relationship quality two years later. For all analyses, the quality of the relationship at Time 1 was controlled, so that only the influence of typical coping on changes in relationship quality from Time 1 to Time 2 would be considered. Such an analytic strategy is extremely conservative, as parents probably coped in ways that were fairly similar even prior to their descriptions of relationship quality at Time 1. However, by showing that coping is related to changes in relationship quality over time, the assumption that patterned family responses to interpersonal stressors have cumulative effects on parent-child relationships over time is supported. Because of the stringent nature of the analyses, the hypotheses formulated were fairly broad in scope. The primary purpose was to test the hypothesis that the 128 interpersonal coping behaviors of one parent would impact the quality of the relationship between the other parent and children in the stepfamily. It was further expected that there would be differences in the effect, depending upon whether it was a parent-child relationship or a stepparent-stepchild relationship. The hypotheses were partially supported by the results. For husbands, wives' coping as a set interacted with the stepgap to explain significant variance in husbands reports of changes in closeness at Time 2. Specifically, it was found that wives who reported greater typical use of confrontation to cope with family stressors had husbands who reported less closeness to their stepchildren at Time 2, controlling for closeness at Time 1, than husbands whose wives reported lower typical use of confrontation. This result suggests a dynamic that may be involved in promoting the disengaged parenting style typical of stepfathers (Hetherington, 1993). As husbands are already both culturally and biologically disposed to retreat from daily involvement with their stepchildren, in a conflictual family atmosphere, husbands may be even more inclined to avoid involvement. For wives, husbands' coping as a set interacted with the stepgap to explain significant variance in changes in tension at Time 2. In particular, the results suggested that when husbands withdraw, it can negatively impact wives' relationships with their own children. Wives whose husbands reported higher levels of withdrawal reported an increase in tension with their own children two years later. One mother described how her husband was negatively affecting her relationship with her daughter. " M y teen-aged daughter is pregnant, and I want to be there for her, but my husband gets really jealous. 129 He is suffocating me, and I need to be available to my daughter. She needs me right now." Husbands' coping as a set did not result in a significant reduction in deviance. This was because there was not significant variance in wives' stepgap slopes to be explained, due to a high degree of consistency in wives' reports of closeness over time. However, the significant interactions between parents' coping and the stegap were still interpreted for the following reasons. First, it is generally agreed that "theoretical considerations are primary in whether a Level 1 coefficient should be conceived as random" (Bryk & Raudenbush, 1992 p. 116). Secondly, because the stepgap slope is a random effect, the variance in the stepgap slopes can only be calculated for those families with sufficient data for estimation. In this case, that would be only those families with 3 or more children under the age of 21, of whom at least 1 must be a stepchild and at least 1 must be the wife's own child. These restrictions resulted in only 39 families for which slope variance could be estimated. Thirdly, the fixed effects of coping on wives' reported closeness with both stepchildren and own children are calculated using all available data (N = 74 wives), and therefore accurately describe between-family effects in the sample under study (Bryk & Raudenbush, 1992). Fourth, for the analyses reported throughout this thesis, maximum likelihood methods of estimation were used so that models with different fixed effects could be compared. This method has a downward bias, which is negatively affected by small sample size (for a discussion of variance estimation in random coefficient models, see Snijders & Bosker, 1999). Therefore, although husbands' coping as a set did not result in a significant reduction in deviance for this model, 130 husbands' interpersonal withdrawal was still interpreted as having an important effect on the stepgap. Husbands who reported greater typical use of interpersonal withdrawal to cope with family stressors had wives who reported a decrease in closeness with their stepchildren over time. As one frustrated stepmother remarked, "The biggest problem I have is my husband's total lack of involvement with his children, and his unwillingness to be a father to them. And nobody accepts me as his children's step-mother, not his kids, not his parents, not his ex-wife, no one." Husbands' withdrawal also had negative implications for wives' relationships with their own children. Husbands who withdraw consistently in response to stress are likely to be the type of individual who does not cope well with any kind of stress. Their lack of proactive coping strategies may preclude the development of a cohesive stepfamily unit, due to either their unwillingness or inability to extend themselves in an effortful way. Alternatively, husbands who withdraw consistently may be reacting to the unexpected complexities of stepfamily life. If their first efforts at coping are not successful, they may begin to withdraw increasingly as the most effective way to avoid family tension. Unfortunately, the results presented here also suggest that wives cannot hope to grow closer to their stepchildren and maintain good relations with their own children without their husbands' active participation in stepfamily life. Taken together, these results suggest that in stepfamilies where wives report using higher amounts of confrontation to cope with family stressors, husbands may withdraw and the result is a decrease in husbands' closeness with stepchildren over time. Further, in stepfamilies where husbands' report more interpersonal withdrawal to cope with 131 family stressors, their wives report a decrease in closeness with their stepchildren and greater tension with their own children two year later. Wives' use of interpersonal withdrawal was also related to their reports of lower closeness with their own children two years later, although the negative impact on wives' stepchildren was negligible. Implications of this Research Theoretical Implications One might be tempted to suggest that evidence of the existence of the stepgap merely upholds "the trite observation that 'blood is thicker than water' " (Pinker, 1997). However, differences in attitudes and behavior toward relatives and non-relatives are not well explained by the majority of theories in social psychology that focus on human interactions. One explanation for these differences is the effect of kinship loyalties. According to evolutionary theory, the added good will one feels towards one's kin affects behavior to the extent that a kind act wil l help a relative to propagate copies of one's genes. If such a biological urge is operating, it is understandable that it wil l influence us to behave in very different ways with our own children than with the children of our sexual partner. It is also likely that such differences will be most evident when we are under stress. The differences in the patterns of interactions for parent-child dyads as opposed to stepparent-stepchild dyads revealed in this research add another dimension to the understanding of the stepgap. It seems unlikely that a stepparent-stepchild relationship can hope to duplicate the intimate, ingrained patterns of rewarding interactions that 132 characterize parent-child communication. Further, the norm of reciprocity predicts that individuals tend to help those who have helped them in the past, and to retaliate against those who have injured them. In parent-child relationships there is a long-term affective bond of an accumulative nature (White, 1994). Stepparents, on the other hand, may come to view their stepchildren as a threat to their own well-being. The diary data indicates that stepparents may indirectly, and perhaps even unconsciously, reject stepchildren's attempts to gain their favor. The act of cooperation involves a state of hopeful vulnerability in which one makes an offer while simultaneously incurring a risk (Sheldon, 1999). The risk, of course, is that the other individual may not reciprocate. When people demonstrate the willingness to take the risk, it is because cooperation wil l provide the best rewards for both parties, i f it can be arrived at and maintained (Axelrod, 1984). When a stepchild offers a conciliatory gesture (e.g. affection and support), it may be in the hope of a rewarding response from a stepparent. However, the results of this research suggest that instead of stepparents offering a reciprocal, cooperative response, they are actually less likely to compromise than they typically would. In terms of the classic "prisoner's dilemma" paradigm, the most rational strategy in a short-term situation is to assume that the other wil l not reciprocate, and therefore to defect in self-defense. For stepparents who are not invested in their relationship with their stepchildren, and do not see it as a long-term situation with anticipated rewards, defection may be in their own best interests. It has been suggested that the idea of reciprocity is at the heart of all stable relationships, and is a basic norm in all social interactions (Thibault & Kelley, 1959). If stepchildren and stepparents are not confident 133 that their attempts to cooperate will be reciprocated, the development of a stable relationship becomes very problematic. Another contribution of this study is its emphasis on the role of "the other parent" in stepparent-stepchild and remarried parent-child relationships. It has been pointed out by a number of researchers that the quality of social relationships may be an important determinant of the coping strategies individuals select (DeLongis & O'Brien, 1990; Dunkel-Schetter, Blasband, Feinstein, & Herbert, 1992; Schreurs & de Ridder, 1997; Pearlin & McCall , 1990). In a stepfamily, the coping strategies employed were expected to have different effects on parents' relationships with their own children and with their stepchildren. It is not difficult to imagine a scenario in which a mother's behavior could affect relationship quality in different ways throughout the stepfamily system. Perhaps the wife in a stepfamily may react negatively to a family conflict. She may become angry and confrontational, or she may withdraw and brood. Due to a sense of loyalty, the mother's children may take her side in the conflict. They may choose to blame their stepfather, and perhaps his children as well, for their mother's upset. The father's children, on the other hand, are likely dealing with loyalty conflicts of their own. They may find in their stepmother's behavior an opportunity to blame their father for his new wife's behavior, and perhaps even for remarrying in the first place. The results of this study emphasize the importance of studying relationship quality as an outcome affected by parental coping in families. They also point to the utility of studying specific strategies in a detailed way to gain insight into both their short-term and long-term effects. 134 Methodological Implications This study makes use of recently developed statistical techniques that help to develop a more holistic view of the stepfamily system. Multilevel techniques for diary data also mitigate some forms of confounding by using participants as their own controls, and using temporal precedence to strengthen causal inference (Affleck, Zautra, Tennen, & Armeli, 1999). The research presented here extends the stepfamily literature by allowing for within-family inferences that can help develop a greater understanding of the complex relationships that often occur in blended families. Another innovation of this research is the use of a stress and coping paradigm to examine the processes behind stepfamily relationships on a day to day basis. The replicability of such patterns may help to refine theories of kinship and remarriage. This research also realizes the promise of a contextualist approach to research; searching for patterns in the data, with an eye to the generation of some new hypotheses, as well as the confirmation of some old ones (McGuire, 1983). The fact that these analyses provided results similar to those in many other stepfamily studies suggests that multilevel statistics are reliable, useful, as well as extremely flexible when faced with the challenge of studying individuals nested in groups. Limitations of the Research There are, of course, a number of limitations inherent in this research. First, one must confront the weaknesses of using a volunteer sample. It is possible that the families in our sample were particularly motivated to participate because of the difficulties they were having with stepfamily functioning. As many of our analyses were limited to those 135 families in which both partners participated, this sample may overly represent those stepfamilies where there was a greater degree of cohesiveness and cooperation between partners. This also has implications for any conclusions drawn about changes over time. The couples who participated in our study may have been more motivated to work out their problems, and thus more likely to resolve major issues over time due to consistent effort. These issues would be of more concern i f the changes over time were uniformly positive, suggesting some sort of regression to the mean. Fathers with custody of their children were probably also oversampled in this study, due to the requirement that there be at least one stepchild in the house one quarter of the time. However, the results, which are quite consistent with literature where sampling was more carefully controlled, need not be questioned too severely on these grounds. Gottman (1990) has noted that oversampling of extreme groups forces a more rectangular distribution and reduces regression to the mean. Reliance on self-report measures is another limitation. Such measures do provide a more feasible method of collecting relatively detailed data on a number of participants, on the other hand. Stepparents' perceptions of their relationships with their stepchildren and their own children are an important area for study, and one that can only be obtained through self-report. Although the perspective of stepchildren is not considered in this study, it is acknowledged that it likely to be different from the parental perspective (Preece, 1993). 136 Closing the Stepgap This research affirms that the stepgap exists, that it is not uncommon, and that it can change over time. Visher and Visher (1990) have suggested that an important characteristic of adults in successful stepfamilies is that they have realistic expectations. In other words, they acknowledge and accept that their family will be different from a first marriage family. Indeed, the most identifiably distressed stepfamilies were those who expected to recreate a nuclear family, preferably a better one than the first one. There is sometimes an unspoken assumption when using a term like "gender gap" or "race gap" or even "stepgap" that such a "gap" should not exist. It is often stated, or at least implied, that any "gap" is due to factors that should be corrected so that the two groups wil l be on a more equal footing. However, when considering the "stepgap", it is necessary to question that assumption further. Would an ideal stepfamily be one where each parent had similar relationships with their own children and with their stepchildren? Some researchers have suggested that in their fervent desire for cohesion and closeness, remarried parents often push children to the point where they actively withdraw (Visher & Visher, 1988). Children may not be comfortable with the levels of closeness their stepparents hope for (Papernow, 1993). It has been consistently found that stepfamilies are not as close as nuclear families (Kennedy, 1985; Pi l l , 1990) and that stepparent-stepchild relationships are not as emotionally close as parent-child relationships (Ganong & Coleman, 1986; Hetherington & Clingempeel, 1992, Hobart, 1989). Many clinicians and researchers assume that stepfamilies tend to become closer over time. However, previous longitudinal studies conducted on stepfamilies have found little empirical 137 support for this (Hetherington & Clingempeel, 1992; Kurdek, 1991). Nonetheless, stepparents continue to try their best to do right by their charges. One woman stated that for her the biggest problem was "Being consistent. I am always trying to treat my son and my stepson equally. You want to try and be equal, but sometimes this is difficult." It may be that well-intentioned efforts to be fair and kind are all that is required for a well-functioning stepfamily. Future research reflecting stepchildren's' point of view on what makes a stepfamily atmosphere most comfortable for them would be provide additional information on this subject. Further, a focus on stepchildren's perceptions would be informative. A second practical message provided by this research is further evidence that children's developmental stage has a significant influence on the quality of their relationships with parents and stepparents. It is important to note that age has a negative impact on all parent-child relationships, not just stepparent-stepchild relationships. Adolescents can be naturally obnoxious, self-centered, impenetrable, moody, and arrogant, and it is easy for a stepparent to take this personally as a function of the divorce and remarriage, rather than normal adolescence (Bray & Harvey, 1995, pg.125). Clinicians working with stepfamilies can encourage stepparents to differentiate between adolescent issues and stepfamily issues, dealing with each appropriately. It may also help stepparents to know that the process of stepfamily integration may take longer with older children. Stern (1978) found that with young children, it took approximately two years for stepfathers to enter the family and achieve a co-management status with their wives. However, in families with older children, Papernow (1984) found that many families needed five to six years to achieve satisfactory integration. For many families, 138 knowing that they are progressing at the normal rate may alleviate some of the anxiety and tension within the stepfamily. From comments made to the interviewers who collected the data used in this study, it appears that many stepparents try very hard to be as kind, supportive, and fair to their stepchildren. One stepmother said, "the biggest problem we have is trying to integrate my stepson into the family. He seems to hold back, I think he feels that i f he becomes part of this family he is being disloyal to his mother." A stepfather remarked, "The most serious family problem that we have right now is trying to get money together to send all the kids (his two stepsons and his own son) to post-secondary education. I feel that this is a top priority, but the rest of the family doesn't necessarily agree." Another woman from the study noted, "We have to pay my husband's wife all this money for child support, even though she is (an employed professional) with a good salary. But my husband's boys are with us most of the time, and we end up spending a lot of money on them while they are here. M y husband doesn't care, he is just happy to be with them, but she doesn't give us anything." These comments suggest that the role of a stepparent can be difficult, expensive, and unrewarding. One stepmother, with no children of her own, stated, "We never had a chance to be on our own before the children came. We disagree a lot on how to raise the children." In addition, another stepfather, who was separated from his wife at Time 2, remarked, "Stepparents are really low on the food chain when the marriage dissolves. I was (my stepson's) father for seven years. I was very emotionally close to him, and now I have no say in what happens in his life." Given the potential for loss of the stepparent-stepchild relationship i f the marriage dissolves, it would seem that having a lower level of 139 commitment to one's stepchildren than to one's own children is an understandable response. Knowledge of the influence of spouses' coping on parent-child and stepparent stepchild relationships is something that might be most useful prior to making a commitment to a relationship with a partner who has children from a previous marriage. It has been remarked that very few couples think seriously about the potential impact of their prospective partners' ways of coping with interpersonal stress on the happiness and well-being of the new family prior to their remarriage or cohabitation (Ganong & Coleman, 1994). This thesis indicates that it is of critical importance for a well-functioning stepfamily that both parents be able to avoid aggressive and hostile ways of handling conflict, remain engaged and open, and promote a cooperative environment in the new stepfamily. "He who has never hoped can never despair." George Bernard Shaw. 140 REFERENCES Affleck, G., Zautra, A . , Tennen, H. , & Armeli, S. (1999). Multilevel daily process designs for consulting and clinical psychology. A preface for the perplexed. Journal of Consulting and Clinical Psychology, 67, 746-764. Ahrons, C. R., & Rogers, R. H . (1987). Divorced families: A multidisciplinary developmental view. New York: Norton. Ahrons, C. R., & Wallisch, L . (1987). The relationship between former spouses. In D. Perlman, and S. Duck (Eds.), Intimate Relationships. Newbury Park: Sage. Aldwin, C M . & Revenson, R.A. (1987). Does coping help? A reexamination of the relation between coping and mental health. Journal of Personality and Social Psychology, 43, 337-348. Alexander, J. F. (1973). Defensive and supportive communication in normal and deviant families. Journal of consulting and Clinical Psychology, 40,223-231. Amato, P.R. & Booth, A . (1996). A prospective study of divorce and parent-child relationships. Journal of Marriage and the Family, 58, 356-365. Amato, R., & Keith, B . (1991). Consequence of parental divorce for children's well-being: A meta-analysis. Psychological Bulletin, 110,26-46, Ambert, A . (1989) Ex-Spouses and New Spouses: A Study of Relationships. Greenwich: JAI Press. Anderson, E. R., Hetherington, E. J., Reiss, D., & Howe, G. (1994). Parents' nonshared treatment of siblings and the development of social competence during adolescence. Journal of Family Psychology, 8,303-320. Axelrod, R. (1984). The evolution of cooperation. New York: Basic Books. Barry, W.A. (1970). Marriage research and conflict: A n integrative review. Psychological Bulletin, 73,41-54. Bateson, Gregory, 1972. Steps to an ecology of mind. New York: Ballantine. Bell, E. C. & Blakeney,R. N . (1977). Personality correlates of conflict resolution modes. Human Relations, 30, 849-857. 141 Belsky, J., Crnic, K. , & Woodworth, S. (1995). Personality and parenting: Exploring the mediating role of transient mood and daily hassles. Journal of Personality, 63, 905-929. Benoit, W. L. , & Benoit, P. J. (1988). Factors influencing the accuracy of verbal reports of conversational behavior. Central States Speech Journal, 39,219-232. Billings, A . G., & Moos, R. H . (1981). The role of coping responses and social resources in attenuating the stress of life events. Journal of Behavioral Medicine, 4, 139-157. Bishop, C. (1990). Coping with rheumatoid arthritis. Unpublished master's thesis. University of British Columbia, Canada. Block, J. (1996). A contrarian view of the five-factor approach to personality description. Psychological Bulletin, 117, 187'-215. Bolger, N . & Zuckerman, A . (1995). A framework for studying personality in the stress process. Journal of Personality and Social Psychology, 69, 890-902. Bolger, N . , DeLongis, A . , Kessler, R.C., & WetMngton, E. (1989). The contagion of stress across multiple roles. Journal of Marriage and the Family, 51, 175-183. Booth, A . , & Amato, P.R. (1994). Parent marital quality, parental divorce, and relations with parents. Journal of Marriage and the Family, 56, 21-34. Booth, A . , & Edwards, J. N . (1992). Starting over: Why remarriages are more unstable. Journal of Family Issues, 13, 179-194. Boswell, J. (1770). The Life of Samuel Johnson. Everyman's Library. Bowman, M . L . (1990). Coping efforts and marital satisfaction: Measuring marital coping and its correlates. Journal of Marriage and the Family, 52, 463-474. Bray, J. H . (1988). Children's development during early remarriage. InE. M . Hetherington & J. D. Arasteh (Eds.), Impact of divorce, single parenting, and stepparenting on children (pp 279-288). Hillsdale, NJ: Erlbaum. Bray, J. H . (1992). Family relationships and children's adjustment in clinical and nonclinical stepfather families. Journal of Family Psychology, 6, 60-68. Bray, J., & Berger, S. H . (1993). Development issues in Stepfamilies Research project: Family relationships and parent-child interactions. Journal of Family Psychology, 7, 76-90. 142 Bray, J. H. , Berger, S. H. , Silverblatt, A . H. , & Hollier, A . (1987). Family process and organization during early remarriage: A preliminary analysis. In J. P. Vincent (Ed.), Advances in family intervention, assessment, and theory (pp. 253-279). Greewich, CT: JAI. Bray, J. H. , & Harvey, D. M . (1995). Adolescents in stepfamilies: Developmental family interventions. Psychotherapy, 32, 122-130. Bray, J. H. , & Hetherington, E M . (1993). Families in transition: Introduction and overview. Journal of Family Psychology, 7, 3-8. Broderick, C. (1993). Understanding Family Process: Basics of Family Systems Theory. Newbury Park: Sage. Broderick, C , & Smith, J. (1979). The general systems approach to the family. In W. R. Burr, F. I Nye, & I L . Reiss (Eds.), Contemporary theories about the family, Volume II (Chapter 3). New York: The Free Press. Bryk, A . S., & Raudenbush, S. W. (1992). Hierarchical linear models: Applications and data analysis methods. Newbury Park Ca: Sage Buchanan, C. M . , Maccoby, E. E., & Dornbusch, S. M . , 1991. Caught between parents: Adolescents'experience in divorced homes. Child Development, 62, 1008-1029. Burrell, N . A . (1995).Communication patterns in stepfamilies: Redefining family roles, themes, and conflict styles. In M . A . Fitzpatrick & Anita L . Vangelisti (Eds.), Explaining family interactions (pp. 291-309). Thousands Oaks: Sage. Buss, D . M . (1992). Manipulation in close relationships: Five personality factors in interactional context. Journal of Personality, 60, 477-499. Canary, D. J., & Cupach, W. R. (1988). Relational and episodic characteristics associated with conflict tactics. Journal of Social and Personal Relationships, 5, 305-325. Canary, D. J., Cupach, W. R , & Messman, S. J. (1995). Relationship Conflict. Thousand Oaks: Sage. Canary, D. J., & Spitzberg, B . H . (1989). A model of perceived competence of conflict strategies. Human Communication Research, 15,630-649. Canary, D. J., & Spitzberg, B . H . (1990). Attribution biases and associations between conflict strategies and competence outcomes. Communication Monographs, 57, 139-151. 143 Carnevale, P. J., & Probst, T. M . (1998). Social values and social conflict in creative problem-solving and categorization. Journal of Personality and Social Psychology, 74, 1300-1309. Carnevale, P. J., & Pruitt, D. G. (1992). Negotiation and mediation. Annual Review of Psychology, 43, 531-582. Carson, R. C. (1991). The social-interaction viewpoint. In M . Hersen, A . E. Kazdin, & A . S. Bellack (Eds.), The Clinical Psychology Handbook, 2nd Edition. New York: Pergamon. Carver, C.S., & Scheier, M.F. (1990). Origins and functions of positive and negative e affect: A control-process view. Psychological Review, 97, 19-35. Carver, C.S., Scheier, M.F. , & Weintraub, J.K. (1989). Assessing coping strategies: A theoretically based approach. Journal of Personality and Social Psychology, 61, 404-412. Chapman, M . (1977). Father absence, stepfathers, and the cognitive performance of college students. Child Development, 48, 1155-1158. Cherlin, A . J . (1992). Marriage, divorce, remarriage (rev. ed.) Cambridge, M A : Harvard University Press. Clingempeel, W.G., Brand, E., & Ievoli, R. (1984). Stepparent-stepchild relationships in stepmother and stepfather families: A multi-method study. Family Relations, 33(3) 465-473. Clingempeel, W. G., Brand, E., & Segal, S. (1987). A multilevel-multivariable-developmental perspective for future research on stepfamilies. In K . Pasley & M . Diinger-Tallman (Eds.), Remarriage and stepparenting today: Current research and theory (pp. 65-93). New York: Guilford. Clingempeel, W. G., & Segal, S. (1986). Stepparent-stepchild relationships and the psychological adjustment of children in stepmother and stepfather families. Child Development, 57, 474-484. Coleman, M . , & Ganong, L . (1990). Remarriage and stepfamily research in the 80s: New interest in an old family form. Journal of Marriage and the Family, 52,925-940. Comstock, J., & Buller, D. B . (1991). Conflict strategies adolescents use with their parents: Testing the cognitive communication characteristics model. Journal of Language and Social Psychology, 10,47-59. 144 Coyne, J. C , & Fiske, V . (1992). Couples coping with chronic illness. In T . J . Akamatsu, J. C. Crowther, S. C. Hobfoll, & M . A . P. Stevens (Eds.), Family health psychology (pp. 129-149). Washington, DC: Hemisphere. Coyne J., C , & Smith, D. A . F. (1991). Couples coping with myocardial infarction: I. A contextual perspective on wives' distress. Journal of Personality and Social Psychology, 6, 404-412. Crosbie-Burnett, M . (1984). The centrality of the step relationship: A challenge to family theory and practice. Family Relations, 33, 459-464. Crosbie-Burnett, M . (1989). Application of family stress theory to remarriage: A model for assessing and helping stepfamilies. Family Relations, 38, 323-331. Dadds, M . R., & Powell, M . B . (1991). The relationship of interparental conflict and global marital adjustment to aggression, anxiety, and immaturity in aggressive and nonclinic children. Journal of Abnormal Child Psychology, 19, 553-567. Daylen, J. (1993). Gender differences in spouses coping with marital tension. Unpublished doctoral dissertation, University of British Columbia. DeLongis, A . & Kessler, R.C. (1989). Unpublished raw data. DeLongis, A . & O'Brien, T. (1990). A n interpersonal framework for stress and coping: A n application to the families of Alzheimer's patients. In M.A.P . Stephens, J.H. Crowther, S.E. Hobfoll, & D.L. Tennenbaum (Eds.). Stress and coping in later-life families (pp. 221-238). Washington, D . C : Hemisphere Publishers. DeLongis, A . , & Preece, M . (2000). Coping skills. In George Fink (Ed.), Encyclopedia of Stress. San Diego: Academic Press. Demo, D.H. , & Acock, A . C . (1996). Family structure, family process, and adolescent well-being. Journal of Research on Adolescence, 6, 457-488. Doherty, W J . & Jacobson, N.S. (1982). Marriage and the family. In B.B. Wolman (Ed.), Handbook of developmental psychology (pp. 667-679). Englewood Cliffs, NJ: Prentice-Hall. Dunkel-Schetter, C , Blasband, D., Feinstein, L . G., & Herbert, T. L . (1992). Elements of supportive interactions: When are attempts to help effective? In S. Spacapan & S. Oskamp (Eds.), Helping and being helped in the real world (pp. 83-114). Newbury Park, C A : Sage. Eisenberg, A . R. (1992). Conflicts between mother and their young children. Merrill-Palmer Quarterly, 38,21-43. 145 Emery, R. E. (1994). Renegotiating family relationships: Divorce, child custody, and mediation. N Y : Guilford. Endler, N.S. & Parker, J.D. A . (1990). Multidimensional assessment of coping: A critical evaluation. Journal of Personality and Social Psychology, 58, 844-854. Finkelhor, D. (1984). Child sexual abuse. New York: Free Press. Fitzpatrick, M . A . (1988). Between husbands and wives: Communication in marriage. Newbury Park: Sage. Fitzpatrick, M . A . , & Ritchie, D. L . (1993). Communication theory and the family. In P. Boss (Ed.), Sourcebook of family theories and methods: A contextual approach. N Y : Plenum. Folkman, S. & Lazarus, R.S. (1980). A n analysis of coping in a middle-aged community sample. Journal of Health and Social Behavior, 21, 219-239. Folkman, S., Lazarus, R.S., Dunkel-Schetter, C , DeLongis, A . , & Gruen, R.J. (1986). Dynamics of a stressful encounter: Cognitive appraisal, coping, and encounter outcomes. Journal of Personality and Social Psychology, 50, 992-1003. Friedland, N . (1990). Attributions of control as a determinant of cooperation in exchange situations. Journal of Applied Social psychology, 20,303-320. Furstenberg, F.F. (1987). The new extended family: The experience of parents and children after remarriage. In K . Pasley & M . Ihinger-Tallman (Eds.), Remarriage and stepparenting: Curent research and theory. New York: Guilford. Ganong, L . FL, & Coleman, M . (1994). Remarried family relationships. Thousand Oaks, CA:Sage. Ganong, L . FL, & Coleman, M . M . (1986). Stepchildren's perceptions of their parents. Journal of Genetic Psychology, 148 (1), 5-17. Glance, N.S., &Ffuberman, B . A . (1993). The outbreak of cooperation. Journal of Mathematical Sociology, 17,281 -302. Glick, P.C. (1989). The family life cycle and social change. Family Relations, 38, 123-129. Gottman, J. M . (1979). Detecting cyclicity in social interaction. Psychological Bulletin, 86, 338-348. 146 Gottman, J. M . (1987). The sequential analysis of family interaction. In T. Jacob (Ed.), Family interaction and psychopathology: Theories, methods, & findings (pp. 463-478). N Y : Plenum. Gottman, J. M . (1990). Complex statistics are not always clearer than simple statistics. A reply to Woody and Costanzo. Journal of Consulting and Clinical Psychology, 58, 502-505. Gottman, J. M . (1994). What predicts divorce? The relationship between marital processes and marital outcomes. Hillsdale, NJ: Lawrence Erlbaum. Gottman, J. M . , & Levenson, R. W. (1992). Marital processes predictive of later dissolution: Behavior, physiology, and health. Journal of Personality and Social Psychology, 63, 221-233. Guisinger, S., Cowan, P., & Schuldberg, D. (1989). Changing parent and spouse relations in the first years of remarriage of divorced fathers. Journal of Marriage and the Family, 51, 445-456. Hetherington, E. M . (1982). Effects of divorce on parents and children. In Michael Lamb (Ed.), Nontraditional Families: Parenting and Child Development, Hillsdale, N . J.: Lawrence Erlbaum. Hemerington, E. M . (1987). Family relations six years after divorce. In K . Pasley 8c M . Ihinger-Tallman (Eds.), Remarriage and stepparenting today: Current research and theory (pp. 185-205). New York: Guilford. Hetherington, E. M . (1991a). Families, lies, and videotapes. Journal of Research on Adolescence, 1, 323-348. Hetherington, E. M . (1991b). The role of individual differences in family relations in coping with divorce and remarriage. In P. Cowan & E. M . Hetherington (Eds.), Advances in family research: Vol.2. Family transitions (pp. 165-194). Hillsdale, NJ: Erlbaum. Hetherington, E. M . (1993). A n overview of the Virginia Longitudinal Study of Divorce and Remarriage with a focus on early adolescence. Journal of Family Psychology, 7, 39-56. Hetherington, E M . , & Anderson, E. R. (1987). The effects of divorce and remarriage on early adolescents and their families. In M . D. Levine and E R. McArney (Eds.), Early adolescent transitions (pp. 49-67). Lexington, M A : Heath. 147 Hetherington, E. M . , Arnett, J. D., & Hollier, E. A . (1988). Adjustment of parents and children to remarriage. In S. A . Wolchik and P. Karoly (Eds.), Children of divorce: Empirical perspectives on adjustment (pp. 67-110). New York: Gardner. Hetherington, E. M . , Bridges, M . , & Insabella, G . M . (1998). What matters? What does not? Five perspectives on the association between marital transitions and children' adjustment. American Psychologist, 53, 167-184. Hetherington, E. M . , & Camara, K . A . (1985). Families in transition: The processes of dissolution and reconstitution. Review of Child Development Research, 7, 398-439. Hetherington, E. M . , & Clingempeel, W. G. (1992). Coping with marital transitions: A family systems perspective. Monographs of the Society for Research in Child Development, 57(2-3, Serial No. 227). Hetherington, E. M . , & Jodl, K . M . (1994). Stepfamilies as settings for child development. In A . Booth & J. Dunn (Eds.), Stepfamilies: Who benefits? Who does not? Hillsdale, NJ: Lawrence Erlbaum. Hobart, C. W. (1987). Parent-child relations in remarried families. Journal of Family Issues, 8, 259-277. Hobart, C. (1988). Perceptions of parent-child relationships in nuclear and remarried families. Family Relations, 37, 175-182. Hobart, C. (1989). Experiences of remarried families. Journal of Divorce, 13, 121-144. Hobart, C. (1991). Conflict in remarriages. Journal of Divorce and Remarriage, 15, 69-86. Hobart, C. (1994). New relationships: Remarriage and stepparent. In L .E . Larson, J. Walter Goltz, & C. W. Hobart (Eds.), Families in Canada: Social context, continuities, and changes. Scarborough, Ont: Prentice-Hall. Holden, G. W., & Miller, P. C. (1999). Enduring and different: A meta-analysis of the similarity in parents' child rearing. Psychological Bulletin, 125, 223-254. Horney, K . (1943). Our inner conflicts: A constructive theory of neurosis. New York: House, J. S., Landis, K . R., & Umberson, D. (1988). Social relationships and health. Science, 241, 540-545. Karney, B . R., & Bradbury, T. N . (1995). The longitudinal course of marital quality and stability: A review of theory, method, and research. Psychological Bulletin, 188, 3-34. 148 Kennedy, G. (1985). Family relationships as perceived by college students from single-parent, blended, and intact families. Family Perspective, 19, 117 -126. Kiecolt-Glaser, J. K. , Malarkey, W. B., Chee, M . A. , Newton, T., Cacioppo, J. T., Mao, H . Y . , & Glaser, R. (1993). Negative behavior during marital conflict is associated with immunological down-regulation. Psychomatic Medicine, 55, 395-409. Komorita, S. S. (1984). Coalition bargaining. Advances in Experimental Social Psychology, 18, 184-247. Komorita, S. S., Parks, C. D., & Hulbert, L . G. (1992). Reciprocity and induction of cooperation in social dilemmas. Journal of Social Psychology, 61, 607-617. Kreft, I., & De Leeuw, J. (1998). Introducing multilevel modeling. Thousand Oaks: Sage. Kurdek, L . (1990). Divorce history and self-reported psychological distress in husbands and wives. Journal of Marriage and the Family, 52, 701-708. Kurdek, L . (1991). Predictors of increases in marital distress in newlywed couples: A 3-year prospective longitudinal study. Developmental Psychology, 27, 627-636. Kurdek, L . , & Fine, M . (1993). The relation between family structure and young adolescents' appraisals of family climate and parenting behavior. Journal of Family Issues, 14, 279-290. Larson, J. FL, & Allgood, S. M . (1987). A comparison of mtimacy in first-married and remarried couples. Journal of Family Issues, 8, 319-331. Larson, R. W., & Almeida, D. M . (1999). Emotional transmission in the daily lives of families: A new paradigm for studying family process. Journal of Marriage and the Family, 6 1 , 5 - 2 0 . Laursen, B . (1993). The perceived impact of conflict on adolescent relationships. Merrill-Palmer Quarterly, 39, 535-550. Lazarus, R.S. & Folkman, S. (1984). Stress, appraisal, and coping. New York: Springer. Maccoby, E. E. (1984). Socialization and developmental change. Child Development, 55,317-328. Mattlin, J.A., Wethington, E., & Kessler, R.C. (1990). Situational deterniinants of coping and coping effectiveness. Journal of Health and Social Behavior, 31, 103-122. 149 McCarthy, J. (1978). A comparison of the probability of the dissolution of first and second marriages. Demography, 15, 345-359. McCrae, R.R. (1984). Situational determinants of coping responses: Loss, threat, and challenge. Journal of Personality and Social Psychology, 46, 919-928. McGuire, W. J. (1983). A contextualist theory of knowledge: Its implications for innovation and reform in psychological research. Advances in Experimental Social Psychology, 16, 1-47. McLanahan, S., & Sandefur, G. (1994). Growing up with a single parent: What hurts, what helps? Cambridge, M A : Harvard University Press. Menaghan, E. G. (1982). Coping with parental problems. Journal of Family Issues, 4, 483-506. Messinger, L . (1976). Remarriage between divorced people with children from previous marriages. A proposal for preparation for remarriage. Journal of Marriage and Family Counseling, 2, 193-200. Mills , D. (1984). A model for stepfamily development. Family Relations, 33, 365-372. Minuchin, P. (1985). Families and individual development: Provocations from the field of family therapy. Child Development, 56, 289-302. Molander, P. (1992). The prevalence of free riding. Journal of Conflict Resolution, 36, 756-771. Montemayor, R. (1986). Family variation in parent-adolescent storm and stress. Journal of Adolescent Research, 1,15-31. O'Brien, T. & DeLongis, A . (1996). The interactional context of problem-,emotion-, and relationship-focused coping: The role of the big five personality factors. Journal of Personality, Special Issue. O'Brien, T. B. , DeLongis, A . , & Campbell, J. (1996). [Stress and coping processes in stepfamilies]. O'Brien, T.B. & DeLongis, A . (1997). Coping with chronic stress: A n interpersonal perspective. In E. Gotlieb (Ed.). Coping with chronic stress. NY:Plenum Press. Orbell, J. M . , van de Kragt, A . J. C , Dawes, R. M . (1988). Explaining discussion-induced cooperation. Journal of Personality and Social Psychology, 54, 811-819. 150 Oshman, A. , & Manosevitz, M . (1976). Father absence. Effects of stepfathers upon psychosocial development in males. Developmental Psychology, 12, 479-480. Paikoff R. L. , & Brooks-Gunn, J. (1991). Do parent-child relationships change during puberty? Psychological Bulletin, 110, 47-66. Papernow, P. (1984). The stepfamily cycle: A n experiential model of stepfamily development. Family Relations, 33, 355-363. Papernow, P. L . (1988). Stepparent role development: From outsider to intimate. In W. R. Beer (Ed.), Relative strangers: Studies of stepfamily processes (pp. 54-82). Totowa, NJ: Rowman & Littlefield. Pasley, K . , & Ihinger-Tallman, M . (1982). Remarried family life: Supports and constraints. In G. Rowe (Ed.), Building family strengths, 4 (pp367-382). Lincoln: University of Nebraska Press. Patterson, G. R. (1979). A performance theory for coercive family interaction. In R B. Cairns (Ed.), The analysis of social interactions (pp. 119-162). Hillsdale, NJ: Lawrence Erlbaum. Patterson, G. R. (1982). Coercive family process. Eugene, OR: Castalia. Patterson, G. R. (1983). Stress: A change agent for family process. In N . Garmezy & M Rutter (Eds.), Stress, coping, and development in children. N Y : McGraw Hi l l . Patterson, G. R., & Dishion, T. J. (1988). Mujltilevel family process models: Traits, interactions, and relationships. In R. A . Hinde & J. Steenson-Hinde (Eds.) Relationships within families: Mutual influences (pp. 283-310). Pearlin, L . I., & Schooler, C. (1978). The structure of coping. Journal of Health and Social Behavior, 19, 2-21. Pearlin, L . I., & McCall , M . E. (1990). Occupation stress and marital support: A description of microprocesses. In J. Eckenrode & S. Gore (Eds.), Stress between work and family (pp. 39-60). New York: Plenum Press. Pi l l , C. J. (1990). Stepfamilies: Redefining the family. Family Relations, 39, 186-193. Pinker, S. (1997). How the mind works. N Y : Norton. Preece, M . (1993). Differences in perspective: Parents' and children's ratings of parent-child interactions and associated childhood adjustment problems. Unpublished senior thesis, University of Los Angeles, California. 151 Preece, M . (1994). Coping with memories of traumatic events. Manuscript in preparation. Preece, M . (1996). Coping with interpersonal stressors: Issues of love and status. Unpublished master's thesis, University of British Columbia, Vancouver, British Columbia, Canada. Preece, M . , O'Brien, T., DeLongis, A . & Campbell, J. (August, 1996b). Personality and coping with marital and parenting stressors. Poster session presented at the American Psychological Association Convention, Toronto, Canada. Preece, M . , O'Brien, T., DeLongis, A . , & Campbell, J. (2000). Agentic and Communal Appraisals: Application to coping in stepfamilies. Manuscript under review. Putnam, L . L. , & Wilson, C. E. (1982). Communicative strategies in organizational conflicts: Reliability and validity of a measurement scale. In M . Burgoon (Ed.), Communication yearbook 6 (pp. 629-652). Beverly Hills, Ca: Sage. Rahim, J. A . (1983). A measure of styles of handling interpersonal conflict. Academy of Management Journal, 26,368-376. Repetti, R. L . (1989). Effects of daily workload on subsequent behavior during marital interaction: The roles of social withdrawal and spouse support. Journal of Personality and Social Psychology, 12, 125-131. Repetti, R .L. (1992). Social withdrawal as a short-term coping response to a daily stressor. In H . S. Friedman (Ed.), Hostility, coping, and health (pp. 151-165). Washington, D . C : American Psychological Association. Repetti, R. L. , & Wood, J. (1997). Effects of daily stress at work on mothers' interactions with preschoolers. Journal of Family Psychology, 11,90-118. Roberts, L . J & Krokoff, 1990. A time-series analysis of withdrawal, hostility, and displeasure in satisfied and dissatisfied marriages. Journal of Marriage and the Family, 52, 95-105. Rosenthal, D. A . , Demetriou, A. , & Efklides, A . (1989). A cross-national study of the influence of culture on conflict between parents and adolescents. International Journal of Behavioral Development, 12,207-220. Ross, R. G., & DeWine, S. (1988). Assessing the Ross-DeWine conflict management message style (CMMS). Management Communication Quarterly, 1,414-429. Rusbult, C. E., Verett, J., Whitney, G. A. , Slovik, L . F., & Lipkus, I. (1991). Accomodation processes in close relationships: Theory and preliminary empirical evidence. Journal of Personality and Social Psychology, 60, 53-78. 152 Rutter, M . (1985). Family and school influences on behavioral development. Journal of Child Psychology and Psychiatry, 26, 349-368, 1985. Santrock, J. W. (1972). The relations of type and onset of father absence to cognitive development. Child Development, 43,255-469. Santrock, J. W., & Sitterle, K . (1987). Parent-child relationships in stepmother families. In K . Pasley & M . Ihinger-Tallman (Eds.), Remarriage and stepparenting today: Current research and theory (pp. 273-299). New York: Guilford. Santrock, J. W., Warshak, R. A . , & Eliot, G. W. (1982). Social development and parent-child interaction in father custody and stepmother families. In M . E. Lamb (Ed.), Nontraditional families: Parenting and child development. Hillsdale, NJ: Erlbaum. Saucier, J. & Ambert, A . (1986). Adolescents' perception of self and of immediate environment by parental marital status: A controlled study. Canadian Journal of Psychiatry, 31, 505-512. Schelling, T. C. (1960). The Strategy of Conflict. Cambridge, M . A . : Harvard University Press. Schreurs, K . M . G., & de Ridder, D. T. D. (1997). Integration of coping and social support perspectives: Implications for the study of adaptation to chronic diseases. Clinical Psychology Review, 17, 89-112. Sheldon, K . M . (1999). Learning the lessons of tit-for-tat: Even competitors can get the message. Journal of Personality and Social Psychology, 77, 1245-1253. Sillars, A . L . (1980a). Attributions and communication in roommate conflicts. Communication Monographs, 47,180-200. Sillars, A . L . (1980b). The sequential and distributional structure of conflict interactions as a function of attributions concerning the locus of responsibility and stability of conflicts. In D. Nimmo (Ed.), Communication yearbook 4 (pp. 217-235). New Brunswick, NJ: Transaction Publishing. Sillars, A . L . (1985). Interpersonal perception in relationships. In W. Ickes (Ed.), Compatible and incompatible relationships (pp. 277-305). New York: Springer-Verlag. Sillars, A . L. , Colletti, S. F., Parry, D., & Rogerts, M . A . (1982). Coding verbal conflict tactics: Nonverbal and perceptual correlates of the "Avoidance-Distributive-Integrative" distinction. Human Communication Research, 10, 317-350. 153 Sillars, A . L. , & Scott, M . D. (1983). Interpersonal perception between intimates: A n integrative review. Human Communication Research, 10, 153-176. Sillars A . L . , & Weisberg, J. (1987). Conflict as a social skill. In M . E. Roloff & G. R. Miller (Eds.), Interpersonal processes: New directions in communication research, 10, 317-350. Sillars, A . L. , & Wilmot, W. W. (1989). Communication strategies in conflict and mediation. In J. A . Daly & J. M . Wiemann (Eds).) Strategic interpersonal communication (pp. 163-190). Hillsdale, NJ: Lawrence Erlbaum. Skinner, B . F. (1983). The behavior of organisms. New York: Appleton-Century-Crofts. Snijders, T.A.B. (1995). Multilevel models for family data. In J.J. Hox, B.F. van der Meulen, J . M . A . M . Janssens, J.J.F. ter Laak, and L.W.C. Tavecchio (Eds.), Advances in Family Research. Amsterdam: Thesis Publishers. Snijders, T. A . B. , & Bosker (1999). Multilevel analysis: An introduction to basic and advanced multilevel modeling. London: Sage Spanier, G. B . (1976). Measuring dyadic adjustment: New scales for assessing the quality of marriage similar dyads. Journal of Marriage and the Family, 38, 15-26. Sprey, J. (1971). On the management of conflict in families. Journal of Marriage and the Family, 33, 722-731. Stafford, L . , & Bayer, C. C. (1993). Interaction between parents and children. Newbury Park,CA: Sage. Statistics Canada (1995). Canadian families: diversity and change, 1995. http:///www.statcan.ca/Daily/English/960619/d960619.htm. Statistics Canada (1997). 1996 Census: Marital status, common-law unions and families. http://www.statcan.ca/Daily/English/971014/d971014.htm Statistics Canada (1998). National Longitudinal Survey of Children and Youth: Changes in the family environment. http://vmw.stotcan.ca/Daily/English/980602/d980602.htm Statistics Canada (1998). Marriages and divorces, 1996. http://ww.statcan.c^aily/English/980129/d80129.htm Steinberg, L . ( 1981). Transformations in family relations at puberty. Developmental Psychology, 17, 833-840. 154 Stern, P. N . (1978). Affiliating in stepfather families: Teachable strategies leading to stepfather-child friendship. Western Journal of Nursing Research, 4, 76-89. Straus, M . A . (1979). Measuring intrafamily conflict and violence. The conflict tactics (CT) scales. Journal of Marriage and the Family, 41, 75-88. Stryker, S. (1972). Coalition behavior. In Experimental Social Psychology, Taylor, S. E., Repetti, R. L. , Seeman, T. (1997). Health psychology: What is an unhealthy environment and how does it get under the skin? Annual Review of Psychology, 48, 411-447. Teachman, J. D. (1986). First and second marital dissolution: A decomposition exercise for whites and blacks. Sociological Quarterly, 27, 571-590. Thibault, J. W., & Kelly, H . H . (1959). The social psychology of groups. New York: Wiley. Thoits, P. A . (1986). Social support as coping assistance. Journal of Consulting and Clinical Psychology, 54, 416-423. Ting-Toomey, S. (1983a). A n analysis of verbal communication patterns in high and low marital adjustment groups. Human Communication Research, 9, 306-319. Touval, S. (1975). Biased intermediaries: Theoretical and historical considerations. Jerusalem Journal of International Relations, 1, 51-69. Visher, E. B. , & Visher, J. S. (1988). Old loyalties, new ties: Therapeutic strategies with stepfamilies. New York: Brunner/Mazel. Visher, E. B. , & Visher, J. S. (1990). Dynamics of successful stepfamilies. Journal of Divorce and Remarriage, 14(1), 3-12. Vuchinich, S., Hetherington, E. M . , Vuchinich, R. A . , & Clingempeel, W. G. (1991). Parent-child interaction and gender differences in early adolescents' adaptation to stepfamilies. Developmental Psychology, 27, 618-626. Wald, E. (1981). The remarried family: Challenge and promise. New York: Family Service Association of America. Weick, K . E. (1971). Group processes, family processes, and problem-solving. In J. Aldous et al. (Eds.), Family problem solving: A symposium of theoretical, methodological and substantive concerns (pp. 3 - 39). Hinsdale, IL: Dryden Press. V 155 White, L . (1994). Stepfamilies over the life course: Social support. In A . Booth & J . Dunn (Eds.), Stepfamilies: Who benefits? Who does not? Hillsdale, NJ: Lawrence Erlbaum. White, L . K. , & Booth, A . (1985). The quality and stability of remarriages: The role of stepchildren. American Sociological Review, 50, 689-698. Whitaker, S., & Bry, B . H . (1991). Overt and covert parental conflict and adolescent problems: Observed marital interaction in clinic and non-clinic families. Adolescence, 26, 865-876. Zi l l , .N. (1988). Behavior, achievement, and health problems among children in stepfamilies: Findings from a national survey of child health. In E. M . Hetherington & J. D. Arasteh (Eds.), Impact of divorce, single parenting, and stepparenting on children (pp. 325-368). Hillsdale, NJ: Lawrence Erlbaum. Z i l l , N . , Morrison, D. R., & Coiro, M . J. (1993). Long-term effects of parental divorce on parent-child relationships, adjustment, and achievement in young adulthood. Journal of Family Psychology, 7, 91 -103 156 APPENDIX A Coping Items for Diary Data Compromise (a = .73) Tried to find a solution that was fair to all involved. Tried to meet the other person half-way. Tried to compromise with other(s) involved. Confrontation (a = .70) Stood my ground and fought for what I wanted. Expressed anger to the person(s) who caused the problem. Tried to get the person responsible to change his or her mind. Interpersonal Withdrawal (a = .75) Withdrew from the other person(s) involved. Gave the other persbn(s) involved the "silent treatment". I sulked. Tried to keep my feelings to myself. Tried to keep others from knowing about the problem or about my feelings. 157 APPENDIX B Statistical Principles behind Multilevel Modeling The following is a description of the statistical principles behind multilevel modeling, the type of analyses used predominantly in this study. This description is taken in large part from an article by Tom A . B . Snijders (1995) on the application of multilevel modeling to family data. Snijders has published extensively on statistical issues in multilevel models (e.g., Snijders & Bosker, 1990; Snijders & Bosker, 1994; Snijders, Spreen, & Zwaagstra, 1994; Snijders & Bosker, 1999). The following equation represents the first level of a simple two level model for the effect of x on Y , with coefficients that differ between families: Y ^ P o j + p.jX^ + Rij Where Yy is the value of the dependent variable P 0 j is the family-specific intercept P y is the family-specific regression slope Xy is the value of the explanatory variable, Ry is the unexplained part ("residual") of the dependent variable Xy. Because the regression slope p,j is family specific, the effect of x u can vary among families depending on characteristics which may be measured or unmeasured. Because this is a random effects model, the analysis does not focus on individual values of the coefficients, but instead focuses on the population of children in remarried families at one 158 level, and the population of remarried parents at another. The random coefficients are split into a fixed part, which is the sample mean, and a random part, which has a mean of zero, and can be written as follows: Poj = Yoo + U 0 i , Pij = Yio + U , j , where Yoo is the population mean of the intercepts, y 1 0 is the population mean of the regression coefficients, UQJ is the group specific part of the intercept, Uy is the group-specific part of the regression coefficient. When these equations are combined, we have the formula for the complete model: Y a = Y«X) + YioXfl + UQ, + U„x s + Rfl In this equation, and y 1 0 are the fixed effects, U 0 j and Uy are the random effects. Model Assumptions It is assumed that the residual at level 1, R^, is statistically independent of the random effects at level 2, U o j and \JV}. It is also assumed that UQJ and U,j have a bivariate normal distribution. The distribution is then characterized by the variances varfUqj) and var(U,j) and the covariance (U^Uy). If vartUoj) = 0 and varfUy) = 0, then the coefficients are the same for all families and the results would be the same as for a one-level multiple regression model. This, however, is rarely the case. More often it is found that only varfJJy) = 0, which means that the regression coefficients do not vary, and the 159 variable x has only a fixed effect. In such a case, where the effects are fixed but the intercept is still variable, the resulting model is called a random intercept model. In some cases, however, both the intercept and the regression coefficient vary, and then explanatory variables can be modeled on both sources of variation. The above model describes three sources of unexplained variance. The random intercept (U 0 j) represents unexplained variance between families in the average value of the dependent variable (Yy). The random slope (U,j) represents unexplained variation in the effect of x on Y between families. The random residual (R^) represents unexplained variation between individuals in the same family. In ordinary least squares regression analysis there is only one residual variance, and the goal is to use explanatory variables to reduce the amount of variance. The goal is similar in multilevel models, but now we have three sources of variance that we wish to reduce by using explanatory variables. Variables at the first level will reduce individual var(Rjj), and i f these explanatory variables also explain differences between family means, var(U0j) will be reduced as well. By including explanatory variables at level two (parental characteristics), var(U0j) can be further reduced, and by including cross-level interactions between higher level and lower level variables (cross-level interactions), var(U,j) can be reduced. The following formulae show how group-level variables are used to explain further variance in U 0 j and U,j. A group-level variable (w )^ describing a family characteristic, can be used to explain variability in both the family-specific intercepts (poj) and the family specific regression coefficients, or slopes (B,j). P o j = Y o o + YoiWj + U o j 160 Pij = Yu> + Y n W J + U , j Then, by combining these expanded level 2 formulae with the formula describing the relations between level-1 variables, we have a more detailed model containing additional fixed effects, but the same three random components. Please note that y^, y 0„ and yu are fixed coefficients. = Yoo + YoiWj + YioXy + Y n W ^ + U 0 j + Various methods have been suggested for determining whether the introduction of a particular explanatory variable, or group of explanatory variables should be retained in the model. In ordinary least squares regression, R squared is most often used to determine whether an explanatory variable or group of explanatory variables reduce the variance in the model to a significant degree. However, in a multilevel model, this becomes a more complex question, because variables introduced at one level, while reducing variability at that level, can simultaneously increase the variability at another level. If a multiparameter test is required, a test of variances and covariances based on the likelihood-ratio test is the most general. Such a test involves the alternative hypotheses Ho:T = T 0 and H 1 : T = T, where T is an arbitrary variance-covariance matrix, and T 0 is a reduced form of T, T is measured by the deviance in the variance-covariance matrix, and is calculated as -2 times the value of the log-likelihood function evaluated at the maximum. Thus, the deviance may be viewed as a measure of model fit: the higher the deviance, the poorer 161 the fit (Bryk & Raudenbush, 1992). To test a composite hypothesis, the deviance is estimated for two models, and these two deviances, D 0 , and D( are compared. The test statistic is computed as follows H = D 0 - D , This statistic has a x 2 distribution with m degrees of freedom, where m is the difference in the number of unique variance and covariance components estimated in the two models. Thus, i f chi-square statistic representing the difference in deviance between the two models is significant, one model can be assumed to be a significant improvement over the other, given the number of degrees of freedom lost. Another rule of thumb that may be used to reach the conclusion that one model is a significant improvement over another is that the difference in deviances between two models should be at least twice as large as the difference in the number of estimated parameters (Kreft & DeLeeuw, 1998). 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0099539/manifest

Comment

Related Items