This is to certify that the dissertation entitled MATERNAL EXPRESSED EMOTION, CHILD BEHAVIOR PROBLEMS AND THE CHILD’S SENSE OF COHERENCE: TOWARDS A RESILIENCE MODEL presented by SHIZUKA SHIMABUKURO has been accepted towards fulfillment of the requirements for the PhD. degree in Family and Child Ecologx W Major Professoys Signature W/i, 910/0 Date MSU is an Affinnative Action/Equal Opportunity Employer LIBRARY Michigan State University .0-I-l-0-.-.-.-C-.-I-C-O-I-I-O-¢-0-0-0-0-.-I-I-I-I-I-O-l-0-0-0-l-O--~— PLACE lN RETURN BOX to remove this checkout from your record. To AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DATE DUE 5/08 K'lProlechresIClRC/Dateoue indd MATERNAL EXPRESSED EMOTION, CHILD BEHAVIOR PROBLEMS, AND THE CHILD’S SENSE OF COHERENCE: TOWARDS A RESILIENCE MODEL By Shizuka Shimabukuro A DISSERTATION Submitted to Michigan State University In partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Family and Child Ecology 2010 ABSTRACT MATERNAL EXPRESSED EMOTION, CHILD BEHAVIOR PROBLEMS, AND THE CHILD’S SENSE OF COHERENCE: TOWARDS A RESILIENCE MODEL By Shizuka Shimabukuro A total of 285 child-mother pairs were recruited from 5 schools in Naha City, Okinawa, Japan (ages 10-13). The goal of the study was to understand better how family environment factors and child resiliency factors jointly contribute to child behavior problems. Family environment factors included maternal depression (Center for Epidemiological Studies-Depression, CESD), positive (Involvement) and negative (Criticism) expressed emotion (EE) between mother and child (Expressed Emotion Adjective Checklist), and family relationship quality (Family Relationship Inventory). Child resiliency factors included Sense of Coherence (SOC) and Self-Esteem (SE). The Internalizing and Extemalizing scales of the Youth Self-report (YSR) were used as outcome measures. Child reports of each of the negative family environment factors were positively related to Internalizing and Externalizing problems, while each of the child resiliency factors was negatively related. This differs from reports from Western studies that maternal Criticism is specifically related to Externalizing behaviors. Higher levels of maternal Criticism and lower levels of Involvement, as reported by the child, were more closely related to girls’ Internalizing and Externalizing problems than was true for boys, imPlinlg that girls were more sensitive or vulnerable to maternal emotionality than boys. Mother reports of her Criticism and Involvement were less strongly related to child behavior problems, with significant correlations only with Internalizing behaviors. Her reports of Criticism and Involvement were not significantly related to child resiliency factors. A significant indirect path was identified from child reports of negative family environment factors to child behavior problems through child’s SOC in an SEM. However, the corresponding indirect path from family factors to child behavior problems through SE was “inconsistent,” enhancing rather than decreasing reported behavior problems. The hypothesis that SOC mediates the impact of stressful family influences was confirmed. The inconsistent effect of SE may reflect strong J apanese-Western differences regarding the construct of self-esteem. A more collectivist society, like Japan, would rely on “jibun” or the importance of self defined in the context of family and within cultural norms. Copyright by SHIZUKA SHIMABUKURO 2010 ACKNOWLEDGEMENTS The process of writing a dissertation was a repetition of challenge and excitement, and recognition of weaknesses and strengths of my own. Thinking back to the past years, I was always helped by people and I have so many appreciations to them that I cannot describe by words in this limited space. Without their help, I could not be able to complete my dissertation. I first would like to express my appreciation to my advisor, Dr. Richard Wampler, who committed to mentor me not only to complete the dissertation project, but also gave me psychological support for the self growth and professional development. He always did his best to provide me his time to deeply discuss the ideas, put his input and feedback, and guide me what I should do next. While I was in Okinawa for data collection, he was generous enough to talk with me over SKPYE on a weekly basis. He was always encouraging, positive, and hopeful. I-Iis strong encouragement, clear advice, and wise suggestions strengthened my belief that I could do it,‘ which I helped me to overcome my anxiety. Afier I returned from Okinawa, he created a working space next to his office for me to focus on the dissertation, which helped me to manage my time, concentrate on analyzing data, and keep my motivation every day. I would also like to emphasize my gratitude that he made lots of effort and energy to edit my dissertation. He patiently and carefully edited with me sitting at his side. I have never been able to accomplish this dissertation project without his intellectual contribution and psychological support; I would hold his spirit of mentorship in me and return what I received to other people in a future. Secondly, I would like to express my appreciation to Dr. Denise Satin Amault who committed to support me from various ways, psychologically, intellectually, and financially for a long period of time. She contributed to the theoretical frameworks to connect theory with practice, which made this research possible. While I worked as a research assistant for 5 years for her research project, she taught me different skills in qualitative research to be a good researcher from a scratch to the completion of the 5- year-long project. I learned survey study, interview, psychiatric interview, theoretical framework to connect theory with practice, data analysis, and an importance of networking with people in a community through the experiences. I learned how complicating and difficult the cross-cultural research was. All of those experiences that could never be learned in a short period of time helped me implement my research in Okinawa. The hours and hours of intellectual conversations with her certainly gave me a lot of excitement in research and became a motivation to learn more in a cross-culture research. Thirdly, I would like to say thank you to Dr. Grifl‘ore and Dr. Desiree Qin who gave me helpful feedback, great intellectual suggestions, and encouragement. The statistical analysis portion and cultural perspective on child development were so essential that each of their input helped me to think about them carefully, which help me to make my dissertation more accurate and meaningful. I am also thankful to my department, Family and Child Ecology and Graduate School to provide me a financial support to complete my dissertation. Fourthly, I would like to say thank you to my parents and siblings. They are always my emotional support. Since you started supporting me to be successful from vi when I was a child, it has been a long time. They always made me feel strong and ease my anxious feelings and comforted the feeling of tiredness. I also say thank you to my sister. She is a person who I cried out when I had mistakes or difficult times throughout the process of writing of my dissertation and while studying in the US. Without all of their support, I would not be able to accomplish this project. Finally, I would like to say thank you to my friends, Chiharu Kato, Grace Chen, Satoko Motohara, Chikako & Hiroshi Tokashiki, and Tomokazu Nagai. It would be much harder than it was if I did not have such good friends like you. Each of you was the outlet for my stress when I was really tired, became low in my motivation and energy, confronted difficulties, and got bored. Thank you so much for your senses of humor, warm and kind words, and cheers that always put a light on my heart. vii TABLE OF CONTENTS LIST OF TABLES ..................................................................... xiii LIST OF FIGURES ..................................................................... xv CHAPTER 1 INTRODUCTION The Experience of Depression ..................................................................... 1 Overview of Internalizing and Extemalizing Behavior Problems ........................................................... 3 Prevalence of internalizing behavioral symptoms (depression) ...... 3 Early onset of internalizing behavior problems ............................... 4 Extemalizing behaviors ................................................................... 4 Internalizing and Extemalizing Disorders in Children in Japan .................................................... 5 Child Development and Family Factors in Behavioral Disorders ............... 6 Family environment and negative behavioral outcomes ................. 7 Expressed Emotion and behavior problems ..................................... 8 Critique of EB research with children .............................................. 9 Theoretical Framework .............................................................................. 10 Ecological Theory and Child Development ................................... 10 Family Systems Theory: Child Development in the Family System ................................................................ 11 Communication Theory: Emotional Interaction in a Family as an Influence on Child Behaviors .......................................... 13 Need for the Research ................................................................................ 15 CHAPTER 2: LITERATURE REVIEW Development in Middle Childhood ........................................................... 19 Family Environment and Childhood Risk ................................................. 20 Introduction to Expressed Emotion ........................................................... 21 Expressed Emotion and Adult Mental Illness ................................ 21 History of research on Expressed Emotion .................................... 21 Defining Expressed Emotion ..................................................................... 23 Categories of EB ............................................................................ 23 Brief Review of Adult Studies of EB ............................................. 25 Children and Maternal EE ......................................................................... 28 Maternal Criticism and Extcmalizing Behavior Problems ............ 28 Specific EE Components (Criticism and Emotional Over- Involvement) as Predictors of Internalizing and Extemalizing Behaviors ............................................................ 31 Criticism as a predictor of internalizing and externalizing disorders ....................................................... 32 viii E01 and child disorders ................................................................. 33 Linking maternal criticism and child problems ............................. 34 Maternal criticism and emotional over-involvement linked to emotional or behavioral disorders .............................. 36 Contribution of Maternal Depression to Children’s Behavior Outcomes: Possible Mediating Effect of EB ......................... 37 EE Research with Japanese Samples ......................................................... 43 Review of Japanese EE Studies ............................................................ 43 Research with adult Japanese samples ........................................... 43 Research with Japanese adolescents .............................................. 46 Resiliency and the Present Study ............................................................... 47 Mechanisms of the Joint Contribution of Environmental Factors and Child Resiliency to Child Behavior Outcomes: Deficit Models vs. Resilience Models .................................................. 48 Diathesis-Stress model and child behaviors .................................. 48 Resilience model and child behaviors ............................................ 50 Internal Resiliency Resources in Children ................................................. 52 Self-competence as a mediator between maternal negative feedback (EB) and childhood adjustment ................... 52 Development and “Sense of Coherence” as a resilience resource ..................................................................... 5 3 EE and resiliency ........................................................................... 57 , Mother-Child Relationships in Japan ......................................................... 58 Mother-child relationships in the macrosystem ............................. 59 Conceptions of child development in Japan .................................. 59 Differing perceptions of “healthy” child development in the US and Japan ................................................................... 63 Construction of the self and self-esteem of children ..................... 65 Research Questions and Hypotheses ......................................................... 68 CHAPTER 3 METHODS Participants ................................................................................................. 73 Children .......................................................................................... 73 Mothers .......................................................................................... 73 Rates of return ............................................................................................ 74 Demographic Characteristics ..................................................................... 74 Procedure-Children’s Data ......................................................................... 75 Recruitment .................................................................................... 75 Data collection ............................................................................... 76 Instruments-Children ..................................................................... 77 Youth Self Report (YSR) ................................................... 77 Self-Liking/Self-Competence Scale-Revised Version (SE) .................................................................. 79 Sense of Coherence (SOC) ................................................ 80 Expressed Emotion Adjective Checklist (EEAC) .............. 81 Family Relationship Index (FRI) ....................................... 83 ix Procedure-Mother’s Data ........................................................................... 84 Questionnaires ................................................................................ 84 Instruments-Mothers ...................................................................... 84 The Center for Epidemiologic Studies Depression Scale (CBS-D) ........................................................................... 85 Analyses ............................. 87 Sample Size .................................................................................... 87 Analysis for Specific Aim 1 (Hypotheses 1.1, 1.2, and 1.3) .................................................. 87 Analysis for Specific Aim 2 (Hypotheses 2.1., 2.2, and 2.3) ................................................. 87 Analysis for Specific Aim 3 (Hypotheses 3.1, 3.2, and 3.3) .................................................. 89 Moderation effect ............................................................... 89 Mediation effect ................................................................. 89 Structural Equation Modeling (SEM) (Hypothesis 4.1) ........................................................................ 90 Missing data treatment for SEM analysis ...................................... 91 Chapter 4: RESULTS Preliminary Analyses ................................................................................. 97 Description of the Variables ...................................................................... 97 Child variables ............................................................................... 97 Maternal variables .......................................................................... 97 Correlations between Variables ................................................................. 97 Child variables ............................................................................... 98 Mother variables ....................................................................... i..L..98 High~Risk and Normal-Risk YSR Groups ................................................. 98 Relationships between Demographic and Independent Variables ............. 99 Child variables ............................................................................... 99 Mother variables .......................................................................... 100 Hypothesis 1. Relationships between Measured Variables ..................... 102 Family environment and YSR score ............................................ 102 Child report ...................................................................... 103 Mother report ................................................................... 103 Child resiliency factor and YSR score ......................................... 104 Family environment and resiliency resources .............................. 104 Child report ...................................................................... 105 Mother report ................................................................... 105 Hypothesis 2. Maternal Criticism and Involvement in Relation to Behavior Problems ........................................................... 105 Relationships between Involvement and Internalizing and Extemalizing Scores ................................... 106 Child report ...................................................................... 106 Mother report ................................................................... 107 Correlations between Child Report of Involvement and YSR Behavior Scores by Gender ..................................... 107 Boy’s report ..................................................................... 107 Girl’s report ...................................................................... 108 Correlations between Mother Report of Involvement and YSR Behavior Scores by Gender .................................................... 108 Mothers of boys ............................................................... 108 Mothers of girls ................................................................ 109 Relationships between Criticism and Internalizing and Externalizing Scores ............................................................... 109 Child report ...................................................................... 109 Mother report ................................................................... 109 Correlations between Child Report of Involvement and YSR Behavior Scores by Gender ............... . ..................................... 110 Boy’s report ....................................................................... 110 Girl’s report ....................................................................... 1 10 Correlations between Mother Report of Criticism and YSR Behavior Scores by Gender .................................................... 110 Mothers ofboys .......... 110 Mothers of girls ................................................................ 110 Gender Differences in Strength of Correlations .......................... 111 Differences by Normal-risk and High—risk Groups ..................... 1 11 Group formation ............................................................... 112 Criticism and risk ............................................................. 113 Involvement and risk ........................................................ 113 MANOVA ........................................................................ 1 13 Criticism and Involvement in High- and Low-Risk Groups ........ 115 Group formation ............................................................... 115 Criticism and risk ............................................................. 115 Involvement and risk ........................................................ 115 MANOVA ........................................................................ 116 Criticism and Involvement as Specific Predictors ....................... 117 Predicting Extemalizing ................................................... 118 Predicting Internalizing .................................................... 118 Summary .......................................................................... I I9 Predicting Internalizing and Extemalizing in High-risk Samples ............................................................ 119 Predicting Extemality ...................................................... 120 Predicting Internalizing .................................................... 121 Summary .......................................................................... 122 Hypothesis 3. Mediating 311d Moder ating Effects of Resiliency Factors ........................................................................... 122 Moderating Effects of Resiliency Factors .................................... 122 Full sample ....................................................................... 123 High-risk samples ............................................................ 124 Moderating Effects of Gender ..................................................... 124 Gender and Involvement ...................................... 124 xi Gender and Criticism ......................................... 125 Mediating Effect of SE and SOC .................................... 126 Mediating Effects of Resiliency Factors on Mother’s Involvement and Behavior Problems ...................... 127 Mediation of SOC on Involvement and Internalizing ...... 12 7 Mediation of SOC on Involvement and Extemalizing 128 Mediation of SE on Involvement and Internalizing and Extemalizing .................................................................... 128 Mediating Effects of Resiliency on Mother’s Criticism and Behavior Problems ............................................................ 128 Mediation of SOC on Criticism and Behavior Problems. 129 Mediation of SE on Criticism and Behavior Problems....129 Mediation of SE on Mother’s Report of Her Own Criticism and Behavior Problems ............................... 130 Hypothesis 4. Testing the SEM .............................................................. 130 Modeling ...................................................................................... 130 Principal Components Analysis of the SE Scale ......................... 136 CHAPTER 5: DISCUSSION AND CONCLUSIONS Demographic Information ........................................................................ 177 Sample Characteristics ................................................................... 178 Gender Differences ........................................................................ 178 Gender differences in Internalizing and Extemalizing ........................................................ 178 Possible sampling bias ..................................................... 179 Gender and self-esteem ....................................... 179 Family Environment Factors, Resiliency Resources, and Internalizing and Extemalizing Behaviors........................... .............. 180 General Findings .......................................................................... 180 Child Perceptions of Maternal EE and Behavior Problems: Comparisons between Normal-Risk and High-Risk Groups of Children ................................................................. 181 Maternal High Criticism and Low Involvement as a Risk Factor ...................................................................... 182 Constructs of Maternal Criticism and Emotional Involvement in Japanese Culture ............................................ 184 Moderation Effects of Gender ..................................................... I86 Mediation Effects and Resiliency Factors ................................... 187 Contributions of Mother’s Perceptions ........................................ 189 Limitations ............................................................................................... 191 Clinical Implications ................................................................................ 194 A APPENDICES ..................................................................................................... 196 BIBLIOGRAPHY ................................................................................................ 228 xii 3.1 3.2 3.3 3.4 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 4.10 4.11 4.12 4.13 LIST OF TABLES Return Rates for the Questionnaire Packets and Consent Rates for Children’s and Mothers’ Participation ........................................ 92 Demographic Information for Children ........................................ 93 Number of Children and Family Structure for Participants (Mother’s Report) ............................................................... 94 Parent’s Employment (Mother’s Report) ..................................... 95 Child Measures by Gender ..................................................... 154 Maternal Measure by Child Gender .......................................... 141 Correlations between Child’s Variables ...................................... 142 Correlations between Child Variables by Gender ........................... 143 Correlations between Mother Variables and Child Outcome Variables .......................................................................... 144 Correlations between Mother Variables and Child Outcome Variables by Gender ............................................................ 145 YSR Normal-risk vs. High-risk by Gender and Grade.......................146 YSR Normal-risk and High-risk Groups: Child Reports of Mother’s Involvement and Criticism .......................................... 147 YSR Normal-risk and High-risk: Mother’s Reports of Her Own Involvement and Criticism of Her Child ...................................... 148 Coefficients for Regression of Criticism and Involvement on Extemality (Full Sample) ....................................................... 149 Coefficients for Regression of Criticism and Involvement on - Intemality (Full Samme) ........................................................ 150 Coefficients for Regression of Criticism and Involvement on Intemality among Extemalizing High-risk Group of Children ............. 151 Coefficients for Regression of Criticism and Involvement on Intemality among Internalizing High-risk Group of Children .............. 152 xiii 4.14 4.15 4.16 4.17 4.18 4.19 4.20 4.21 4.22 4.23 ' Report of Mother’s n the Relationship between Child’s and YSR Internalizing 153 OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO Mediation Effect of SOC 0 Report of Mother’s Involvement Behavior cores ....................... the Relationship between Child’s and YSR Extemalizing 154 OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO Mediator Effect of SOC on Report of Mother’s Involvement Behavior Scores ........................ Relationship between Child’s ect of SE on the YSR lntemalizing 155 Mediator Eff Report of Mother’s Involvement and Behavior Scores OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO Iationship between Child’s ct of SE on the Re and YSR Extemalizing 156 ther’s Involvement OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO Mediator Effe Report of Mo Behavior Scores OOOOOOOOOOOOOOOOOOOOOOOOOOOOOO Relationship between Child R Internalizing 157 OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO ffect of SOC on the Mediator E Mother’s Criticism and Y8 Report of Behavior Scores OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO Relationship between Child’s Extemalizing 1 58 ooooooooooooooooooooooooooooooooo QC on the Mediator Effect of S Criticism and YSR Report of Mother’s Behavior Scores ................................ ationship between Child’s ntemalizing 159 OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO E on the Rel Mediator Effect of S Criticism and YSR I Behavior Scores ............................... ionship between Child’s xtemalizing OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO E on the Relat Mediator Effect of S Criticism and YSR B Report of Mother’s Behavior Scores ............................... ionship between Mother’s Report of Behavior Scores ............... 161 Mediator Effect of SE on the Relat Her Own Criticism and YSR Internalizing Correlations between YS Component Analysis ............ ooooooooooooooooooooooooooooooooooooooooooo ..... xiv LIST OF FIGURES 1.1 Conceptual Map for Study ......................................................... 18 2.1 Proposed Structural Equation Model for Study ................................. 72 3.1 SEM Analysis to Test the Mediating Effect of Child Coping Resources ............................................................................ 96 4.1 Child Report of Mother’s Involvement by Family Structure ............... 163 4.2 Child Report of Mother’s Criticism by Family Structure. ........ ' .......... 164 4.3 Child Report of Involvement toward Mother by Family Structure. . . ........165 4.4 Mother’s Family Relationship Inventory by Family Structure .............. 166 4.5 Moderating Effect of Gender on the Relationship between Mother’s Involvement and Extemality ........................................ 167 4.6 Moderating Effect of Gender on the Relationship between Mother’s Criticism and Intemality ............................................. 168 4.7 Moderating Effect of Gender on the Relationship between Mother’s Criticism and Extemality ............................................. 169 4.8 Child’s SOC Mediates the Relationship between Child 3 Report of Mother 5 Involvement Directed Toward Child and Child’s Behavior Problems ................................................................ 170 4.9 Child’s SE Mediates the Relationship between Child’s Report of Mother’s Involveme t Directed Toward Child and Child’s Behavior Problems ................................................................ 171 4.10 1 Child’s SOC Mediates the Relationship between Child’s Report of Mother’s Criticism Directed Toward Mother and Child’s Behavior Problems ................................................................ 172 4.11 Child’s SE Mediates the Relationship between Child’s Report of Mother’s Criticism Directed Toward Mother and Child’s Behavior Problems ............................................................... 173 4.12 Full Model Showing All Standardized Path Coefficients .................... 174 XV 4.13 SEM showing predictor variables and significant pathways between latent variables .................................................................. 175 4.14 SEM excluding child report data showing predictor variables and ' significant pathways between latent variables .............................. 176 xvi CHAPTER I INTRODUCTION The Experience of Depression Common wisdom holds that childhood is the time when children expand their world into one full of joy, happiness, excitement, and curiosity. For some children, this idealistic expectation is simply unreasonable. Problems arising during childhood can have lifelong consequences, and childhood experiences and conditions have been associated with many adult disorders (Marsh & Dozois, 2003). Achenbach (1991), in the Child Behavior Checklist, defined “Internalizing” behavior problems as a combination of social withdrawal, somatization, and depression/anxiety. Further, “Extemalizing” was defined as a combination aggression and delinquent behaviors. The early onset of such internalizing and externalizing behavior problems in childhood has been found to be a risk leading toward more serious maladaptive outcomes in the future (Cicchetti & Toth, 1998; Mun, Fizgerald, Von Eye, Puttler, & Zucker, 2001). For example, depression can be an overwhelming experience for children and adults. Rollo May, one of the fathers of existential psychology, stated that “Depression is the inability to construct a future” (May, 1969, p. 243). While depression is bad enough for adults, it is especially poignant when the depressed person is a child. Instead, Takashi scribbled: “I am stupid,” “I am slow,” “Everybody wants me to die,” “Nobody likes me,” “I am ugly,” “I am dumb,” “I better die,” “my mother doesn’t like me,” and “I am like an ugly monkey.” He was screaming inside, but was externally silent about his psychological pain. how he is doing there. His self-description shows the complexity of children’s emotional and social lives. The goal of this research is to examine the complex relationships among family environment, resiliency resources, and risk. 2 Prevalence of internalizing behavioral symptoms (depression). According to the US Surgeon General’s report (2009), 10% to 15% of children and adolescents in the US report some symptoms of depression. Approximately 5% of children between the ages of 9 and 17 have been diagnosed with depression at some time in their lives. Compared with the l-year prevalence rate of 5.3% in US adults, the l-year prevalence (MDD) has been reported for adolescents (15% to 20%) (Cicchetti & Toth, 1998). Taken 3 together, the implication of these results is that issues of childhood depression are elusive and complex. Early onset of internalizing behavior problems. An early onset of internalizing behavior problems in childhood is predictive of various negative long-term developmental outcomes. After a diagnosis of MDD in childhood, there is a 40% probability of a recurrence in 2 years, rising to 70% in 5 years. The depressive episodes of about 60% to 70% of children diagnosed with MDD persist into adulthood. In general, MDD precedes alcohol or substance abuse (Sanford et al.,1995). Further, approximately 25% to 34% of children and adolescents who were diagnosed for depressive disorders have attempted suicide (Birmaher et al., 1996; Kovacs, 1996, 1997). The development of social-cognitive and interpersonal skills and maintenance of the attachment bond between parent and child are especially vulnerable to the negative effects of childhood and adolescent depression (Kovacs, 1997). Kovacs described the situation as “. .. [depressed children] are removed from the normal matrix of socialization” (1997, p. 289). Extemalizing behaviors. Extemalizing behaviors are “the most common form of mental health problems in children” (Denharn, Workman, Cole, Weissbrod, Kendziora, & Zahn-Waxler, 2000, p. 24) with a wide range of symptoms. Prevalence rates for externalizing disorders have varied from 2% to over 15% of the population. Extemalizing 4 behaviors are generally categorized into two major types, inattention and hyperactivity on one hand and aggression and conduct problems on the other. Approximately 9% of boys and 2% of girls have symptoms of conduct disorder, and approximately 3 to 7% of children have attention-deficit hyperactivity disorder (ADHD) (American Psychiatric Association, 2005; Hinshaw, 1992). The early onset of externalizing behavior problems, like aggression in childhood, is a precursor of antisocial behaviors (Moffltt, 1993) and antisocial personality disorder (Zucker, Ellis, Fitzgerald, & Bingharn, 1996; Mun et al., 2001). Children with externalizing behaviors commonly have difficulties to regulating emotions and jeopardize the opportunities to build a successful interpersonal relationship (Denham et al., 2000). Therefore, it is important to identify the early experiences that increase the risk of developing internalizing and externalizing behavior problems in children. Internalizing and Extemalizing Disorders in Children in Japan. In Japan, as in the West, childhood internalizing disorders, specifically depression, were neglected for a long time because of theoretical assumptions that children could not experience depression. In more recent studies, an increasing number of reports about Japanese children suffering from various social problems, including social withdrawal, school refusal, bullying, and/or eating disorders, have appeared. In Japan, the prevalence rate of 5 Attention Deficit-Hyperactivity Disorder (ADHD) was reported to be about 7.7%, somewhat higher than the prevalence rate of 4 to 6% in the US (Satake, Yamashita, & Yoshida, 2004). According to a large-scale study among 2,453 Japanese children and adolescents (6 to 15 years old), about 11% of Japanese elementary-school children scored as depressed, based on Birleson’s Depression Self-Rating Scale for Children (DSRS), a self- administered depression scale (Denda, Kato, Kitagawa, & Koyama, 2006). Within the larger sample of 6-15 year olds, approximately 15% of 10 and 11 year old children were at risk of depression (scores above the cut-off point of 15 on DSRS). Another Denda study (2007) was conducted with 3,331 Japanese children and adolescents (6 to 15 years old). The results classified 7.8% of elementary school-aged children and 22.8% of junior high school-aged children as depressed, again based on self-report of symptoms. However, few of these children actually had been diagnosed or treated for depression (Denda, 2007). The symptoms characteristic of depression in Western cultures also were reported by those Japanese children and adolescents were extreme fatigue or loss of energy, loss of interest or pleasure, impaired concentration, insomnia, social withdrawal, and anorexia or weight loss (Denda, Sasaki, Asakura, Kitagawa, & Koyama, 2003). Child Development and Family Factors in Behavioral Disorders 6 The progress of children’s behavior problems is related to both family environment, which supports the child’s developmental needs, and developmental chronology (Bronfenbrenner, 1979). Healthy emotional, social, and self development in middle childhood depends heavily on emotionally positive interactions between caregiver (usually, the mother) and child (Cole, Michel, & Teti, 1994). Bronfenbrenner’s Ecological Theory emphasizes the importance of the relationship between mother and child in a family as an influence on the ongoing process of child development. Within the context of family environment, the mother-child interaction with emotional exchange structures a child-specific dyadic context and is a fundamental dimension of the family environment. It has powerful effects on a child’s cognitive and socioemotional development (Bronfenbrenner, 2005). The mother—child dyadic relationship provides the opportunity for the child to learn “interactive skills and a concept of interdependence, an important step in cognitive development” (p. 57). The patterns of the dyadic interaction in the family are applied to the other relationships outside the family context as well because child can use the emotional competence that was gained in the dyadic relationships in the family. Family environment and negative behavioral outcomes. The relationship between family environment and child’s negative behavioral outcomes has been 7 confirmed by many studies based on different theoretical approaches. F amily environment studies have identified as important the parents’ childrearing strategies and behaviors (Belsky, 1984), the child’s attachment style (Ainsworth, 1967; Ainsworth & Bell, 1970; Ainsworth, Blehar, Waters, & Wall, 1978; Bowlby, 1960, 1975, 1981, 1982, 1988), coercive parent interaction patterns (Snyder, 1995), and parental acceptance- rejection (Rohner & Britner, 2002). The present research will examine mother-child communication patterns as they relate to child behavioral problems. Expressed Emotion and behavior problems. Expressed Emotion (EE) is a construct that was defined by Brown and Rutter ( l 966) and operationalized in the Camberwell Family Interview (CFI, Vaughn & Leff, 1976) or the F ive-Minute Speech Sampler (FMSS, Magna-Amato et al., 1986). Three EE components have been identified: Criticism, Hostility, and Emotional Over-involvement. The construct of EE is thought to reflect the emotional atmosphere of the family, specifically the mother and child relationship. Maternal expressions of extreme Criticism and/or Emotional Over- Involvement directed toward the child are a risk marker for the child’s negative outcomes, fer example, depression, anxiety, anorexia, obsessive compulsive, and conduct disorders, and aggressive behaviors (Huguelet, Fayre, Binyet, Gonzales, & Zabala, 1995; Marom, Munitz, Jones, Weizman, & Herrnesh, 2005; McCreadie, Robertson, Hall & Berry, 1993; 8 Nelson, Hammen, Brennan, & Ullman, 2003). The application of the BE construct has been expanded to different medical conditions, including diabetes, asthma, epilepsy, and obesity management. Emotionally negative attitudes (Criticism and/or Emotional Over- Involvement) from a key relative create an unbearably stressful family environment that may exacerbate the behavior and emotional problems of children, some of whom are already be vulnerable (Hooley & Gotli, 2000; Hooley & Parker, 2006). Critique of EE research with children. Despite strong evidence of the correlation between a high level of EB and children’s behavior disorders, there are some arguments against the simple conclusion that high-BE causes relapse, emotional disorders, or children’s behavior problems. First, most of the studies in the BE literature have been limited to samples of clinically-referred children or adults with psychological disorders. Even though the importance of high levels of maternal BE in childhood disorders is acknowledged, few studies of EB have been done among school-aged children sampled from a school or community setting, and none of these studies have been conducted in Japan. Second, most of the studies with children have been based on a pathological model focusing on children’s vulnerability or predisposition to EE, rather than resiliency. It is still unclear how an individual child’s resiliency alters the relationship between the 9 construct of EB and child negative outcomes. Factors have been found to exist between perception of mother-child emotionality toward each other and child adjustment problems (e.g., Gomez, Gomez, DeMello, & Tallent, 2001; Toth, Cicchetti, & Kim, 2002; Kim & Cicchetti, 2004), but the possible mitigation by child’s resiliency factors of negative family environment factors on child’s psychological and behavioral problems has not been examined in the research literature. Theoretical Framework This study was guided by three theories: Bronfenbrennor’s Ecological theory, Family Systems theory, and Communication theory to understand the relationship. between context and children’s behavior problems. The process of child development occurs within the context of the system of relationships that form his or her environment. Family Systems theory focuses on the characteristics of the family as a context for development, but in many ways is consistent with Ecological theory (see Figure 1.1). Communication theory focuses on the patterns of communication in a family, some of which create a potential risk for children’s psychological and behavioral behavior problems. Ecological Theory and Child Development Ecological theory views the child’s context as a set of nested structures. Human 10 development is shaped by experiences in multiple settings involving multiple systems over the course of that person’s life up (Bronfenbrenner, 1979, 2005). Children’s characteristics are produced by multiple interactions with different parts of the environment, in Bronfenbrenner’s terms the microsystem, mesosystem, exosystem, and macrosystem. The microsystem is the most immediate context in which the child is exposed to and becomes active with certain roles and with face-to-face interpersonal relationships. Home, classroom, and peer group are the typical examples of a microsystem. Each child experiences different patterns of activities with certain roles through mutual interpersonal relations in various settings or microsystems. The child’s own unique physical and mental characteristics affect the deveIOpment in a way that child’s behaviors become the part of the process of interaction. Thus, characteristics of child’s behaviors are produced in the interactions in a setting as a whole, rather simply being “the sum of its parts” (Bronfenbrenner, 1979, p. 109). Each microsystem may be affected by events or persons in the other larger systems, but the fundamental experiences for a child are in her/his microsystems. This research focuses on the microsystem of the child’s ecological system. Family Systems Theory: Child Development in the Family System ll Family Systems theory helps understand the child’s developmental psychopathology in ways that the child lives in a dynamics of the family (microsystem) (Patrick & Cicchetti, 2004). Family Systems theory seeks to describe individual development by focusing on how the individual’s unique role in the family creates patterns and characteristics among relationships with individuals within and between subsystems in the family (siblings, parents, couple, parent-child). These subsystems are seen as functioning as a whole to make a “family.” Family systems theory emphasizes ’ relationship structures, interpersonal boundaries, power distributions, and communication patterns (Minuchin, 1985). A child’s adjustment and possible maladaptive behaviors cannot be understood without looking at meanings of behaviors within the context of interactions where the child responds (Cicchetti & Howes, 1991; Daies & Forrnan, 2002). Family systems theory views a child’s behaviors in an Open system, meaning that each family member, including the child, functions as an ongoing transactional interaction (Cicchetti & Tucker, 1994; Minuchin & Fishman, 1981) between “an active changing organism in a dynamic changing context” (Patrick & Cicchetti, 2004, p. 478). It is critical to pay attention to the transactional characteristics of interaction created by both child’s and mother’s l2 perceptions and sequencing behaviorsas influences on the child’s internalizing or/and externalizing behavior difficulties. Because child development is a process characterized by a series of changes in cognitive, linguistic, social, emotional, and physical experiences within the environment, the child’s development in the family system is also characterized by “hierarchical transformation from transactional feedback loops involving the biopsychosocial characteristics of family members, structural processes in the family” (Patrick & Cicchetti, 2004, p. 479). The meanings of a child’s behavior problems may be a reflection of the dimensions of interplay in the unique family system. This study examines the mother’s and the child’s perceptions of their emotional interactions. Communication Theory: Emotional Interaction in a Family as an Influence on Child Behaviors Communication theory describes the impact of unhealthy communication patterns between individuals within the microsystems on child’s emotional and behavioral adjustment. Communication patterns and the emotional attitude of an important family member toward a target family member contribute importantly to the individual’s psychological health. For example, unbearable negative communication patterns between mother and child cause psychological confusion and stressors in a family. Such an 13 environment elicits or reinforces problematic behaviors in the target child. The communication is a reciprocal interaction and also a child-specific, non-shared environmental factor contributing to the experiences of that specific child (Caspi et al., 2004). Communication theory was developed based on observations of family interactions by a group of family therapists of patients with schizophrenia (Bateson, Haley, Weakland, 1956). They introduced the concept of the double bind, describing a pathological family communication pattern of patients with schizophrenia in a family context. The double bind message includes two conflicting messages that create psychological confusion, rather just a simple contradiction (Nichols & Schwartz, 1998). The patient’s schizophrenic behaviors are seen as a part of the whole family’s attempt to adjust or maintain homeostasis in order for the family to function and preserve its equilibrium. The bizarre behaviors of patients with schizophrenia are seen as a product of confusing communication patterns. As a consequence of these confusing patterns of communication, the child grows Up unskilled in the ability in determine what people really mean and unskilled in the ability to relate with others (Nichols & Schwartz, 1998). When children have such a dynamic relationship in the family, they are less likely to receive positive supports and 14 encouragement from their parents, hindering the child’s self-esteem, self-confidence, and psychological coherence or the ability of comprehend, manage, and make meaning life events. When these communications are unclear or carry conflicting messages confusion and stress occur in children. A child’s sense of coherence is shaped by the powerful force of repeated interactions with a person in an important relationship (Antonovsky, 1976; Cole, Martin, Powers, & Truglio, 1996; Conley, Haines, Hilt, & Metalsky, 2001). Need for the Research More children are being diagnosed with depression because of the realization that children show depression in ways that differ from adults. Children with the early onset of internalizing disorders tend to be impaired in important areas of functioning, (Kovac & Devlin, 1998) and these disorders have devastating impacts on the lives of the child and his/her family. Children who develop internalizing disorders in childhood are more likely to develop other problems in adolescence and adulthood. The problem is magnified because many children who have internalizing disorders are untreated in community samples (Kovac & Devlin, 1998). lntemalizing disorders damage a child’s life because they impede normal development and create problems that do not dissipate with time. The family’s response to the internalizing and externalizing problems may be dysfunctional, affecting everyone in the family as the family organizes around the 15 problems. Thus, the consequences of such disorders in childhood to the child and her/his family cannot be minimized. Even though the high rate of depression in children in Japan is known, and there are clear indications that intervention and prevention programs for both children and families are needed, no study has been done to examine the relationship between children’s internalizing and externalizing problem behaviors and familial environment and children’s resilient capacities. Moreover, most of the research on the relationship between a relative’s expressed emotion (EE) and mental illness or mood disorders has been done among adults, primarily in Western countries. This study is a first step in filling the gap in our understanding of maternal EE and childhood internalizing and externalizing behavior problems in the Japanese culture and understanding the reciprocal relationships between mother and child behaviors. Finally, the results of this study will help inform the various behaviorally-focused interventions by providing a better understanding of the role of parent-child interactions in children’s behavior problems. Although it is important to improve the mother’s parenting skills, the emotional and cognitive components of the negative parent-child interactions and familial environment need to be changed as well, addressing the 16 bidirectional dynamic between mother and child interactions (Coville, Miklowitz, Taylor, & Low, 2008). Thus, the study will examine the parent-child dynamics that create and maintain a negative family emotional environment, and their interaction with child’s resilience in the context of internalizing and externalizing behavior problems. In summary, the purpose of this study is to examine the relationships among family environment including mother- child perception of EB, children’s resiliency factors, and children’s psychological and behavioral problems among Japanese school- aged children sampled from Japanese public schools. 17 $35 new man. 333280 . N .~ 3sz D 350 BSmhoouog . 803ng All: 8:309:00 8:03:00 mo omnom m All. . 53mm . \ macaw noon Eoouwofiom 18 839$ anthem Henchman “meanings: CHAPTER II LITERATURE REVIEW Development in Middle Childhood Middle childhood, defined as ages 6 to 12 (Richardson, 2005; Santrock, 2000; Zembar & Blurne, 2008), is an important time in child development. Children in middle childhood develop academic and social skills not found in younger children. In terms of Erikson’s developmental stage theory (1963), the child must resolve the conflict between industry (competence) and inferiority (incompetence, failure) in middle childhood. Resolving this developmental conflict includes building a strong sense of self- competence with positive self-esteem, self-efficacy, and self-concept, as well as developing an internal locus of control. This sense of self-competence is the foundation for the development of a sense of coherence as an adolescent and adult (Antonovsky, 1987) Achieving industry or self-competence includes the ability to persist in mastering a skill or completing a project over increasingly extended periods of time. Elementary school-aged children focus on the acquisition of skills and competence in different areas of development, including acquiring basic academic knowledge and skills (reading and mathematics) and basic interpersonal skills (cooperation. following group norms and 19 rules). As these skills and competencies are achieved, they allow the children to decrease their degree of dependency on parents. Children are encouraged and expected to behave with greater autonomy (Richardson, 2005). Although children in middle childhood are reaching out to peers and other adults, the role of the family in providing support and encouragement is predictive of a successful resolution of this developmental phase. Research has shown that the family emotional environment has an impact on the child’s development of a sense of industry. Parents continue to function as the child’s primary resources for emotional security and as providers of social support during middle childhood (Cicchetti, 1996: Crittenden & Ainsworth, 1989; Richardson, 2005). When the emotional environment is not supportive, the risk of the child displaying dysfunctional and negative behavior outcomes, including suicide attempts, is increased (Asarnow, 1992; Asarnow, Carlson, & Guthrie, 1987; Sroufe, 1997). In a family characterized by frequent mutual negative emotional interactions between child and parent or between parents, joint family activities are restricted. The child’s opportunity to learn appropriate relationship management through observation or experience cannot occur (Bronfenbrenner, 2005). In such an environment, Opportunities to develop emotional and social competencies are shut off for the child. Family Environment and Childhood Risk 20 The elementary school-age child is tasked with finding ways of coping with different situations and adjusting to different contexts. By the time s/he has entered school, the child has developed a relationship with her/his primary caretaker(s) that serves,'to an extent, as a model. for other relationships (Ainsworth, 1967; Ainsworth & Bell, 1970; Ainsworth, Blehar, Waters, & Wall, 1978; Bowlby, 1960; 1975, 1981, 1982, 1988). Based on these primary dyadic relationships, the child can acquire self- competence and sense of coherence—skills, emotional knowledge, values, and meaning-“that make her/him resilient in the face of stressors. When such a relationship is dysfunctional, the child may develop emotional symptoms and/or behavior problems. Introduction to Expressed Emotion Expressed Emotion and Adult Mental Illness History of research on Expressed Emotion. Research on Expressed Emotion (EE) as a predictor of relapse and rehospitalization in schizophrenia has a long history in adult psychiatry (Brown & Rutter, 1966). In the literature, “Expressed Emotion” or “EE” refers to expressions of criticism, l'restility, and emotional over-involvement on the part of family members toward a targeted member,, a child or spouse. BE captures an important dimension of the relationslups between the targeted family member s and his/her relatives (HOOIey & Teasdale, 1989). In general, positive expressions of emotion have been 21 assumed to be an asset for the targeted person, while negative expressions of emotion create a risk for the targeted person. However. when there is extreme emotional over- involvement, even if there are positive expressions of emotion, the targeted, person is placed at further risk. The concept of EB was developed by Brown and Rutter (1966) based on their clinical observations that the families of patients with schizophrenia. These families often described the patient in ways that were quite negative. The more negative the description of the patient by family members. the more likely the patient was to relapse and be rehospitalized. They hypothesized that family member of the patient put psychological pressure on the family member with schizophrenia and that this stress precipitated the relapse (Leff & Vaughn, 1985). In the Brown and Rutter (1966) study, the husband had been diagnosed with schizophrenia. The interviewer met the wife and husband separately, and in a second interview, they were seen together. Later, 30 couples who had been interviewed initially were seen a second time to check the reliability and validity of the measures. Each of the interviews lasted 34 hr. In addition to the self~reports of feelings in the semi-structured interview, actual expressions of positive and negative feelings including tone of voice (cg, sarcasm, gesture, and facial expression) were accounted for in the assessment of the family member’s emotional attitudes directed toward the family member with schizophrenia. Studies have examined the associations between EE by important family members and the risk of relapse in different disorders. The application of the BE construct has been expanded to different psychological and physical disorders, including depression, diabetes, asthma, epilepsy, and obesity management. Other studies have tested theoretical speculations about the meaning of the EB construct and the degree to which high EE predicts relapse over a period of years (Marom et al., 2005; H uguelet et al., 1995; McCreadie et al., 1993). Defining Expressed Emotion Categories of EE. 1 hree categories of negative EE have been identified (Vaughn & Leff, 1976): Criticism of the targeted person, Hostility directed toward the targeted person, and Emotional Over-Involvement (E201) with the targeted person. Each of these elements potentially plays a different role in increasing the risk of relapse or the onset of the disorder. However, typically. only Criticism and E01 are used in EE research because Hostility and Criticism are strongly correlated (Vaughn & Leff, 1976) These dimensions were operationalized by developing the Camberwell Family Interview (C F1). The C F1 is an extensive, standardized, structured interview with 23 important family members, typically, the patient’s spouse or parent. Criticism reflects relatives’ expressions of unfavorable, dissatisfying, and regretful emotions, or desire for things to be different in regard to the patient or the patient’s illness. However, overtly critical expressions are not the only ways the CF 1 scoring defines a statement as Criticism. Even though they do not explicitly blame the patient, the following statements are the examples of comments of mothers towards children with mental illness that would be scored as Criticism: “I’d rather he didn’t lie to us" or “I wish that he could hold down a job~~any job would do‘” (Leff& Vaughn, 1985, p. 38'). Hostility includes expressions of dislike or rejection of a patient, and is characterized with comments that attack the patient as a person rather than his/her behaviors because of hist'her mental disorder. 'l‘be following statement is an example of comments coded as Hostile: “He’s not, any benefit to himself or any benefit to society or any benefit to the family situation” (Leff & Vaughn, 1985, p. 41). Lastly, Emotional Over-Involvement. or EOI includes expressing a great many worries about the patient in an extremely enmeshed or symbiosis-like relationship. Mothers show self-sacrificing behaviors and over-protection of the child. They present exaggerated emotional responses, such as extremely intense anxiety directly related to the patient’s welfare, and often sacrifice having a life of their own in order to devote their 24 lives to caring for the patient. The behaviors of 1301 were found to be characteristic of families of patients with depression, anxiety, and other psychiatric illnesses. The following statements would be coded as 15.01: “I felt terrible~1 felt my whole world was shattered. . .. I’ve spent many a time crying, wondering what went wrong almost every day” (Leff & Vaughn, 1985, p. 45). All of these concepts must be understood within their cultural context. For example, criticism must be understood from the standpoint of the cultural norm and values related to criticism. In Japanese culture, criticism may be part of a group-oriented model of self improvement. Likewise, evaluations about whether a person is emotionally over-involved or enmeshed depends on cultural normative standards about the appropriate levels of involvement. In the following review, research highlighting the links between criticism and E01 will be presented, followed by details about Japanese cultural norms related to parenting. Brief Review of Adult Studies of ER Consistently, research has shown that higher level of maternal or spousal EE is a predictor of relapse or poor outcome among adult patients with a wide range of mental and physical disorders. including schizophrenia (Marom et a1, 2005; Huguelet et al., 1995; McCreadie et 311,, 1993), unipolar depression (Hooley, Orley, & Teasdale, 1986; 25 Hooley & Teasdale, 1989; Vaughn & Leff, 1976; Hooley & Licht, 1997; Kamal, 1995), bipolar disorder (Miklowitz, Goldstein, Nuechterlein, Snyder, & Mintz. 1988; Miklowitz, Goldstein, Richards, Simoneau, & Succath, 2003; Vaughn & Leff, 1976;), borderline personality disorder (Hoffman et al., 2005; Coville et al., 2008), obsessive compulsive disorder (Chambless, Bryan, Aiken, Stelketee, & Hooley, 2001), anxiety disorders (Charnbless et al., 2001), and eating disorders (Butzlaff& Hooley, 1998; Leff & Vaughn, 1985; Hedlund, F ichter, Quadflieg. & Brandi, 2003). Cross-sectional Studies have formd the relationship between specific type of disorder and EE. For example, individuals diagnosed with depression are more sensitive to Criticism than individuals with schizophrenia (Hooley et al., 1986; Hooley & Gotlib, 2000', Hayhurst, Cooper, Paykel, Vearnals, & Ramana, 1997), individuals diagnosed with obsessive-compulsive disorder or agoraphobia were more likely to relapse when they lived in family with high level of Hostility (Chambless & Stekete, 1999; Chambless, Floyd, Rodebaugh, & Stelketeee, 2006), and the relatives of individuals diagnosed with anxiety disorders were found to be more emotionally over-protective (1301) about patients’ difficulties. Furthermore, longitudinal studies have reported there are relationships between the number and/or lengths of hospitalization and the important relative’s higher level of EB (Cnicsm or E01) (Marom et al., 2005; Huguelet et al., 1995; 26 McCreadie et al., 1993; Schulze, Hornung, Stricker, & Buchkremer, 1997). A longitudinal study has shown that the relative’s EE status (high or low) is stable over 5 years in the majority of relatives (63% of the relatives of 32 schizophrenic patients) (McCreadie et al., 1993), reflecting a more fixed attitude toward the patient (Huguelet et aL,l995) In short, high levels ofEE, especially high levels of Criticism, are a prognostic indicator of the course of disorders among adults across a broad spectrum of psychological disorders and chronic physical health problems. Criticism/Hostility is more strongly related and EOI is less strongly related to relapse rates, poorer outcomes of intervention programs for treating disorders, and the course of development of disorders. That is, the predictive power of EB levels in family members of persons with a variety of psychological and medical has been studied both cross-sectionally and longitudinally with consistent results. The BB construct may be a proxy for relatives’ attributional style and their beliefs about how voluntary the symptoms of the disorder are (Brewin, MacCarthy, Duda, & Vaughn, 1991.; Hooley, 1998). Relatives tend to attribute their own personal problems or difficulties to themselves, and assume this is true for others. Relatives with high EB are less flexible and have a low tolerance for family members’ behaviors or difficulties 27 (Barrowclough & Hooley, 2003; Hooley & Campbell, 2002; Hooley & Hiller, 2000). Even though there are inconsistent results between studies with clinically-referred and community samples, many studies have reported enough evidence to suggest that a high level of relatives’ EOI is related to relapse in anxiety disorders. Longitudinal studies demonstrated that either high EE or high level of Criticism by parents or spouse was significantly related to the readmissions and a longer hospitalization, compared with low EE or a low level of Criticism. Children and Maternal EE Maternal Criticism and Extemalizing Behavior Problems Studies of maternal EE have explored children’s behavior problems. A significant relationship between high maternal Criticism and externalizing problems in children has been found by many studies across countries and cultures (Stubbe, Zahner, Goldstein, & Leckman, 1993; Peris & Baker, 2000). One study compared mothers’ level of Criticism among three groups of children ages 6-1 1: 30 children referred for conduct disorder, 30 children for emotional disorders. and 30 children in a control group (Vostanis, Nicholls, 8: Harrington, 1994). Mothers of children referred for conduct disorder or emotional disorders were rated as having significantly higher EB, compared with mothers of non- referred children. Among the three groups, a high level of maternal Criticism 28 discriminated the group of children referred for conduct disorder .from children with emotional disorders and the control group. The level of mother’s Warmth was also significantly different in the three groups. Mothers of children with conduct disorder or psychiatric disorders showed the least Warmth, and mothers of children in the control group showed the most (Vostanis et al.). Stubbe and her colleagues U993) conducted a study among 108 children (6-11) and their mothers. The children’s diagnoses varied: 72.4% had no diagnosis, 14.8% had been diagnosed with a disruptive behavior disorder (ADD, ODD, or CD), and 9.2% carried an anxiety-depressive disorder (e. g., overartxious disorder, obsessive compulsive disorder, Major Depression Disorder). Among 108 mothers, 31 (23.3%) mothers were evaluated as high EE, W'lth 4i. '3‘0 of these 31 mothers characterized as high 1313 Criticism and 52% as high Emotional Over-Involvement. However, only 24% of their children carried a formal diagnosis (Stubbe et al.). However, more than half of children (56.1%) with high EE mothers (based on high levels of Criticism and/or EOI) showed one or more diagnosable conditions, compared with only 18.9% of children with low EE mothers. When only those mothers with high levels of Criticism were considered, approximately 75% of their children met criteria for one or more disorders (based on DSM-lll-R). In contrast, 70% of children who lived with mothers who showed high levels EOI were 29 diagnosed with anxiety-depreSsit'e conditions exclusively and had no co-morbid externalizing behaviors. Furthermore, a longitudinal study confirmed the significant relationship between mother’s high Criticism and young children’s later externalizing behavior problems. A longitudinal Study with children and mothers (N =~' 91) examined the power of maternal EE assessed when their children were in preschool to predict children’s disruptive behaviors in lst grade (n =7 48) and DSM-IV diagnoses in the 3rd grade (r2 = 69) (Peris & Baker, 2000). The original sample was skewed in that children with parents who specifically reported externalizing or internalizing symptoms were overrepresented. The disruptive behaviors were measured by the mothers’ report on the Child Behavior Checklist (CBCL) and the lSt grade teacher’s report on the Teacher Report Form of the CBCL (TRF). Mothers‘ high levels Criticism, but not her level of EOI, at preschool was related significantly to children‘s externalizing problem behaviors in the lSt grade. A majority (72%) of preschool children with externalizing behavior problems had mothers classified as high (Criticism 2 years earlier. However, a child’s internalizing problems were not related to the mother’s EE status. In the 3rd grade, 69 of the original sample 0191 children were assessed for DSM- I V disorders. Of the 69, 35 met criteria for one or more diagnoses, reflecting the skew in 30 the recruiting procedure. Of the children whose mothers were rated as high EE at preschool, 64.3% met DSM-I'V criteria for Attention Deficit/Hyperactivity Disorder (AD/HD). Maternal stress and children’s behavior problems at preschool independently explained 17.7% and 16% of the variance, respectively, for the externalizing behavior problems at 3rd grade. Maternal EE rating at the time when children were in preschool explained an additional 7.7% of the variance. Specific EE Components (Criticism and Emotional Over-Involvement) as Predictors of Internalizing and Extemalizing Behaviors There are many studies strongly indicating the relationship between high levels of EB Criticism and externalizing problems in children and adolescents, as well as several studies demonstrating the relationship between high EB Criticism and internalizing problems (Asamow, Tompson. Woo, & Cantwell, 2001; Hirshfield et al., 1997; McCarty, Lau, Valeri, & Weisz, 2004; McCarty & Weisz, 2002; Stubbe et al., l993; Vostanis et al., 1994). High EOI has been reported to be specific to anxiety disorders in Western and Japanese studies (Chambless & Stcketee, 1999: Stubbe et al., 1993; Yoshida, 2001; Hirshfield, Biederman, Brody, F araone, & Rosenbaum, 1997). Only one EE study conducted among Japanese adolescents reported a significant relationship between high maternal E01 and eating disorders in adolescents (Yoshida, 2001). The results suggest 3l that the specific maternal attitude represented by a higher level of Criticism or higher level of EOI may be more stressful to children who are particularly vulnerable to a specific disorder, anxiety disorders vs. Oppositional Defiant Disorder for example (Asarnow, Godstein, Tompson, Guthrie, 1993). Criticism as a predictor of internalizing and externalizing disorders. McCarty et al. (2004) conducted a study among 252 children and adolescents (7-17) referred to a clinic. They examined (a) the relationship between maternal EE status and general behavioral characteristics of morhers directed towards their children, and (b) the specific relationship between EE components, Criticism and E01, and children’s diagnoses. Within the sample of children, many were diagnosed with more than one disorder. Behavior disorders were most commonly diagnosed in this sample (disruptive behavior disorders, 52.1%; AD/HD, 37.0%), although over a third of the sample were diagnosed with depressive disorders (33.3%) and/or anxiety disorders (37.6%). Mothers with high BB were more likely to make antagonistic and negative comments, and express disgust toward their children, compared to mothers with low or marginal levels of EB. In addition, low EE mothers interacted with their children with less harshness than high EE mothers. 32 Similar results were found in an earlier study (McCarty & Weisz, 2002). Mothers who were very critical of their children tended to perceive their children as having more externalizing behaviors or as being more problematic. However, these very critical mothers did not perceive their children as having internalizing behavior problems. The mother’s focus on her child’s externalizing behaviors (aggression, high-risk or delinquent behaviors) would be reflected in her more extreme ratings of externalizing behaviors at the cost of observing the internalizing behaviors (withdrawal, somatization) that her child also might be displaying. EOI and child disorders. There is contradictory evidence as to whether EOI is specific to particular behavior problems in children. Some studies found the relationship between relatives” high E01 and internalizing behavior problems, especially anxiety disorders (Hirshfield etal., 1997; Stubbe et al., 1993', Chambless & Steketee, 1999', Yoshida, 2001), but other studies failed to find the relationship (McCarty & Weisz, 2002', McCarty et al., 2004). In two different studies by McCarty and her colleagues, mothers’ high level of B01 was related neither to the children’s internalizing nor externalizing behaviors. These findings contradict those of Stubbe et al. (1993). In Stubbe et al.’s sample, 70% of children who lived with mothers showing high levels of EOI were diagnosed with anxiety-depressive conditions exclusively and had no co—morbid 33 externalizing behaviors. The difference in results may lie in the samples: the children in the two McCarty studies were all clinically referred, whereas only 28% of the Stubbe et al. sample had been referred to a clinic. In addition, McCarty et al. (2004) suggest two reasons: low construct validity of diagnoses with children and adolescents who are clinically referred and the difficulty in detecting maternal behaviors corresponding to EOI. Similar concerns can be raised about issues of diagnosis and definitions of criticism and emotional over-involvement in the Japanese culture. Linking maternal criticism and child problems. There are several possible explanations for the relationship between maternal EE and childhood emotional and behavior problems. “The combination of particular child vulnerabilities and EE attitudes, however, may lead to poor outcomes” (Asarnow, Tompson, Hamilton, Goldstein, & Guthrie, 1994, p. 130) reflects the stress-diathesis model. That is, negative maternal attitudes expressed as high EE create stresses for children who are especially vulnerable to some childhood disorders, such as mood, eating, and anxiety disorders (Asarnow et al., 2001, Butzlaff & Hooley, 1998; Coiro & Gottesman, 1996;. Hooley & Gotlib, 2000', Hirshfeld et al., 1997', Nelson et al., 2003). For example, Hirshfeld et al. reported that the behavioral inhibition of children suspected to be at risk because their biological family members were outpatients with anxiety or depressive disorders was strongly related to 34 maternal Criticism. The development of socio-emotional competence may be negatively influenced through such intense EE interactions (Seifer, Baldwin, & Baldwin, 1992). A second explanation is that Criticism in an interactional pattern occurring on a daily basis can be interpreted as a form of social threat. Family member’s critical remarks and psychological, cognitive, and physical symptoms displayed by people who are vulnerable to depression are linked through the nervous system. Criticism is especially difficult to handle for people who are vulnerable to depression (Hooley & Gotlib, 2000; Hooley. Gruber, Scott, Iiiller, & Yurgelun-Todd, 2005). A psychosocial event and maternal criticism are so stressful or overwhelming to the target person that he or she reacts with biobehavioral symptoms (Hooley et al., 2005). Neurological evidence of the impact of criticrsm was found in a study by Hooley et al. (.2005). Adult. patients with unipolar depression were asked to listen to two different audiotapes . One tape was the critical remarks by the patient’s own mother and the other was her remarks praising the patient. The outcome variable was the change in the activity of the dorsolateral prefrontal cortex (DLPFC), a brain region known to be central to the integration of cognitive and emotional information. The level of activation in DIPFC of patients who had a history of major depressive episodes significantly decreased after they listened to their mothers” criticism. The activation level in DLPFC stayed the same as they listened to their 35 mothers’ praise. The control groups showed the same activation levels for both criticism and praise (Hooley et al.). This study provides a neural model for relapse in depression. Maternal criticism and emotional over-involvement linked to emotional or behavioral disorders. A third possible explanation for the relationship between maternal EE and emotional or behavioral disorders comes from the Double Bind theory. In this conceptualization, the depressed child is trapped between the mother’s high level of Criticism and her EOI with the child. Based on the Double Bind theory, when their mother expresses both high levels of Criticism and E01, the child is unable to respond to the critism and the E01 at the same time, creating a kind of paralysis . Mothers with high EOI or both high E01 and Criticism make more disturbing statements (Strachan, Goldstein, & Miklowitz, 1986) or more confusing, ambiguous, and unclear statements (Hubschmid & Zemp, 1989). The child is told she is doing everything wrong and, at the same time, told how much szhe is loved and cherished by the morher (Nichols & Schwartz, 1998). This kind of environment leads to psychological confusion. Transactional theory: Reciprocal negative exchanges. A fourth explanation comes from Transactional theory (Sameroff & Chandler, 1975). The mother is influenced by her child’s difficult behaviors and becomes very negative toward the child. In response to the child’s behavior, the mother becomes irritable and is increasingly critical of the child. 36 Reciprocally, when the child is criticized by the mother, s/he reacts negatively to the parent (McCarty et al., 2004), further fueling the cycle. In this environment, the child is likely to show increasing levels of both internalizing (depression, withdrawal. somatization, anxiety) and externalizing (aggression and delinquency). Contribution of Maternal Depression to Children’s Behavior Outcomes: Possible Mediating Effect of EE Research evidence makes it clear that maternal depression is one of the most important predictors of negative developmental outcomes and dysfunction for children . (Feng et al., 2009:. Buehler & Welsh, 2009; Hirshfeld et al., I997; McKee et al. 2008). When the mother is depressed, all aspects of parenting are limited—supervision, teaching, communication. monitoring. etc. (e.g., .DeGarmo, Patterson, & Forgatch, 2004). Compared to children of mothers with no history of depression, children of mothers who are clinically depressed or who have a history of depression are at high risk for having internalizing and externalizing behavior disorders (Davies, Dumpenci, & Windle, 1999; Schwartz, Dorer, Beardslee, Lavori, & Keller, 1990). One of the potent processes affecting these children is that a depressed parent is less likely to provide supportive communication (responsiveness and connectedness to the child) and behavioral control (regulation of the child’s behavior through firm and consistent 37 discipline). Interaction between mother and child becomes negative, characterized as less praise, less effective supervision, less consistent discipline, but more conflict, reflecting criticism, rejection or withdrawal, and/or disengagement. The research literature also demonstrates that children of mothers with depression also are more likely to be vulnerable to emotional or behavioral disorders. In terms of the Diathesis-Stress model, there are several potential sources of stress when living with a mother with depression. There is evidence that a mother with depression has less control of her negativity toward her child. Cognitively, she tends to perceive her child as having more externalizing behaviors than another observer would find (McCarty et al., 2004). Depressed mothers are more likely to “perceive causes of their children’s negative behavior as stable, personal/idiosyncratic and controllable by the child and simultaneously showed. a greater tendency to perceive themselves as a cause of their children’s negative behavior" (McCarty et al. ,2004, p. 90’). This formulation has been shared by other researchers (Bolton et al., 2003; Fergusson, Lynskey. & Horwood, I992; Hamish, Dodge, & Valente, 1995). A negative view of her child is expressed in critical and hostile remarks and behavior. At the same time, the depressed mother is more likely to blame herself for problems she sees in the child. To compensate, she may become emotionally over-involved with the child, becoming intrusive and overly pretective. 38 Also when a mother is depressed, she lacks the energy to provide proper care, discipline, support, and supervision of her child. Multiple effects of this lack of energy are evident in the child developing disturbances of secure attachment (Ainsworth et al., 1989), becoming more likely to adopt a coercive style in interacting with the mother (Patterson, 1982), moving through the neighborhood and community without supervision, and engaging in inappropriate and antisocial behaviors (DeGarmo & F orgatch, 2005; Patterson, Reid, & Dishion, 1992). In this situation, the child also fails to develop a sense of self-competence and uses a negative cognitive style to process life events. From a behavioral perspective, this pattern of simultaneously blaming their child and perceiving themselves as the cause of the problems is consistent with the pattern seen in mothers of adult patients with Other disorders (cg, schizophrenia) (Barrowclough & Hooley, 2003). This “child-blaming” (Bolton et al., p. 242) is related to harsher parental disciplinary responses as well (Joiner & Wagner, 1996; Smith-Slop & O’Leary, 1998), something that would be expected to increase the child’s negative behaviors. Three BF. studies included mother’s depression as a variable in examining the linkages among depression, EB, and children’s behavior problems. Bolton et al’s study (2003) examined the relationship among a mother’s depression. her EE, and attributional Style and her child’s behavior problems with 61 mothers and their children. Morhers who 39 scored higher on Beck Depression Inventory showed higher levels of EB (either Criticism or EOI), but lower warmth. Mothers with high EE and low EE showed significant differences in the pattern of their attribution of their children’s behavior problems. Depending on the specific EE element, attribution varied among mothers with high EE. Mothers with high Criticism were more likely to believe that child could (and should) control his or her behavior and attributed the child’s behavior problems to child him/herself. On the other hand, mothers with high EOI tended to perceive their children’s problems as their fault, blaming themselves and saying that they failed to control children’s negative behaviors. Maternal depression was a significant predictor of high maternal EB. Further, maternal EE mediated between mother’s depression and mother’s rating of externalizing problems for children. Another study (Nelson et al., 2003) focused on the association between mother’s depression or EE and the behavior problems of 800 15-year-old adolescents. Mother’s high BE Criticism partially mediated between mother’s depression and children’s behaviors and functional impairment. Mother’s depression and maternal Criticism were also independent predictors of children’s behaviors and functional impairment. B01 was not a predictor of either the adolescents’ behaviors or funcrional impairment. In a third study, the relationships among maternal Criticism, maternal depression, and externalizing behavior problems in I94 early adolescents aged 11-12 were examined (Frye & Garber, 2005). The new finding in this study was the “child-effect model” (p. l): adolescent externalizing behaviors in 6Ih grade significantly predicted maternal Criticism in 8th grade. EOI in the 8th grade was not predicted significantly by the 6'h grade externalizing behaviors. In summary, mothers with high levels of depressive symptoms give more critical and hostile comments (reflecting to a high level of Eli Criticism) and less warmth to their children because they perceive the difficulties as in children. i.e., the child’s internal or personal problems. Mothers with depression and high levels ofE OI blame themselves for the child’s problems. Both groups of mothers tend to rate children’s behaviors more negatively. but both mother’s depression and high EB levels also are independently predictive of children’s behavior outcomes. Three factors, mother's depression, her EE. and her causal attribution combine to influence her child’s behavior. However. how the mother’s depression and E13 are related or how these two factors interact together as negative contributors has not been clarified adequately. There have been debates about the relationship between the construct of EB and maternal depression, and EE. constructs have been. seen as a proxy for the symptoms of maternal 4i depression. In one study, maternal depression predicted all the child’s problems, including internalizing and externalizing problems and functional impairment (Nelson et al., 2003). However, the evidence for such a strong relationship between BB and parental depression has been contradicted by the results of a study by McCleary and Sanford (2002) who found that depression and maternal BB Criticism independently predicted children’s externalizing behavior problems and functional impairment. Two models have been proposed to explain the relationships among maternal depression, BB, and children’s behavior problems, one is additive, the other meditational. The additive model proposes that children of mothers who had/have past or current depression and high BB are three times more likely to have emotional behavior problems than children of mothers who only had/have past or current depression with low or medium levels ofBB (Schwartz et al., 1990). Consistent with this additive model is the finding that EB criticism and maternal depression independently contributed to children’s externalizing behaviors (McCleary & Sanford, 2002). On the other hand, the meditational model argues that maternal BB is a mediator for the relationship between maternal depression and child behavior problems (Nelson et al., 2003). One study has been conducted to examine the influence of BB on children’s behavior problems by sampling monozygotic twins (age 5) to eliminate biological 42 differences as a confounding variable (Caspi et al., 2004). The result showed that there is a longitudinal, systematic effect of different maternal BE on the behavior problems of monozygotic twins. The twins displayed different behavior problems that reflected the different characteristics of matemal BE directed toward each twin. Also, differences in behaviors in twin siblings at age 5 were stable at least to age 7. Although monozygotic twin siblings carried the same genetic makeup and grew up in the same family environment (Bronfenbrenner, 1979), each twin was differently influenced by maternal emotional attitudes (Caspi ct al.. 2004). This suggests that the maternal EB creates only a part of the overall family environment, but has an independent effect on children’s behavior problems (Ilirshfeld et al., 1997). Maternal BB is a child- speciflc aspect of mothers’ cmorional attitude toward the target individual child and independently affects the presence, course and outcome of children’s behavior problems (Bolton et al., 2003). EE Research with Japanese Samples Review of Japanese EE Studies Research with adult Japanese samples. In Japan, as in the rest of the world, the maJOI'ity of BB studies have been focused on the relationship between the relatives’ BB status and the patient’s risk of relapse. Studies have used the CPI or the FMSS translated 43 & Inoue, 1995), mood disorders (Mino, Inoue, Shimodera, & Tanaka, 2000; Mino et al., 2001; Uehara, Yokoyama, Goto, & Ihada, 1996; Tanaka et al., 1995; Mino et al., 1997). depressive symptoms (Mino et al., 1998), social functioning (Inoue et al., 1997) and, uniquely, the deterioration in symptoms of dementia (Nomura et al., 2005). Results of these Japanese studies with the CF I or F MSS have paralleled those of Western studies. The four published papers examining family members’ BB status and adult patients with schizophrenia are a series based on the same 52 adult patients with schizophrenia and their relatives. The relatives included parents ( father and/o.r mother), Spouses (wives or husbands), and other important household members. Findings were consistent with Western BE. studies. The key relatives’ EB status is strongly related to the risk of relapse in adult patients. The first BB study in Japan demonstrated that the 9- month relapse rate was significantly higher in patients who were from high BB families than those from low EB families ( T anaka et al., 1995). Using data from the same sample, Mino et al. (I 998) examined the changes in Symptom levels and relatives’ BB status at a 9-month follow-up after discharge from the 44 hospital. Specifically, they examined the association between relatives’ EE status and changes in the “negative symptoms” levels (emotional withdrawal, motor retardation, expected activities and free-time activities) was significantly higher when relatives were in the high BF. group (Inoue et al., 1997). Followed by Tanaka‘s study, Mino et al. (1997) conducted a 2-year follow-up study to examine the relationship between BE and the risk of patients‘ relapse over 2 years. Similar to Western studies, 71% of patients who lived with family members with high EE status relapsed, but only 37% of patients who lived with family members with There are two studies examining the association between EF. and the course of 32 Patients with mood disorders (Mino et al., 2001 )and 40 patients 45 with major depressive disorder (MDD) (Uehara et al., 1996). Across two studies, key relatives’ high level of EB was significantly related to a high relapse rate in patients. Mino et al. found a remarkably low level of Criticism in the families with mood disorders, compared to families of patients with schizophrenia. Uehara et al. suggested that a previous history of depressive episodes added another risk factor for relapse. lmportantly, the frequency of Criticism was significantly lower in Japanese families as compared with families of patients with mood disorders in Western countries. Mino et al. concluded that a lower cut-off point was needed to evaluate high- or low-BE in Japanese samples, reflecting the cultural differences. Research with Japanese adolescents. There is only one study with Japanese adolescents and their mothers (Yoshida, 2001). The study examined the characteristics of the families of 25 adolescents with eating disorders using the CFI. The mean age of the adolescents was 18.6 years. Of the 25, 16 were diagnosed with anorexia nervosa and 9 were diagnosed with bulimia nervosa. Results showed that 44% of the mothers had high levels of EOI. Mothers were more likely to sacrifice themselves in attempts to protect their child. When compared with families of patients with schizophrenia and mood disorders, mothers of adolescents with eating disorders showed significantly higher levels of EOI. When the frequency of Criticism was compared across the three groups, mothers 46 of adolescents with eating disorders had much lower levels of Criticism than did relatives of patients with schizophrenia. Resiliency and the Present Study It is clear that high EE, especially Criticism, in key relatives presents an increased risk of both the onset of psychological and behavioral disorders (Doane, West, Goldstein, Rodnick, & Jones, 1981: Schwartz et al., 1990) and relapse after stabilization (Brown & Rutter, 1966). However, it is important to note that not every adolescent at risk of schizophrenia because of family biological factors and critical parents actually becomes schizophrenic, nor do all children whose parents have a history of depression or other mood disorders actually develop such disorders. Such resilience in the face of serious challenges is a matter of great interest. One possible explanation for the resilience these children and adolescents show may lie within the, child. There is a complex dynamic of interacrion between environment and the individual child which leads to different developmental courses and outcomes (Bronfenbrenner, i. 979). Children who experience Similar adversities in life show different developmental outcomes (multifinality), but on the other hand, children who live in extremely different. environmental conditions develop the same or similar outcome (equifinality). The resilience model focuses more on the positive side of development and 47 helps to understand how some children living in stressful environment maintain their firnctioning and master their developmental tasks. Resilience can be understood as either an outcome characterized by particular patterns of functional behavior despite risk or a dynamic process of adaptation to a risk setting. In an adverse setting. multiple risk factors and protective factors interact with one another (Olsson, Bond, Burns, Vella-Brodrick, & Sawyer, 2003). Resilience as an outcome is exemplified by good mental health, maintained functional capacity, and social competence. On the other hand. resrlience as a process focuses the mechanisms or processes of the path led toward the certain outcome or of acts that modify the impact of a risk setting. That is, resilience is the developmental process by which young people successfully adapt. The mechanism of resulience occurs in a process of interactions of risk and protective mechanism in a risk setting (Olsson et al.). In this model, resilience is a cognitive-emotional resource. These resources have been hypothesized to include self- esteem and sense of coherence. Mechanisms of the Joint Contribution of Environmental Factorsand Child Resiliency to Child Behavior Outcomes: Deficit Models vs. Resilience Models Diathesis-Stress model and child behaviors. The Diathesis-Stress model considers the potential contribution of the interaction between individual vulnerability 48 (including heredity. cognitive, and psychosocial vulnerabilities) and environment to the development of problems, and “it provides an important heuristic for the formulation of research questions, while at the same time providing a conceptual structure within which the meaning ot‘research findings can he evaluated" (Richters & Weintraub, 1990, p. 70). For example. depression in children is seen as the result of the combination of stresses and vulnerabilities. Their interactions trigger children’s internalizing or extemalizing behaviors. There is a genetic predisposition to depression which will be expressed only if the environment. is sufficiently stressful. The Diathesis-Stress model has been also used to explain the relationship between “ maternal EE and children’s behavror problems (internalizing and externalizing behavior problems) (l-looiey 8: Gotlib, 2007; Miklowitz, Goldstein, Falloon, & Deane, 1984). Subjected to frequent maternal criticism and expressions of hostility. children who are vulnerable to emotional disorders and/or behavioral disorders are more likely to perceive those mother’s negative emotions as stress. Overwhelmed by maternal Eli, they develop Symptoms efrnood disorders and anxrety ("acting in”), and/or behavior problems (“acting out”). Because of the. nature and importance of the relationship with the mother and because of various vulnerabilities, the child is neither free nor able successfully to challenge the messages the mother is sending. Thus. children who have a pattern of 49 negative information processing (cognitively negative attribution) as a result of high levels of maternal EEand who live in a stressful environment are more likely to develop depressive disorders and/or externalizing behavior problems (e. g., Kwon & Laurenceau, 2002; Turner & Cole, 1993; ,Hankin, Abramson, & Silar. 2001 ). However. the Stress-Diathesis model does not give us the final word about the development of psychological disorders and behavior problems because it does not adequately explain the exceptions to the development of maladjustment in children (Richters & Weintraub. 1990). Knowing that a mother is depressed does not automatically mean that her son or daughter fails to develop a positive sense of competence or engages in extreme internalizing or externalizing behaviors. Therefore, this research is guided by a resiliency model that can be used to expiain both unhealthy and healthy outcomes. Resilience model and child behaviors. in the 1970s, a different way of thinking about child development in unfavorable circumstances was gradually achieved through studies of people with disorders like schizophrenia. Researchers started to focus on people with schizophrenic disorder who functioned well in social settings. They were competent in meeting their responsibilities at work, in social reiations, and in marriage (Luther, Cicchetti. & Becker. 2000). Corresponding to this new trend of investigation of 50 positive outcomes of adult patients With schizophrenia, children of mothers with psychological disorders were also studied. Many children maintain their functioning despite their high~risk status, and researchers increasingly examined individual variations in response to adversity. Resilience refers to manifested competence in the context ofsignificant challenges to adaptation or development (Masten & Coatswor’th, 1998). Resilience is defined as “the process of, capacity for. or outcome of successful adaptation despite challenging or threatening circumstances” (Masten, Best & Garmezy, l99l ). Previous studies investigated sue resilience in children who were under multiple adverse conditions. such as socioeconomic disadvantage (Garmezy, 1995, Rutter, 1979; Werner & Smith, 1992), parcntai mental illness (Matser. & Coatsworth. 1998), maltreatment (Cicchetti & Rogosch, l997; Cicchetti, Rogosch. lynch, & Holt, 1993), chronic illness (Masten. 1994). and catastrophic life events (O’Cougherty—Wright, Masten, Northwood. & Hubbard, 1997). Compared to the Diathesis-Stress model (.h'leehl. 1962; Rosenthal, 1963). the resilience model helps to understand individual variations in response to risk factors (Jenkins, 2008: Rutter, 1990; 1993): “a dynamic process encompassing positive adaptation within the context of significant adversity” (Luthar et al., 2000, p. 543). The 51 main characteristics of the resilient child in middle childhood are positive interpersonal relationship in a peer group, self-management otbehavior, and academic performance. It has been widely reported that relationships with caring. socially—responsible adults and higher levels of intellectual functioning in the Child are important elements leading a child to be resilient. Better intellectual abilities play a role in allowing the child to process adverse events in more positive ways to allow for the use of a variety of coping strategies (hopefulness vs. hopelessness, solvable problem vs. insolvable) (Masten,l994). Internal Resiliency Resources in Children Self-competence as a mediator between maternal negative feedback (EE) and childhood adjustment. Self—competence affects not only a child‘s construction of self- concept and sense of coherence, but also plays a role to link the environment (context) and the processes of development of children and adolescents. Negative self- conceptualization is a risk for depressive symptoms because those who see themselves incompetent are more likely to view their world negatively (lacquer, Cole, & Searie. 2004', Cole, Martine, Pecxe. Seroczynski. & Fier, 1999‘). On the other hand, when children acquire a strong sense of self-competence, they are more likely to process difficult situations successfully. Achieving an internalized sense of self-competence is 52 limited by excessive parentall negative feedback, panicularly negative maternal expressed emotion (Brown & Rutter, 1966; Vaughn & Leff, 1976), because children take information about themselves from the environment, “particularly feedback from significant others, as: they construct beliefs about their competencies” (Jacquez et al.. 2004, p. 355). The mother’s perception of her child‘s competence, communicated verbally and non-verbally to her child, is strongly related to the child’s experience of self-competence (Cole, Martin, & Powers, 1997; Cole, Jacqucz, & Maschman, 2001). In middle childhood, children are influenced heavily by their mothers" feedback. The feedback conveys to the child what the morhcr believes the child can accomplish in overcoming day—to-day challenges, which in turn is related to negative representational models of the self (Cicchetti, 1996; Rohner & Britner, 2002). Development and “Sense of Coherence” as a resilience resource. The concept of “Sense of Coherence” was developed within the Salutogenic Model that focuses on the 1While the father’s role in developing the child’s self-competence is acknowledged Widely, the information on parent-child. interactions in the research literature is overwhelmingly based on mother'child data. coping styles of persons who remain healthy in a stressful environment, as opposed to focusing on persons who respond to stress with increased risk of sickness or disease (Antonovsky, 1987‘). The term “salutogenesis,” emphasizes behaviors or factors that are health-promoting rather than focusing on the pathogenic origins ofdisease or poor coping behaviors (Wolff & Ratner, 1999). According to Antonovsky, a person’s sense of coherence is a global c0ping or resilience resource (Johnson, 2004). Children who develop a strong sense of self-competence in childhood lay the foundation for an adolescent and. adult “sense of coherence” (e.g., Glanz. Masltarinec, & Carlin, 2005', Johnson, 2004; Wolff & Rather, 1999). Coherence (or sense of control of one’s own life) is the ability to perceive stressors as manageable, meaningful, and comprehensible, instead of being overwhelmed and hopeless in the face of such events. Comprehensibility is described as the person’s cognitive ability to cope with a stressor. That is, to see the stressor as a problem that can be solved. Manageability is defined the individual’s willingness to use instrumental coping skills to deal with the stress. When the stressor occurs and the person perceives it a comprehensible or solvable, s/he acts to resolve the stressor. Meaningfiilness means that the person has motivational energy when facing stress instead of fleeing the situation or becoming paralyzed. Thus. a person with a strong sense of coherence does not avoid thinking about the stressor, might S4 engage his/her social network for support or change his/her diet to a healthier one, and is willing to address the stressor because s/he believes that a positive outcome is possible (Antonovsky, 1987). The sense of coherence develops in the context the child’s experiences of her/his family (Wolff & Rattler, 1999) and in the wider contexts of school, peer relationships, etc. A strong sense of coherence in an adult is positively related to the experiences in childhood and adolescence that are shaped by “structured role relationships within the family and the emergent personality disposition" (Sagy & Antonovsky, 2000. p. 164,). Thus, early experiences set the stage for a strong or weak sense of coherence that is well established before young adulthood (Glanz et al., 2005). Children exposed to unresolved and uncontrollable traumatic situations (e.g., severe physical abuse, sexual abuse. bombings, war) would be predicted to have a very weak sense of coherence or a high degree of learned helplessness (Maier & Seligman, 1976). In contrast, children who are encouraged and assisted in solving problems or dealing with difficult or traumatic situations would be predicted to have a strong sense of coherence. Active engagement and participation in decision making and emotional closeness to family members are crucial for child to develop a strong sense of coherence (Johnson, 2004, p. 421). The result is the development of a basic trust in life and in oneself during 55 middle childhood. Child-rearing patterns, the family’s ways of socializing the child, influence the development of sense of coherence in a way that leads to the development of meaningfulness (Wolff & Ratner, 1999). For example, when a child engages in activities with the family, the nature of their responses to the child creates a set of experiences that give the child a sense of the “possible.” Severe punishment. or criticism of the child for his/her behavior in these activities decreases meaningfulness. and the child comes to see her/himself as incompetent. Conversely, with positive interactions in the family enVironment, the child can demonstrate self—resourcefulness and knowledge in other contexts, and expects to have positive relationships with others. Sense of Coherence theory is consistent with Bronfenbrenner’s (1979) view that child development occurs through multiple and continuing interactions with environments. There are many ways to describe the attributes of successful children. They are willing to try new things because of a sense of self-efficacy (“I can do it”); they cooperate in play with other children (“I am responsible for my behavior”) because of an internal locus of control and because of a positive self-concept (“I am likeable, others accept me”); and they see problems as solvable, accept failure as temporary, and continue to attempt to master a task (developing sense of coherence). Thus, it may be hyporhesized 56 that the positive family environment helps children develop a strong sense of coherence. In turn, the sense of coherence will affect the child’s general health and behaviors (including internalizing and externalizing behaviors). . Children who fail to resolve this stage of development have intense feelings of inferiority and incompetence in both academic achievement and in developing positive peer relationships (Nelson et al., 2003). Regardless of their prior development, it is important for children to be supported and encouraged by others during this time (Zembar & Blume, 2009). A positive relationship with the child‘s primary caregiver (typically, the mother) is an important source of support for the child. When the relationship is limited by continuing conflict or physical/psychological neglect and"or abuse, the development of the child’s sense of coherence is stunted (e. g.. DeGarmo & Forgatch, 2005; Forgatch & DeGarmo, 1999; F orgatch, Patterson, DeGamio, & Beldavs, 2009) EE and resiliency. Many studies of Expressed Emotion have examined the relationship between maternal emotional attitudes and children’s internalizing and externalizing behavior outcomes without considering child’s resiliency factors. So, even though there is a relationship between maternal EE and children’s negative behavior problems, it does not mean that “the variable has an effect in the absence of other 57 variables, even though it sounds as if that is what it means” (Rutter, 1990, p. 184). Rutter said that “the study of protective processes could throw light on what is involved” (p. 184). .A model of resilience can guide research that examines complex interactions (processes) among protective factors and risk factors. Such research would include different dimensions of contexts. as well as the course of positive or negative outcomes. A model of resilience moves away from a simple model of maternal Criticism and E01 as the cause of children’s negative outcomes. For example, continuing negative experiences in a social context impede positive development of internal resilience in children and adolescents, such as self—competence and a growing sense of coherence. Social context is defined as "a set of interpersonal conditions, relevant to a particular behavior or disorder and external to, but shaped and interpreted. by, the individual child" (Boyce et al., 1998. p. 146). Even one context (e.g., family) consists of multiple dimensions (Boyce et al.), and each of these dimensions interacts with one another in meaningful ways (Bronfenbrenner, 1979). Mother-Child Relationships in Japan The present research addressed the relationships among maternal depression, maternal EE. child resilience factors, and behavroral problems for children in Japan. It is 58 necessary to consider the differences between Japanese and American families in terms of child rearing strategies, behavioral eXpectations, and normative parent and child behaviors. These cultural differences could very well affect the choice of instruments used or the interpretation of the results of the study. Mother-child relationships in the macrosystem. The meanings of the interactions in the dyadic relationship between mother and child are constructed by the culture, subculture, or other macrosystem structures in which the family and the dyadic relationship are embedded (Bronfenbrenner, 2005). The macrosystem consists of the overarching pattern of micro-, meso-, and exosystems characteristic of a given culture, subculture, or other extended social structure, with particular reference to the developmentally instigative belief systems, resources, hazards, lifestyles, opportunity structures. life course options, and patterns of social interchange that are embedded in such overarching systems.(Bronfenbrenner, 2005, p, 101 ). Conceptions of child develoriment in Japan. Researchers have studied Japanese mother-child relationships to understand the socioemotional development of Japanese children (Shwalb, Shwalb, & Shoji, 1996). The formal study of mother-child relationships in Japan began in the l7" century. but much of the contemporary research 59 in Japan hasfocused on comparisons with data from Western studies. For example, American researchers reported that Japanese mothers were “physically closer and more soothing toward their babies” (Shwalb eta1., p. 171) when compared to American mothers. Caudill and Weinsrein (cited in Shwalb et al., p. 171 .) interpreted those child rearing behaviors to mean that “Japanese mothers ‘produced’ less active infants." Chao (1994) has argued that, in. the early stages of child development, the mother in East Asia “provides an extremely nurturing environment for the child by being physically available and by promptly attending to the child’s every need. When children reach school age, the mother provides the support and drive for them to achieve in school and to ultimately meet the societal and familial expectations for success” (p. 11 12). Establishing the mother-child relationship in Japan is influenced by Confucian principles to a large extent (Chao, 1994, Kojima, 1986; Shwalb et al., 1994). The basic idea is that a person is defined by his/her relationships with others. Ideal relationships are structured to define the role of each person in the relationship (child-mother, child-father, husband-wife) and the need for harmonious relationships is emphasized. in a relationship, each person must strive to maintain harmony, based on her/his role in the relationship and level of responsibility for the relationship. 60 Confucian concepts are the basis for the ways in which a mother interacts with her child. Japanese mothers’ attitudes, especially for educating or parenting a child, are also influenced by the popularly~accepted Japanese the theory of the child. There is a general belief that all children are. born with great potential and abilities. Further, children are, in fact, quite similar to one another at birth in temis of their innate characteristics and intellectual abilities (T akata, 198 7). These potential abilities cannot be actualizcd without proper support and guidance in the child’s early years, provided mainly by the mother. Any individual differences that come to exist in children are attributed to external factors, such as the environment, but also including the mother’s efforts with her child. Even though there is a tendency to look for the external factors to account for individual differences in children, there is a general belief among Japanese that the child i s an “autonomous learning organism" (Kojima. 1986, p. 322). Two things are emphasized in child rearing as children move beyond infancy: to value living in harmonious human relationships and to contribute to society through the exercise ofhonesty, hard work, and patience. With the idea of child as an “autonomous learning organism,” mothers are encouraged not to be controlling of the learners (child’s) behavior, but to allow the child to learn to regulate behavior by himself or herself. Japanese mothers” socializations with their children are aimed to cultivate a child 61 who is “sunao,” understood to mean “authentic in intent and cooperative in spirit” (Holloway, 1988; Shimahara, 1986). Such a child has mastered social skills, including emotional maturity, obedience, and social courtesy. In contrast, American mothers expect their children to be. verbally assertive, independent. and uniquely individual (Kitayama, Markus, Matsumoto, & Norasakkunkit, 1997; Markus, & Kitayama, 199l; Takata. 1995). Japanese mothers also put great importance on the harmonious relationship between mother and child. Japanese mothers are informed by their culture to avoid excessive praise of their children because of a concern that their children would become arrogant and disobedient, resisting discipline from their parents, a violation of harmonious human relationship (Kojima, 1986). On the other hand, Japanese mothers may be intensely involved with their children. and may derive a significant amount of life satisfaction from their maternal role. These cultural difference may influence the accuracy of Eli instruments to capture Criticism or EOI The emphasis on cquipotentiality (all children have the same innate abilities and capacities) and harmonious relationships in roles (obedience. social courtesy, maturity in behavior) leads to an emphasis on effort by both Japanese mothers and children. Working hard in school is seen as the primary factor determining academic performance, with less emphasis on the child’s ability (Holloway. 1988). The scholastic achievements of 62 Japanese children are frequently attributed to this strong cultural emphasis on commitment to work hard and perseverance at tasks. Differing perceptions of “healthy” child deveIOpment in the US and Japan. To expand relationships with others is one of the important developmental tasks for all children in middle childhood. However, there is a gulf between the US and Japan as to what are healthy behaviors in middle childhood. American parents believe that child's social initiative, assertiveness, and emotional expressiveness are the part of the signs of maturation. It is desirable for American children to individuate and assert themselves as they grow (Rothbaum, Pott, Azuma, Miyake, & Weisz, 2000). The conflicts between mother and child are natural and to be expected in the process ofemotional development in the US. In contrast, Japanese parents believe that proper socioemotional development in children should be accompanied by an increase in the ability to accommodate others and to obey them. For a Japanese tnother, a child who is assertive is sometimes interpreted as being immature (Rothbaum et al.; Lebra, 1994). The differences in the strategies that American and Japanese parents use to shape a child’s behaviors also reflect the differences in beliefs of what healthy chiid . development looks like. The strategies of American parenting include more direct control attempts, Characterized by more commands, attempts at coercron, punishments and 63 rewards, and other ways of displaying the parent’s authority. This corresponds to the respect for the development of each child’s assertiveness and autonomy. American mothers model in their own behavior and explicitly emphasize to their children that it is important to have their own minds and articulate their positive and negative emotions directly. On the other hand, Japanese parents attempt to avoid direct confrontations and contests of will. The strategies Japanese parents are likely to use are indirect expressions of disapproval, such as refusing to speak to the child, being apparently indifferent to her/him, or shunning the child (Aznma, 1996, Jonson, 1993‘) Japanese parents use “indirect and psychological methods to control their children, reasoning, gurlt and anxiety induction, shaming, modeling, and appealing to the child's feelings and desires" (Rothbaum et ai., 2000). These. communications would not necessarily be picked up as BE by W estem standards. _ American mothers model in their own behavior and explicitly emphasize to their children that it is important to have their own minds and articulate their positive and ncgativc emotions directly. Japanese mothers encourage their children to have empathy and receptivity to others, almost to read the other's mind. The emotional verbal exchanges that occur between Japanese parents and children are likely to be quite indirect ' ' ' w‘ ,- q . ~ ' n 4 and may be difficult for the child {or a non-Asran observer) to interpret or u. ttletstand .he 64 parents’ intentions (Azuma, I994; Minami & McCabe, 1995). Instruments based on Western beliefs about parenting and proper child behavior may not fit Japanese families. Construction of the self and self-esteem of children. The different ways of a “self” is construcred in the US and Japan have been noted in the literature (Taniguchi, 2005). The notion of selfis Constructed within the Japanese cultural system values, emphasizing the importance of “maintaining, affinning, and becoming part oi'signiticant social relationships” (Kitayama et al., 1997). The development of selfproceeds. based on which aspects of the self the cultural system is organized to foster and promore. The cultural characteristics. such as individualism vs. collectivism, shape the beliefs and values of how people improve and develop the self. People in different cultures take different ways to maintain and enhance an overall evaiuation of the self. The self in Japan is understood to be a part of the family members in addition to individual. That is, the self (‘ffubun”) does not exist without the group/family to which that the person belongs. In the US. the self is perceived to exist by itself (Rosenberger, 1992, Tani guchi). Thus, "self” has very different meanings in the IFS and Japan, an individual t US) versus an individual in the context of family and community (Japan). The construction of the self he tvvecn the US and Japan is different. almost directly the opposite. European Americans focus on positive self-relevant information, “self- 65 enhancement,” that values positive characteristics and abilities for personal success. On the other hand, in a collectivist culture like Japan. people use “self-criticism.” to understand, or at least explain, personal successes in terms of effort or luck and to account for failure in terms of a lack of ability or talent (Kitayama et al., l. 997; Holloway, 1988). In traditional Japanese culture, people are sensitive to negative sclllrelevant information. These cultural differences may be relevant to the impact of maternal criticism on self-esteem Based on the Japanese beliefs about children, the Japanese emphasize effort (Holloway, 1988; Holloway, Kashiwagi, Hess, & Azuma, 1986) over ability. The word “competence" is not defined solely as unusual talent or genius. but also as the capacity for hard work and persistence (l-iolloway). Japanese mothers emphasize commitment and perseverance more than American mothers because Japanese mothers believe that the notion of effort includes a positive orientation toward the intrinsic benefits of such persistence (Holloway). Those ideas tend naturally to structure the interaction pattern between mother and child as a tmnsactional system where mother and child pursue a goal together and both are more critical of the self. (Kojima, 1986) Thus, compared to children in the US, Japanese children emphasize their weaknesses and make more negative internal attributions than children in the is 66 However, thisself-criticism is not necessarily an indication of low self-esteem or something to be avoided or overcome; rather, it has positive social and psychological consequences (Kitayama & Masuda, 1995; .Markus & Kitayama, 199T ). The mother’s values, beliefs, and expectations for the child. provide a framework for the mother—child relationship, leading the child to become a functional member of a collective society. Thus, measures of self-esteem that work well in Western cultures may net be as valid in Eastern cultures. Japanese mothers tend to emphasize on the child’s effort as a key factor for achievement or success, rather than lack, of ability (Hayami, 1981). Because of this, they also tend to focus more on unrealized abilities. This focus may result in more critical comments toward the child, rather than praising himzher for accomplislunems. Hayami (1984) found a strong relationship between amount of the effort by the child and her/his feelings of pride in a successful performance. However, the question of how much the child’s self—criticism and mother" s verbal and nonverbal communications emphasizing effort and perseverance influence. child behavior problems remains unclear at this point. In summary, the role of Criticism and the meaning of emotional involvement in terms of child self-esteem will reflect the culture. There may be a unique combination of Criticism and Emotional Over-involvement in Japanese families that. will not be assessed 67 in the present study. It is beyond the scope of this research to identity the cultural nuances of these communication patterns. Therefore, for the purposes of this study. an adjective checklist assessing the Criticism component of EB can be used. However, there are no culturally relevant assessments of [501. Research Questions and Hypotheses Specific Aim 1: Examine the relationships among maternal depression. maternal Criticism and Involvement, family relationships. children’s internal resiliency factors (self-esteem and sense of coherence), and children’s internalizing and externalizing behavioral problems H”; There Will be a significant relationship between a negative family environment and child behavior problems. HL 7,: There will be significant negative relationships between child resiliency resources (SE and SOC) and child behavior problems. H; 3: There will be a significant relationship between a negative family environment and child resiliency resources. ()8 Specific Aim 2: Examine whether EE components, either high levels of Criticism or Involvement(. are associated differentially with the Internalizing or Extemalizing behavior problems. Hm: There are positive relationships between a high level of Criticism and a low level of Involvement and child Internalizing and Extemalizmg Scores. H228: There will be significant differences in the child perception of her/his mother’s Criticism and Involvement between children in normal-risk and high-risk behavior problem groups. Hub! There will be significant differences in the mother‘s perceptions of her own Criticism and Involvement directed toward her child between children who are. in normal-risk and high-risk behavior problem groups. “new: Mother and child perceptions of mother’s Criticism specifically predict child Extemalizing behavior problems. “2.2mm; Mother and child perceptions of mother’s Involvement specifically predict child’s Internalizing behavior problems. H233: Mother and child perception of mother’s Criticism specifically predict Extemalizing behavior problems in 3 hi gh-risk Extemalizing group. 69 H231): Mother and child perceptions of mother’s Involvement specifically predict Internalizing behavior problems in a high-risk Internalizing group. Specific Aim 3: Examine the moderation effect and mediation effect ofchildren’s Sense of Coherence (SOC) and Self-Esteem (SE) in the relationship between maternal Criticism and Involvement and child behavior problems H34: Child Sense of Coherence (SOC) and Self-Esteem (SE) moderate the relationship between family environment factors and lntemalizing and Extemalizing behavior problems. H33: Gender moderates the relationship between child perceptions ofmother‘s Involvement and Criticism and Intern alizing and Extemalizing behavior problems. H333; Child Sense of Coherence (SOC) and Seif~Esteem (SE) med iatc the relationship between child report of mother’s Involvement and. Internalizing or Extemalizing problems. H331): Child Sense of Coherence (SOC) and Self-Esteem (SE) mediate the relationship between child report ofmorher’s Crrtrcrsm toward the child and Internaiizing or Extemalizing problems .70 Specific Aim 4: To test the model. fit with the data with. Structural Equation Modeling. H“: An adequate model can be developed that describes the relationships among family environment, child resiliency, and behavior problems (Figure 2.1). Child Outcomes Maternal Distress Internal izing Family Relationship Climate Maternal / . . Criticism “WWW / \\ . \ Maternal Emotional / Over-Involvement \ / \ \\ l/ \ / \ \\ ,1 \ \ ,l/ \ ‘\\\ // ‘\ / Sense of Self-Esteem Coherence Child Resilience Factors Figure 2.1. Proposed structural equation model for study. 72 CHSPTER III METHODS This study used a cross-sectional survey design with a sample of 5th and 6th-grade children and their mothers in Okinawa Japan. The study investigated the relationships among family relationship factors, child’s resilience factors, and child’s psychOlogical and/or behavioral problems. Children were recruited from 5 public elementary schools in Okinawa. The survey for children was conducted in classroom settings with the principal’s permission and the cooperation of the teachers in the 5“1 and 6'h grade classrooms. Participants Children. There were 285 participants in this study (136 boys and 149 girls). F ifih and 6th grade Japanese children and their mothers were recruited from 5 elementary schools in Naha City, Okinawa, Japan. The researcher obtained positive permission from the child’s mother for herself and/or her child to participate in the study. The researcher also obtained an assent fi-om the child before administering the child questionnaire. Mothers. Among 469 returned questionnaires, 80 mothers who consented to fill out the questionnaires for themselves did not consent to their child’s participation, and 73 110 mothers who consented to their child’s participation to the study did not consent to their own participation. The mother’s average age was 41.7 years old. Rates of return. Table 3.1 summarizes the rates of return for the questionnaire packets sent home to the mother, the rate of positive consent for participation in the survey for children and mothers (based on the number of packets returned to school), and the actual participation rates of children and mothers (based on the consent rate). The final number of mothers and children dyads participating in this study was 285, and all analyses are based on this dyadic sample. Demographic Characteristics The demographic characteristics of the participants are presented in Table 3.2. The majority of children in the 5th grade were 11 years of age; the majority of those in the 6th grade were 12. The family characteristics are presented in Table 3.3. According to national Japanese census data, the average number of children in a family is 1.34. However, the number of children in the participating Okinawan families was larger, with 57% of families having 3 or more children in the home. The majority of children (84.9%) lived with both parents (including those living with both parents in extended families). In keeping with traditional Japanese culture, 25.7% of these two-parent families lived with grandparents in a three-generation household. A majority of mothers and virtually all 74 fathers were employed outside the home (Table 3.4). The fathers were more likely to have higher status employment, to have full-time employment, and less likely to be “contract” (non-permanent) employees than the mothers. Procedure-Children’s Data Recruitment. Once the participating schools had been identified, sealed and code-numbered envelopes were sent home via the 5th and 6th graders to be delivered to their mothers (or primary female caretaker). The, envelope included (1) a description of the proposed study, (2) a consent form including three requests for consent, and (3) a copy of the mother’s questionnaire (Appendix A). The first consent request was for the mother to allow the child to complete a packet of questionnaires at school. The second consent request was for the mother to indicate her willingness to complete a questionnaire packet of her own. The third consent request was for the mother to be approached by the investigator to be interviewed. A second envelope, addressed to the investigator and with the same number as the original envelope, was included to return the mother’s questionnaire with the signed consent form. The data from a child and mother in the same family were assigned the same number with letters C and M added, respectively. This precaution was taken for protection of the privacy of the participants and to be able to combine the two records for 75 the data analysis. Each set and 6th grade homeroom had a sealed box with a slot in it to allow the children to return the envelopes. Numbering the return envelope allowed the investigator to determine the gross return rate (percentage of all envelopes returned, Table 3.1). The consent form permitted a count of the number of mothers giving positive permission for their child’s participation and for their own willingness to fill out a questionnaire themselves and/or be interviewed. Data collection. Two weeks were allowed for mother’s questionnaires to be returned to the box placed in the classroom, and teachers were asked to encourage that their students return the envelopes even if the mother was denying consent for participation in any phase of the study. After two weeks, the envelopes were collected by the investigator, and a list was constructed of the children with permission to participate. Questionnaires with numbers that matched the numbers on the consent form were prepared and administered in a classroom in the school building at the end of the school day. Children without positive parental consent were dismissed. At the time of administration, the investigator gave a brief presentation and asked the children to assent to participation (Appendix B). Children who did not assent were dismissed at that point. Questionnaires were distributed after collecting the assent forms from children. F orty- five minutes was allotted for the child to complete the questionnaire. 76 Instruments-Children The measurements used in this study for children were the Youth Self Report (YSR), the child and adolescent version of the Achenbach Child Behavior Checklist (CBCL) forms, the two-dimensional (Self-Liking and Self—Competence) Self-Esteem Scale (SE), the Sense of Coherence Scale (SOC), the Expressed Emotion Adjective Checklist (EEAC), and the Family Relationship Inventory (FRI) (Appendix C). Youth Self Report (Y SR). The Japanese version of the Youth Self Report (YSR) was used for this study (Achenbach, 1991, 2000; Itani et al., 2001). The Japanese YSR consists of 112 items. The 112 items are answered on a 3-point Likert scale ranging from 0 (Not True), 1 (Somewhat or Sometimes True), and 2 (Very True or Often True). To create the Japanese YSR from the English YSR, English native speakers proficient in Japanese translated each item. After the English version of YSR was translated into Japanese, it was back-translated into English by Japanese-English translators for quality assurance (Itani et al.; Tejima et al., 1994; Tejima et al., 1995; Tejima et al., 1996). Norms for Japanese children 11-17 are well established, as are satisfactory indicators of reliability and validity of the Japanese version (Itani et al.). Japanese children (1 1-17) report fewer lntemalizing and Extemalizing behaviors, in general (Itani etal., 2001) than US and other Western samples. Therefore, the cut-off 77 points to differentiate normal, borderline clinical, and clinical groups for the Japanese version of the YSR are set differently from the American standardized cut-off points. In US samples (Achenbach, 1991), the cut-off point for the “normal” or nonclinical group is a T-score of less than 67 (94th %—ile), 67-70 (95 - 98th %-ile) for the borderline clinical group, and over 70 (over 98th %-ile) for the clinical group. In Japanese samples, a T-score over 59 (84th %-ile in US samples) is used as the cut-off point for discriminating between normal-risk and borderline clinical groups, and over 63 (90% in US samples) is used as the cut-off point for the clinical group for lntemalizing and Extemalizing scales (Itani et a1; Tejima et al., 1994; Tejima et al., 1995; Tejima et al., 1996). This study used the cut- off points that Itani and her colleagues identified in their study. Children with Internalizing or Extemalizing scores above 59 were considered to be in high-risk lntemalizing or Extemalizing behavior problem groups. In the Itani et al. (2001) study, Cronbach alpha values for 7 of the 8 YSR scales (except Thought Problems) were found to have acceptable internal consistency (Cronbach a—Intemalizing scales: Social Withdrawal, a = .74; Somatic Complaints, (1 = .67; and Anxiety/Depression, a = .83; Extemalizing scales: Delinquent Behavior, (1 = .67; and Aggressive Behavior, (1 = .89; Scales not used on lntemalizing or Extemalizing scales: Social Problems, (1 = .74, Attention Problems, (1 = .78; Thought 78 Self-Liking/Self-Competence Scale-Revised Version (SE). The SE is a 16-item self-report scale assessing two distinct aspects of self-esteem: self-liking and self- items assessing self-competence and 8 items assessing self-liking. Self-competence refers to the perceived positivity or negativity toward one’s self as an inner resource of power and efficacy. Self-liking refers to a perceived evaluative experience of one’s self as a social object, recognizing a good or bad person in one’s self and an overall sense of self- Worth as an individual with social significance. Examples of items on the Self- Competence scale are “I am highly effective at the things I do” and “I wish I were more skillful in my activities (reverse scored)” Examples of items on the Self-Liking scale are “It is sometimes unpleasant for me to think about myself (reverse scored)” and “I never doubt my personal worth” (Tafarodi & Swann, 1995a). The two subscales use a 5-point Likert scale ranging from 1 (absolutely applies to me) to 5 (doesn ’I apply to me at all). Higher scores on the SE are taken to mean that the child has higher self-esteem. 79 According to BIOWn’s study (2008), the Cronbach alpha was .89 for all 16 items. The internal reliabilities of the Self-competence and the Self-liking scales were reported as .80 and .82, respectively (Brown). Cronbach alpha was .82 for the 16 items for the present study. variable that provides a ‘psychological stress-resistance resource” (Hass & Graydon, Graydon, 2009; Togari & Yamazaki, 2005; Yamazaki). The SOC has been related to a healthy adaptation to stress in school age-groups (Hass & Graydon, 2009; Lundberg, 1997; Torsheim, Aaroe, & Wold, 2001 ). In Torsheim et al.’s study, the level of SOC explained 39% of the variance in subjective health complaints among 1 l-year old children, and it is seen as a resilience factor in school-aged 8O children. Lundberg hypothesized that a high level of SOC modulates the negative influences of stress factors from a conflicted environment in the family or at school. The SOC scale has l3-items and utilizes a 5-point Likert scale ranging from 1 to 5. Responses for most questions vary from 1 (Very Ofien) to 7 (Never). The responses for a few questions are 1 (Like it a lot) to 5 (Don ’t like it all). Scores at the lower end of the scale represent more negative responses. Example questions are “How often do you have the feeling that you don’t really care about what goes on around you?” (Meaningfulness), “How often does it happen that you don’t quite understand your own feelings and ideas?” (Comprehensibility), and “How ofien has it happened that people whom you counted on disappointed you?” (Manageability). Cronbach alpha was .85 in the study among American early adolescents ages 11, 13, and 15 years (Torsheim, Aaroe, & Wold, 2001). Cronbach alpha for the Japanese SOC scale was found to be over .80 (Togari & Yamazaki, 2005). This same study also provided construct validity support. In the present study, the internal reliabilities of the meaningfulness (5 items), comprehensiveness (5 items), and manageability (3 items) scales were .65, .70 and .50, respectively. Because of these relatively low reliability measures only the total score was used (Cronbach a = .81 using all 13 items). Expressed Emotion Adjective Checklist (EEAC). The EEAC (Friedman & 81 Goldstein, 1993) is a self-report checklist that was developed as a brief measure of EB. Each section consists of 10 positive and 10 negative adjectives chosen to assess the two dimensions of the Expressed Emotion: Criticism and Emotional Over-Involvement (EOI). The first set of 20 items assesses the levels of maternal Criticism and E01 over the last 3 months. In the second 20 items, the child rates his/her own behaviors toward his/her mother (10 positive and i 10 negative EE adjectives) over the same time period. Each adjective is rated on 8-point Likert scale ranging from 1 (never) to 8 (always). Examples of the positive adjectives are accepting, considerate, fiiendly, and loving. Examples of the negative adjectives are angry, mean, rude, and hostile. The total scores for negative and positive adjectives conceptually represent the level of Criticism and E01. Because the EEAC was not available in Japanese, each English adjective in EEAC was translated into Japanese by the investigator. ext, a panel of three English- speaking Japanese graduate students living in the US reviewed the translation, comparing the English word to the Japanese translation. A second panel of three different English- speaking Japanese graduate students living in the US back-translated the Japanese items into English to confirm the accuracy of the translation. Hooley (2007) has criticized the EEAC as providing a poor match to the CPI results, and the results of the present study bear this out (see Chapter IV). It would appear 82 that the items intended to measure EOI are actually a measure of the emotional involvement in the relationship between parent and child. Therefore, from this point forward, Involvement will be used in place of E0]. In the present study, Cronbach alpha was .90 for the 10 child perception of mother Criticism items and.83 for thelO child perception of mother’s Involvement items. In the second 20 items, the child rated his/her own behaviors toward his/her mother (10 positive and 10 negative EE adjectives) over the same time period. Cronbach alpha was .91 for the 10 items asessing the child’s perception of his/her own Criticism of the mother and .84 for the 10 items assessing the child’s perception of his/her own Involvement in the present study. Family Relationship Index (FRI).The Family Relationship Index (FRI) is a short version of Family Environment Scale (FES; Moos, 1974; Moos & Moos, 1981) with a 12-item, true-false, self-rating scale. The FRI focuses on three dimensions to assess global family functioning: Cohesion, Expressiveness, and Conflict. Originally, those three dimensions were included in the IO-dimension F ES. The Cohesion items capture the level of commitment, help, and support among family members. The Expressiveness items assess the level of open communication characterized as acting openly and expressing feelings directly. The Conflict items measure the level of openness 83 to express anger, aggression, and conflict among family members. Example questions for Cohesion, Expressiveness, and Conflict are “Family members really help and support one another,” “F amily members often keep their feelings to themselves,” and “We fight a lot in our family,” respectively. The total score on the three subscales was used in the present study, with Conflict items reverse scored. A Cronbach alpha of .62 was found for the F RI items, a value that would be considered marginally acceptable. The Japanese version of the FRI has been used for investigating the characteristics of families and family functioning of Japanese breast cancer patients (Ozono et al., 2001; Fujio, 2003). Among adult Japanese breast cancer patients and their families, Fujio (2003) found evidence both for the reliability and validity of the FRI. Procedure-Mother’s Data Questionnaires. The questionnaire packet was delivered to the mother by her child. The questionnaire packet included a set of instructions, the mother’s questionnaire (coded to match the child’s identifying number), and a consent form. After the mother signed the consent form to agree to her participation, she completed the mother’s questionnaire. When completed, the mother enclosed the questionnaire and the consent form in the sealed envelope. The questionnaires were returned to the school by the child and placed in the box in the classroom. 84 Instruments-Mothers The instruments used in this study were the Center for Epidemiologic Study . Depression Scale (CESD), Expressed Emotion Adjective Checklist (EEAC), and Family Relationship Index (FRI) (Appendix C). The descriptions of the F arnily Relationship Index and the Expressed Emotion Adjective Checklist were presented in the children’s Instruments section. Internal reliabilities for these two instruments are as follows: FRI, Cronbach a = .52 for mothers; EEAC: mother’s perception of her Criticism of child, Cronbach a = . 84; mother’s perception of her Involvement toward child, Cronbach a = . 83; mother’s perception of child’s Criticism of her, Cronbach a = . 88; mother’s perception of child’s Involvement toward her, Cronbach a = . 82 The Center for Epidemiologic Studies Depression Scale (CES-D). The CESD scale includes 20 items with 4 subscales: depressed mood, positive affect, somatic activity, and interpersonal relations (Radloff, 1977). To emphasize the current state of mind, the questions include “How often in the past few weeks did you...” The CESD is a self-rating Likert scale with answers ranging from 1 (Never) to 5 (Very Often; 5 -7 days in a week). The example questions are “I felt that I was just as good as other people (reverse scored)” and “I felt lonely.” Only the total CESD score was used in the present study. The final CESD score ranges from 0 to 60, with a higher score indicating greater 85 impairment. People with a total score of 16 or higher are typically identified as a depressiVe ‘case.’ The CESD has been widely used in various Asian countries, including Japan. Researchers have critiqued the CESD from a construct validity point of view, pointing out that the East Asian beliefs and practices around the experience and expression of positive emotions may falsely inflate the depression scores on a variety of self-report depression screening instruments (Cho & Kim, 1998; Iwata & Buka, 2002; Iwata & Roberts, 1996; Iwata, Saito, & Roberts, 1994;1wata et al., 1998; Noh, Kasper, & Chen, 1998). Asian adults suppress the expression of positive affect, even though the responses to negative symptom items are comparable between groups (Iwata et al., 1998). The internal consistency of the scale significantly improved when the original positive affect items were revised to negatively-worded items (Iwata, Saito, & Roberts, 1994; Iwata, Roberts, & Kawakami, 1995). Iwata and colleagues developed and tested the CESD-R (Korean) version that reverses the wording of the four positive affect items, and found that these modifications improve the accuracy of these instruments with East Asian populations. Cronbach alpha after the revision was .92 (Iwata et al., 1998). The Japanese translation of the CESD—R was used in this present study. Cronbach alpha was .94 for the CESD in the present study. 86 Analyses Sample Size The sample size was 285 child-mother pairs. Of 483 possible pairs, 198 questionnaires were missing either the child’s report (88) or the mother’s report (1 10). In addition, 9 of the 285 children did not complete the YSR, reducing the sample size to 276 for any analysis involving Internalizing or Extemalizing scores. Analysis for Specific Aim 1 (Hypotheses 1.1, 1.2, and 1.3) Zero-order correlations were used to test relationships between familial factors. (mother’s CESD-R and FRI scores), mother’s and child’s perception of EB (Criticism and Involvement), child’s resilience factors (child’s scores on the SOC and SE), and child YSR lntemalizing and Extemalizing behavioral problem scores. Analysis for Specific Aim 2 (Hypotheses 2.1., 2.2, and 2.3) Zero-order correlations were used to examine the relationships between child and mother reports of Criticism and Involvement toward the child (112,1). In addition, the same relationships were examined by child gender and by mothers of boys vs. mothers of girls. Groups were formed to examine relationships between high and low EE scores and YSR normal-risk and high-risk scores under Hypotheses 2.2a and 2.2b. YSR high- 87 risk groups had Extemalizing or lntemalizing scores above 59. To identify high and low EE levels, the upper and lower 25th percentiles of the scores in each EEAC subscale, Criticism and Involvement, were used to identify high level and low level groups. First, xz-square tests were used to examine group differences between the two levels of Criticism and the two levels of Involvement (EEAC) and the child’s risk level of both lntemalizing and Extemalizing problems. Then, Multivariate Analyses of Variance (MANOVAs) were used to examine whether there were significant differences in the child’s perception of her/his mother’s Criticism and Involvement between children in normal-risk and high-risk Internalizing and Extemalizing groups. Hypothesis 2.2a examined the relationship between child’s perception of mother’s Criticism and Involvement and lntemalizing and Extemalizing behavior problems. Hypothesis 2.2b examined the relationship between mother’s perception of her own Criticism and Involvement and the child’s risk level of Internalizing and Extemalizing behavior . A multiple regression analysis was used for Hypothesis 2.2c. In the first step, child perception of mother’s Criticism and Involvement and mother perception of her own Criticism and Involvement were entered as predictors of lntemalizing or Extemalizing scores. In the second step, the Extemalizing or lntemalizing score was entered as a control for the correlation between lntemalizing and Extemalizing. The 88 child’s lntemalizing behavior problems (i.e., Social Withdrawal, Somatization, Anxiety/Depression) co-occurred with Extemalizing behavior problems (i.e., Aggression, Delinquency). Analysis for Specific Aim 3 (Hypotheses 3.1, 3.2, and 3.3) Moderation effects. Hierarchical regression procedures were used to test Hypotheses 3.1a and 3.1b. The goal was to determine if there were moderating effects of the child’s self resources (SOC and SE) on the relationship between family environment variables (mother and child report of Criticism and Involvement, CESD-R, mother and child FRI scores) taken one at a time and lntemalizing and/or Extemalizing behaviors. This analysis followed three steps: (1), each variable was entered as a predictor of the child’s level of YSR Internalizing or Extemalizing, (2) child’s SOC and SE scores were entered, and (3) the interaction terms (e. g., CESD-R by SOC) were entered. The same procedure was used to test the moderating effect of child gender on the relationship between the mother’s variables and YSR lntemalizing and Extemalizing scores. Mediation effects. Multiple regression analysis was used to test Hypotheses 3.2, i.e., whether the child’s SOC and SE mediated the relationship between the mother’s variables and YSR lntemalizing and Extemalizing scores. Steps were as follows: (1) a 89 significant zero-order correlation was confirmed between each predictor variable (mother and child report of Criticism and Involvement, CESD-R, mother and child FRI scores) and each outcome variable (lntemalizing or Extemalizing scores). (2) A significant zero-order correlation was confirmed between each predictor variable and each of the putative mediating variables (SOC or SE). For this step, the putative mediating variable became an outcome variable. (3) The predictor variable and mediator variable were entered simultaneously as predictor variables. The initial correlation between the predictor variable and outcome variable must be significantly decreased for a mediation effect to be established. With complete mediation, the relationship between the predictor variable and the outcome variable will become zero (Baron & Kenny, 1986) Sobel’s test was used to confirm whether the indirect (mediation) effect on the dependent variable through the mediator variable was significant. Sobel’s test calculates the product of the direct path from the predictor variable on dependent variable through the mediating variable using the formula, square root of bzsa2 + azsb2 + sazsbz, where a and b are unstandardized regression coefficients and so and sb are their standard errors (Baron & Kenny, 1998). Structural Equation Modeling (SEM) (Hypothesis 4.1) 90 SEM was used to test Hypothesis 4.0, examining the fit of the overall path model. The sample size for SEM analysis was 265 child-mother pairs’ data. Missing data treatment for SEM analysis. Because the software used for the SEM analyses requires that there be no missing data, missing data were treated as follows. Missing data in YSR outcome variables were considered as missing completely at random (MCAR) because the amount of missing data was trivial, less than 1% (9) of the data. The researchers made memos anonymously when children did not complete the YSR questions in the last section of the questionnaire booklet. The reasons these 9 children could not complete the YSR section were either running out of time, accidentally skipped the page, or leaving earlier for a private reason. Thus, the researcher made the judgment that there was no particular pattern of missing data dependent on the values or the observed data (Rubin, 1976). Thus, listwise deletion was utilized for treating missing data. The literatures suggest that “when the data are MCAR there is little difference in the estimation bias for listwise deletion, pairwise deletion and maximum likelihood” (Carter, 2006). 91 Table 3. 1 Return Rates for the Questionnaire Packets and Consent Rates for Children ’s and Mothers ’ Participation (dyad pairs, n = 285) 5th 6th Packets Distributed Total Return Rate (%) n Grade Grade 527 518 1045 58.1% 607 Consent Rate Consent to Survey n (% of forms returned) Children 77.3% 469 Mothers 62.6% 380 Consent to Interview Mothers 16.3% 99 92 Table 3.2 Demographic Information for Children Girls Boys Total %-age n = 149 n = 136 in in (52.3%) (47.7%) Category Sample Age (% in Category) 10 21 (52.5%) 19 (47.5%) 40 (100%) 14.3% 11 76 (54.3%) 64 (45.7%) 140 (100%) 50.0% 12 52 (54.2%) 44 (45.8%) 96 (100%) 34.3% 13 0 (0%) 4 (100%) 4 (100%) 1.4% Grade (% in Category) 5th Grade 77 (52.0%) 71 (48.0%) 148 (100%) 51.9% 6th Grade 72 (52.6%) 65 (47.4%) 137 (100%) 48.1% Note: Only 280 mothers reported their child’s age 93 Table 3.3 n=149 n‘l36 M 6 (53.3%) 14 (46.7%) 30 (100%) 10.6% 2 43 (46.7%) 49 (53.3%) 92 (100%) 32.8% 3 62 (55.9%) 49 (44.1%) 111 (100%) 39.2% 4 or more 28 (56.0%) 22 (44.0%) 50 ( 100%) 17.8% Family Structure W 9 (53.6%) 77 (46.4%) 166 ( 100%) 59.2% Single Mother 15 (39.5%) 23 (60.5%) 38 ( 100%) 13.6% Two Parents in 41 (56.9%) 31 (43.1%) 72 (100%) 25.7% Extended Family Single Mother in 2 (50.0%) 2 (50.0%) 4 ( 100%) 1.4% Extended Family \ Note: Only 283 mothers reported the number of children in the home; 280 "reported family structure. 94 Table 3.4 Parent’s Employment (Mother ’s Report) Girls Boys Total %-age n = 147 n = 134 in In Mothers (n = 281) (52.3%) (47.7%) Category Sample Not employed outside home 43(54.4%) 36(45.6%) 79(100%) 28.1% Employed outside home 104(51.5%) 98(48.5%) 202(100%) 71.9% Girls Boys Total %-age n=132 n=116 in InFull Fathers (n = 247) (%) ( %) Category Sample Not employed outside home 0(0.0%) 3(100%) 3(100%) 1.2% Employed outside home 131(53.7%) 113(46.3%) 244(100%) 98.8% Note: There were 34 single-mother households; thus, the number of fathers identified is 247. 95 Depression Her Own Involvement Sense of Coherenc Meaningfulness ” Manageability , ll ’ Comprehensibility l . Extemalizing . lntemalizing 7 ¥ r Delinquency Aggression Withdrawal‘] LSomatization Anxrety — Depression Fig. 3.1. SEM Analysis to Test the Mediating Effect of Child Coping Resources. Mother’s Perceptions Child’s Perceptions Mother’s Involvement Family Functioning Self- Competence Self-Liking fl 96 Chapter IV RESULTS Preliminary Analyses Description of the Variables Child variables. Okinawan Japanese 5‘h and 6'“ grade elementary students (girls, n = 140; boys, n = 136) were participants, along with their mothers. Table 4.1 summarizes the results for each variable by gender and for the full sample of 276 children. Dependent variables for children included total scores for Sense of Coherence (SOC), Self-Esteem (SE), Family Relationship Inventory (FRI), and their subscales. In addition, the child’s report of positive and negative emotional expressions to and from the mother served as dependent variables. The outcome variables for this study were Youth Self-Report (YSR) lntemalizing and Extemalizing scores. Maternal variables. Maternal variables are summarized in Table 4.2. Dependent variables included the mother’s FRI, the Center for Epidemiological Studies’ Depression Sca1e(CES-D), and her report of positive and negative emotional expressions to and from her child (Expressed Emotion Adjective Checklist, EEAC). Correlations between Variables 97 Child variables. Correlations among child variables are presented in Table 4.3 for the sample of 276 children, and in Table 4.4 by gender. An alpha level of .005 was selected to partially control for the large size of the sample. However, for child variables, virtually all correlations were significant (p < .001). Therefore, Cohen’s (1988, 1992) large effect criterion of r E: .500 was used to identify important relationships between pairs of variables. . Mother variables. Correlations among mother variables and between mother variables and the child outcome measures are presented in Table 4.5 for the entire sample and, by gender, in Table 4.6. Cohen’s (1988, 1992) large effect criterion of r _>_ .500 also was used to identify important relationships between pairs of variables. High-Risk and Normal-Risk YSR Groups The YSR norms developed for Japanese samples were used to establish cut-off points for children’ s behavior problem scores. Two groups were established: a mum]- risk group (T-score < 59) and a high-risk group (T-score 2 59).The prevalence rates of Internalizing and Extemalizing behaviors in the normal-risk and hi gh-risk group are presented in Table 4.7. Of the 55 children who scored in at least one of the high-risk groups, 19 (34.5% of children in either high-risk group, 7.1% of the full sample) scored in both high-risk groups. 98 xz-tests were used to test the independence of high- and normal-risk lntemalizing and Extemalizing groups by gender and by child age 5. There was no significant group difference by gender between the high- and normal-risk groups (lntemalizing, {(1) = 0.0, p = .992; Extemalizing , {(1) = 1.36, p -= .244). Further, there was no significant group difference in the risk level of lntemalizing or Extemalizing behaviors by the age of the child (10 and 11 year old children vs. 12 and 13 year old children), lntemalizing, {(1) = .18, p =-- .667; Extemalizing, 78(1) = .42, p = .517. A MANOVA comparing T-scores for lntemalizing and Extemalizing scores by age group and gender also was not significant, age, Wilks’ it = .991, F (2, 266) = 1.164, p = .314, r72 = .009, power = .254; gender, Wilks’ it = .996, F (2, 266) = .560, p = .572, 272 == .004, power = . 142), and none of the univariate ANOVAs was significant. Relationships between Demographic and Independent Variables Child variables. A MANOVA with gender, age, family constellation, and number of children as predictor variables and the child variables as dependent variables was carried out. A significant effect was found only with gender, Wilks’ it = .919, F (l, 271) = 3.047, p < .005, n2 = .081, power = .937. Significant differences by gender were found for self-esteem only (Table 4.1), with girls scoring significantly lower than boys, F (l, 271) = 601.424, p < .005, n2 == .033, power = .824. The interaction between gender 99 and family structure was significant for the child report of mother’s Involvement, F (2, 271) = 3.524, p < .05, 772 == .028, power = .653 (Figure 4.5), and Criticism, F(2, 271) = 4.273, p < .05, 42 = .033, power = .742 (Figure 4.6). The interaction between gender and family structure was significant for the child report of Involvement directed toward his or her mother, F(2, 271) = 4.049, p < .05, if = .032, power = .718 (Figure 4.7). This interaction is presented graphically in Figures 4.1- 4.3. Boys reported more maternal Involvementand girls report less when living with a single mother (Figure 4.1). Boys living with both parents in an extended family reported more Criticism from their mothers than when they were living with two parents without an extended family or with a single mother (Figure 4.2). Girls reported the most criticism when living with a single mother. Boys living in a two parent family without an extended family reported less Involvement toward the mother (Figure 4.3), but girls reported the least positive expressions toward their mothers when living with a single mother (Figure 4.3). Mother variables. A MANOVA with child gender, child age, family . constellation, and number of children as predictor variables and the mother variables as dependent variables was carried out. There was no significant main effect. However, there was a significant interaction of gender and family structure, Wilks’ it = .917, F (12, 100 271) = 1.788, p < .05, r72 = .044, power = .898 (Table 4.2). The interaction between gender and family structure was significant for the mother’s FRI score, F(2, 271) = 4.887, p <= .01 , n2 = .038, power = .801. Mothers of girls in single parent families reported that the family relationship (FRI) was most negative (Figure 4.4), and, in contrast, mothers of boys reported the most positive family relationship in single families. These findings emphasize the different status and roles of boys and girls in single-parent families. Although there was no main effect for gender, univariate AN OVAs indicated that the child’s perception of the mother’s criticism differed significantly by gender, with girls reporting significantly more perceived maternal criticism than boys, F (1 , 271) = 3.992, p < .05, n2 = .016, power == .512. Also in univariate ANOVAs, the child’s perception of maternal criticism differed by family type, with children in two-parent families living with extended family reporting significantly more perceived maternal criticism, F (2, 271) = 3.927, p < .05, n2 = .031, power = .703. Mother’s reported criticism directed toward her child also differed significantly by family structure. Both single mothers and mothers living in extended two-parent families reported that they were more critical of their child than mothers living in two-parent families, F (2, 271) = 4.993, p < .01, n2 = .039, power = .810. Hypothesis 1. Relationships between Measured Variables 101 Specific Aim 1: Examine the relationships among maternal depression, maternal Criticism and Involvement, family relationships, children’s internal resiliency factors (self-esteem and sense of coherence), and children’s internalizing and externalizing behavioral problems. Hm: There will be a significant relationship between a negative family environment and child behavior problems. Family environment and YSR score. The elements of the family environment were assessed for both the child and his or her mother. The child’s behavior problems were assessed by the level of the Internalizing and Extemalizing YSR scores. The child’s report included the FRI score and the child reports of criticism and the positive expressions between her/himself and the mother (Expressed Emotion Adjective Checklist, EEAC). Scores on the EEAC Criticism scale were predicted to be positively correlated with scores on the YSR scales; conversely, scores on the EEAC Involvement scale and FRI were predicted to be negatively correlated with the YSR scores. The mother’s data included maternal depression (Center for Epidemiological Studies, Depression Scale, CBS-D), FRI, and her report of Criticism and Involvement between herself and her child (EEAC). Scores on the CES-D and EEAC Criticism were predicted 102 to be positively correlated with scores on the YSR scales; conversely, scores on the EEAC Involvement and the FRI were predicted to be negatively correlated with the YSR scores. Child report Hypothesis 1.1 was confirmed regarding the child. The child’s report of the positive family relationship and positive expressions between her/himself and mother were significantly and negatively correlated with both the YSR lntemalizing and Extemalizing scores (Table 4.4). The reports of Criticism between the child and the mother were each significantly and positively correlated with YSR Internalizing and Extemalizing scores. Thus, when child perceived the family relationship more positively, the level of his/her internalizing and externalizing behaviors was lower than when his/her perception was negative. Mother report. Hypothesis 1.1 was partially confirmed for the mother’s reports. Mother’s reports of Criticism between herself and her child were significantly positively correlated with the child’s lntemalizing scores (Table 4.5). The YSR Extemalizing score was positively correlated with her report of her child’s Criticism of herself. However, the mother’s report of her depression, the family relationships, the mother’s positive expressions toward her child, and her child’s positive expressions toward her were not significantly correlated with the YSR scores. In fact, the strength of the relationships 103 between all the mother’s variables and child outcomes were significantly weaker than those found for the relationships between the child variables and the Internalizing and Extemalizing scores (Fisher’s z-test, all 23 > 1.96. p < .05). H1, 2: There will be significant negative relationships between child resiliency resources (SE and SOC) and child behavior problems. Child resiliency factor and YSR score. Sense of Coherence (SOC) and Self— Esteem (SE) scores were both significantly and negatively correlated with both lntemalizing and Extemalizing scores on the YSR (Table 4.3). However, the SOC score was significantly more strongly correlated with both lntemalizing and Extemalizing scores than the child’s SE score (Fisher’s z-test, both zs > 1.96. p < .05). In fact, when SOC and SE were entered as predictor variables in a stepwise multiple regression, only SOC entered as a predictor for both lntemalizing and Extemalizing. H13: There will be a significant relationship between a negative family environment and child resiliency resources. Family environment and resiliency resources. The same mother and child variables used in Hypothesis 1.1 to describe the family environment were correlated with 104 the child resiliency measures (SOC and SE). The results are parallel to those of Hypothesis 1.1. Child report. Significant positive correlations were found between each of the positive family environment measures (FRI, EEAC Involvement between the child and her/his mother) and the SOC and SE scores. The more positive the family environment reported by the child, the stronger the resiliency resources. The child report of his/her own Criticism toward mother and of mother’s Criticism of him/her were both significantly negatively correlated with SOC and SE scores. Mother report. There was no significant correlation between any of the mother’s scores (FRI, CES-D, EEAC Criticism and Involvement between mother and child) and the child’s SOC (all absolute values of rs <.110, ps >.065). In contrast, all correlations between the child’s self esteem and the mother’s scores were significant (all absolute values of rs >.117, ps <.05), with the exception of mother report of her positive expressions toward the child (r = .101 , p = .093). As was true for Hypothesis 1.1, the correlations between the mother’s variables and child resiliency were significantly smaller than those between the child’s variables and his/her resiliency resources (Fisher’s z-test, all 23 > 1.96. p < .05). Hypothesis 2. Maternal Criticism and Involvement 105 in Relation to Behavior Problems Specific Aim 2: To examine whether EE components, either high levels of Criticism or Emotional Over-Involvement (Involvement), are associated differentially with the Internalizing or Extemalizing behavior problems. H“. There are positive relationships between a high level of Criticism and a low level of Involvement and child lntemalizing and Extemalizing Scores. Relationships between Involvement and Internalizing and Extemalizing Scores Child reports. Child reports of mothers’ positive emotional expressions (Involvement) were significantly and negatively related to both lntemalizing and Extemalizing scores (Table 4.3). Similarly, child reports of their own Involvement directed toward their mothers were significantly and negatively related to their reports of their own lntemalizing and Extemalizing behavior problems. Hypothesis 2.1 was not confirmed for Involvement, based on child report. In fact, the reverse was true. Children who reported that their mothers expressed less positively 106 or that they were less positively emotionally connected to their mothers tended to have more lntemalizing and Extemalizing behavior problems. Mother reports. Hypothesis 2.1 was also not confirmed for Involvement based on the mother’s report (Table 4.5). Mother reports of Involvement directed toward her child were significantly and negatively related to child lntemalizing score, i.e., children whose mothers reported they were less positively connected to their children tended to have more Internalizing behavior problems. Mother report of Involvement was not significantly correlated with the child Extemalizing score. How much the mother reporting her child expressing positive emotions toward herself did not relate to either lntemalizing or Extemalizing scores. Correlations between Child Report of Involvement and YSR Behavior Scores by Gender Boys’ reports. Table 4.4 presents the correlations between child report of Involvement and child YSR behavior scores separately by gender. Boys’ reports of mother’ s Involvement were significantly and negatively related to boys’ lntemalizing behaviors, but not to Extemalizing behaviors. Boys’ own Involvement directed toward their mothers was not significantly related to either lntemalizing or Extemalizing behavior problems. 107 Girls’ reports. Among girls, there was a significant and negative correlation between girls’ reports of their mothers’ positive expressions and both lntemalizing and Extemalizing behavior problems. Further, girls’ own Involvement directed toward their mothers was significantly and negatively related to both lntemalizing and Extemalizing behavior scores. Thus, Positive emotional connection between mother and child were negatively related to child Internalizing behavior problems of Japanese children, both boys and girls. However, Involvement was significantly related to Extemalizing scores only in girls. Correlations between Mother Report of Involvement and YSR Behavior Scores by Gender Mothers of boys. When the child was a boy, mother report of her Involvement with her son was significantly and negatively related to child lntemalizing and Extemalizing behaviors. Mother reports of child Involvement with them herself were not significantly related to either YSR behavior problem scores. The result was consistent with the child’ 3 report in that the emotionally positive attitude (Involvement) from the mother was significantly related to child behavior problems, rather than the child’s Involvement with the mother. 108 Mothers of girls. When child was a girl, there was no significant relationship between mother’s report of her own or her daughter’s Involvement and either lntemalizing nor Extemalizing scores (Table 4.6). Relationships between Criticism and lntemalizing and Extemalizing Scores Child reports. Child reports of mother Criticism were significantly and positively related to both lntemalizing and Extemalizing scores (Table 4.3). Similarly, child reports of their own Criticism directed toward mother were also significantly and positively related to their reports of their own lntemalizing and Extemalizing behavior problems. Hypothesis 2.1 was confirmed for Criticism. Children who reported that their mothers showed more Criticism and/or that they showed more Criticism toward their mothers tended to have more lntemalizing and Extemalizing behavior problems. Mother reports. Mother reports of the child’s Criticism directed toward her were significantly and positively related to child’s both Internalizing and Extemalizing YSR scores (Table 4.5). Mother report of her own Criticism directed toward her child was significantly and positively related to the child’s lntemalizing behaviors, but not to Extemalizing behaviors. Hypothesis 2.1 was partially confirmed for the Criticism based on mother’ 3 report as well. 109 Correlations between Child Report of Involvement and YSR Behavior Scores by Gender Boys’ reports. There were significant and positive relationships between boys’ reports of their mothers’ Criticism and both lntemalizing and Extemalizing behavioral problems in boys (Table 4.4). Similarly, boys’ reports of their own Criticism directed toward their mothers were significantly and positively correlated with boys’ lntemalizing and Extemalizing behavior scores. Girls’ reports. The relationships between girls’ reports of Criticism were parallel to those of boys. Correlations between Mother Report of Criticism and YSR Behavior Scores by Gender . Mothers of boys. When a child was a boy, both the mother’s own Criticism toward her child and her child’s Criticism toward her were significantly and positively related to child’s lntemalizing behavior score, but not to the Extemalizing behavior score (Table 4.6). Mothers of girls. When the child was a girl, mother reports of her own Criticism directed toward the child were significantly and positively related to her daughter’s Internalizing behavior scores. Mother reports of her daughter’s Criticism toward her was 110 significantly and positively related to her daughter’s lntemalizing and Extemalizing YSR scores, in contrast to the result found with mothers of boys. Gender Differences in Strength of Correlations Fisher’s z-test was used to compare the correlations between child variables by gender. Data from girls was significantly more strongly correlated for 9 pairs of variables: Sense of Coherence (SOC)-mother Involvement toward child (positive correlation, +), SOC-child Involvement toward mother (+), Family Relationship Inventory (F Rl)-mother Involvement toward child (+), FRI-child Involvement toward mother(+), mother Involvement toward child-mother Criticism of child (negative correlation, -), mother Involvement toward child-Extemalizing (-), mother Criticism toward child-Extemalizing (+), mother Criticism toward child-Internalizing (+), and child Involvement toward mother-Extemalizing (-). None of the pairs of correlations showed a stronger relationship for boys. Differences by Normal-risk and High-risk Groups [12.2.2 There will be significant differences in the child perception of her/his mother’s Criticism and Involvement between children in normal-risk and high-risk behavior problem groups. 111 verses high-risk Internalizing and Extemalizing groups, in terms of the child’s EXpressed Emotion Adjective Checklist (EEAC). Conversely, a low level of Criticism or Involvement was defined as a score in the lowest quartile of the respective EEAC scale. Because a xz-test was planned, only the highest and lowest quartiles for Criticism or Involvement, rather than all 4 quartiles, were selected to limit the degrees of freedom and to emphasize group differences Criticism and Involvement were analyzed separately because each captures an essential dimension of the emotional attitudes that are hypothesized to be specifically related to a certain disorders or behavior problems (Hooley & Teasdale, 1989). Child’s YSR lntemalizing and Extemalizing scores were classified into two groups (normal-risk and high-risk groups) based on Japanese YSR cut- Off points of T-scores greater than 59 (Table 4.7). It should be noted that 19 children were in both high-risk groups, making it more likely that if an indicator of group independence for, say, Internalizing were significant, the indicator for the corresponding behavior scale, Extemalizing , would be significant as well. 1 12 Criticism and risk. f-tests indicated that the normal- and hi gh-risk groups for both lntemalizing and Extemalizing behavior problems were different in their perceptions of their mothers’ Criticism directed toward themselves , Internalizing, x 2 (l) = 22.51, p < .001; Extemalizing, 752(1) = 29.64, p < .001. Children in both the high-risk groups perceived more Criticism from their mothers than children in the normal-risk groups (Table 4.8). Involvement and risk. xz-tests indicated that there was a significant difference between the normal-risk and high-risk groups in both lntemalizing and Extemalizing scores in their perceptions of their mothers’ Involvement with them, Internalizing, x2 (1) = 25.216, p < .001; Extemalizing , {(1) = 16.051, p < .001. In contrast to the effect of Criticism, children in both high-risk groups perceived less Involvement from their mothers (Table 4.8). MAN OVA. Data from the full sample of children were included in a MANOVA comparing Criticism and Involvement scores by the groups of children in normal-risk and high-risk groups for lntemalizing and Extemalizing behavior problems. The MAN OVA confirmed the significant differences seen in the more limited sample of high- and normal-risk groups and very high and very low Criticism and Involvement groups. There were significant differences between normal-risk and high-risk groups in child’s l 13 perceptions of her/his mother’s Criticism directed toward her/him, lntemalizing, Wilks’ l. --- .954, F(2, 269) =-- 6.531, p < .005, I72 = .046, power = .906; Extemalizing, Wilks’ it .935, F (2, 269) = 9.390, p < .001, r,’ = .065, power = .978. Univariate ANOVAs indicated that the child’s perceptions of mother’s Criticism were significantly different between the normal-risk and high-risk groups for both lntemalizing and Extemalizing behavior problems, Internalizing, F (l, 269) = 8.665, p < .005, n2 = .031, power = .835; Extemalizing, F (1, 269) = 18.760, p <.001, if = .065, power = .991. Children in the high-risk group reported more Criticism. Furthermore, child’s perceptions of mother’s Involvement directed toward child were significantly different between the normal—risk and high-risk groups, lntemalizing, F(1,269)= 11.123, p < .001, r72 = .039, power = .914; Extemalizing, F(1, 269) =4 4.033, p <.05, r72 = .015, power = .516. Children in the high-risk group reported less Involvement. Hm: There will be significant differences in the mother’s perceptions of her own Criticism and Involvement directed toward her child between children who are in normal-risk and high-risk behavior problem groups. 114 Criticism and Involvement in High- and Low-Risk Groups Group formation. This analysis is parallel to that with the child perception of his/her mother’s Criticism and Involvement. The highest and lowest quartiles of mother’s report of her own Criticism or Involvement toward her child each were used to establish two groups of children. The two groups of YSR high-risk or normal-risk children were then used to compare mother reports of Criticism and Involvement. Criticism and risk. The differences in mother’s perceptions of her own Criticism and Involvement directed toward her child between the normal-risk and high-risk groups of children are presented in Table 4.9. The children with high levels of Internalizing had mothers who reported more criticism of the child than the normal-risk group, but the two groups were not different when high levels of Extemalizing were considered, lntemalizing, x’ (1 )= 7.32, p < .01; Extemalizing, x2 (1 )= 1.14, p = .286 (both with Yates Continuity Correction). The result for lntemalizing and Criticism is consistent with the result seen with the child’s report of her/his mother’s level of Criticism; however, there was no significant group difference associated with different levels of Criticism and Extemalizing behaviors. Involvement and risk. There was no significant difference in the distribution of high and low Involvement groups by normal—risk and high risk lntemalizing groups; 1 15 however, there was a significant difference between high and low Involvement groups for normal-risk and high-risk Extemalizing groups, lntemalizing, x2 (1 )= 2.95, p --= .086; Extemalizing, x2 ( l)= 3.92, p < .05 (both with Yates Continuity Correction). As was true for the child report of mother’s Involvement, mother’s report of greater Involvement was associated with lower risk of high scores on the YSR Extemalizing scale. MAN OVA. Data from the full sample of children were included in a MAN OVA comparing mother’s Criticism and Involvement Scores by the groups of children in normal-risk and hi gh-risk groups for Intemalizmg and Extemalizing behavior problems. The MAN OVA results were consistent with the results of the xz—tests. The MAN OVA yielded no significant overall difference for YSR risk groups in terms of either Criticism and Involvement, Internalizing, Wilks’ it = .982, F (2, 269) = 2.414, p = .091, r72 = .018, power = .484; Extemalizing, Wilks’ h = .995, F(2, 269) = .740, p = .478, r72 = .005, power = .175. Univariate AN OVAs indicated that the mother’s report of her Criticism toward her child was significantly different between the normal-risk and high-risk groups for lntemalizing, but not Extemalizing, behavior problems, lntemalizing, F (l , 269) = 4.814, p < .05, 772 = .018, power = .589; Extemalizing, F(1, 269) = .043, p = .835, if < .001 , power == .055. Children in the high-risk group reported more Criticism. 116 The mother’s report of her Involvement directed toward her child was not significantly different between the normal-risk and high-risk groups, lntemalizing, F ( l , 269) = .621, p = .431, n2 = .002, power == .123; Extemalizing, F (1, 269) = 1.374, p = .242, r,’ = .005, power = .215. Criticism and Involvement as Specific Predictors Two multiple regression analyses were conducted to examine the relationships between mother and child perceptions of her Criticism or Involvement and the child’s Internalizing and Extemalizing behavior problems. Four predictors, mother reports of her own Criticism and Involvement directed toward her child and child reports of his/her mother’s Criticism and Involvement directed toward him/her, and gender were entered first in the regression equation. Because lntemalizing and Extemalizing behavior problems co-occurred in children, one (lntemalizing or Extemalizing) of the two behavior problem variables was also entered to control the effect when the other behavior variable was predicted. Hue“) Mother and child perceptions of mother’s Criticism specifically predict child Extemalizing behavior problems. 117 Predicting Extemalizing. When only the four parent/child Criticism/Involvement variables were included to predict Extemalizing scores, a significant R2 was obtained, with only the child report of mother’s Criticism yielding a significant Bovalue (see Table 4.10 for coefficient values), R2 = .216, F (4, 268) = 18.414, p < .001. When the score for YSR lntemalizing was entered in the regression equation, the R2 increased to .421 (R’,,.,,,. =—- .205, F.,,.,,,, (1, 267) = 94.751, p < .001). Significant B-values were obtained for the YSR lntemalizing score and the child perception of her/his mother’s Criticism (Table 4.10). Therefore, only the child perception of mother’s Criticism was a significant predictor (positive) of child Extemalizing behavior problems, whether or not the lntemalizing score was included in the regression equation. H2342) Mother and child perceptions of mother’s Involvement specifically predict child’s lntemalizing behavior problems. Predicting lntemalizing. When only the four parent/child Criticism/Involvement variables were included to predict lntemalizing scores, a significant R2 was obtained, with child reports of mother’s Criticism and Involvement yielding significant B-values (see Table 4.11 for coefficient values), R2 = .199, F (4, 268) = 16.642, p < .001. When the score for YSR Extemalizing was entered in the regression equation, the R2 increased to .210, Fchange (1, 267) = 94.751, p < .001). However, only the child .409 (chhange = 1 18 perception of her/his mother’s Involvement and the YSR Extemalizing score were related significantly to lntemalizing behavior problems (Table 4.1 1). Therefore, only child perception of mother’s Involvement was a significant predictor (negative) of child’s lntemalizing behaviors when the Extemalizing score was included in the regression equation. Summary. Thus, the hypothesis regarding Criticism as a factor in predicting Extemalizing behavior problems was confirmed. Higher levels of Extemalizing behavior were predicted by child reports of mother’s Criticism, even controlling for the Internalizing score. The hypothesis regarding Involvement as a factor in predicting lntemalizing behaviors was not confirmed. Higher levels of lntemalizing behavior problems were predicted only by lower levels of child perception of Involvement from her/his mother. Mother’s perception of her Involvement was not a significant predictor. Gender did not predict either behavior problem. Predicting lntemalizing and Extemalizing in High-risk Samples H23... Mother and child perception of mother’s Criticism specifically predict Extemalizing behavior problems in a high-risk Extemalizing group. 119 Predicting Externallty. When only the four parent/child Criticism/Involvement variables were included to predict Extemalizing scores of children in the high-risk Extemalizing group (boys, n = 14: girls, 11 = 23), a significant R2 was not obtained with any of the variables, only the four parent/child Criticism/Involvement variables, R2 = .202, F(4, 32) = 2.020, p =- .115; including lntemalizing, R2: .202, Fem. (1, 31) = 114, p = .240. Table 4.12 presents the for coefficient values. Hm. Mother and child perceptions of mother’s Involvement specifically predict lntemalizing behavior problems in a hi gh-risk lntemalizing group. A group of children were identified whose lntemalizing scores placed them in a high risk category (boys, n = 21: girls, 11 = 24). The multiple regression procedure used for the full sample was repeated to examine the relationships between mother and child perceptions of Criticism and Involvement and the child’s lntemalizing and Extemalizing behavior problems. Four predictors, mother reports of her own Criticism and Involvement directed toward her child and child reports of his/her mother’s Criticism and Involvement directed toward him/her were entered first in the regression equation. Because Internalizing and Extemalizing behavior problems co-occurred in children, one 120 (lntemalizing or Extemalizing) of the two behavior problem variables was also entered to control the effect when the other behavior variable was predicted. Predicting Internalizing. When only the four parent/child Criticism/Involvement variables were included to predict lntemalizing scores, a significant R2 was obtained, with mother reports of own Involvement and child reports of mother’s Criticism yielding significant B-values, (see Table 4.13 for coefficient values), R2 = .414, F (4, 40) = 7.056, p < .001. When the score for YSR Extemalizing was entered in the regression equation, the R2 increased to .470 (Rim, = .056, am. (1, 39) = 4.109, p < .001). However, when Extemalizing was included, mother perception of her own Involvement and child perception of mother’s Criticism still were significantly related to child lntemalizing behavior problems. YSR Extemalizing score was also related to lntemalizing behavior problems (Table 4.13). Therefore, the hypothesis of a positive relationship between mother’s perception of her Involvement and child’s lntemalizing behavior problems was not confirmed. How mothers perceived their own Involvement toward their child was a significant predictor (negative) of child’s lntemalizing behavior problems, even though mother’s perceptions were not related to lntemalizing behavior problems in the full sample population (Hypothesis 2.2c[1], above). Child perceptions of Criticism from their 121 mothers were also significant predictors (positive) of lntemalizing behavior problems for children in the high-risk group, when the Extemalizing score was controlled. Summary. Thus, the hypothesis regarding Involvement as a factor in predicting Internalizing behavior problems was partially disconfirrned with respect to mother’s report of her own Involvement. In fact, the results were the reverse of what was predicted. In contrast to the results testing Hypothesis 2.2c (2), mother’s perception of her own Involvement was a predictor (negative) of lntemalizing behavior problems. Child report of Criticism from mother and Extemalizing behavior problems were also significant positive predictors of lntemalizing behavior problems. The hypothesis regarding Criticism as a factor in predicting Extemalizing behavior problems of the children in the Extemalizing high-risk group was not confirmed. There was no evidence to support a relationship between children’s Extemalizing behavior problems and mother or child report of mother’s Criticism. Hypothesis 3. Mediating and Moderating Effects of Resiliency Factors I Specific Aim 3: Examine the moderation effect and mediation effect of children’s Sense of Coherence (SOC) and Self-Esteem (SE) in the relationship between maternal Criticism and Involvement and child behavior problems L Moderating Effects of Resiliency Factors 122 H“: Child Sense of Coherence (SOC) and Self-Esteem (SE) moderate the relationship between family environment factors and lntemalizing and Extemalizing behavior problems. Full sample. Regression analyses were conducted to examine the relationship between each of the family environment variables, child’s SOC or SE, and child’s lntemalizing and Extemalizing behavior problems. Family environment variables were child reports of mother’s Criticism and Involvement (EEAC), child report of the family relationship (FRI), and maternal depression (CBS-D). The analysis procedure was conducted as follows. As the first step, one of the continuous predictor variables and one of the moderator variables (SOC or SE) were centered to eliminate multicollinearity effects between the predictor and moderator. As the second step, the predictor variable (family environment) and the purported moderator variable (SOC or SE) were entered simultaneously to predict Internalizing or Extemalizing scores. As the third step, the interaction term was entered (e.g., FRI x SOC). No significant moderation effects were found; that is, none of the interaction terms was significant in the regression equation. Child SOC and SE were significantly and negatively correlated with both lntemalizing and 123 Extemalizing scores, beyond the contribution of the main effects of the family- related predictors. Either the predictors were significantly associated with the child’s lntemalizing or Extemalizing symptoms, independent of the moderator variable, or the family factors lost the power of prediction because of a stronger association between Internalizing or Extemalizing and the moderating variables, child’s SOC and SE. That is, entering the child’s resilience variables did not change the relationship between family factors and Japanese child’s behavior problems. High-risk samples. The same analyses were conducted among the group of 55 children in the combined high-risk lntemalizing/Extemalizing group. There was no significant moderation effect for child SOC and SE. Moderating Effects of Gender H33: Gender moderates the relationship between child perceptions of mother’s Involvement and Criticism and lntemalizing and Extemalizing behavior problems. Gender and Involvement. The same regression analyses were ' conducted to examine the moderation effect of child’s gender. There was no moderation effect of gender on the relationship between child reports of 124 mother’s Involvement and Internalizing behavior problems. For Extemalizing, the interaction term, child’s report of his or her mother’s Involvement by gender, was significant for child Extemalizing behaviors, B = -. l 79, R2 = .014, p < .05 (Figure 4.5). Further, child perceptions of his or her mother’s Involvement significantly and negatively predicted child Extemalizing behaviors, [3 == -.297, R2 = .088, p < .001. However, there was no main effect for gender, 13 = .066, R2 = .000, p = .255. The result indicated that girls were more affected by mother’s Involvement than boys. ‘ Gender and Criticism. Gender also moderated the relationship between child’s perception of his or her mother’s Criticism and both lntemalizing and Extemalizing behaviors. The interaction term, mother’s Criticism by gender, was significant for both lntemalizing and Extemalizing: lntemalizing, B = .339, R2 = .018,p = .014 (Figure 4.6); Extemalizing, B = .389, R2 = .024, p < .005 (Figure 4.7). Child perception of mother’s Criticism significantly and positively predicted child’s lntemalizing behaviors, B = .407, R2 = .166, p < .001, and Extemalizing behaviors, 13 = .433, R2 = .187, p < .001. There was no main effect for gender in predicting either lntemalizing, B = .020, R2 = .000, p = .724, or Extemalizing, [3 = .056, R2 = .003, p = .307, behaviors. Girls were more vulnerable to mother’s Criticism than boys. 125 Mediating Effects of SE and SOC The analyses examined whether each of the child’s resiliency factors, SOC and SE, mediated the relationship between each family environment factor and Internalizing and Extemalizing behaviors. The procedures recommended by Kenny and colleagues (Baron & Kenny, 1986; Frazier, Tix, Barron, 2004) were followed. Only the significant mediation effects mediating the relationship of family environment variables and child Internalizing and Extemalizing behaviors are presented in this section. The Tables 4.14 - 4.22 and Figures 4.8 - 4.12 correspond to the descriptions of the result in this section. Sobel’s test was used to evaluate the effect of the mediator variable. Sobel ’3 test is calculated as the product of the direct path from the independent variable on dependent variable through the mediating variable with the formula, 2 = the square root of bzsaz + azsb2 + sazsbz, where a and b are unstandardized regression coefficients of the path from the independent to mediating variable and the path from mediating variable to the dependent variable respectively. The sa and sb are the standard errors of the each path (Baron & Kenny, 1986; Patrick & Bolger, 2002). The 2 score must exceed the critical value of ”1.96 required for p < .05. 126 Mediating Effects of Resiliency Factors on Mother’s Involvement and Behavior Problems H33, Child Sense of Coherence (SOC) and Self-Esteem (SE) mediate the relationship between child report of mother’s Involvement and Internalizing or Extemalizing problems. Mediation of SOC on Involvement and lntemalizing. Mediation effects of each of the child’s resiliency factors (SOC and SE) for the path from child’s report of mother’s Involvement to Internalizing were examined. First, the child’s report of mother’s Involvement toward the child (the predictor) was regressed onto the child’s Internalizing behavior score (outcome variable) (Step 1). Next, SOC (the hypothesized mediator) was regressed on the mother’s Involvement directed toward child (Step 2). Third, child Internalizing behavior score was regressed on both SOC and the mother’s Involvement directed toward child. The regression coefficient between mother’s Involvement and lntemalizing behavior scores was reduced from -.364 to -.143 after child’s SOC was entered, but the path was still significant (Figure 4.8, Table 4.14). The SOC was found to be a partial mediator in the relationship. Sobel’s test continued that the degree of reduction from -.364 to -.143 was significant (2 = -6.04, p < .01). 127 Mediation of SOC on Involvement and Extemalizing. The same procedure was used to examine the mediation effect of child SOC between the child report of mother’s Involvement and Extemalizing. The direct effect from mother’s Involvement directed toward child and child’s Extemalizing behavior scores dropped from -.297 to -.099 (us). The SOC completely mediated the relationship. Sobel‘s test confirmed that the drop from -.297 to -.099 was significant (2 = 4.35, p < .001). (Table 4.15; Figure 4.8). Mediation of SE on Involvement and Internalizing and Extemalizing. Similarly, Self-Esteem (SE) partially mediated the path from child report of mother’s Involvement and child’s Internalizing and Extemalizing behavior problems. The mediation effects are presented in in Figure 4.9 and Tables 4.16 and 4.17. Sobel’s test confirmed the indirect effect of SE mediating child report of mother’s Involvement and child’s lntemalizing (z = -2.97, p < .001) and Extemalizing (z = -2.73, p < .001) behavior problems. Mediating Effects of Resiliency on Mother’s Criticism and Behavior Problems H333, Child Sense of Coherence (SOC) and Self-Esteem (SE) mediate the relationship between child report of mother’s Criticism toWard the child and lntemalizing or Extemalizing problems. 128 Mediation of SOC on Criticism and Behavior Problems. Mediation effects of each of the child’s resiliency factors (SOC and SE) for the path from child’s report of mother’s Criticism to behavior problems were examined. The regression coefficient between mother’s Criticism directed toward child and child’s lntemalizing behavior scores dropped from .407 to .188 (Table 4.18 and Figure 4.10). Sobel’s test confirmed that the degree of reduction from .407 to .188 was significant (2 = 6.17, p < .001). The same procedure was used to examine the mediation effect of child’s SOC between the child’s report of mother’s Criticism and Extemalizing scores. The direct effect from mother’s Criticism directed toward child and child’s Extemalizing behavior scores dropped from .433 to .258 (Table 4.19 and Figure 4.10). Sobel’s test confirmed that the degree of reduction from .433 to .258 was significant (2 = 5.42, p < .001). The mediation effect is presented in). Mediation of SE on Criticism and Behavior Problems. Similarly, SE partially mediated the relationship between child report of mother’s Criticism and child’s Internalizing and Extemalizing behavior problems. The relationship between Criticism and lntemalizing dropped from .407 to .188 (Table 4.20 and Figure 4.11:), and between Criticism and Extemalizing dropped from .433 to .25 8 (Table 4.21 and Figure 4.11). Sobel’s test confirmed the significant partial mediation effect of SE on the path from 129 mother’s Criticism and lntemalizing (z = 3.23, p < .001), and Extemalizing (z = 2.57, p < .001) behaviors. Mediation of SE on Mother’s Report of Her Own Criticism and Behavior Problems. The relationship between mother’s report of her own Criticism and lntemalizing Behavior Problems was partially mediated by SE. The direct effect of mother’s own Criticism on child’s lntemalizing behavior scores dropped from .188 to .141 (Table 4.22 and Figure 4.12). Sobel’s test confirmed that the drop was significant (z -—- -2.73, p < .001). Hypothesis 4. Testing the SEM Specific Aim 4: To test the model fit with the data with Structural Equation Modeling. Modeling 1143: An adequate model can be developed that describes the relationships among family environment, child resiliency, and behavior problems. The SEM was constructed to reflect the predicted mediation paths from child and mother perceptions through SOC and SE to lntemalizing and Extemalizing diagrammed in Figure 3.1. Because the AMOS-18 software used for the SEM analyses does not allow missing data, the “Listwise” method was used to exclude missing data. Those missing data were assumed as a missing completely at random (MCAR) because they did not 130 depend on any demographic variables or any of the measured variables. Accordingly, 20 of the 285 cases were not included because of a missing score as follows: YSR (n = 9), SOC (1), mother’s FRI (l), child’s EEAC (5), and mother’s EEAC (4). Four test statistics are reported for the models: the chi-square ()8), the fper degrees of freedom (CMIN/DF), the comparative fit index (CPI), and the root-mean- square error of approximation (RMSEA). f is inflated by the sample size, and would be expected to be significant, implying a poor fit. Therefore, the other three test statistics were evaluated. The CMIN/DF allows for a correction of the xz-valueby the degrees of freedom, and values in the range of 2 to 1 or 3 to 1 indicate acceptable fit between the hypothetical model and the sample data (Carmnines & McIver, 1981). CFI compares the fit of the estimated model to the independence model. CF 1 values of .90 or higher are accepted as indicating a good-fitting model. There is some disagreement as to the acceptable values for the RMSEA. Brown and Cudeck (1993) argue that RMSEA values of .05 or less indicate a good model fit, and limit their range to 00-05. However, Hu and Bentler (1999) states that RMSEA values below .05 indicate a “good” fit for a model, values above .10 indicate a “poor” fit, and values between .05 and .10 an “adequate” fit. 131 First, the measurement model was tested to confirm the adequate fit with the data (omitting error terms and correlations). The measurement model provided acceptable fit to the data, f (104, n == 265) = 259.186, p < .001; CMIN/DF = 2.492; CFI = .903; and RMSEA = .075. Figure 4.12 depicts the full model examining the relationships among the latent variables of mother’s perceptions, defined as her reports of her level of depression (CES- D), her own Criticism toward and Involvement with her child (EEAC), and her view of family relationships (FRI); child’s perceptions, defined as his/her reports of his/her mother’s Criticism and Involvement; the two resiliency factors (Sense of Coherence and Self-Esteem); and, as outcome measures, child lntemalizing and Extemalizing behaviors. The model predicted that negative perceptions of the mother and child (mother’s depression, child and mother reports of high levels of mother’s Criticism or low levels of mother’s involvement, and poor family relationships) would be negatively related to both SOC and SE, as resiliency factors (Antonovsky, 1987; Jacquez et al., 2004; Johnson, 2004) and as well as directly and positively related to Internalizing and Extemalizing behavior problems. Further, the model predicted that both SE and SOC Would be negatively related to Internalizing and Extemalizing problem behaviors. Further, the full 132 model tested whether SE and/or SOC would have a mediating effect on child and/or mother perceptions. The full SEM that included the direct and indirect effects from mother and child perceptions, child’s resiliency factors, and lntemalizing and Extemalizing symptoms proved to be an adequate fit (Figure 4.12) with f (105, n = 265) = 260.557, p < .001; CMTN/DF = 2.481, CFI = .903 ; RMSEA = .075, 90%-ile confidence interval = 064-086. Significant pathways were found for both direct and indirect effects from child perceptions to lntemalizing and Extemalizing behaviors. Child SOC and SE partially mediated child’s perceptions to child’s both lntemalizing and Extemalizing behavior symptoms (Figure 4.132). SOC and SE combined mediated 42% of child perceptions to child’s lntemalizing behavior symptoms and 19% of child’s perceptions to child’s Extemalizing symptoms (Shrout & Bolger, 2002). An alternative SEM that allowed only direct effects (no mediation pathways) also had an acceptable fit with the data: 1’ (49, n = 265) = 128.197, p < .001; CMTN/DF = 2.616; CF I = .908; RMSEA = .078, 90%-ile confidence interval = 062-095. As was true 2For simplicity in showing the significant pathways in Figure 4.12, Figure 4.13 omits error terms, correlations between error terms, and nonsignificant path coefficients. 133 for the model described in Figures 4.12 and 4.13, there was no significant direct between other perceptions and lntemalizing and Extemalizing behavior symptoms. Similarly, when indirect pathways were removed from this alternative model, the pathways between child perceptions were significantly and positively related to both Internalizing and Extemalizing behaviors. Although the two models had statistically adequate fit with the data, the full model, because it includes both direct and indirect effects gives more information about the mechanism of the relationship among family factors, resiliency factors, and behavior symptoms. Although the data are cross-sectional and causality cannot be assumed, this information about indirect pathways can help to identify the critical components of interventions, e.g., finding ways to increase SOC presumably would reduce both lntemalizing and Extemalizing problems for all children, but especially those in suboptimal home environments. In both the full and alternative models, the direct and indirect pathways from mother’s perceptions to lntemalizing and Extemalizing behaviors were not significant. However, a model that removed the child’s perceptions, but left those of the mother also provided an adequate fit for the data. This mother-only model had an acceptable fit with the data: )(2 (68, n = 265) = 174.224, p < .001; CMIN/DF = 2.562; CF I = .912; RMSEA = .077, 90%-ile confidence interval = 063-091. 134 The result with the mother-only model demonstrated significant pathways to both child’s resiliency factors and behavior symptoms. Because child’s perceptions as a latent variable had a stronger relationship with child’s resiliency factors, SOC and SE, it accounted for the variances of child’s resiliency factors to a large extent. In effect, when both mother’s and child’s perceptions were taken into consideration in the model, mother’s perceptions became hidden or nonsignificant. The child’s negative perceptions in the family were negatively related to both SOC and SE scores. As expected, child’s SOC was negatively related to child’s both lntemalizing and Extemalizing behaviors, functioning as a mediator to reduce the negative effects from family environment on child behavior problems. However, unexpectedly, the latent variable, “Self-esteem,” was positively related to child’s both Internalizing and Extemalizing behavior symptoms. SE, by itself, was negatively correlated with both lntemalizing and Extemalizing behaviors (Table 4.3) and was a mediator between mother’s Criticism and Involvement in a simple mediation model (Figures 4.9 and 4.11). In the SEM, SE created a suppression effect that inconsistently increased the effects of negative child perceptions of the family environment to both Internalizing and Extemalizing behaviors (MacKinnon, Krull, & Lockwood, 2000). This inconsistent effect of SE was found when the mother and child perceptions were omitted 135 from the SEM, indicating that it was not the result of the presence of either of these latent variables. Principal Components Analysis of the SE Scale Because of the puzzling results of the SEM analyses, a varimax principal components analysis limited to 3 components was conducted with the 16 SE scale items. The three components accounted for 49.3% of the variance in the items. Three subscales were constructed based on two criteria to select an item for a scale, (a) a minimum component loading of .5 and (b) no loading above .3 on any other component. Seven SE items (#3 2, 3, 5, 9, 12, 14, and 15[reworded positively]) loaded on the first component (the component accounted for 30% of the variance) made up a scale labeled “Positive Self-image.” Items were all positively worded (e.g., “I am highly effective at the things I do, I am secure in my sense of self-worth”). The second scale, “Negative Self—image,” was made up of 3 items (#5 1, 6, and 7) from the second component (11.5% of the variation). Items were all negatively worded (e.g., “ It is sometimes unpleasant for me to think about myself"). The third scale, was made up of 3 items (#3 10, 13, and 16) from the third component (7.7% of the variance), and was labeled “Self-criticism.” Items were all worded in negative ways (e.g., “I wishI were more skillful in my activities”). Of the 3 remaining items, 2 had positive loadings above .3 on Positive Self-image and Self- 136 criticism (# 4, “I am almost always able to accomplish what 1 try for, “ and #11, “I never doubt my personal worth”), and #8 had a positive loading on'Negative Self-image and Self-criticism (“At times, I find it difficult to achieve the things that are important to me”). Table 4.23 presents the correlations between the 3 component-based subscales and Internalizing and Extemalizing scores. Paradoxically, all 3 subscales are significantly and negatively related to Extemalizing and Internalizing scores. Further, the strength of the correlation between Intemality and Positive Self-image or Negative Self-image is significant, with the Negative Self-image-Internality correlation significantly stronger. l37 Table 4.] Child Measures by Gender Girls (n = 140) Boys (n = 136) Total (N = 276) Measure M Med. M Med. M Med. (SD) (SD) (50} Sense of Coherence 27.3 28.0 28.1 29.0 27.7 28.0 (6.0) (4.8) (5.5) Meaningfulness 8.6 9.0 8.7 9.0 8.6 9.0 (1.9) (2.0) (2.1) Manageability 8.5 9.0 8.8 9.0 8.6 9.0 (2.2) (2.0) (2. 1) Comprehensibility 10.2 1 1.0 10.7 1 1.0 10.4 11.0 (2.8) (2.3) (2.6) Self-Esteem 46.2 46.0 48.7 47.5 47.4 47.0 (8.3) (9.0) (8,7) Self-Liking 23.6 24.0 25.5 25.0 24.5 24.0 (5.1) (5.5) (5.3) 138 Table 4.1 (cont’d) Girls (n = 140) Boys (n = 136) Measure M Med. M Med. Self-Competence Family Relationship Inventory Cohesion Expressivity Conflict Child Involvement toward mother (SD) 22.5 22.0 (4.2) 8.7 9.0 (2.2) 2.9 3.0 (0.9) 2.4 3.0 (1.1) 2.9 3.0 (1.0) 59.5 61.0 (16.7) 139 Total (N=276) M Med. (SD) (SD) 23.2 23.0 22.9 22.9 (4-6) (4.4) 8.2 8.4 8.4 9.0 (2.2) (2.2) 2.7 3.0 2.8 3.0 (1.0) (1.0) 2.4 3.0 2.6 3.0 (1.2) (1.1) 3.0 3.0 3.0 3.0 (0.9) (0.9) 58.6 59.5 59.1 60.5 (15.0) (15.9) \ Table 4.1 (cont’d). Girls (n = 140) Boys (n = 136) Total (N = 276) Measure M Med. M Med. M Med. (SD) (SD) (SD) Child “MM 26.2 24.0‘ toward mother (11.1) (11.8) (11.4) Mother Involvement 63.9 68.0 63.3 65.0 63.6 67.0 toward child (13.4) (12.7) (13.0) Mother criticism 23.1 21.0 22.9 20.5 23.0 21 .0 toward child (11.1) (11.4) (11.2) YSR Total (T) 51.8 51.0 50.1 50.0 50.1 50.0 (10.2) (9.8) (9.7) YSR lntemalizing 50.3 49.0 49.8 50.0 50.1 50.0 (T) (10.1) (9.2) (9.7) YSR Extemalizing 51.8 49.0 49.0 49.0 49.6 49.0 (7) (10.2) (9.8) (9.9) 140 Table 4. 2. Maternal Measures by Child Gender. F Girls (n = 140) Boys (n = 136) Total (N = 276) Measure M Med. M Med. M Med. (SD) (SD) (SD) Family Relations 'W9A 100* Inventory ( 1 .6) (2. 1) (1 .9) Maternal 4.9 3.0 5.5 2.0 5.2 2.0 Depression (7.2) (8.5) (7.8) Mother Involvement 65.4 67.0 66.2 66.2 65.7 68.0 toward child (12.0) (10.5) (11.4) Mother criticism 26.1 25 .0 25.4 23.3 26.0 24.0 toward child (11.0) (11.0) (11.0) Child Involvement 63.8 65 .0 63.6 58.0 63.9 66.0 toward mother (12.3) (12.6) ( 12.0) Child criticism 23.9 22.0 23.1 24.0 23.6 21.0 toward mother (10.1) (1 1.7) (10.9) 141 .992 .33 60:00 don. N xv waxes 2a moumm “coho owcaq .80. v Q .Eeomfiwmm 2m maoufloboo =< ”Bo an Sn- 3.. SN. own- com- In- waamaeoam a? .H - Sm. RN- Se. eon- one- 3.. 3n- manaeoé mm»: - me... So. new- on..- me..- ewe... doses Base 58an .0 - - ea? 8e. o3. 8m. New. sees Ease beamed m - on? N3..- 2%.. on..- 2% Ease shouts dose: .m - 8e. 3m. 8o. 2% ease 3328a dose: d - e9. Ea. bangs asaeouaom Queen .0 - an. sooaofiom .m . 00:20:00 mo endow H m o m m a o m < 2:282 .m :3 .. .Vm . .- 2.03. 36 .- mg. 58 .- wafififiSE M940 :mwvr Ivon- Ecomr :emm. .....o....- 220 2958 832...... .2. .«o tome. 3883). m :w _ mr Room. tea.- ..omm. 220 BESS 3.3.68 .0: mo tome. 38032 .m - tommr : _ mm. :omvr .5638 230 «0 tome. BESS). .Q - rm“.- :2... 3.3.8.. 2.5 .o :82 gases. .0 - :mom... 38m comma-Egon 35032 .m - , beach: afimcouflom Emma-m 38.832 .< I 0 m m G U m < , 0:532 3mm n 2» «Saute; 2.68.5 3.25 .38 .333..an his: 28.53 35:33-80 .h .4 fianm 144 .23 5mm: 05 S com-«Eon E .me MO 20522 Sm 638 05 mo .2“: .533 Ba 5 3283a 98 when we $5508 “on 8333.200 ”802 232 .22 .528 5%. N a 228 pa 82:. 802... 023 3.23.. a 8222.8 .fiouuau-Sa .25. v a... .25. v 2: - :22. .5. .82 as. 2:.- §. 98.. wafiaeo.xm .22» .m :22. - -2~. :82.- - .2: 28.- 2a. 8. .- 223822 23.0 «2. .22. - :22..- .22. :2? :03. :02.- 220 2.33 5.35 E 2 :22 282.2 a 3.. us.- ..2? - :3..- Fan. :8? :2... 2.32 205232 E .8 :32 353.2 .m .22. 2:. 3m. «2.- - :29- :m :.. :5..- amguo 220 .6 :22 283.2 d .35.- ..8: :3..- :22. :24..- - :22.- ..E. 228292 220 2 was 382.2 .0 3.- §. :.2.~. :2": 22 :2”.- - :2“...- 23 82.828 353.2 .m 22. m3. :3..- :2.~. Fan- :8... LR.- - 23822 2288232 baa-.2 3832 .< m 0 m m a o m < "832mb .8 .8338; 2:830 3.25 ~25 338.82 gmfioxw 2833 3833230 6 .V Sac-m 145 Table 4. 7. YSR Normal-Risk vs. High-Risk Groups by Gender and Grade. T Normal High Normal High Total Total Gender Risk Risk Risk Risk T 123 24 147 124 , 23 129 Girls _ (83.7%) (16.3%) (100%) (84.4%) (15.6%) (100%) 108 21 129 115 14 147 Boys (83.7%) (16.3%) (100%) (89.1%) (10.9%) (100%) 231 45 276 239 37 276 Total (87.3%) (16.3%) (100%) (86.6%) (13.4%) (100%) Grade fl T 147 27 174 152 22 174‘ 5th Grade (84.5%) (15.5%) (100%) (87.4%) (12.6%) (100%) 80 17 97 82 15 97 6th Grade (82.5%) (17.5%) (100%) (84.5%) (15.5%) (100%) 146 fi Criticism. YSR Internalizing YSR Extemalizing Mother’s Normal-Risk High-Risk Normal-Risk High-Risk Involvement Group Group Group Group toward the Child Low Involvement 46 (42.2%) 24 (85.7%) , 54 (47.0%) 16(72.7%) High Involvement 63 (57.8%) 4 (14.3%) 61 (53.0%) 6 (37.5%) Risk Group Total 109 ( 100%) 28 ( 100%) 115 ( 100%) 22 ( 100%) YSR lntemalizing YSR Extemalizing Mother’s Normal- High-Risk Normal-Risk High-Risk Criticism toward Risk Group Group Group the Child Group Low CRIT 69 (59.0%) 1 (5.0%) 68 (59.6%) 2 (8.3%) High CRIT 48 (41.0%) 20 (95.0%) 46 (40.4%) 22 (91.7%) RiskGroupTotal 117(100%) 21 (100%) 114(100%) 24(100%) ‘— 147 Table 4. 9 YSR Normal-risk and High-risk Groups: Mother ’s Reports of Her Own Involvement and Criticism of Her Child Measure YSR lntemalizing YSR Extemalizing Mother’s Normal High-Risk Normal High-Risk Involvement Group Group Group Group Toward Her Child Low Involvement 50 (47.2%) 16(69.6%) 53 (47.3%) 13 (76.5%) High Involvement 56 (52.8%) 7(30.4%) 59 (52.7%) 4 (23.5%) Risk Group Total 106 ( 100%) 23( 100%) 112 ( 100%) 17 ( 100%) ‘ Internalizing Extemalizing ‘ Mother’s Criticism . Normal High-Risk Normal High-Risk“ of Her Child Group Group Group Group Low CRIT 60 (54.1%) 4( 19.0%) 57 (50.9%) 7 (359V~ High CRIT 51 (45.9%) 17 (81.0%) 55 (49.1%) 13 (65.0%) Total 111(100%) 21 (100%) 112 (100%) 20 (100%) * 148 Table 4.10 B Error [3 t-test p Mother reported Involvement -.O82 .052 -.096 -l .589 .113 Mother reported Criticism -.066 .055 -.074 -I.197 .233 Child reported Mother Involvement -.020 .052 -.026 -.3 74 .709 Child reported Mother Criticism .404 .063 .447 6.374 .001 W B Error 0 t-test p M Mother reported criticism -.094 .048 -.105 -1.969 .050 Child reported Mother Involvement .053 .046 .069 1.151 .251 Child reported Mother criticism .280 .056 .310 5.001 .001 T -score lntemalizing .51 l .052 .506 9.734 .001 149 Table 4.11 Coeflicients for Regression of Criticism and Involvement on Intemality (filll sample). Model 1 B Std. Error B t-test p Mother reported Involvement -.051 .052 -.O60 -.988 .324 Mother reported criticism .054 .056 .061 .979 .328 Child reported Mother Involvement -. 141 .052 -.187 -2.693 .008 Child reported Mother criticism .242 .063 .270 3.818 .001 Model 2 B Std. Error 13 t-test p Mother reported Involvement -.009 .045 -.011 -.202 .840 Mother reported criticism .088 .048 .100 1.844 .066 Child reported Mother Involvement -.l31 .045 -.174 -2.906 .004 Child reported Mother criticism .035 .059 .039 .602 .548 T-score Extemalizing .513. .053 .517 9.734 .001 150 Table 4.12 Coeflicients for Regression of Criticism and Involvement on Intemality among Extemalizing High-Risk Group of Children. T Mode] l B Std. Error 13 t-test p Mother reported Involvement -.l31 .120 -.246 -1092 .283 Mother reported criticism .014 .099 .031 .142 .888 Child reported Mother .010 .081 .026 .120 .905 Involvement Child reported Mother criticism .112 .098 .268 1.149 .259 T Model 2 TX}; Std. Error 13 t-test T Mother reported Involvement‘ -.136 .1 19 -.25 7 -l.146 .2? Mother reported criticism .002 .099 .005 .021 .983 Child reported Mother .017 .081 .047 .214 .832 Involvement Child reported Mother criticism .087 .099 . .207 .876 .388 T -score Extemalizing .123 .103 .206 1.197 .240 151 Table 4.13 Coeflicients for Regression of Criticism and Involvement on Intemality among Internalizing High-Risk Group of Children. Model 1 B Std. Error 13 t-test p Mother reported Involvement -.124 .048 -.330 -2.608 .013 Mother reported criticism .076 .054 .184 1.404 .168 Child reported Mother .060 .056 .172 1.068 .292 Involvement . Child reported Mother criticism .217 .070 .516 3.111 .003 ‘ Model 2 B Std. Error ‘73 t-test T Mother reported Involvement -.109 .046 -.291 -2.356 .0? Mother reported criticism .088 .052 .212 1.669 .103 Child reported Mother .048 .054 .139 .888 .380 Involvement Child reported Mother criticism .154 .074 .366 2.079 .044 T -score Extemalizing .139 .068 .271 2.027 .050 152 Table 4. [4. Mediator Eflect of SOC on the Relationship between Child’s Report of Mother ’s Involvement and YSR Internalizing Behavior Scores Testing steps in mediation model B SE 13 95% CI ,6 Testing Step 1 _ Outcome: YSR Internalizing score Predictor: child report of mother’s -.271 .042 -.353,-.188 -.364*" Involvement toward child TEesting Step 2 (Path a) Outcome: SOC Predictor: child report of mother’s .176 .023 .132, .221 .420m Involvement toward child Testing Step 3 (Paths b and c’) Outcome: YSR lntemalizing SCOI'C Mediator: soc -.934 .095 -1.120,-.747 -.526"“ Predictor: child report of mother’s -.106 .040 -.184, -.028 -. 143 Involvement toward child Note. CI = confidence interval; SOC = Sense of Coherence. ‘p < .01; ”P < .005; "'P < .001. 153 Table 4.15. Mediator Eflect of SOC on the Relationship between Child’s Report of Mother ’s Involvement and YSR Extemalizing Behavior Scores Testing steps in mediation model B SE B 95% CI )6 Testing Step 1 (Path c) Outcome: YSR Extemalizing score Predictor: child report of mother’s -.225 .044 -.311,-.139 -.297‘" Involvement to child Testing Step 2 (Path a) Outcome: SOC Predictor: child report of mother’s .176 .023 .132, .221 .420m Involvement to child Testing Step 3 (Paths b and c’) Outcome: YSR Extemalizing score Mediator: SOC -.853 .103 -1.056,-.651 -.472“" Predictor: child report of mother’s -.075 .043 Involvement to child -.l60, .010 -.O99 Note. CI = confidence interval; SOC = Sense of Coherence. ‘P < .01; "p < .005; "’p < .001. 154 Table 4. 16. Involvement and YSR Internalizing Behavior Scores Testing steps in mediation model B SE B 95% CI ,8 Testing Step 1 (Path c) Outcome: YSR Internalizing SCOI'C Predictor: child report of mother’s -.271 .042 -.353, -.188 -.364" Involvement to child Testing Step 2 (Path a) Outcome: SE Predictor: child report of mother’s .337 .034 .269, .405 .504" Involvement to child Testing Step 3 (Paths b and C’) M Outcome: YSR lntemalizing SCOI‘C Mediator: SE -.221 .071 -.361,-.080 -.l99” Predictor: child report of mother’s -.196 .048 -.290, -.103 -.264m Involvement to child Note. CI = confidence interval; SE = Self-Esteem. ’p < .01; "p < .005; "‘P < .001. 155 Table 4.1 7. Testing steps in mediation model B SE B 95% CI ,6 Testing Step 1 (Path CM Outcome: YSR Extemalizing score Predictor: child report of mother’s -.225 .044 -.311, -.139 -.297" Involvement to child Testing Step 2 (Path a) Outcome: SE ## Predictor: child report of mother’s .337 .034 .269, .405 .504 Involvement to child Testing Step 3 (Paths b and c’) Outcome: YSR Extemalizing SCOI'C Mediator: SE -.213 .075 -.360,-.067 -.l89" It Predictor: child report of mother’s -.153 .050 -.251, -.055 -.202 Involvement to child fl ‘ Note. CI = confidence interval; SE = Self-Esteem. " p < .005; "”P < .001. 156 Table 4.18. Mediator Efi'ect of SOC on the Relationship between Child Report of Mother ’s Criticism and YSR Internalizing Behavior Scores TTesting steps in mediation model B S Testing Step 1 (Path 0) Outcome: YSR Internalizing Predictor: child report of mother’s .352 criticism to child E B 95% CI .04 00 .258, .446 Testing Step 2 (Path a) .407". Outcome: SOC Predictor: child report of mother’s criticism to -.212 child .026 -.263, -.161 -.436’" Testing Step 3 (Paths b and c’) Outcome: YSR lntemalizing Mediator: SOC -.895 Predictor: child report of mother’s . 1 62 criticism to child ‘ .095 -1.082, -.709 .046 .071, .253 -.504"" 0*! .188 Note. CI = confidence interval; SOC = Sense of Coherence. ‘P < .01; "p < .005; "’p < .001. 157 Table 4.19 Mediator Eflect of SOC on the Relationship between Child’s Report of Mother ’s Criticism and YSR Extemalizing Behavior Scores E Testing steps in mediation model B SE B 95% CI ,6 Testing Step 1 (Path c) Outcome: YSR Extemalizing Predictor: child report of mother’s m , , , , .380 .048 .286, .475 .433 cr1t1c13m to chlld Testing Step 2 (Path a) Outcome: SOC Predictor: child report of mother’s m -.212 .026 -.264, -.161 -.436 criticism to child Testing Step 3 (Paths b and 0’) Outcome: YSR Extemalizing Mediatorr SOC -.725 .100 -.923,-.528 -.402‘“ Predictor: child report of mother’s .226 .049 .130, .323 .258’" criticism to child Note. CI = confidence interval; SOC = Sense of Coherence. ‘P < .01; "p < .005; "‘p < .001. 158 Table 4. 20. Effect of SE on the Relationship between Child’s Report of Mother ’s Criticism and YSR Internalizing Behavior Scores Testing steps in mediation model B SE B 95% CI ,8 Testing Step 1 (Path c) Outcome: YSR Internalizing score Predictor: child report of mother’s m . . . . .352 .048 .258, .446 .407 cr1t1Clsm to child Testing Step 2 (Path a) Outcome: SE Predictor: child report of mother’s -.264 .043 -.349,-.178 -.340"” criticism to child Testing Step 3 (Paths b and c’) ‘ Outcome: YSR Internalizing SCOI'C Mediat°f=SE -.243 .064 -.368,-.118 -.218‘" Predlctor: child report of mother s .288 .049 .191, .385 .333... criticism to child Note. CI = confidence interval; SE = Sense of Coherence. ’P < .01; "p < .005; ”‘p < .001. 159 Table 4.21. Mediator Eflect of SE on the Relationship between Child’s Report of Mother ’s Criticism and YSR Extemalizing Behavior Scores Testing steps in mediation model B SEB 95% CI 15’ T TTe$ting Step 1 (Path c) T Outcome: YSR Extemalizing score Predictor: child report of mother’s .380 .048 .286,.475 .433‘" criticism to child TTe$ting Step 2 (Path a) Outcome: SE Predictor: child report of mother’s criticism to -.264 .043 child Testing Step 3 (Paths b and c’) -.349, -.178 -.340‘" Outcome: YSR Extemalizing score Mediator: SE -.184 .065 -.311,-.056 -.162" Predlctor: child report of mother s .332 .050 .233, .431 .377... criticism to child Note. CI = confidence interval; SE = Sense of Coherence. ’p < .001; ”p < .005; "p < .001. 160 Table 4. 22. Effect of SE on the Relationship between Mother ’s Report of Her Own Criticism Directed toward Child and YSR Internalizing Behavior Scores T Testing steps in mediation model B SE B 95% CI )6 Testing Step 1 (Path c)A Outcome: YSR lntemalizing score Predictor: mother report ofher own .167 .053 063,271 .188" Criticism to child Testing Step 2 (Path a) Outcome: SE Predictor: mother report of her own -.124 .047 Criticism to child Testing Step 3 (Paths b and c’) -.217, -.031 -.155" Outcome: YSR Internalizing score Matrimony; -.350 .064 -.476,-.224 -.3l3"" Predictor: mother report of her own .125 .051 Criticism to child .025, .225 .141‘ Note. CI = confidence interval; SE = Self-Esteem. ’p < .01; "p < .005; ”*p < .001. 161 Table 4. 23. Correlations between YSR Scales and SE Subscales Derived fiom Component Analysis. T-SCORE T-SCORE Scale EXTERNALITY INTERNALITY Positive Self-image -.208m «197" Negative Self-image -.338m -.480m Self-criticism -.231m -.223m "'10 < .0013 “p < .002 162 —- BOY --- GIRL 62 _ 58. Child’s Report of Mother’s Involvement 56' Two parent Two parents + I Single mother/w/wo family Extended family Extended family Family Structure Fig. 4.1. Child Report of Mother’s Involvement by Family Structure - 163 Gender .5 es -BOY U .5 .u g ‘6 .s 2 r- ~ .2. . e s g m E 24 P E U 22 1'" 7 T TI 1 , Two parent Two parents + Single mother/w/wo family Extended family Extended family Family Structure Fig. 4.2. Child Report of Mother’s Criticism by Family Structure 164 .11:- iii . --flul..lrll h 0 '5, 5‘: 62.5 ~ T BOY 1: . --- GIRL i 3 . g 60.0 - 0 5 i: ' g: ,5: 57.5 _ ‘E‘ E. S" 55.0 — “a 2. 6 52.5 — _ I _ l , ' Two parent Two parents + Single mothedw/wo family Extended family Extended family Family Structure Fig. 4.3. Child Report of Involvement Toward Mother by Family Structure 165 166 § 9.8 1 Gender g . — BOY : ..... Z '5 9.5 - GIRL “I 5‘ E 2;: 9.3 _ t; if “a 9.0 _ b w '5 e E 8.8 _ Two parent Two parents + TSingle mother/w/wo famrly Extended family Extended family Family Structure F rg 4.4. Mother 8 Family Relationship Inventory by Family Structure 66 L Gender - —BOY ---GIRL 54 .. .5. E all; 52 E” g a. N 5:1 “a E 5 48 _ 46 _ ' Lower 25 50th 75th Upper 25th percentile percentile percentile percentile Child ’8 Report of Mother’s Involvement Figure 4.5. Moderating effect of gender on the relationship between mother’s Involvement and Extemality. 167 66 - Gender in- BOY --- GIRL 54 .— E «E 2 a 52 . i '11 g 50 _ “a g a: U 48 _ 46 t. ‘ Lower 25 * 50th 75th Upper 25th percentile percentile percentile percentile Child’s Report of Mother’s Involvement Figure 4.6. Moderating efl’ect of gender on the relationship betweenmother’s Criticism and Intemality. 168 Gender 6° l. -— BOY --- GIRL ' 55'" - Z’. '3 a .E 0 m an l 25 5° ' a i l't' [I] “a g 45 -‘ a U 4 4o _ I j I I Lower 50th 75th Upper 25 percentile percentile 25th percentil - percentile Child’s Report of Mother’s Criticism Directed toward Child Figure 4.7. Moderating effect of gender on the relationship between mother’s Criticism and Extemality. 169 lntemalizing Behaviors (.333'”) Mother‘s m . . . -.218 Cntrcrsm . ward SE “(gm -340 l ' ..162” (.377“) Extemalizing .433’" Behaviors Figure 4.11. Child’s SE mediates the relationship between child’s report of mother’s Criticism directed toward mother and child’s lntemalizing and Extemalizing behavior problems. Note. The regression weights in the parentheses are new standardized coefficients (beta weight) before the mediator is added to the model. 'p < .01; "p < .005; "‘p < .001. 173 1 . Depression Mother’s 0.486 0.759 Criticism 1 Mother’s Child’s , Her Own 63 Perception - —o.779 . . . Perce tlo , Crrtrcrsm . p 11 Mother 8 _ .68 Involvement 1 -0.650 @ Her Own 1 Involvement '5 6 Family Functioning @ a 1 Family 1 Functioning . 064 ”'5“ -.06 -0.659 a $1 1 Sense Of 0-12 0.473 0.104 1 34 SCIf- 914 Coherence Esteem \ 0.727 .812 .822 Self- Meaningfulnes 0.72 Competence ' ' Self-Liking 912 Manageability \ -0.935 -1985 0.745 0.697 1 @ Comprehensibility 6 @ ’L'J e20 1 Extemalizing Internalizing 1 e21 0.785 0.788 0.481 0.559 0.878 Anxiety Delinquency Aggression Withdrawal Somatlzatlon Depression e e] Figure 4.12. Full model showing a]! standardized pa th coefi‘icien ts. 174 Depression ' i T» Mother‘s Her Own - _ l Criticism Perceptions Her Own l_Involvement *— Family Functioning Sense of Coherence 0.473 Meaningfulness Manageability 0.935 -108 Comprehensibility Extemalizing Delinquency l Aggressicfl Withdrawal Somatization Figure 4.13. SEM showin latent variables. Significant nega significant positive pathways, obliquely. 175 g predictor variables and si Mother’s Criticism Chi ld's _ , .. . Perceptions Mother’s Involvement Family Functioning -0.659 Self—Esteeem p i Self- Competence rx Self-Likin «‘10 3 g Q‘ o Internalizing Anxiety Depression gnificant pathways between tive pathway coefficients are shown horizontally; Depression Her Own Criticism Her Own Involvement Family Functioning Sense of’ Coherence [ Meaningfulness Manageability Comprehensibility Mother‘s f Perceptions -0.183 *0-247 -1.191 Extemalizing Delinquency Aggression tort“ -O.321 Self-Esteeem Self- Competence Self-Liking J +0693 lntemalizing Withdrawal Somatization Anxiety Depression Figure 4.14. SEM excluding child report data showing predictor variables and significant path ways between latent variables. Significant negative path way coefiicien ts are shown horizon tall y; significant positive path ways, obliquely. 176 CHAPTER V DISCUSSION AND CONCLUSIONS Demographic Information Sample Characteristics Several findings of interest emerged regarding the characteristics of children who participated in the present study. Compared to normative Japanese samples (Itani et al., 2001; Tejima et al., 1994; Tejima et al., 1995; Tejima etal., 1996), only 2% of the present sample should be in the clinical range on each YSR scales (i.e., above 98th %-ile), and only 4% of the present sample should be in the borderline clinical range (i.e., between the 94m and 98th %-ile). However, the prevalence rates for lntemalizing behavior problems were 8.3% (clinical range) and 8.0% (borderline range), with girls less likely to be in the borderline range (6.1% vs. 10.1% for boys) and more likely to be in the clinical range (10.2% vs. 6.2% for boys). The prevalence rate for lntemalizing behaviors in the clinical range is similar to that reported for a large sample of elementary school Japanese children by Denda (2007), using the Birleson Depression Scale (7.8%). The prevalence rates for Extemalizing behavior problems were 8.3% (clinical range) and 5.1% (borderline range), with girls more likely to be in the borderline range (6.8% vs. 3.1% for boys) and slightly more likely to be in the clinical range (8.8% vs. 7.8% for boys). It is a matter of concern that 19 of the children in the sample (6.9%) had both lntemalizing and Extemalizing scores in the clinical or borderline clinical range, making up 42.2% of the high-risk Internalizing group and 51.4% of the high-risk Extemalizing group. Internalizing and Extemalizing scores are not independent. The positive relationship between the two scales (r = .593) in the present study allows the prediction that children 177 with high—risk scores on one scale would have high-risk scores on the other scale. However, these 19 children would be placed in a double risk category because of their high scores on both YSR scales. It would seem that these children would be prime candidates for interventions either at school or through community therapists. Gender Differences Gender differences in lntemalizing and Extemalizing. Although more girls (absolute number and percentage) than boys had symptoms in the high-risk range on both Internalizing and Extemalizing , no statistical difference was found between boys and girls in scores on either scale or in the gender distribution between borderline clinical and clinical groups. Further, there was no difference in behavior problems by age. These results differ from most previous studies showing that boys have more Extemalizing behavior problems than girls (American Psychiatric Association, 2005; Rothbaum & Weisz, 1994), and girls start showing more depression-like symptoms than boys around age of 10 (Angold and Rutter, 1992). It is worth noting that norms for the YSR have typically been assumed to be parallel to those of the CBCL (e.g., Itani et al. 2001); however, it is possible that this is not the case in Japanese samples. Parents of children and adolescents, ages 6-17, in 12 cultures reported in Internalizing and Extemalizing behavior symptoms (Crij nen, Achenbach, & Verhulst, 1997). However, parents across cultures reported that girls have more symptoms than boys on somatic complaints and anxious/depressed (lntemalizing symptoms), but fewer symptoms on attention problems, delinquent behavior, and aggressive behavior (Extemalizing symptoms). Crijnen et al. do not provide information on gender differences by age. The absence of gender or age differences in the present 178 study suggests that the gender difference in 11 and 12 year-old Japanese (85% of the children in this study) has not emerged yet. Possible sampling bias. It is important to remember that the present participants were volunteers recruited from public schools. Variation by family structure and age was limited. Most of the participants came from two-parent families, and only a few of children were 10 or 13 years old. There is also a disadvantage in that 45% of children eligible for the study and 64% of mothers did not participate in the study. Overall, 285 mother-child pairs were formed from the data, representing 27.3% of all possible participants. This is a limitation of the study in that the characteristics of the missing pairs are? not reflected to the results. Future research will examine any differences between the children in the present sample versus the children whose mothers consented for their children but did not participate themselves, and the mothers who completed the survey and the mothers who completed the survey but did not consent for their child to participate. Gender and self-esteem. The predicted gender difference was found in the level of self-esteem: girls had significantly lower self-esteem than boys, consistent with the findings in many other studies. The lower self-esteem in Japanese girls may be related to cultural practices that suppress the development of self-esteem for girls. In the period of middle childhood, social relationships become more complicated, and gender role expectations are intensified (Bolognini, Plancherel, Bettschart, & Halfon, 1996). It becomes harder to maintain the sense of self in the peer group and other relationships because the development of self-concept or self—esteem relies to large extent on success in maintaining a sense of group acceptance, fitting in with others, and maintaining 179 interpersonal harmony with social roles (Kitayama et al., 1997; Markus & Kitayama, 1991). In the single-mother family, those cultural and social expectations would be expected to impact in an additive manner on girls’ self-esteem. Considering that Japanese girls are expected to take more responsibility for helping out their mothers (Kojima, 1986), single mothers were reported by their children to be more critical of their daughters than sons. Single mothers reported that their daughters expressed more criticism than their sons, and their daughters reported more maternal criticism and less positive involvement. The result reflects the reciprocal negative interaction between mother and daughter, each expressing more criticism and less involvement with the other in a single-mother home. Family Environment Factors, Resiliency Resources, and Internalizing and Extemalizing Behaviors General Findings The predicted relationships among the variables in the full sample were found for child reports in terms of zero-order correlations. Children who perceived a negative family environment were more likely to have a higher score on lntemalizing and Extemalizing behaviors. Those children were also more likely to have a lower self-esteem and less well—developed sense of coherence. Girls showed higher correlations between maternal emotional expressions and other variables (e.g., between higher Criticism and more Internalizing) than boys. This difference reflects both family structure and the reality that mothers and daughters in most cultures have more frequent and more intense interactions than mothers and sons. Compared to child’s perceptions, mother perceptions were less 180 related to children’s self-report of SE and SOC, and lntemalizing and Extemalizing behaviors. Child Perceptions of Maternal EE and Behavior Problems: Comparisons between Normal-Risk and High-Risk Groups of Children There were significant differences between normal-risk and high-risk groups in child perceptions of his or her mother’s Criticism and Involvement directed toward him/herself. Children in the high-risk group perceived their mothers as being more critical and hostile and showing less positive involvement than children in the normal-risk group. BE studies have found that mothers of children with behavior problems engage in more antagonistic and negative comments and expression of disgust (i.e., criticism), compared to mothers with low or marginal levels of EB (Mch et al., 2004; McCarty & Weisz, 2002). The results of the present study are consistent with the earlier studies and confirm the connection between high maternal Criticism and increased lntemalizing and Extemalizing behaviors. The reciprocal negative perception of Criticism between mother and child was found in the high-risk group of children and their mothers. Children who live in a family environment with high levels of maternal Criticism would be predicted to have a negative information processing style (cognitive diathesis) and to be less sensitive to positive comments fiom their mothers. Such children would be more likely to develop behavior problems (e.g., Kwon & Laurenceau, 2002; Turner & Cole, 1993; Hankin, Abramson, & Silar, 2001). Although it was hypothesized that mother and child perceptions of EB would have different relationships to that child’s behavior problems, the child reports of mother’s 181 Criticism were related to both lntemalizing and Extemalizing behaviors. The mother reports of her own Criticism were related only to Internalizing behaviors. The results of the present study are consistent with the results of many studies indicating the relationship between high EE Criticism and Extemalizing and Internalizing problems (Asamow et al., 2001; Hirshfield et al., 1997; McCarty et al., 2004; McCarty & Weisz, 2002; Stubbe et al., 1993; Vostanis et al., 1994). In the full sample, a lower level of maternal Involvement was predictive of both Internalizing and Extemalizing behaviors. However, the level of Involvement was more strongly related to lntemalizing than Extemalizing behaviors. There was a significant difference, as well, in child perception of mother’s Involvement between normal-risk group and high-risk groups of children. High-risk children reported less Involvement. There was a significant difference in boy’s Internalizing behaviors between the two risk groups, but girls in the high-risk group and those in the normal-risk group did not differ in terms of lntemalizing behaviors. Further, a negative relationship between mother’s Involvement and boy’s lntemalizing behaviors was found. This finding is important because, although girls are more vulnerable to depression (Abela, 2001; Abela & Payne, 2003; Abela &Taylor,, 2003; Angold & Rutter, 1992; Cicchetti & Toth, 1998; Hankin et al., 2001; Hankin & Abramson, 2001; Holsen, Kraft & Vitterso, 2000), boy’s internalizing behaviors tend to be overlooked. Further investigation of the expression of Internalizing behaviors (somatization, withdrawal, depression/anxiety) in Japanese boys would be useful. Maternal High Criticism and Low Involvement as a Risk Factor 182 Mother and child perceptions of maternal Criticism and Involvement were investigated as predictors of child’s lntemalizing and/or Extemalizing behaviors in the full sealple. Mothers evaluated their own Criticism and Involvement directed toward their child, and children evaluated their mother’s Criticism and Involvement. After controlling for Extemalizing behavior scores, only the child reports of mother’s Involvement was a significant predictor of lntemalizing behaviors. When the same variables were used to predict Extemalizing behaviors while controlling for Internalizing scores, only the child perception of mother’s Criticism was a significant predictor. ’ In the high-risk lntemalizing group, the results were somewhat different. With Extemalizing scores controlled, mother reports of her own Involvement with her child and her child’s perception of her Criticism both predicted a higher score on lntemalizing symptoms. Next, predictors of Extemalizing behavior problems in the Extemalizing high- risk group of children were assessed. No significant predictor was found when lntemalizing scores were controlled. In summary, research is needed to further examine the effects of mother Criticism and Involvement on Internalizing and Extemalizing behaviors. The full sample results follow expected lines with child reports of low levels of Involvement related to Internalizing behaviors and high levels of Criticism related to Extemalizing behaviors. However, in the high-risk Internalizing group, the mother’s own report of her Involvement and the child’s report of her Criticism were related to lntemalizing. This finding suggests the mothers of these high risk children realized that'they were less involved with their children. Further, the absence of significant predictors of Extemalizing in the high-risk Extemalizing group suggests a positive feedback 100p in 183 which the child’s delinquent or aggressive behaviors feed back to produce still more problematic behaviors without reference to maternal behaviors. While maternal reports of Criticism toward the child were a weak, but significant, predictor of Internalizing behaviors, they accounted for less than 4% of the variance in Internalizing behaviors. This was also true for maternal reports of the child’s Criticism of herself. Only mother reports of her child’s Criticism of herself was a significant predictor of her child’s Extemalizing; again a weak but significant correlation was found (< 2% of variance in common). That is, although mother reports of her own criticism of the child . were linked to the child’s depression-related symptoms (Internalizing), the mother’s experience of criticism from her child was the only maternal variable linked the child’s acting-out behaviors (Extemalizing). Cultural issues may be important here as well. Although a Japanese mother is expected to love and support her child, she is also expected to provide very explicit criticism to guide her child, and would be expected to be quite reactive to any negative behavior directed toward her from the child. The Constructs of Maternal Criticism and Emotional Involvement in Japanese Culture The validity of a measure in one culture is always an important issue when the measure was created in different culture. Mother Involvement on the EEAC originally was aimed to assess the mother’s excessive worries about the child in an extremely enmeshed or symbiotic-like relationship. Using other assessments (CFI or FMSS), high level of Emotonal Overinvolvement has been reported to be related specifically to anxiety disorders in both Western and Japanese studies (Chambless & Steketee, 1999; Stubbe et al., 1993; Yoshida, 2001; Hirshfield et al., 1997). However, the present study did not produce this result. The construct validity of EOI is sensitive to cultural values and .184 beliefs. The normative standard in Japanese culture for a mother’s involvement in her child’s life differs from Eur0pean-American culture because the Japanese mother’s emotional expressions are established in the culture imperative that defines a mother’s responsibilities, beliefs, and expectations for emotional closeness in a relationship. The Japanese conception of a child’s development is that great potential characteristics and abilities are given equally to each newborn baby (Takata, 1987). These potential gifts are actualizcd through the mother’s proper support and guidance. Japanese mothers accept that it is their responsibility to provide the support and motivation for their children to achieve in school and, ultimately, to meet the societal and familial expectations for success. Reciprocally, Japanese mothers also gain satisfaction by devoting themselves to their child’s success which structures a physically and emotionally close mother-child relationship. Maternal expression of Criticism in Japan is also established to teach and maintain the cultural norms and values. For example, “Don’t be so arrogant, there are many people who are much better than you, so keep working” is a fiequently-repeated slogan for mothers. In Japanese society, being hard worker, showing effort and perseverance, and maintaining group harmony through acceptance of social roles and conforming to norms are all very important values that children need to learn in this age period. Thus, criticism for a Japanese child and mother is for self-improvement and does not convey a negative message directed toward child. Mother-child interaction is also different by the age of the child. The way a mother responds to her child in middle childhood would be very different than the way she would respond to an adolescent or adult child with a psychological diagnosis (Mino 185 et al., 2001; Nomura et al., 2005; Shimodera et al., 2002; Yoshida, 2001). The norms for appropriate interaction between mother and child must be considered based on child’s development of chronology as well (Bronfenbrenner, 197 9). Research is need to establish better instruments to assess Japanese mother-child interactions. Moderation Effects of Gender Moderation analysis allows the researcher to identify relationships between predictor variables and outcomes that are stronger for people with differing characteristics (Frazier et al., 2004). Different levels of Sense of Coherence or Self- Esteem did not moderate the relationships between different levels of family environment and behavior outcomes in either child and mother reports. However, the effects of mother’s low Involvement and Criticism on child’s outcome behaviors were different between girls and boys. Girls were more strongly influenced than boys by mother’s Involvement and Criticism (child’s report). Specifically, when mothers were more positively involved, girl’s Extemalizing behaviors were significantly reduced. When mothers were disengaged with girls, girls’ Extemalizing behaviors started accelerating, while boy’s Extemalizing behaviors kept increasing linearly with decreasing levels of maternal Involvement. Mother Criticism (child’s report) was related differentially to both lntemalizing Extemalizing behavior problems. The direction of the relationship between mother Criticism and behavior outcomes was the same between genders: the higher mother’s Criticism was, the more severe the lntemalizing and Extemalizing problems were. However, compared to boys, the magnitude of the influence from mother’s Criticism was stronger for girls. 186 In summary, Japanese girls’ and boys’ lntemalizing and Extemalizing behaviors change at different rates in relationship mother Involvement and Criticism. Girls were more negatively influenced than boys by the lowest level of mother Involvement and highest level of Criticism. This is consistent with research showing that girls are more sensitive and vulnerable to mother’s negative remarks and warm encouragement in this age range. Mediation Effects and Resiliency Factors The present study addressed the likely mechanisms through which family environment factors relate to lntemalizing and Extemalizing behavior problems. The constructs of Sense of Coherence (SOC) and Self-Esteem (SE) as child resiliency factors were hypothesized to mediate between negative family environment factors and Internalizing and Extemalizing behaviors. According to a series of regression analyses, SOC and SE served as mediating variables, accounting for a significant portion of the direct effect between family environment and child behavior outcomes. Because the child report of mother’s Criticism and Involvement were still significant predictors of child behavior outcomes after the mediating variable was introduced, SOC and SE served as partial mediatos for the relationship. SOC and SE appeared to protect children from lower levels of Involvement and higher levels of Criticism. Even though children are affected by negative family environments, children with higher levels of SOC and SE seemed to have better ways to handle these negative effects. Child’s SOC and SE contributed as protective factors in a dynamic process of adaptation to a risk setting. 187 However, the model fit analysis using SEM demonstrated that SOC and SE have different effects on the relationship between family environment and behavior problems. To the extent that a causal model is assumed, SOC absorbed the effects of the child’s perception of negative family environment factors and protected the child from having lntemalizing or Extemalizing behaviors. Children with a better-established sense of coherence are cognitively and emotionally capable of understanding the nature of problems and are more willing to confront them. As such, SOC is a construct that cuts across many cultures, Eastern and Western. In order to cope with a problem, the child must be able to comprehend the prOblem (culturally-specific), find meaning in it (culturally-specific), and find a way to manage the problem (also, culturally-specific). The Japanese translation of the SOC scale necessarily incorporates the Japanese cultural concepts reflecting these skills. On the other hand, child SE contributed to increasing lntemalizing and Extemalizing behaviors. In zero-order correlation and simple mediation tests, SE was a negative predictor and mediator of the effect of maternal Criticism or Involvement on YSR scores. Similar findings would be expected when only these simple regression models are used. However, when included in the SEM, total SE score was an “inconsistent” mediator. Statistically speaking, the correlations between child’s SCO and SE and between lntemalizing and Extemalizing behaviors might caused problems in estimating SE that actually heightened the scores on behavior problems. Self-esteem in Japanese culture does not necessarily correspond to the Western description of self-esteem as how well individuals evaluate their overall worth as a person based on the positive aspects of the individual self (Rosenberg, 1979', Harter, 188 1999). In Japan, self-criticism is a tool for the self-improvement, and a willingness to be critical of one’s self is taken as an individual strength. Japanese children’s self-esteem will be more closely related to how well they are connecting with key persons, e.g., mother, because the value of the self is established by being a part of the family. Western self-esteem scales based on individual positive self-appraisal do not capture the Japanese sense of self or “jibun” which is created under a cultural imperative that self is found through relationships with others. Western self-esteem scales are missing the contextualization in place, event, and social group (Rosenberger, 1989). The results of the principal components analysis, used in the construction of the new self-esteem subscales confirmed this argument. Contrary to the Western image of negative self-image and self-criticism as factors that would diminish self-esteem, the negative self-image and self-criticism were related to less, not more, lntemalizing and Extemalizing behaviors in this sample. However, because the total score for the SE scale was used,these items actually were subtracted from the total SE scale. The analysis suggest very strongly that they should be added instead. More research is needed, examining how these important cultural elements regarding self-esteem among Japanese children are achieved through effort, perseverance, commitment, modesty, cooperation, obedience, and conformity to family and social group norms. Contributions of Mother’s Perceptions Mother perceptions of the family environment were not related significantly to either child resiliency or behavior problems when they were included in the SEM along with child perceptions. However, further SEM analyses showed that there were both direct of mother’s perceptions on lntemalizing and Extemalizing behavior outcomes and 189 indirect effects through SE and SOC. It is not surprising that child perceptions of a negative family environment have a stronger relationship with behavior problems. The children were describing their own perceptions and behaviors, some of which would not be obvious to their mothers. It would be interesting to have mothers complete the Japanese version of the CBCL and compare those results to the children’s responses on the YSR. The discrepancies between CBCL and YSR results would point to areas where mother and child reports might differ on other measures. In Japanese culture, children see themselves differently from European American children in terms of a construction of self. For Japanese children, a harmonious interrelationship with a great commitment to the expected roles may become an advantage to build a good sense of self because the sense of self is recognized by finding meaning through the group (Rosenberger, 1989). Positive emotions come from being part of a group and relating harmoniously (Kitayama, Markus, & Kurokawa, 2000). Children learn to vary their behavior based on context. That is, for Japanese children, presenting others with a social-self (including self-criticism), rather than an individual- self (working for self-enhancement), has “positive social and psychological consequences” (Kitayama et al., 1997, p. 1246). In middle childhood, Japanese children may have difficulties in situations that require achieving some balance between individual-self and social-self. Striving for autonomy and accommodation are competing skills for Japanese children, and achieving harmony is a value that may take precedence of achieving autonomy. Therefore, researchers must understand the construct of self-esteem to measure and carefirlly interpret child behavior problems in the context of the family 190 environment. Child behaviors in Japan can never be understood without contexualization in regard to place, relationship, and social group as well. Limitations main islands of Japan without some reservations because the samples were recruited in Okinawa. Although Japanese in ethnicity, culture, and values, Okinawa has a strong identity established by a long history of being isolated by distance from the main islands of Japan. Okinawans may be more likely to emphasize the importance of social networks and family support systems, to hold a strong sense of identity as an islander, and to be committed to involvement with the community. Some cultural differences do exist between “Northerners” and the population of Okinawa, as evidenced by the finding that Okinawan families were more likely to have large families. Although there is substantial agreement between the results of this study and previous studies in regard to the effect of maternal criticism and positive emotional expressions, care must be taken not to over-generalize to other Asian populations or to Western populations. The ways children interpret their mothers’ verbal and nonverbal communications differ between East and West and among Asian cultures. Although the schools selected for the study were not a random sample of Okinawan schools, they are reasonably representative of Okinawan schools. Duplicating this research with clinical samples where it is more likely to find extreme levels of EB will help in better understanding the relationship between family environment and children’s resiliency factors and behavior problems. 191 The validity of the EEAC as a measure of EOI is questionable (Hooley & Parker, 2006). It was developed to be a less time-consuming alternative to the CFI or F MSS, but it does not correspond well with those two measures. Therefore, it will be necessary to use the F MSS to assess the range of emotional involvement statements in Japanese mothers to establish a baseline. As noted, the use of the CFI with non-adults is limited to a single study on older Japanese adolescents with eating disorders that used the CF I (Yoshida, 2001 ). Yoshida reported that the mothers of these adolescents had high levels of EOI, but were less likely to have high levels of Criticism than mothers of adult patients with schizophrenia or mood disorders. However, the meaning of Yoshida’s results is not clear. Accepting Western standards for scoring EOI or Criticism in Japanese populations may not be useful; only further research can provide an answer. The present study was cross-sectional in nature, and, therefore, cannot draw definitive conclusions regarding cause-and-effect relationships between family environment and child behavior problems. It was not possible to include other, possibly important factors, such as the father’s experience in the family and the child’s peer relationships. This study was focused on the mother-child relationship in the context of family. Other research has shown that father’s positive involvement with the child is significantly related to decreased child behavior problems (Dunham et al., 2000). The construct validity of the measurements needs to be considered when using any scale developed in different cultures and among specific samples. The SE scale (Self- Linking/Self-Competence) was selected because it was thought to be more sensitive to the cultural differences in view of the self. The measurement assesses the overall evaluation of oneself as a source or agent and includes self-efficacy and autonomy on one 192 hand and overall sense of worth as an individual with social significance on the other. However, the present study provides clear evidence that culturally important elements of the Japanese social self are missing or mis-scored in the scale. Another measurement problem was the Criticism scale of the EEAC (Expressed Emotion Adjective Checklist). In Japan, self-criticism is used as a tool for self- improvement, and mothers also use criticism in the interactions with her child. However, the meaning and purpose of mother’s criticism are practiced under the rules of the culture. There are also difficulties in detecting mother’s involvement (emotional, physical, and social) to establish a normative standard. Further, the Youth Self Report has been translated into Japanese and is widely used. Different cut-off points have been identified for Japanese children. As is true for using the YSR in Japanese children, maternal expressed emotion (EE) among Japanese samples may need different criteria to distinguish a high or low level of Criticism, reflecting the different ways Criticism is displayed. The present study is the first step for understanding maternal EE and child’s development, and will be followed by an analysis of the interviews with mothers in Okinawa already collected. Further, the study relies on self-report measures to assess the variables of interest. Multiple measurements by different informants (e.g., teacher’ 3 report) to assess the behavior problems would allow for triangulation of the child reports. Similarly, observing mother-child interactions would allow for triangulation with the self-report data. Finally, only a single instrument developed in American samples initially was used to assess each variable; for example, the YSR was used to evaluate the children’s behaviors. Because of their origin in Western cultures, their fit with Japanese cultural norms can be questioned. 193 Clinical Implications There is an extensive research literature showing the correlation between maternal depression and child behavior problems, as well between maternal BE and child Internalizing and Extemalizing behaviors. However, as shown in this study, there are gaps between mother and child perceptions of family relationships and their mutual patterns of Criticism and Involvement. In the present study, mother reports were not associated strongly with either resiliency factors or behavior problems. In contrast, child reports of maternal behaviors had strong correlations with the child’s sense of coherence and self—esteem and with Internalizing and Extemalizing behaviors. Family therapy could help mother and child to minimize their perception discrepancies and come to a better understanding of their differences. Secondly, individual, group, and family therapies with activities or exercises to enhance the level of sense of coherence can be helpful to encourage the child to adopt I more appropriate strategies to the needs of different contexts. Further research regarding the sense of coherence is needed with children who have disadvantages in their lives, but who function well. How to enhance a sense of coherence in at-risk children is an important challenge for future research. Thirdly, there were differences in perceptions of mother’s Criticism and Involvement not only between normal-risk and high-risk children but also between boys and girls. In some aspects, high-risk children would be expected to be less responsive to maternal attempts at positive emotional connection, the result of feeling unfairly criticized and/or ignored by the mother. Girls were much more vulnerable to maternal negative emotionality and behaviors. Therapy interventions need to take those gender 194 differences and similarities into consideration when working with families, and explore the social and cultural expectation differences for boys and girls with mother and child. It is very important to acknowledge that changes in maternal or child behavior may not result in changes in the perception of mother or child because of prior negative experiences. This lack of responsiveness need to be considered before attempting to rebuild the mother-child relationship and change the perceptions of children and mothers about each other’s willingness to connect more positively. Importantly, the influences from cultural values need to be reflected in therapy practice. 195 APPENDICIES -¥Ebfl7y#—t%¥ (Child’s Questionnaire) SVfiVMEki 196 bfrfcwtl’ilifilcom'cfii't tfigirtbb iotUrw . Etta.bTKLthAKfioflbéfightltflbDiiW? b‘C'bJKbb «12(396 3:382’53926 ioTcO’Jtln . 9&1“. %fotTCODEEUDEP'C’t'Cbli/aérb bfcfitfebiboto k'Ctrffififi V‘< E3233‘E‘5E5 bibfibtrw $<%5Eb7£b‘ .aata.earaaaraawraeoswaommaaermaaw 5%45’15731326 C k112i?) 0 if??? t‘C'bJKbZi «12(326 £3396 iofco’fl‘ .hata.b€bk§5fiflwwmwekfiu.Efiihfirmwmbm savanna: tram arm? screams J:6 tirh’ébé ioTcfitéo 15 1$< - 197‘: L11. $51523 Akifi‘JKI/TCC 1: Kfififlfio 16 li< ' 3‘37”: Lli. fifiébllfiL’G‘BD‘B 5‘6, 17 15 < - 197‘: L11. tritium. 18 1£< ‘ 7137‘: L11. Rlébfiikfiéfl'ia 19 li< ° 397': L11. %fiébllfllx’ffififlg 9' “—o 20 lif< ° bf: L11. Bléfiullifl/‘Cfigifi E10, 202 A“. \ \ E E ‘7V5H1EE (Mother’s Questionnaire) iVfiVMfik$ 203 7V7~5K3E2wtfiwtfiAEEméTFEEEEKE76E205E1 KBwTfifiéniTo HTUD’EFEKR’E’K < 7‘53 1, x, 1. 2. %EEwfifi fifiébwfifi $8505 _____ ififiififiilcow'CEJfliiélfifilfi‘xy?L’C<7‘:fé1no [jfiifi [JEEEE [3:EEEE %fifditfimtmii’m D111» 1:] mini FanagitAmA.ETHEEEEEOchc<fiéwO CD BEE (2.) EtiE (1137155) ®E1iE (%a’atifi) (3) 21 535E @ %03161 56E11f51w0 1:111!» Elm/wt [111/‘1 tEEiti/Sré‘wlf. %Tliiéfififié’O'GlfiA/E1i.$1.1:‘E‘1\7‘:1\:&72110%DE 1 2 3 4 5 6 7 s 2.6, 6 fk’blifllCfiL’CEb‘R’Dz‘fiibéo 1 2 3 4 5 6 7 8 7 %EMitAmtsfiawamowfiga-ao 1 2 3 4 5 6 7 s 8 $8511.5116E77551Ur010160 1 2 3 4 5 6 7 8 9 %ggmjugfigog 1 2 3 4 5 6 7 s 10 fc‘f‘isli.fLUchbll—Efifhilttof11h01fi°éLo I 2 3 4 5 6 7 8 13 %Ebmjéfiiwfii, 1 2 3 4 s 6 7 s 14 581.11.71.1zocxarca. 1 7 3 4 5 6 7 8 15 $81311.naasnw::m:44§&ri, 1 2 3 4 5 5 7 8 207 l6 $811311. 11.1136fo 17 \BWBL'C 17 \6. 1 2 3 4 5 6 7 3 17 $81311. 722171156. 1 2 3 4 5 6 7 8 18 587511. fttcfltfiitaxflna l 2 3 4 5 6 7 8 19 $8511. $1.1:S61L’CEHBE7‘50 1 2 3 4 5 6 7 8 20 31-31611. EL1CS'1‘L‘CIE3705E1‘. 1 2 3 4 5 6 7 8 551i. Eff-3.735 03373727120)? E‘iglfidfi'éfifigé’fiw‘flj L'Ckéi'C < 7’52‘51‘0 37373272755? 3’11 kifi‘i‘éflfi. 809.1: 5 itfisbzli'fiifi LETM. é< J: i? 23% 1‘04) .Ebtrm % 5 .E. 5 ‘6 5 .E. 5 1. $1.11. 5158 E®:&>&§1j‘]\n~cm5° 1 2 3 4 5 6 7 3 2 71.13:. %E‘btfimmtcno’cmzfiMargrm 1 2 3 4 5 6 7 8 6° 3 711:. %k’EtcfiL'cigo—cwzpo 1 2 3 4 5 6 7 8 4 $1.11. %E‘tficfié’cmé, 1 2 3 4 5 6 7 8 5 71.1.1.sarcomaaoavaamm 1 2 3 4 5 6 7 s 2.11%.. 6 $1.11.%Emmrycgmaaoggo—cguc 1 2 3 4 5 6 7 8 106° 7 $L1i.¥-EEUJBEEU\€:J:O 1 2 3 4 5 6 7 8 8 7111.¥&*131:E7365'cr;6. 1 2 3 4 5 6 7 8 9 $1.11. %Ebl:lfi&o< 1 2 3 4 5 6 7 s 10 711.11. %Ehwtb1z—-$fifi1zfréo 1 2 3 4 5 6 7 8 11 $L11.¥E‘Etwo1;1¢fi fififimufififé,Ltmuwbafi7V7—1Lfiégsfifimw— PéiwfifiuAnf,fi¥5iEfiLTxa5xuiEéhtEW%mi fiiiTéwiTJ: 5Liéb<fififimfibi¢o Elfifi’éua 12B 22 El 0K1650 > EE%EK%MT6:2K6§LthEwt%é,7V5~1%E&,% fifi8ii§11§é§uz1fi7z - 212 ¥£Bwafii%kifigfififi 6%m«0188®@§% :07y7—ba,bak®@%&%é&ifi®%%éflfiféck EEWKfibn57V5—1fificfobatmfluéfifiébkw %fiwfifi,ifimowT,%ttflaféfisz5:&%@UT WélkfiohfwgfifibbiTo7V7—Lfififlfiflwfifil %EA%MTélkfi,%fitfifimfl&bé:kflfiéifogi t