1 MARRIED KOREAN WOMEN'S RESPONSES TO ... - CiteSeerX

0 downloads 0 Views 614KB Size Report
guiding theories used in the model by pointing out critical issues. ...... Many abused women do not perceive themselves as being abused ...... In contrast, a small sample size can mask the effect of large specification ... principal investigator; data will be stored in a lock drawer for five years for future ...... Education:______ Yrs.
1

MARRIED KOREAN WOMEN’S RESPONSES TO DOMESTIC VIOLENCE WITHIN THE FRAMEWORK OF SOCIO-CULTURAL CONTEXT By Myunghan Choi _________________________________________ Copyright © Myunghan Choi 2004

A Dissertation Submitted to the Faculty of the COLLEGE OF NURSING In Partial Fulfillment of the Requirements For the Degree of DOCTOR OF PHILOSOPHY In the Graduate College of Nursing THE UNIVERSITY OF ARIZONA 2004

2

The University of Arizona  Graduate College of Nursing As members of the Final Examination Committee, we certify that we have read the dissertation prepared by Myunghan Choi entitled Korean Women’s Responses to Domestic Violence Within the Framework of Socio-Cultural Context and recommend that it be accepted as fulfilling the dissertation requirement for the Degree of Doctor of Philosophy

______________________________________ Linda R. Phillips, PhD, RN, FAAN

_______________ Date

______________________________________ Sandra Cromwell, PhD, RN

_______________ Date

______________________________________ Kathleen Insel, PhD, RN

_______________ Date

______________________________________

_______________ Date

______________________________________

_______________ Date

Final approval and acceptance of this dissertation is contingent upon the candidate’s submission of the final copies of the dissertation to the Graduate College. I hereby certify that I have read this dissertation prepared under my direction and recommend that it be accepted as fulfilling the dissertation requirement.

Dissertation Director: Linda R. Phillips, PhD, RN, FAAN

Date

3

STAEMENT BY AUTHOR

This dissertation has been submitted in partial fulfillment of requirements for an advanced degree at The University of Arizona and is deposited in the University Library to be made available to borrowers under rules of the Library. Brief quotations from this dissertation are allowable without special permission, provided that accurate acknowledgment of source is made. Requests for permission for extended quotation from or reproduction of this manuscript in whole or in part may be granted by the copyright holder.

SIGNED: ______________________________

4

ACKNOWLEDGMENTS

Many individuals have contributed to the completion of this study. First, I want to thank Dr. Phillips, my dissertation chair who consistently supported me and provided her profound knowledge to complete this study. Her suggestions and comments regarding theory and research methodology throughout the chapters made me think. Dr. Phillips was phenomenally insightful and had a significant impact on the quality of the dissertation. Dr. Cromwell, my theory professor and a committee member contributed guiding theories used in the model by pointing out critical issues. Her short messages and notes improved the quality of my dissertation. Dr. Insel, my methodology professor, as well as my personal life mentor, fostered building my knowledge about Structural Equation Modeling. She has remained by my side through the years as not only a member of committee but also as a sister, helping me in finding solutions to my financial difficulties. I also would like to thank Laurie and Jim Carey, as my extended family who have taken care of my youngest daughter, Hosu when I was in my office struggling, writing, and revising dissertation chapters. Their love and thoughtful care is unforgettable. I would like to extend my deepest gratitude to Sandy McGinnis who has read this dissertation several times and did wonderful proofreading. She was always there with a smile whenever I needed her help. I also want to remember the support and encouragement of Dr. Rousseau who called me a phenomenal woman and encouraged me to continue to be a strong mother of two girls; and a Professor, Connie Trice who supported and encouraged me from the beginning of my first year until I finished the study. I must express my appreciation to my colleagues, especially, Beth Bonham, Mary Doyle, and Liz Greenberg who showed their love and kindness during the study.

5

Lastly, I want to thank my father, brothers, and their families who offered long distance support and concern during the PhD program. I also thank Jean who turned out to be a wonderful daughter, caring and helping others as a leader (10th grade) of the International Student Association at her school and Hosu who is very independent at her schoolwork, got on the honor roll at 7th grade, and never failed to catch the school bus getting up in early morning at 5 A.M. I am very grateful to everyone who supported me in this study. I love them all so much.

6

DEDICATION

This study is dedicated to several individuals. First, to my mother Wolsoon Kim, who is suffering from early stage dementia, arthritis, and Hwabyung. She raised four children successfully in the situation of domestic violence and has been a victim of domestic violence in the socio-cultural context even though she has often expressed her desire to leave her husband. I remember her crying “without your father I would live happily, enjoyably, and cheerfully.” Her choice to stay and be a mother to four children made her a vulnerable Confucian woman whose role was restricted to being a biological being. She has been a wonderful bridge by allowing us to see her husband abusing her motherhood and womanhood. Second, this study is dedicated to the many victims of domestic violence at the hands of their husbands for their contributions to making this study a reality. Many victims who were shaped as mothers, wives, sisters, daughters, and daughtersin-law in Korean society are still struggling to get family and social support and want to be independent from their abusers. They are almost invisible in the socio-cultural context, hoping to get out of the prevailing domestic violence.

7

TABLE OF CONTENTS

LIST OF TABLES............................................................................................................ 13 LIST OF FIGURES .......................................................................................................... 15 ABSTRACT...................................................................................................................... 16 CHAPTER ONE ............................................................................................................... 18 Scope of Problem.............................................................................................................. 20 Background of Domestic Violence from a Korean Perspective ....................................... 23 Confucianism ................................................................................................................ 24 Confucianism and Abuse. ......................................................................................... 25 Confucianism and Hwabyung. ................................................................................ 27 Confucianism and Abuse Intolerance.. ..................................................................... 29 Significance of Problem.................................................................................................... 31 Statement of Purpose ........................................................................................................ 33 Research Question ............................................................................................................ 34 Summary ........................................................................................................................... 34 CHAPTER TWO .............................................................................................................. 35 Overview of Model ........................................................................................................... 36 Theoretical Perspective................................................................................................. 36 Theoretical Framework................................................................................................. 38 Theoretical Model......................................................................................................... 39 Theoretical Underpinning and Hypothesis ................................................................... 42 Age and Power.......................................................................................................... 43 Education and Power. ............................................................................................... 43 Income (employment) and Power............................................................................. 44 Cultural Theory............................................................................................................. 45 Psychological-Relational Power and Collectivism................................................... 45

8

Psychological-Relational Power and Traditional Family Ideology.. ........................ 46 Social Exchange Theory ............................................................................................... 48 Psychological-Relational Power and Marital Satisfaction........................................ 48 Patriarchal Theory......................................................................................................... 50 Psychological-Relational Power and Attitude Toward Power Ascription. .............. 51 Summary ........................................................................................................................... 53 CHAPTER THREE .......................................................................................................... 54 Methodology ..................................................................................................................... 54 Research Design................................................................................................................ 54 Correlational Research Design ..................................................................................... 54 Target Population and Selection Criteria.......................................................................... 55 Data Collection Sites......................................................................................................... 55 Protection of Human Subjects .......................................................................................... 58 Data Collection Methods .................................................................................................. 59 Instruments........................................................................................................................ 60 Measurement Instruments From Literature .................................................................. 60 Collectivism Scale. ................................................................................................... 60 Traditional Family Ideology Scale............................................................................ 61 Marital Satisfaction Scale. ...................................................................................... 61 Attitude Toward Power Ascription Scale. ................................................................ 62 Hwabyung Scale. ...................................................................................................... 63 Newly Developed Instruments...................................................................................... 64 Korean Women’s Abuse Screening Tool (KWAST). .............................................. 64 Korean Women’s Abuse Intolerance Scale (KWAIS).............................................. 64 Instrument Development................................................................................................... 66 Preliminary Qualitative Interview……………………………………………………..66 Selection of Conceptual Model..................................................................................... 68 Explications of Objectives for Measurement Instruments............................................ 68 Development of Blueprint............................................................................................. 69 Construction of Measurement Instruments................................................................... 71 Item Set ..................................................................................................................... 71 KWAST. ............................................................................................................... 71 KWAIS................................................................................................................... 71 Scoring Rules ............................................................................................................ 72 KWAST. .............................................................................................................. 72

9

KWAIS................................................................................................................... 73 Cultural Equivalence..................................................................................................... 73 Determining Relevance and Function of Domestic Violence. ................................. 73 Translation of Instruments. ....................................................................................... 74 Newly Developed Instrument.. .............................................................................. 74 Validity Assessment...................................................................................................... 76 Face Validity. ....................................................................................................... 76 Content validity..................................................................................................... 76 Testing Translated Instruments................................................................................. 78 Re-Evaluation of Process.......................................................................................... 81 Procedure For Data Analysis ............................................................................................ 82 Assumptions of SEM.. .............................................................................................. 83 Procedure of Analysis. .............................................................................................. 84 Summary ........................................................................................................................... 85 CHAPTER FOUR............................................................................................................. 86 Results............................................................................................................................... 86 Description of Sample................................................................................................... 86 Description of Korean Abused Women........................................................................ 87 Age, Education, and Income..................................................................................... 87 Religion..................................................................................................................... 89 Health Status and Living Situation. .......................................................................... 89 Korean Women’s Abuse Screening Tool. ................................................................ 90 Korean Women’s Abuse Intolerance Scale .............................................................. 92 Collectivism. ............................................................................................................. 94 Traditional Family Ideology. .................................................................................... 96 Marital Satisfaction................................................................................................... 97 Attitude Toward Power Ascription........................................................................... 98 Hwabyung Scale ....................................................................................................... 99 Summary ......................................................................................................................... 102 Testing of Psychological Properties................................................................................ 103 Reliability Assessment................................................................................................ 103 KWAST & KWAIS. ............................................................................................... 104 Collectivism Scale. ................................................................................................. 105 Traditional Family Ideology Scale.......................................................................... 105 Marital Satisfaction Scale... .................................................................................... 105 Attitude Toward Power Ascription Scale. .............................................................. 105 Hwabyung Scale. .................................................................................................... 105 Validity Assessment.................................................................................................... 106

10

Correlations............................................................................................................. 108 Confirmatory Factor Analysis................................................................................. 109 Theoretical Assumptions. ................................................................................... 111 Missing Data ................................................................................................................... 114 Impact of Missing Data............................................................................................... 117 Missing Data and SES. ........................................................................................... 117 Missing Data and SEM.. ......................................................................................... 119 Mechanisms of Missing Data ...................................................................................... 117 Missing Completely at Random (MCAR). ............................................................. 120 Missing at Random (MAR) .................................................................................... 120 Nonignorable (NI)................................................................................................... 120 Missing Data Analysis ................................................................................................ 121 Missingness by Location. ....................................................................................... 121 Missingness by Age. ............................................................................................... 122 Missingness by Religion. ........................................................................................ 122 Missingness by Health Status.. ............................................................................... 123 Missingness by Living Status. ................................................................................ 123 Missingness by Income Reporting.......................................................................... 124 Missingness by Predicted by Model Variables....................................................... 125 Missing Data Analysis of Income by Little’s MCAR Test..................................... 127 Treatment of Missing Data ……………………………………………………….…121 Full Information Maximum Likelihood……………………………………………128 Test of Normality Assumption.................................................................................... 130 Implications of Normality....................................................................................... 130 Normality Test.. ...................................................................................................... 130 Summary ......................................................................................................................... 132 Structural Equation Modeling......................................................................................... 133 Hypothesis................................................................................................................... 134 Overall Hypothesis.................................................................................................. 134 Specified Relations in The Model........................................................................... 135 Measurement Model ................................................................................................... 136 Model Summary of Measurement Model. .............................................................. 136 Summary of Parameters of Measurement Model. ................................................. 137 R2 of Measurement Model ...................................................................................... 137 Measurement Model Testing .................................................................................. 139 Structural Model ......................................................................................................... 141 Model Summary of Structural Model. .................................................................... 141 Summary of Parameters Structural Model.............................................................. 141 Standardized Regression Weights of Structural Model. ......................................... 142 Multiple Squared Correlations of Structural Model ............................................... 145 Structural Model Testing. ....................................................................................... 146

11

Summary ......................................................................................................................... 149 CHAPTER FIVE ............................................................................ ……………………150 Summary and Discussions of Findings........................................................................... 150 Research Question 1. .................................................................................................. 150 What are the psychometric properties of two measurement instruments?. ............ 150 Research Question 2. .................................................................................................. 152 What is the explanatory power of a theoretical model?.......................................... 152 What are the relationships among abuse, Hwabyung and abuse intolerance?........ 154 Abuse and Abuse Intolerance.............................................................................. 154 Abuse and Hwabyung. . ...................................................................................... 155 Hwabyung and Abuse Intolerance.. .................................................................... 156 Psychological-Relational Power and Abuse Intolerance.. ................................. 156 Summary.. ........................................................................................................... 157 What are the relationships among psychological relational power, abuse, and Hwabyung? ............................................................................................................. 158 Psychological-Relational Power and Abuse....................................................... 158 Psychological-Relational Power and Hwabyung. .............................................. 159 Summary. ............................................................................................................ 159 What aspects of psychological relational power mainly contribute the abuse intolerance? ............................................................................................................. 160 Psychological-Relational Power and Collectivism. ........................................... 160 Psychological-Relational Power and Traditional Family Ideology................... 160 Psychological-Relational Power and Marital Satisfaction. ............................... 161 Psychological-Relational Power and Power Ascription.. .................................. 163 Summary.. ........................................................................................................... 164 What are the relationships of variables within the framework of socio-structural and psychological relational power? ............................................................................. 164 Psychological-Relational Power and Socio-Structural Power........................... 164 Summary. ............................................................................................................ 165 Limitations ...................................................................................................................... 166 Sample Selection Bias................................................................................................. 166 Translation Issue ......................................................................................................... 167 High Cronbach's Alpha ............................................................................................... 167 Missing Data ............................................................................................................... 168 Multivariate Normality ............................................................................................... 169 Chi-square ................................................................................................................... 170 Theoretical Re-specification of Model ....................................................................... 172 Topics for Further Study................................................................................................. 174 Implications for Nursing ................................................................................................. 175

12

Conclusion ...................................................................................................................... 177 APPENDIX A................................................................................................................. 178 APPENDIX B ................................................................................................................. 179 APPENDIX C ................................................................................................................. 182 APPENDIX D................................................................................................................. 183 APPENDIX E ................................................................................................................. 185 APPENDIX F.................................................................................................................. 187 REFERENCE.................................................................................................................. 188

13

LIST OF TABLES Table 1..………………….………….……………………………………………….67 Table 2………….…………………….………….…………………………………..70 Table 3………….…………………….………….…………………………………..75 Table 4………….…………………….………….…………………………………..78 Table 5………….…………………….………….……………………………….….78 Table 6………….………………….………….………………………...…………...79 Table 7………….…………………….………….……………………………….….79 Table 8………….…………………….………….…………………………...….…..80 Table 9………….…………………….………….……………………………….….87 Table 10-13……………………………………………………………………….….88 Table 14-15.………….…………………….………….……………………………..89 Table 16………….………………………………………………………………..…90 Table 17-18………….…………………………………………………………….…91 Table 19………….……………….………………………………………………….92 Table 20………….………………………………………………………………..…94 Table 21………….………………………………………………………………..…96 Table 22………….……………………………………………………………….….97 Table 23………….………………………………………………………………..…98 Table 24………….…………………………………………………………….….…99 Table 25………….………………………………………………………….…..….100 Table 26………….………………………………………………………….….…..101 Table 27………….………………………………………………………….…..….106 Table 28………….………………………………………………………….….…..109 Table 29…………….…………..……………………….……………………….....110

14

Table 30…..……..…….………….……………………….………………………....112 Table 31…..……..…….………….……………………….………………………....112 Table 32….……….………………..….……………….…………………………….116 Table 33………….….………….……………………….…………………………...121 Table 34..………….…………………………………….……………………….......122 Table 35.………….…………………………………….…………….………….......122 Table 36.………….………………………………….….………………….….…... .123 Table 37.………….………………………………….….…………………..……... .123 Table 38.………….…………………………………….……………….…………...124 Table 39…...…………….………………………………….……………….….….. .125 Table 40…...…………….………………………………….……………….….….. .125 Table 41…..……….…………………………………….…………………….…......126 Table 42…..……….…………………………………….…………………….…......128 Table 43…..……..…….…………………………………………………..………....131 Table 44…...…………….………………………………….……………….….….. .131 Table 45……..…….………………………………………………………..…….... .132 Table 46……..…….………………………………………………………..……......133 Table 47…..…..……….…………………………………………………..………....137 Table 48…..…..……….…………………………………………………..………....137 Table 49…...…………….………………………………….……………….….….. .137 Table 50…..…..……….…………………………………………………..………....138 Table 51…...…………….………………………………….……………….….….. .138 Table 52……..…….………………………………………………………...….…....140 Table 53…..…..……….…………………………………………………………......141 Table 54…...…………….………………………………….……………….….….. .141 Table 55……..…….……………………………………………………….…..….....142 Table 56……..…….……………………………………………………….…..….....143 Table 57……..…….……………………………………………………….…..….....146 Table 58……..…….……………………………………………………….…..….....148

15

LIST OF FIGURES Figure 1…………………………………………………………………………......40 Figure 2……………………………………………………………………………..72 Figure 3……………………………………………………………………………..72 Figure 4……………………………………………………………………………..73 Figure 5……………………………………………………………………………113 Figure 6……………………………………………………………………………127 Figure 7……………………………………………………………………………139 Figure 8……………………………………………………………………………144

16

ABSTRACT

Problem and Background. Domestic violence against married women has persisted throughout Korean history. In Korean society, domestic violence is culturally sanctioned. Cultural approval inspires and reinforces the use of violence in the family, particularly against women and undergirds traditional family ideology, which has made Korean women a vulnerable population prone to suffer from Hwabyung. Purpose. First, this study was designed to test the psychometric properties of two newly developed instruments, the Korean Women’s Abuse Screening Tool (KWAST) and the Korean Women’s Abuse Intolerance Scale (KWAIS) (Research question 1). Second, this study was designed to test a theoretical model that predicts married Korean women’s responses to domestic violence (Research question 2). Of particular interest were severity of abuse, Hwabyung, abuse intolerance, psychological relational power, and sociostructural power. Design and Method. Ex post facto research is designed to shed light on causal relationships retrospectively linking a phenomenon (domestic violence) that occurred in the past and focusing on presently occurring outcomes to understand relationships among abuse, Hwabyung, and abuse intolerance within the framework of socio-structural and psychological-relational power. Sample. The sample consisted of 184 married Korean women who self-identified as having been physically, psychologically, sexually, or financially abused by their husbands during their marriages. All women were 20 years of age or older. Subjects were

17

recruited from one of three geographic locations (Seoul, Pusan, and Daejon, South Korea) and four different sites including a psychiatric clinic, women’s shelters, home health care centers and the community. Results. For research question 1, the construct validity of the KWAST and KWAIS was partially supported by R2 (.59) and the value of CFI (.967). For research question 2, women’s abuse intolerance (R2 = .86) was impacted directly by abuse (r = .18, p < .01), Hwabyung (r = .09), and psychological relational power (r = .77, p .05 Close fit test CFI* .80 -.90 .90 - .95 > .95 RMSEA > .10 .08 -.10 < .05 Hoelter’s CN < 70 > 200 * Choice of index CFI: Comparative fit index; RMSEA: Root mean square error of approximation Hoelter’s CN: Hoelter’s critical N

111

Theoretical Assumptions. Theoretical assumptions were used to construct the latent factor and considered the following: 1) the theoretical relationships between abuse and four indicators; 2) the structural relationship between abuse and abuse intolerance; and 3) the model testing. The model testing must be interpreted in terms of how it clarifies the construct validity of the measure, that is, how consistent is the evidence with the theoretical construct and relations. Relative to the theoretical relations, the theoretical construct (abuse) and indictors (severity of abuse) was assessed using squared multiple correlations. Squared multiple correlations (R2) was calculated by multiplying standardized regression weights (factor loadings) (Table 30 & 31: Figure 5). R2 known as squared multiple correlation estimates were calculated by squaring standardized regression weights. R2 is the relative amount of variance of a dependent variable (or a factor) explained by explanatory (or observed) variables (Joreskog, 1993). About 73% of the variance in physical abuse was accounted for by the factor, abuse. This indicated that physical abuse was reported more by selfreported women than mental, sexual, or financial abuse. About 42% of the variances in psychological and sexual abuse were respectively accounted for by abuse. About 35% of the variance in financial abuse was accounted for by abuse (Table 31) (Figure 5).

112

Table 30. Standardized regression weights (Factor loadings) Variable Severity of physical abuse Severity of psychological abuse Severity of sexual abuse Severity of financial abuse Abuse

Estimate Abuse Abuse Abuse Abuse Abuse intolerance

.86 .65 .65 .59 .77

Relative to structural relations, the relationship between abuse and abuse intolerance was examined using standardized regressions and squared multiple correlations. R2 showed that 59% of the variance in abuse intolerance was accounted for by abuse (Table 31) (Figure 5). This indicated that women who self-reported as having been abused had moderate propensities or desires to leave their abusive husbands. Table 31. Squared multiple correlations Variable Physical abuse Psychological abuse Sexual abuse Financial abuse Abuse intolerance

Estimate .73 .42 .42 .35 .59

113

Physical Abuse

.86**

R 2 = .73

Psychological Abuse

.65**

R 2 = .42

Abuse

.77**

.65** Sexual Abuse R 2 = .42

.59**

Financial Abuse R 2 = .35 Chi-square (χ2) = 16.888; df = 5; p = .005; CFI = .967 ** p < .001 based on the value of C.R. > 1.96 Figure 5. A structural model for confirmatory factor analysis

Abuse Intolerance R 2 = .59

114

Relative to the modeling testing, chi-square value as an indictor for the exact fit test and goodness-of-fit indexes (e.g., CFI, RMSEA, and Hoelter’s CN) as indicators for the close fit test were assessed. In SEM, the null hypothesis (H0) is that the postulated model fits the sample data (Byrne, 2001). The χ2 is a measure of overall fit of the model to the data. The χ2 measures the discrepancy between the fitted covariance matrix and the sample covariance matrix (Byrne, 2001). The higher the value of chi-square, the greater the discrepancy between the model and the sample data. When the chi-square is nonsignificant at the p-value of .05, it indicates the model fits the sample data (model was accepted based on the theoretical relationships). The chi-square value of 16.888 with 5 degrees of freedom was significant at the .05 level, indicating the model did not fit the data (Table 32). When the model does not fit the data (exact fit test failed), the close fit test should be conducted as an alternative index to construct validity (Hu & Bentler, 1999) because χ2 is sensitive to the sample size and is believed to be unrealistic to determine the fitness in SEM (Byrne, 2001; Hu & Bentler, 1995). The CFI measures the goodness of fit of the hypothesized model and is influenced by Eigenvalues, correlation matrix, and residuals. The CFI ranges from zero to 1.00 and indicates a good fit when the value of the CFI is greater than or equal to .95 (Hu & Bentler, 1999). The CFI provides a measure of complete covariation in the data and thus should be the index of choice for a close fit test (Bentler, 1990). The CFI plays a major role in estimating fit indexes and is considered as the index of choice for the close fit test (Bentler, 1990). The value of CFI (.967) for these data indicated a good fit (Table 32).

115

The RMSEA measures how the model fits data by taking into account the error of approximation in the population (Byrne, 2001). The RMSEA is based on a non-centrality parameter. If the value of chi-square (χ2) is less than the degrees of freedom (number of distinct sample moment – number of distinct parameters to be estimated), then RMSEA is set to zero. Therefore, a good model has an RMSEA of .05 or less. Models in which RMSEA is .10 or more have a poor fit. The value of RMSEA (.11) for these data indicated a poor fit (Table 32). Hoelter’s Critical N estimates a sample size that would be sufficient to yield an adequate model fit for the χ2 test (Hu & Bentler, 1995). Hoelter’s CN should only be computed if the chi-square is statistically significant to assess whether the sample size is sufficient to estimate parameters. The value of χ2 in this study was statistically significant. Therefore, the value of the Hoelter’s CN was assessed. Hoelter (1983) recommended that the value in excess of 200 is indicative of a model that adequately represents the sample size while less than 75 indicates that the sample size is not sufficient to yield an adequate model fit for χ2 test. The value of Hoelter’s CN (164) for these data at the .05 level indicated that the size of sample (n = 184) was not satisfactory (Table 32). The exact fit test failed by rejecting the null hypothesis (chi-square = 16.888; df = 5; p = .005) that the model fits the sample data. χ2 has limitations when evaluating model fit since good models may not fit the data. Therefore, CFI for the close fit test that takes a more pragmatic approach was used. Overall, the model indicates a good fit for the sample data as shown by the value of CFI = .967. The construct validity of abuse and abuse intolerance was partially supported by R2 and the value of CFI (Table 32).

116

Table 32. Goodness of fit indexes Variable Exact Fit Test Chi-square (χ2) Df p-value Close Fit Test CFI* RMSEA PCLOSE Hoelter’s CN * Index of Choice

Estimate 16.888 5 .005 .967 .11 .03 164

117

Missing Data

The impact of missing data (Enders, 2001a; Schafer, & Graham, 2002) has been discussed in several studies. Examining missing data was critical for this study because of potential impacts including: 1) it could lead to biased estimates; 2) it could lead to incorrect variances; and 3) it changes the sample size due to eliminating cases with missing data. In this study, data on education (n = 3) and income (n = 60) were missing. The impact of missing data on education and income was described in terms of socioeconomic status (SES) and structural equation modeling (SEM) followed by a discussion of the mechanisms for dealing with missing data, strategies for missing data analysis, alternative methods to treat missing data, and the normality test. Impact of Missing Data Missing Data and SES. Age, education and income are associated with socioeconomic status and these variables are assumed to influence health behavior (Green, 1970; Anderson & Armstead, 1995). In this study, socioeconomic status was considered to influence the relative social position of the women that affected her attitudes, beliefs, or values about domestic violence (Green, 1970; Anderson & Armstead, 1995). In the study, age, education, and income were expected to influence responses to domestic violence. For example, the younger the woman, the higher the woman’s education level, and the higher the woman’s income, the stronger the woman’s psychological relational power, which directly influenced her propensity or desire to leave her abusive husband (Campbell, 1989; Campbell et al., 1998). Knowledge of

118

socioeconomic status optimizes predictions about people’s actions. Knowledge of socioeconomic status also enhances analyses and evaluations of relationships, the effects of programs and the effects of other independent variables on health behaviors and decision-making (Green, 1970; Anderson & Armstead, 1995). According to Green, when an individual’s educational information is missing, “an attempt should be made to estimate educational level on the basis of all other available information, including occupation, income, age, formal affiliations, and membership in organizations” (Green, 1970. p. 817). However, the questionnaires used in this study were not designed to obtain information on these characteristics. Therefore, it was not possible to estimate the missing data about educational level. In addition, with so few cases (n = 3), no further analysis was done except Little’s MCAR test (L. Phillips, personal communication with, July 30, 2004). Structural relational power characterized by age, education, and income was hypothesized to directly influence psychological relational power, which impacted on married Korean women’s propensities or desires to leave the abusive husbands. A major amount of missingness on income (n = 60) was important because: 1) only partial data were available; 2) the variance of the income variable could be underestimated; and 3) the attenuation of the relationship between socio-structural power (age, education, and income) and psychological relational power could be generated. Nonresponses on income may have been caused because the respondents refused to answer, did not know, did not remember or had no access to income.

119

Missing Data and SEM. Missing data has important implications for structural equation modeling. If there are missing data: 1) it can hide true values that are meaningful and the missing data are not likely to effectively predict the outcome variable in a model (Allison, 2003); 2) missing values not only mean less efficient estimates because of the reduced sample size, but also because the standard complete data methods cannot be easily used to analyze the data (Schafer & Graham, 2002); and 3) possible biases exist because respondents are often systematically different from the nonrespondents. These biases are difficult to eliminate since the precise reasons for nonresponse are usually not known (Rubin, 1987). The nature of the bias depends on the mechanisms causing data to be missing. As a result, 1) a greater discrepancy between the model and sample data could exist; and 2) a probability of rejecting the model could exist. Therefore, the section that follows focuses on describing mechanisms of missing data, missing data analysis, and alternative methods to treat missing data. Mechanisms of Missing Data Little and Rubin (2002) argued that missing data mechanisms are crucial because the impact of the missing data on the results of statistical analysis depends on the mechanism which caused the data to be missing. Causes for missing data fit into three classes: 1) Missing Completely at Random (MCAR); 2) Missing at Random (MAR); or 3) Nonignorable missing data (NI). These classes are important to understand because the problems caused by missing data and the solutions to these problems are different for the three classes.

120

Missing Completely at Random (MCAR). MCAR means that the missing data mechanism is unrelated to the values of variables in the data set (Little & Rubin, 2002; Allison, 2003; Byrne, 2001; Wayman, 2003). In this study, the income variable could be considered to be MCAR, if income did not depend on the value of other variables or the income variable was independent of both observed variables and unobserved variables. Missing at Random (MAR). MAR means that the missing values are related to other measured variables but not to the underlying level of the trait being measured (Enders, 2001a). It is possible that abused women are more likely not to report income than women who are not abused. Therefore, two groups (reported income and non-reported income group) were compared to see the difference of the underlying level of the trait (abuse intolerance). Nonignorable (NI). Missing data is considered to be NI when the missingness is nonrandom and the missingness is not predictable from any one variable in the data set. For example, some abused women did not report income due to other reasons that were not included in the data set. So the missingness on income was not predicted from any one variable in the data set. Then the misisngness on income could be considered to be NI.

121

Missing Data Analysis The data set (n = 184) for this study had missing values on two variables including education and income. The missing data on education was not a significant proportion of all cases and therefore could be easily handled in SEM with Full Information Maximum Likelihood (FIML: an approach to the estimation of simultaneous equations, which uses all information of the observed data). In contrast, if missing data on income (n = 60) is not MCAR, FIML cannot be used to handle missing data. Therefore, missing data on income was assessed using different methods: 1) demographic information; 2) linear regression; and 3) Little's MCAR test. Missingness by Location. Thirty-two participants (53.3%) who did not report their income were from shelters, 16 (26.7%) were from home healthcare centers, 11 (18.3%) were from the community and 1 (1.7%) was from the psychiatry department (Table 33). Participants from shelters were more likely not to report their income than women from other locations as shown by chi-square (8.493), df (3) and p-value (.037) (Table 34). Table 33. Missingness on income by location Variable Shelters Home Health Care Centers Community Psychiatry Department Total

n 32 16 11 1 60

% 53.3 26.7 18.3 1.7 100.0

122

Table 34. Chi-square between income and location Income Location Not Report Report % (n) (n) Community 11 22.0 39 Shelter 32 45.1 39 Psychiatry Department 1 25.0 3 Homecare Center 16 27.1 43 Total 60 32.6 124 Chi-square = 8.493; df = 3; p-value = .037

Total %

n

%

78.0 54.9 75.0 72.9 67.4

50 71 4 59 184

100 100 100 100 100

Missingness by Age. The majority of women (53.4%) who did not report their income were aged 31-50 while 6 (10%) were aged 51-60, 9 (15%) were aged 61-70, 7 (11.7%) were aged 71-80, 4 (6.7%) were aged less than 30, and 2 (3.3%) were aged greater than 81. Missingness on income by age was not statistically significant (Table 35). Table 35. Missingness on income by age Variable

n % 4 6.7 ≤ 30 31-40 16 26.7 41-50 16 26.7 51-60 6 10.0 61-70 9 15.0 71-80 7 11.7 2 3.3 ≥ 81 Total 60 100.0 No statistically significant difference by Chi-square

Missingness by Religion. Thirty-one (51.7%) who did not report income were Christian, 12 (20%) were Buddhist, and 17 (28.3) reported no religion (Table 36). There was no statistically significant difference by religions participants were affiliated.

123

Table 36. Missingness on income by religion Variable n % Catholic 13 21.7 Christian 18 30.0 Buddhist 12 20.0 No religion 17 28.3 Total 60 100.0 No statistically significant difference by Chi-square

Missingness by Health Status. Nineteen women (31.7%) who did not report income self-reported health status as “fair,” 18 (30%) as “bad,” 16 (26.7%) as “good,” and 7 (11.7%) as “excellent.” About 62% (61.7) women who did not report income reported their health status as “fair” or “bad” (Table 37). There was no statistically significant difference by health status. Table 37. Missingness on income by health status Variable

n % Excellent 7 11.7 Good 16 26.7 Fair 19 31.7 Bad 18 30.0 Total 60 100.0 No statistically significant difference by Chi-square

Missingness by Living Status. Thirty-one women (51.7%) who did not report their income stayed in shelters, 19 (31.7%) stayed with their husbands and children, 9 (15%) were with their husbands only, and 1 (1.7%) was with her parents (Table 38). There was no statistically significant difference by living status.

124

Table 38. Missingness on income by living status Variable With husband With husband and children With husband’s parents With her parents Shelters Total

n 9 19 0 1 31 60

% 15.0 31.7 0 1.7 51.7 100.0

No statistically significant difference by Chi-square None of these variables indicated that there was a statistical association between the variable and report of income except location. This indicated that women from shelters were more likely not to report income than women from other sites. This will be discussed in detail in Chapter 5. Therefore, missing data of income could be considered to be MAR. Missingness by Income Reporting. Two groups (income reported and income nonreported group) were compared using t-test to see whether there was a statistically a significant difference of the underlying level of trait (abuse intolerance) (Table 39). The mean score of abuse intolerance in the income reported group was 50.9 (S.D = 15.25) while the mean score of abuse intolerance in the income non-reported group was 50.1 (S.D = 15.86) (t = .321, df = 182, p-value = .749). Therefore, the missingness on income was not related to the underlying level of trait (abuse intolerance) and thus income could be considered to be MAR.

125

Table 39. Missingness on income and abuse intolerance Variable Income report group Income non-report group

Abuse intolerance (M ± S.D) 50.86 ± 15. 25 50.08 ± 15.86

t

df

p-value

.321

182

.749

Missingness Predicted by Model Variables. All the variables in the model were used to predict missingness on income using linear regression. Respondents who did not report income were considered to be a dependent variable. Other variables including age, education, collectivism, traditional family ideology, marital satisfaction, power ascription, abuse, Hwabyung, and abuse intolerance were treated as predictors. The enter method of linear regression analysis was used. The adjusted R square was .076. About eight percent (7.6%) of the variance of missingness on income was explained by the predictors of age, education, collectivism, traditional family ideology, marital satisfaction, attitude toward power ascription, abuse, Hwabyung, and abuse intolerance (Table 40). Age was the only statistically significant predictor (Beta = -.259; t = -2.502; p = .013) of missingness on income (Table 41). Therefore, missingness on income was not affected by other variables in the model except age. Table 40. Model Summary R R Square Adjusted R Square S.E of the Estimate .349 .121 .076 93.94 Dependent Variable: Income Predictors: (Constant), age, education, collectivism, marital satisfaction, traditional family ideology, attitude toward power ascription, abuse, Hwabyung, and abuse intolerance. S.E: Standard Error

126

Table 41. Regression analysis Variable (Constant)

S.E 96.321 .607 .603 1.922 3.108 2.540 1.421 2.153 .228 1.141

Standardized Coefficients Beta

t .627 -2.502 .081 .584 .840 -.363 1.906 -1.357 -.544 .636

Age -.259 Education .006 Collectivism .119 Marital Satisfaction .127 Traditional Family Ideology -.059 Attitude Toward Power Ascription .182 Abuse -.141 Hwabyung -.047 Abuse intolerance .114 Dependent Variable: Income Predictors: (Constant), age, education, collectivism, marital satisfaction, traditional family ideology, attitude toward power ascription, abuse, Hwabyung, and abuse intolerance. * p-value is significant at the .05 level (2-tailed). S.E: Standard Error

p-value .531 .013* .936 .560 .402 .717 .058 .176 .587 .526

Using a liner regression, age was the only predictor to affect missingness on income. Therefore, missingness on income related to age was analyzed using Little’s MCAR test to confirm that missingness on income is not related to age. Figure 6 denoted missingness on income and age. The missingness on income was coded as “99.” The missingness on income was widely distributed over age (Figure 6). This figure did not confirm whether the missingness on income is MCAR. Therefore, the Little’s MCAR test was conducted.

127

200

150

Income 100

50

Age

0 20 on income and30 age

40

50

60

70

80

90

Age Figure 6. Income and age Missing Data Analysis of Income by Little’s MCAR Test. Missing data on education and income were analyzed using an Expectation Maximization (EM) algorithm. EM algorithm is an iterative optimization method to estimate unknown parameters given measurement data (expectation and maximization) (Dempster et al., 1977). The expectation step computes the expected value of the complete data log likelihood given the current parameter values. The maximization step substitutes the expected values for the missing data obtained from the expectation step and maximizes the likelihood function as if no data were missing to obtain new parameter estimates (Enders, 2001b). According to Graham et al. (1996), EM provides the accuracy and efficiency of covariance matrix parameter estimates under the conditions of non-normality. Based on the missing data analysis using the EM algorithm the Little's MCAR test resulted in acceptance of the null hypothesis that the value of income was MCAR (Chi-

128

square = 4.145, df = 2, p = .126) (Table 42). Therefore, missingness on education and income was treated as MCAR. Table 42. EM Estimated Statistics Mean Covariance Education Income Education Income (n = 181) (n = 124) (n = 181) (n = 124) 11.4 89.6 Education 25.8 Education Income 229.5 13788.1 Income Little's MCAR test of mean, covariance, and correlation Chi-square = 4.145, df = 2, p = .126

Correlation Education Income (n = 181) (n = 124) 1.0 .4 1.0

Treatment of Missing Data. In using the Analysis Moments Structure (AMOS) program to confirm or test models, complete data are required for the probability density (Byrne, 2001; Schafer & Graham, 2002). Since missingness on income was confirmed to be MCAR by different missing data analyses, Full Information Maximum Likelihood (FIML) could efficiently handle missing data. Therefore, FIML is discussed with the advantages and disadvantages under the condition of MCAR. Full Information Maximum Likelihood. The FIML is an approach to the estimation of simultaneous equations and is a theory-based approach to the treatment of missing data (Wothke, 2000). FIML assumes multivariate normality and maximizes the likelihood of the model given the observed data. FIML uses all the information of the observed data, including information about the mean and variance of missing portions of a variable, given the observed portion of other variables (Wothke, 2000; Smallwaters, 2004). For example, in this study, there were some variables with data for all 184 cases whereas data

129

about income was available for only 124 cases. The fitting function for FIML is computed by summing all the individual fit functions. In this study, use of the FIML method allowed use of all 184 cases. When there is a large amount of missing data, it can be best to use a FIML approach to estimation (Schafer & Graham, 2002). FIML provides simultaneous and statistically efficient estimation of all direct and indirect effects among the latent variables measured with errors (Raudenbush & Sampson, 1999). According to Wothke's study (2000), estimation of a model is dramatically better with FIML under MCAR. Enders (2001a) also reported that FIML estimates were unbiased and more efficient than the other methods yielding the lowest proportion of convergence failure and providing near optimal Type I error rates. Therefore, FIML is most efficient and least biased under MCAR (Wothke, 2000). FIML parameter estimates are substantially efficient because 1) model rejection rates from FIML were relatively uninfluenced by the missing data rate under MCAR (Enders, 2001a; Enders & Bandalos, 1999); 2) FIML reduces bias (Little & Rubin, 1987; Schafer, 1997); and 3) FIML provides consistent and efficient parameter estimates and standard errors under MCAR (Bollen, 1989; Allison, 2003). Based on missing data analysis and the missing data mechanism, the FIML method was the best option for dealing with missingness on income in the structural equation modeling.

130

Test of Normality Assumption Implications of Normality. Before testing a hypothesized model using SEM, multivariate normality test of variables in the model is important. SEM assumes that the sample data follows a multivariate normal distribution. If this assumption is violated, 1) the chi-square goodness of fit tests can produce inaccurate assessment; and 2) tests of parameters may be biased. Therefore, for the estimation of the model, the multivariate normality test should be examined (Chou & Bentler, 1995). Normality Test. The importance of non-normality of data was examined using the Shapiro-Wilk W Test. The Shapiro-Wilk W test is designed for the sample size less or equal to 2000, and determines the normality of each variable in the model by computing the Shapiro-Wilk statistic, W. distribution and p-value (Shapiro & Wilk, 1965). W is considered as the correlation between given data and their corresponding normal scores, with W = 1 when the given data are perfectly normal in distribution. When W is significantly smaller than 1, the assumption of normality is not met. Shapiro-Wilks W is recommended for small and medium samples up to n = 2000. The Shapiro-Wilk W Test was conducted using Statistical Analysis Systems (SAS, Version 9). The Shapiro-Wilk W Test was conducted to test the univariate normality of age, education and income. For age, the Shapiro-Wilk W was .93 and the null hypothesis that data follow normal distribution was rejected (p = .0001) (Table 43). For education, the Shapiro-Wilk W was .91 and the null hypothesis was also rejected (p = .0001). For income, the Shapiro-Wilk W was .74 and the null hypothesis was rejected (p = .0001).

131

The Shapiro-Wilk W Test indicated that age, education, and income were not univariately normally distributed by rejecting the null hypothesis respectively (Table 43). Table 43. Univariate Shapiro-Wilk W Tests of age, education, and income

Normal Parameters(a, b) Shapiro-Wilk W p-value a Test distribution is Normal. b Calculated from data.

Age (n = 184) 51.74 ± 16.66 .93 .0001

Education (n = 181) 11.04 ± 5.58 .91 .0001

Income (n = 124) 89.91 ± 119.09 .74 .0001

The Shapiro-Wilk W Test also indicated that collectivism (Shapiro-Wilk W = .92, pvalue = .0001), traditional family ideology (Shapiro-Wilk W = .91, p-value = .0001), and marital satisfaction (Shapiro-Wilk W = .91, p-value = .0001) were not normally distributed (Table 44). However, the data for attitude toward power ascription were normally distributed as shown by acceptance of the null hypothesis that the data followed normal distribution (Shapiro-Wilk W = .99, p-value = .3008) (Table 44). Table 44. Univariate Shapiro-Wilk W Tests of collectivism, marital satisfaction, traditional family ideology, and attitude toward power ascription Collectivism (n = 184) Normal Parameters (a, b) 22.73 ± 10.37 Shapiro-Wilk W .92 p-value .0001 a Test distribution is Normal. b Calculated from data.

Traditional Family Ideology (n = 184) 11.45 ± 6.19 .91 .0001

Marital Satisfaction (n = 184) 8.47 ± 4.75 .91 .0001

Attitude Toward Power Ascription (n = 184) 25.58 ± 6.56 .99 .3008

132

The data for abuse (Shapiro-Wilk W = .94, p-value = .0001), Hwabyung (ShapiroWilk W = .97, p-value = .0001), and abuse intolerance (Shapiro-Wilk W = .83, p-value = .0001) were not normally distributed (Table 45). Table 45. Univariate Shapiro-Wilk W test of age, education, and income

Normal Parameters (a,b) Shapiro-Wilk W p-value a Test distribution is Normal. b Calculated from data.

Abuse (n = 184) 7.82 ± 4.70 .94 .0001

Hwabyung (n = 184) 79.07 ± 36.93 .97 .0030

Abuse intolerance (n = 184) 50.60 ± 15.41 .83 .0001

Summary All observed variables in the data set were examined by Shapiro-Wilk W Test to determine the univariate normality in the data set. Only attitude toward power ascription was normally distributed. If non-normality is indicated for one or more of the variables, multivariate normality can be rejected because it is known that all univariate marginal distributions of a multivariate normal distribution are univariate non-normal distribution. The data set did not follow multivariate normal distribution. The assumption of multivariate normality was failed before testing the model. Failure of testing assumption does not mean that there is no reason to test the model perhaps due to sample characteristics. These sample characteristics regarding non-normal data will be described in detail in Chapter 5. In the next section, results of model testing using SEM based on hypotheses will be discussed.

133

Structural Equation Modeling

Structural Equation Modeling (SEM) is a powerful tool for testing two models (a measurement model and a theoretical model) simultaneously by pictorially enabling a clearer conceptualization of the theory under study (Byrne, 2001). The hypothesized model testing was conducted using AMOS 5. The procedure was: 1) developing the hypotheses; 2) testing the measurement model (model summary, parameter estimation, and interpreting the outcome); and 3) testing the structural model (model summary, parameter estimation; and interpreting the outcome). Criteria used to test the model were squared multiple correlations (R2), chi-square, CFI, RMSEA and Hoelter’s CN (Table 46). Table 46. Criteria used for testing structural model Criteria Theoretical relations Squared multiple correlations Model testing Exact fit test Chi-square value p-value Close fit test CFI* RMSEA Hoelter’s CN * Choice of index

Poor fit

Moderate fit

< .05 .80 -.90 > .10 < 70

Good fit

> .05 .90 - .95 .08 -.10

> .95 < .05 > 200

134

Hypothesis Overall Hypothesis: The model of Korean women’s responses to domestic violence fits the sample data Sub-hypotheses: 1. Abuse intolerance is accounted for by psychological relational power, abuse and Hwabyung. 2. Abuse intolerance is indirectly accounted for by psychological relational power through abuse and Hwabyung. 3. Abuse intolerance is indirectly accounted for by abuse though Hwabyung. 4. Hwabyung is accounted for by psychological relational power and abuse. 5. Abuse is directly accounted for by psychological relational power. 6. Psychological relational power is a latent factor measured by collectivism, traditional family ideology, marital satisfaction, and attitude toward power ascription. a.

Psychological relational power has an inverse relationship with collectivism

b. Psychological relational power has an inverse relationship with traditional family ideology c. Psychological relational power has an inverse relationship with marital satisfaction. d. Psychological relational power has a positive relationship with attitude toward power ascription

135

7. Psychological relational power is influenced by age, education, and income. a. Age has an inverse relationship with psychological relational power. b. Education has a positive relationship with psychological relational power. c. Income has a positive relationship with psychological relational power Specified Relations in The Model. 1. The greater the Hwabyung symptoms, the stronger women’s propensities or desires to leave their abusive husbands. 2. The greater the abuse, the stronger women’s propensities or desires to leave their abusive husbands. 3. The greater the abuse, the greater women’s Hwabyung symptoms. 4. The greater psychological relational power, the stronger women’s propensities or desires to leave their abusive husbands. a. The more individualistic the women, the stronger the psychological relational power. b. The less the women’s traditional family ideology, the stronger the psychological relational power. c. The greater the women’s marital conflict, the stronger the psychological relational power. d. The greater the women’s attitude toward power ascriptions, the stronger the psychological relational power. 5. Younger age is related to greater psychological relational power. 6. More education is related to greater psychological relational power.

136

7. Higher income is related to greater psychological relational power. The model testing was accomplished in two steps. First, a measurement model testing was assessed. Second, a structural model testing was attempted. Measurement Model The measurement model defines the relationship between the construct and the indicators. For this study, the measurement model comprising psychological relational power and four indicators including collectivism, traditional family ideology, marital satisfaction, and attitude toward power ascription were tested. Testing the measurement model is an important preliminary step in the analysis of the full latent variable model. If a measurement model does not fit the sample data, testing the structural model is pointless because the measurement model will not allow the structure to work. Therefore, the measurement model is first estimated in the model testing and the correlations or covariance matrix between constructs play a major role in estimating the coefficients between constructs (Figure 7). This measurement model constitutes loadings (the effect of latent variable on the measure: Standardized regression weights) and error variances (the variance in the measure not explained by the latent variable). Model Summary of Measurement Model. The model contained the following variables: 1) observed endogenous variables (collectivism, traditional family ideology, marital satisfaction, and attitude toward power ascription); and 2) a latent variable (psychological relational power). There were a total of 27 variables including 9 variables, 4 observed variables, 5 unobserved variables, 5 exogenous variables, and 4 endogenous variables (Table 47).

137

Table 47. Variable summary of measurement model Variable Number of variables in the model Number of observed variables Number of unobserved variables Number of exogenous variables Number of endogenous variables Total

n 9 4 5 5 4 27

Summary of Parameters of Measurement Model. There were total 17 parameters including 5 fixed parameters and total 12 unlabeled parameters (Table 48). Table 48. Summary of parameters of measurement model Variable Fixed Labeled Unlabeled Total

Weights 5 0 3 8

Covariances 0 0 0 0

Variances 0 0 5 5

Means 0 0 0 0

Intercepts 0 0 4 4

Total 5 0 12 17

The number of data points was 14. The number of distinct parameters to be estimated was 12. Therefore, the degrees of freedom (df = 2) were calculated from the number of distinct sample moments (data points) minus the number of distinct parameters to be estimated (Table 49). Table 49. Notes for measurement model Variable Number of distinct sample moments (data points) Number of distinct parameters to be estimated Degrees of freedom (df)

n 14 12 2

138

R2 of Measurement Model. The theoretical relationship between a construct (psychological relational power) and indicators was assessed using squared multiple correlations (R2) (Table 50 & 51). Ninety-six percent (96%) of the variance associated with collectivism was explained by psychological relational power. Seventy-nine (79%) of the variance associated with traditional family ideology was explained by psychological relational power. Seventy (70%) of the variance associated with marital satisfaction was explained by psychological relational power. Forty-three (43%) of the variance associated with attitude toward power ascription was explained by psychological relational power (Table 51) (Figure 7). Table 50. Standardized regression weights of measurement model Variable Collectivism Traditional Family Ideology Marital Satisfaction Attitude toward Power Ascription

← ← ← ←

Psychological Relational Power Psychological Relational Power Psychological Relational Power Psychological Relational Power

Estimate -.98 -.89 -.84 .65

Table 51. Squared multiple correlations (R2) of measurement model Variable Collectivism Traditional Family Ideology Marital Satisfaction Attitude toward Power Ascription

← ← ← ←

Psychological Relational Power Psychological Relational Power Psychological Relational Power Psychological Relational Power

Estimate .96 .79 .70 .43

139

Psychological Relational Power

-.98**

-.89**

Traditional Family Ideology

Collectivism

2

-.84**

2

R = .96

Marital Satisfaction

R = .70 e2

Power Ascription R2 = .43

2

R = .79 e1

.65**

e3

χ2 = 3.527; df = 2; p = .171 ** p < .001 based on the value of C.R. > 1.96 Figure 7. Measurement model of psychological relational power

e4

140

Measurement Model Testing. The chi-square as an exact fit test indicator suggested that the measurement model fits the sample data as shown by the non-significant chisquare (Chi-square = 3.527; df = 2; p-value = .171). As a result of the p-value (.171), the null hypothesis that the model fits the sample data was accepted (Table 52). The goodness of fit indexes as close fit indexes indicated that the measurement model fits the sample data as shown by the value of CFI (.997), RMSEA (.065) and Hoelter’s CN (478) (Table 52). The measurement model passed both the exact fit test and the close fit test. Therefore, the construct, psychological relational power explained properly the variance associated with each of four indicators including collectivism, traditional family ideology, marital satisfaction, and attitude toward power ascription. Table 52. Goodness of fit indices of measurement model Variable Exact Fit Test Chi-square (χ2) Df p-value Close Fit Test CFI* RMSEA PCLOSE Hoelter’s CN

Estimate 3.527 2 .171 .997 .065 .305 478

141

Structural Model Model Summary of Structural Model. The model contained the following variables: 1) observed endogenous variables (collectivism, traditional family ideology, marital satisfaction, attitude toward power ascription, abuse, Hwabyung, and abuse intolerance); 2) observed exogenous variables (age, education, and income); 3) unobserved endogenous variable (psychological relational power); and 4) unobserved exogenous variables (error terms). Therefore, there were 19 variables in the model, 10 observed variables, 9 unobserved variables, 11 exogenous variables, and 8 endogenous variables (Table 53). Table 53. Variable summary of structural model Variable Number of variables in the model Number of observed variables Number of unobserved variables Number of exogenous variables Number of endogenous variables Total

n 19 10 9 11 8 57

Summary of Parameters Structural Model. Forty-five parameters existed in the model including 9 fixed parameters and 36 unlabeled parameters (Table 54). Table 54. Parameter summary of structural model Variable Fixed Labeled Unlabeled Total

Weights 9 0 12 21

Covariances 0 0 3 3

Variances 0 0 11 11

Means 0 0 3 3

Intercepts 0 0 7 7

Total 9 0 36 45

142

Twenty-nine degrees of freedom were calculated from 65 data points minus 36 distinct parameters to be estimated (Table 55). Table 55. Notes for model of structural model Variable Number of distinct sample moment (data points) Number of distinct parameters to be estimated Degrees of freedom

n 65 36 29

Standardized Regression Weights of Structural Model. Standardized regression coefficients known as regression weights, path coefficients, or factor loadings were computed using maximum likelihood estimates (Table 56). Standardized regression weights represent the independent contributions of each observed variable to the prediction of endogenous variables. This highest standardized regression weight was -.97 indicating that the relationship between psychological relational power and collectivism was strongly negative. In other words, the higher the score of collectivism the less women had the psychological relational power.

The second highest standardized

regression weight was -.89 between psychological relational power and traditional family ideology. Thus, the higher the score on traditional family ideology, the less women had psychological relational power. The third highest standardized regression weight was -.85 between psychological relational power and marital satisfaction. This indicated that the higher the score on marital satisfaction the less women had psychological relational power. The fourth highest standardized regression weight was .77 between psychological relational power and abuse intolerance, indicating that the higher the score of abuse intolerance the more women had psychological relational power. Overall the standardized

143

regression weights in the model were high in relation to psychological relational power (Table 56) (Figure 8). Standardized regression coefficients were used to describe the direction and strength of the relationship between the construct (psychological relational power) and variables. Table 56. Standardized regression weights of structural model Variable Psychological relational power Psychological relational power Psychological relational power Abuse Hwabyung Hwabyung Attitude toward Power Ascription Traditional family ideology Marital satisfaction Collectivism Abuse intolerance Abuse intolerance Abuse intolerance

← ← ← ← ← ← ← ← ← ← ← ← ←

Age Education Income Psychological relational power Psychological relational power Abuse Psychological relational power Psychological relational power Psychological relational power Psychological relational power Psychological relational power Hwabyung Abuse

Estimate -.70 -.17 -.06 .62 .07 .37 .66 -.89 -.85 -.97 .77 .09 .18

144

-.33

-.73

.42

Age

Education

-.70**

Income

-.06

-.17

Psychological Relational Power -.97**

Collectivism R2 = .94

-.89**

Traditional Family Ideology

R2 = .33

-.85**

.77**

R2 =.80

.66**

Marital Satisfaction R2 = .73

Attitude Toward Power Ascription

R2 = .44

.62**

.07 Hwabyung

Abuse .37**

2

R2 = .38

R = .18 .09

.18**

Abuse Tolerance R2 = .86 Chi-square = 174.618; df = 29; p = .000 ** p < .001 based on the value of C.R. > 1.96 Figure 8. Structural model of Korean women’s responses to domestic violence

145

Multiple Squared Correlations of Structural Model. Multiple squared correlations (R2) were calculated by squaring the standardized regression weights (Table 57). The value of R2 of abuse intolerance (.86) suggested that 86% of the variance of abuse intolerance was accounted for directly by abuse, psychological relational power and Hwabyung and indirectly by psychological relational power through abuse and Hwabyung. The value of R2 of Hwabyung (.18) suggested that 18% of the variance of Hwabyung was accounted for directly by psychological relational power and abuse. The value of R2 of (.38) of abuse suggested that 38% of the variance of abuse was accounted for by psychological relational power. The construct, psychological relational power, explained 94% of the variance associated with collectivism, 80% of the variance associated with traditional family ideology, 73% of the variance associated with marital satisfaction, and 44% of the variance associated with attitude toward power ascription. Thirty-three percent (33%) of the variance of psychological relational power was explained by age, education, and income (Table 57). Multiple squared correlations were assessed to test the hypothesis, which explained the relative contribution of each indicator to outcome variable.

146

Table 57. Multiple squared correlations of structural model Variable Psychological relational power Collectivism Traditional family ideology Marital satisfaction Attitude toward power ascription Abuse Hwabyung Abuse intolerance

Estimate .33 .94 .80 .73 .44 .38 .18 .86

Structural Model Testing. The hypothesis of the structural model was that the hypothesized model of Korean women’s responses to domestic violence fits the sample data. For the estimation of the model, FIML was conducted. Most variables except attitude toward power ascription were not normally distributed. Therefore, the robustness of estimation should be considered before estimating the model using AMOS. The estimation of parameters is based on the FIML. The FIML method assumes: 1) the sample is very large; 2) the distribution of observed variables is multivariate normal; 3) the hypothesized model is valid; and 4) the scale of observed variables is continuous (Byrne, 2001). The χ2 test measures the extent to which all residuals in ∑- ∑ (θ) are zero. The null hypothesis

(H0)

postulates

that

specification

of the

factor

loadings,

factor

variances/covariances, and error variances for the model under study are valid (Bollen, 1989; Byrne, 2001). In other words, the model fits the sample data. The test of the H0 model yielded a χ2 value of 174.618, with 29 degrees of freedom and a probability of less than .000 (p < .05), suggesting that the fit of the data to the hypothesized model was not

147

entirely adequate. This test statistic indicated that the hypothetical model did not fit the sample data (Table 58). However, the sensitivity of the likelihood ratio test and the χ2 distribution tend to be substantial when the sample size is large or small, which is critical to the obtaining of precise parameter estimates. According to MacCallum et al., (1996), small sample size tends to inflate the value of χ2. Chi-square statistical significance testing is driven by degrees of freedom involving the number of elements in the sample covariance matrix and the number of parameters to be estimated. A poor fit based on a small sample size may res+ult in non-significant χ2, accepting the null hypothesis that the data fit the model (Marsh, Balla, & McDonald, 1988; Marsh & Balla, 1994). However, the result of this study indicated a significant χ2 and thus the model was rejected unlike Marsh, Balla, & McDonald (1988)’s study. In addition, when data become increasingly non-normal, the χ2 value derived from ML becomes excessively large. When sample sizes are too small in spite of multivariate normality, the ML yield inflated χ2 values (Byrne, 2001). Therefore, the χ2 value has proven to be unrealistic to determine the fitness in most SEM empirical research (Byrne, 2001). Researchers have recommended goodness of fit indexes as a more pragmatic and unique approach to the evaluation process of model fit testing (Hu & Bentler, 1995; Marsh, Balla, & McDonald, 1988). Goodness of fit indices depend on the sample size (Bollen, 1986) and they are appropriate to estimate parameters when the size of sample is sufficient (Marsh, Balla, & McDonald, 1988). Goodness of fit is a measure of the relative amount of variance and covariance. CFI measures the quality of model fit (Bentler, 1992). The value for CFI

148

ranges from 0 to 1, and is derived from the comparison of a hypothesized model with the independence model. Value for CFI closes to .95 indicates a good fit (Byrne, 2001; Hu & Bentler, 1999). CFI was used as the best index of choice (Bentler, 1990). The value of CFI (.893) indicated the model as a poor fit. RMSEA is one of the most informative criteria in covariance structure modeling (Byrne, 2001) and tests how well the model fits the population covariance matrix. A value of RMSEA less than .05 indicates a good fit, values as high as .08 represent reasonable errors of approximation in the population, and values greater than .10 indicate a poor fit (Hu & Bentler, 1995; MacCallum et al., 1996). The value of RMSEA (.166) of the model indicates a poor fit. The value of the Hoelter’s CN index is recommended to be at least 200. The value of Hoelter’s CN (45) indicated that the sample size was not sufficient to yield non-significant chi-square. Results from the estimation of model testing yielded a χ2 value of 174.618, a CFI of .893, an RMSEA of .166, and a Hoelter’s CN of 45 (Table 58), indicating a poor model fit. Overall, the model of Korean women’s responses to domestic violence in socio-cultural context failed both the exact and close fit test using SEM. Table 58. Chi-square test Variable Exact Fit Test Chi-square (χ2) Df p-value Close Fit Test CFI* RMSEA Hoelter’s CN

Estimate 174.618 29 .000 .893 .166 45

149

Summary

This chapter described the results of data analysis including descriptive information regarding the sample, the reliability and validity testing of the measurements, missing data analysis, normality test and testing of the hypothesized model of Korean women’s responses to abuse. The hypothesized model of Korean women’s responses to domestic violence did not fit the sample data, rejecting the null hypothesis. This means the model was not correct under the consideration of exact and close fit indexes. Some questions and issues were raised during the data analysis. These issues will be addressed in Chapter 5.

150

CHAPTER FIVE

This study had two purposes. The first was to examine psychometric properties of two measurement instruments (KWAST & KWAIS). The second purpose was to test a theoretical model (a measurement and a structural model). Within the second purpose, of specific interests were: 1) relationships among abuse, Hwabyung, and abuse intolerance; 2) aspects of psychological-relational power contributing the abuse intolerance; and 3) the relationships of socio-structural and psychological-relational power. In order to bring the results and the discussion presented in Chapter 4 to a conclusion, this chapter addresses summary of findings, discussion of findings in relation to existing literature and theoretical background, limitations, topics for further study, conclusion, and implications.

Summary and Discussions of Findings

Research Question 1. What are the psychometric properties of two measurement instruments?. The psychometric properties of two measurement instruments were measured. Reliability was measured using Cronbach’s alphas. Validity was measured using confirmatory factor analysis. Cronbach’s alpha was .91 for KWAST and .98 for KWAIS. High Cronbach’s alphas suggest high internal consistencies among items.

151

Confirmatory factor analysis was used to assess whether the two instruments measured what they purported to measure. The Korean Women’s Abuse Screening Tool was designed to measure four dimensions of abuse. The Korean Women’s Abuse Intolerance Scale was designed to measure women’s propensities to leave their abusive husbands. A total of 59% of the variance of abuse intolerance was accounted for by abuse. However, the confirmatory factor analysis using SEM led to rejection of the null hypothesis. The negative result could be explained by several factors. First, the sample size (n = 184) was not sufficient to yield a non-significant chisquare as shown by Hoelter’s CN (164). Small sample sizes tend to inflate the value of chi-square due to the discrepancy between the model and the sample data. Second, the theory that abused Korean women have propensities or desires to leave their abusive husbands could be incorrectly specified in that it may not have accounted for the complexity of women’s beliefs and cultural values. In the Korean cultural context, divorce brings a family shame or disgrace. Often abused women do not receive support from their families of origin since after marriage they belong to their husbands’ families. As a consequence, many women know that they do not have a place to go to avoid their abusive husbands. If a woman leaves her abusive husband, she will likely be blamed by her husband’s family and treated badly. Given that fact, it is possible many abused women choose to stay in abusive situations rather than to leave. Adding even more complexity is the relationship between abuse and mental health within the Korean context. Often, women endure abusive situations for the sake of family harmony no matter what sacrifices are required (Min, 1989; Min et al., 1993). As a result, they may

152

suffer from mental illnesses such as, serious depression, and Hwabyung due to long lasting suppression and anger. Through treatment for these illnesses, some abused women receive positive reinforcements and a concomitant reduction in abuse. Therefore, some abused women may have stronger reasons to stay rather than to leave their abusive husbands. The validity test of two measurements (KWAST & KWAIS) measuring the relationship between abuse and abuse intolerance was partially achieved by the close fit test. The failed exact fit test of the two measurements impacted on the conclusion of the model testing. Research Question 2. What is the explanatory power of a theoretical model?. The theoretical model of Korean women’s responses to domestic violence was characterized by theoretical ideas about socio-structural and psychological relational power based on the following suppositions. First, the model was designed to offer a dynamic view of power and responses to domestic violence in a Korean cultural context. Second, it was designed to describe a Korean woman as a rational decision maker who weighs various costs and benefits of the power bases available to her. Third, the model was designed to describe possible consequences of the influence of power by describing the associations, directions, and strengths of the relationships of variables in the model. The results of the study suggested that, as depicted by the model, Korean women’s propensities to leave their abusive husbands were directly influenced by abuse, Hwabyung, and psychological relational power, and indirectly influenced by

153

psychological relational power through abuse and Hwabyung. Women had strong propensities or desires to leave their abusive husbands when they identified that they were frequently abused and had higher levels of psychological relational power. Within the model, women’s abuse intolerance was most strongly predicted by psychological relational power (r = .77; p < .001) followed by abuse (r = .18; p < .001) and Hwabyung (r = .09). When women were individualistic, had lower levels of traditional family ideology, experienced lower levels of marital satisfaction, had strong levels of power ascription, they had strong propensities or desires to leave their husbands. Psychological relational power strongly explained Korean women’s propensities to leave their abusive husbands. The relationship between age, education, and income and psychological relational power also contributed to the magnitude of the relationships. Among age, education, and income, age (r = .70; p < .001) had the strongest explanatory power in influencing psychological relational power. Issues of power and its influence have occupied the minds of behavioral scientists for decades (Glassman, 2003; Zamarripa, Wampold, & Gregory, 2003). Many of the theoretical insights about power relations and their influences have served as the groundwork for the more empirically oriented approach to explain the relationship between power and decision making processes in different cultural contexts (King, 1981; Forbes et al., 1999). Power in this study was viewed as the potential ability of a woman to influence her response to domestic violence. Based on findings from this study, power influences and forms the basis for responses to domestic violence encountered by women living in a Korean cultural context.

154

What are the relationships among abuse, Hwabyung and abuse intolerance? Abuse and Abuse intolerance.

Abuse had a positive relationship with abuse

intolerance (r = .18; p < .001), indicating that more abuse was related to stronger propensities or desires to leave abusive husbands. The weaker than expected relationship may be related to women’s current resources or the power bases they possess (Hansen, 1987). Resources can be money, time, family or social support, or relationships. Women likely estimate the costs of leaving based upon their current resources. Leaving may be more expensive than staying in the abusive situation. Women may calculate and consider alternatives to marriage in terms of maximizing rewards (e.g., attention from family members) and minimizing costs (e.g., financial burden, stress, time, or energy) in their exchanges (Blau, 1964; Homans, 1964; Chadwick-Jones, 1976). Abused women may also consider balancing rewards and costs based on comparisons (the evaluation of rewards and costs based on her own criteria of what she deserves) (Thibaut & Kelley, 1959). Therefore, while the model performed as predicted for abuse and abuse intolerance, the relationship was not as strong as expected. This suggests more complexity to the relationship than specified in the model (MacDonald, 1981; Hansen, 1987; Thibaut & Kelley, 1959). Cultural aspects probably contributed to the relationship of domestic violence and women’s propensities to leave their abusive husbands (Chon et al., 1997; Gelles, 1976; Kim, 1998b; Kim, 1997e). Korean culture may have played a unique role in influencing married women’s propensities to leave their abusive husbands. However, it is important to recognize that even in Western culture, more abuse may not

155

necessarily lead to propensity for women to leave abusive relationships (Han, one or two references here would be good. I suspect Campbell has something to say on this topic). Abuse and Hwabyung. Abuse had a positive relationship with Hwabyung (r = .37; p < .001), indicating that more abuse was related to more Hwabyung symptoms. The relationship between abuse and Hwabyung was moderate and this could be explained by cultural acceptance of domestic violence. Many married women tolerate abuse due to cultural views that have persisted throughout Korean history. Cultural values dictate that Korean women’s anger and rage must be suppressed. This generalized oppression of women’s anger and rage is believed to lead to women’s developing Hwabyung (Lin, 1983). Hwabyung known as “anger syndrome” or “culture-bound mental illness” arises from various sources including having lived in poverty, had little opportunity for education, experienced sudden death of a spouse or son, been forced to act illegally, or been abused. Being abused or the sudden death of a son mainly contributes to women’s developing Hwabyung (Chon, et al., 1997; Min & Lee, 1989 & Min et al., 1993). According to Min & Lee and Min et al. (1993), 81% of participants with Hwabyung were married and 71% of women developed Hwabyung due to domestic violence, extramarital affairs, and their husbands’ alcoholism. Women with Hwabyung suffer from chest stuffiness, a pushing up sensation in the chest, and inside unease and boiling, which leads them to feel that they are going to die. These symptoms were due to the suppression and somatization of anger resulting from a progressive chronic state of resignation (“chenyum”) (Min et al., 1993; Lin & Cheung, 1999; Kim et al., 1996). Several studies as

156

well as data in this study support that there is a moderate correlation between abuse and Hwabyung (Min et al., 1993 & Min & Lee, 1989, Kim et al., 1996). Hwabyung and Abuse Intolerance. Hwabyung had a positive relationship with abuse intolerance. However, Korean women’s propensities to leave their abusive husbands were not influenced by Hwabyung as shown by the magnitude (r = .09) and the non-significant regression weights. This non-significant relationship between Hwabyung and abuse intolerance could be explained by Korean cultural values. Women with Hwabyung have been shown to strive to minimize disruption and show more concern about family matters than personal interests and desires (Triandis, 1995; Lin, 1983; Min & Lee, 1989; Min et al., 1993). Many women with Hwabyung survive by “enduring” as the only way of life (Park et al., 2002). In addition, according to Min et al. (1993), women with Hwabyung said that they were suffering Hwabyung and significant others confirmed the attribution. The acknowledgement by others, especially the husband, gives women with Hwabyung special status in the family and may actually reduce the abuse. Therefore, having Hwabyung may mitigate against woman desiring to leave. Psychological-Relational Power and Abuse Intolerance. The magnitude of the relationship between psychological relational power and abuse intolerance was .77 (p < .001). This indicated that the higher the levels of psychological relational power the more desire the women had to leave their abusive husbands. Women who perceived domestic violence as a deviant behavior were more likely to have higher levels of psychological relational power and so had strong propensities to leave their abusive husbands.

157

The relationship between psychological relational power and abuse intolerance was strong as shown by the magnitude (r = .77). This could be explained by women’s perceptions of domestic violence in Korean culture. Psychological relational power was defined as women’s cognitive or emotional capacity that influences their perceptions of domestic violence as deviant behavior. Psychological relational power is influenced by women’s beliefs, values, and perspectives, which are shaped by culture. Culture guides the way in which social relations of women are structured, shaped, experienced, and understood (O’Hagan, 1999). Women develop knowledge, beliefs, morals, and habits as a result of interactions with members of society and this leads women to develop different abilities to recognize or perceive domestic violence, which ultimately influence women’s perceptions of power (Taylor, 1871). These findings are similar to those of Morash et al. (2000), Noller & Feeney (2002), Markowitz (2001), and Osmond & Thorne (1993) who found that in Western society the termination of relationships is influenced by women’s power, which is related to resources, satisfaction, and gender roles. Campbell & Soeken (1999) and Gelles (1976) reported that women’s termination of marriage is influenced by their power they perceive. Several studies have also supported the relationship between power and the perception of abuse (Abrams, 2003; Campbell & Soeken, 1999; Gelles, 1976). Summary. Abuse intolerance was predicted by abuse, Hwabyung, and psychological relational power. A total of 86% of the variance of abuse intolerance was accounted for by these three variables. Korean women’s abuse intolerance was strongly influenced by

158

women’s psychological relational power followed by being physically, psychologically, sexually, or financially abused and the cultural bound syndrome, Hwabyung. What are the relationships among psychological relational power, abuse, and Hwabyung? Psychological-Relational Power and Abuse. Psychological relational power had a positive relation with abuse (r = .62; p < .001), indicating that the greater the psychological relational power the more women reported that they had been abused. The relationship between psychological relational power and abuse was strong as shown by the magnitude (r = .62). This also could be explained by women’s perceptions of domestic violence. This would make sense because psychological relational power is related to the cognitive or emotional capacity that influences a woman’s perception of domestic violence as a deviant behavior. Women who had high levels of psychological relational power were more likely to perceive abuse as a violent behavior than those who had lower levels of psychological relational power. Abuse was measured based on women’s ability to recognize elements of their own situations in the definitions provided. This recognition was influenced by women’s perceptions, as structured by their experiences, knowledge, and education. Experiences strengthen women’s psychological relational power (Campbell et al., 1998). Women perceive domestic violence as a violent behavior as they experience and increase knowledge. A total of 38% of the variance of abuse was accounted for by psychological relational power.

159

Psychological-Relational

Power

and

Hwabyung.

The

relationship

between

psychological relational power and Hwabyung was .07, indicating that Hwabyung was not influenced by psychological relational power. Psychological relational power did not predict Hwabyung as shown by the non-significant regression weights. According to Min & Lee (1989), Min et al. (1993) and Park et al., (2001), Hwabyung was reported most commonly among women with low levels of education and income. Women with higher levels of psychological relational power are more likely to be educated and have higher incomes than women with lower levels of psychological relational power. Women with higher levels of psychological relational power tend not to manifest Hwabyung symptoms. This may be explained by women’s power and coping strategies. Women with higher power were more likely to cope with Hwabyung symptoms than women with lower power (Min et al., 1993; Part et al., 2001). Therefore, women with higher levels of psychological relational power were less likely to manifest Hwabyung symptoms than women with lower levels of psychological relational power. Hwabyung was not influenced by women’s psychological relational power as shown by the magnitude of relationship. Summary. A total of 18% of the variance of Hwabyung was accounted for by psychological relational power and abuse, indicating that Hwabyung was moderately influenced by abuse but not influenced by psychological relational power.

160

What aspects of psychological relational power mainly contribute to abuse intolerance? Psychological-Relational Power and Collectivism. The relationship between psychological relational power and collectivism was -.97 (p < .001). Women with higher levels of psychological relational power were more likely to have lower levels of collectivism (individualists). The relationship between psychological relational power and collectivism was strong as shown by the magnitude (r = -.97). Collectivism is related to women’s lower levels of idiocentric values. Women with idiocentric values (individualists) value their own success and de-emphasize family harmony and integration (Hofstede, 2001). Women with individualistic values are more likely to be independent of others or family than women who embrace collectivism (Matsumoto, 1996; Yamaguchi, 1994; Triandis, 1999). According to Hofstede (1994), women who strongly ascribe to individualistic ideology are more likely to show power in their social relations. As a result, women with individualistic values had higher levels of psychological relational power, which eventually influenced their propensities to leave their abusive husbands. A total of 94% of the variance of collectivism was accounted for by psychological relational power. The relative amount of the variance of collectivism was the highest among the four indicators.

161

Psychological-Relational Power and Traditional Family Ideology. The relationship between psychological relational power and traditional family ideology was -.85 (p < .001). Women with higher levels of psychological relational power were more likely to show lower levels of traditional family ideology. The relationship between psychological relational power and traditional family ideology was strong as shown by the magnitude (r = -.85). This could be explained by women’s traditional beliefs and values in power relations. In the Korean cultural context, traditional family ideology emphasizes the predominant role of husband in family matters (Levinson & Huffman, 1955; O’Hagan, 1999). Women with low levels of traditional beliefs and roles with regard to family matters were more likely to have high levels of psychological relational power. This may be true because women with high levels of traditional beliefs and roles with regard to family matters were more likely to stick with family and less likely to value their rights. Women with high levels of power value women’s equal rights and de-emphasize women’s obedience to their husbands (Levinson & Huffman, 1955: Langer & Rodin, 1976). Women with higher levels of psychological relational power were less likely to be obedient to husband and thus more likely to show lower levels of traditional family ideology. A total of 80% of the variance of traditional family ideology was accounted for by psychological relational power. The relative amount of the variance of traditional family ideology was second highest among the four indicators.

162

Psychological-Relational Power and Marital Satisfaction. The relationship between psychological relational power and marital satisfaction was -.89 (p < .001). Women with high levels of psychological relational power were more likely to show low levels of marital satisfaction, indicating that women with high levels of marital conflict were more likely cognitively or emotionally to perceive domestic violence as a deviant behavior in their marriages than those women with high levels of marital satisfaction. The relationship between psychological relational power and marital satisfaction was strong as shown by the magnitude (r = -.89). This could be explained by social exchanges in their marriages. Women with lower levels of marital satisfaction are more likely to identify the specific behaviors of abusive husbands (Noller & Feeney, 2002; Pinkley, & Northcraft, 1994). Women with lower levels of marital satisfaction are more likely to establish strong expectations about the type of the relationship and desires that the couples are experiencing as rewards (Gelles, 1972 & 1976; Bramlett & Mosher, 2001; Rogers, 2003). According to Noller & Fitzpatrick (1993), when marital expectations are violated, the conflict is more likely to be increased. When a woman is in marital conflict, she analyzes the reality (contextual characteristics of the situation) behind the power, and this leads to a more negative evaluation toward her abusive husband and subsequently impacts on her perception of domestic violence (Langer & Rodin, 1976). With her increasing power, a woman tends to modify her original needs in the situation of marital conflict (Sarantakos, 2002). When the modified needs are not acceptable, the marital conflict tends to be increased and this leads domestic violence in marriages. The stronger

163

the power obtained by women, the more the women perceive domestic violence as a deviant behavior. Women in higher levels of marital conflict are more likely to be abused as the conflict leads women to risk escalating the abuse (Klein & White, 1996). Therefore, women in a marital conflict situation are more likely to perceive domestic violence as a deviant behavior. A total of 73% of the variance of marital satisfaction was accounted for by psychological relational power. Psychological-Relational Power and Power Ascription. The relationship between psychological relational power and power ascription was .66 (p < .001). Women with high levels of psychological relational power were more likely to have high levels of power in decision-making. This indicates that women with higher levels of psychological relational power believe that they have strong personalities in decision-making. The strong relationship between psychological relational power and power ascription could be explained by the perception of gender role. According to Daly (2001) and Raven (1992), power ascription (gender role identity) is influenced by an individual’s personality or characteristics. Women with low levels of power ascription have low levels of decision-making power and emphasize stereotypical gender roles (Osmond & Thorne, 1993; Blumberg & Coleman, 1989). Women with high levels of power ascription have high levels of decision making power and emphasize non-traditional female roles and de-emphasize the culturally sanctioned roles of the female (Blumberg & Coleman, 1989; James & Dallacosta, 1973; Osmond & Thorne, 1993). Therefore, women with higher levels of psychological relational power are more likely to show higher levels of

164

power ascription, indicating that they believe that they have strong personalities in decision-making. A total of 44% of the variance of power ascription was accounted for by psychological relational power. Summary. Psychological relational power predicted 94% of the variance of collectivism, 80% of the variance of traditional family ideology, 73% of the variance of marital satisfaction, and 44% of the variance of power ascription. Among the four indicators, collectivism was the strongest explanatory indicator of psychological relational power that influenced women’s propensities to leave their abusive husbands. What are the relationships of variables within the frameworks of socio-structural and psychological relational power? Psychological-Relational Power and Socio-Structural Power. Socio-structural power is the capacity to influence a situation or an event in the socio-cultural context. Sociostructural power was measured by age, education, and income and was hypothesized to influence psychological relational power. Age, education, and income had common effects on psychological relational power and influenced the levels of responses to abuse indirectly through psychological relational power. The indicators (age, education, and income) influenced women’s propensities to leave their abusive husbands through psychological relational power. The relationship between age and psychological relational power was -.73 (p < .001). Younger women had higher levels of psychological relational power. Many studies have supported this relationship (Blumberg & Coleman, 1989; Gelles, 1972 & 1976; Bramlett & Mosher, 2001; Rogers, 2003).

165

Less educated women had higher levels of psychological relational power as shown by the inverse relationship (r = -.17) and the correlation was not significant. This finding is ignorable because the value of C.R (critical ratio) was less than 1.96. The relationship between income and the power was also inverse as shown by r = -.06. However, the correlation was not significant and ignorable (C.R < 1.96). Among relationships between socio-structural power and age, education, and income, the relationship between sociostructural power and age supported the hypothesis. Summary. Perception of domestic violence depended on socio-structural power. Age significantly influenced women’s perceptions of abuse. A total of 33% of the variance of psychological relational power was accounted for by age, education, and income. Sociostructural power consisting of age, education, and income shared common effects on psychological-relational power. Among the component elements, age had the strongest explanatory power to predict psychological relational power. The relationship between psychological relational power and age supported the hypothesis.

166

Limitations

Several limitations influenced the interpretation of the findings related to testing psychometric properties and testing the model. Limitations included sample selection bias, translation issues, high Cronbach's alpha, missing data, multivariate non-normality, failure to accept the fit of the Chi-square test, and theoretical re-specification of the model. Sample Selection Bias Participants in this study were recruited from a psychiatry department, shelters, home healthcare centers, and two communities. Women from the psychiatry department had higher Hwabyung symptoms than women from other recruitment sites. Women from shelters were temporarily separated from their husbands and had a higher probability of divorcing their husbands than women living in the communities. Women from shelters were more likely to recognize that they had been abused and more likely to tend to think leaving their abusive husbands. The outcome variable, abuse intolerance measuring women’s propensities to leave their abusive husbands was possibly influenced by including these women. In addition, it is important to acknowledge that non-abused women were excluded from the sample. This selection bias had the effect of altering the estimation of parameters by attenuating relationships among variables in the model testing.

167

Translation Issue Translation of measurement instruments is a vital aspect of construct validity and cross-cultural equivalence. The translation of measurement instruments from the literature was done as carefully a possible to avoid measurement error. For example, the strategies or using simple sentences, selecting active and easy sentences, and avoiding hypothetical phrasings were used during the translation process. However, selecting appropriate terms that are culturally relevant is not easy. Choosing inappropriate terms may be a source of measurement error that arises in the process of translation. In addition, back translation using the bilingual technique of existing measurement instruments from literature was not done for instruments from the literature. Specific considerations of translation methods are necessary in the formation of validated questionnaires for cross-cultural research (Danielian, 1969). High Cronbach's Alpha High Cronbach's alpha indicates high internal consistency of items of a scale. Cronbach's alphas of KWAST and KWAIS were .91 and .98 respectively. These high alphas indicated that a set of items measured a single uni-dimensional latent construct. Therefore, there was evidence that the items measured the same underlying construct. The high Cronbach’s alphas, however, also suggested that there were overlaps (redundancy) in content among some of the items. This could have been true because of overlaps in content or participants’ response biases to the test items. Relative to overlaps in content, several factors may be important. First, three culturally relevant items were added to the final revision of KWAIS. These added items

168

were possibly redundant. Second, derived and modified scales from each original scale contained possibly redundant items because: 1) the preliminary measurement was designed to be overly inclusive; and 2) selected items from the original scales were chosen to some degree because of their relationships to abuse intolerance. Relative to vocabulary problems, translation procedure related to choosing inappropriate words in the translation process could be associated with redundancy. This redundancy issue would be resolved by reducing the number of items in a future study. Missing Data Missing data are a frequent and particularly difficult problem in analyzing data using SEM. First, missing data reduce efficiency because missing data typically lower the statistical power of hypothesis tests and the precision of estimation (Little & Rubin, 2002). Second, standard complete data methods could not be immediately used to analyze the data in SEM (Raaijmakers, 1999). Missing data mechanism was determined to effectively analyze the data of the study. Education and income were determined as MCAR by Little’s test. The data was analyzed using FIML under the consideration of MCAR. FIML required no imputation and produced unbiased parameter estimates and standard errors using all information in the data set to properly estimate parameters. However, theoretically the imputation using FIML could be wrong because the likelihood is computed for the observed portion of each case’s data and then accumulated and maximized. Missing data was a limitation in testing the theoretical model.

169

Multivariate Non-normality Multivariate normal distribution in the dataset is considered an important assumption in SEM to estimate the corresponding population correctly in the statistical sense. This assumption requires Maximum Likelihood estimation to produce appropriate estimates and reasonable sample size of at least 200 observations (Enders, 2001b; Byrne, 2001; Bollen, 1989). The maximum likelihood estimation assumed the distribution of observed variables was multivariate normal. When the data do not follow multivariate normal distribution, the estimation tends to inflate the value of chi-square (Bollen, 1989; Byrne, 2001). As a result, a model may be rejected due to non-normal multivariate distribution. Normal distribution is important in most cases because, philosophically speaking, it represents one of the empirically verified elementary truths in measuring the general nature of reality. If the sample size was large enough and randomly selected among married women, the data would be normally distributed and thus the data set would be within the shape of multivariate normality, the goodness of fit indexes would well fit the data. However, the sample data were derived from designated places including shelters, home health care centers, a psychiatry department and communities. Those people participated in the study were self-identified as being abused. Those subjects constituted a convenient sample. Therefore, failing multivariate normal distribution is expected. Handling non-normal and missing data in the structural equation modeling is the most important issue in testing overall model fit and testing significance of individual parameter estimate values. These tests assume that the fitted structural equation model is

170

true and the data used to test the model follow multivariate normal distribution in the population from which the researcher drew the sample data. The accuracy of Chi-square statistic can be compromised given violation of the distributional assumption. Violations of both the distributional and structural assumptions are common in reality and often unavoidable. Thus, the violations can potentially lead to seriously misleading results. As a result, the Chi-square test statistic of overall model fit was possibly inflated and the standard errors used to test the significance of individual parameter estimates were deflated. Therefore, multivariate non-normality in the structural model was one of the limitations. Chi-square Two models, a measurement and a structural model were tested. The measurement model fit the sample data while the structural model did not. The goodness of fit index for the measurement model was acceptable, showing that the indicators (collectivism, traditional family ideology, marital satisfaction, and power ascription) adequately measured the intended construct (psychological relational power). This accepted measurement model allowed the structure to work. However, the structural model did not fit the sample data. There are several statistical and theoretical reasons for a model to fail the exact fit test. First, the hypothesized model may not be well grounded in theories that are consistent with the real world. This could be explained by the magnitude of some relationships in a model. For example, only 18% of the variance of Hwabyung was accounted for by psychological relational power and abuse. The relative contribution of

171

Hwabyung as a predictor for abuse intolerance was only .09. Therefore, Hwabyung did not predict abuse intolerance in the model. Women who suffered from Hwabyung did not have propensities to leave their abusive husbands. When Hwabyung was deleted from the model, the model fit the sample data. Further study using model modification is needed. Two, the lack of sample size was a factor that was consistently related to the outcome of Hoelter’s CN. In SEM analysis, the sample size is critical to test the model properly. A total of 184 was not sufficient to yield the chi-square statistics. Failure of the exact fit test means that the data matrix does not truly match the model matrix. Mulaik (SEM dialogue, Aug 3, 2004) stated, “Classical statistical hypothesis testing is an inappropriate procedure for selecting a restrictive structural model for multivariate data.” Matching up with the real world is a good basis for defending the exact fit test. However, this is not sufficient reason in itself to reject the model because there is no evidence that there is a correlation between passing the exact fit test and having the model match the actual phenomenon. Many researchers argue that a failed chisquare test is not evidence that the hypothesized model is non-existent (SEM dialogue, Aug 3, 2004; SEM dialogue, Dec 6, 2003). In addition, there is not a good a priori reason to believe that virtually all-passing models are sensitive enough to prove the model fits the data. Therefore, the chi-square cannot distinguish between theoretically correct and theoretically incorrect models, but it does provide one piece of information about whether the hypothesized model perfectly fits in the population (SEM dialogue, Dec 6, 2003). The chi-square test is not a measure of how close a model is to the real world. It does not test free parameters, conceptual identification of factors, or the existence of alternative

172

models that are as good as, or better than the hypothesized model. Statistical results are not the only evidence to consider in determining whether the model is correct. Other evidence such as actual knowledge of the phenomenon being measured and cultural beliefs and values in the Korean context should be considered to practically and theoretically understand the model. An incorrect model does not necessarily mean the model does not represent the state of reality. Therefore, significant relationships including associations, directions, and magnitudes relevant to Korean culture were discussed. Theoretical Re-specification of the Model Relative to the theoretical re-specification of the model, several issues should be discussed. First, socio-structural power and its component elements should be dealt with. Age, education, and income were component elements of socio-structural power and were considered with socio-economic status predicting women’s power. The socioeconomic status impact on domestic violence has been extensively discussed (Byrne et al., 1999; Gelles, 1976; Braun, 1995; Devine et al., 1988). Studies reported that victims of domestic violence are likely to be of younger age, with lower levels of income and education, and were unemployed (Baron & Straus, 1988; Gelles, 1976; Grana, 2001). Others studies point out that controversy exists on the relative influence of cultural and economic factors on domestic violence (Braun, 1995; Byrne et al., 1999). These studies measured socio-economic status using poverty levels in the family. A well-structured poverty index may well predict women’s propensities to leave their abusive husbands.

173

However, for this study, women’s socio-economic status was measured based on self-reported age, education, and monthly income regardless of their employment status. For example, women’s income levels were not related to poverty status or their husbands’ income level. Korean women traditionally have been less educated and thus more frequently unemployed than men. Inequality between women and men is more likely to occur in a society in which traditional social patterns are in flux (Baron & Straus, 1989). According to Straus (1994), social conditions such as community change, social bonds, social support system, and government policy are possible variables to be considered in the relationship between the incidence of victims and domestic violence. Therefore, re-conceptualizing socio-economic status in the Korean cultural context is important to measure women’s power, which predicts their propensities to leave abusive husbands. Second, the conceptualization of psychological-relational power was a limitation in terms of predicting relationships between psychological-relational power and Hwabyung. The relationship between psychological-relational power and Hwabyung was not significant. This relationship could be understood by Korean cultural values and this was discussed in earlier pages. Third, the relationships among Hwabyung, abuse, and abuse intolerance should be re-specified in the model as the relationships were not significant and this also was discussed in earlier pages.

174

Topics for Further Study

Throughout the discussion of the study in Chapter 4, several issues were raised in statistical and theoretical sense that warrant further study. First, the perception of abuse and abuse intolerance needs to be further studied because of the merits of the findings. Second, the perception of Hwayung and its relationship to abuse intolerance, as well as those factors that may influence abused women with Hwabyung to stay or leave their abusive husbands needs to be further studied. Third, the construct, psychological relational power in relation to collectivism, traditional family ideology, marital satisfaction, and power ascription needs to be further studied in a cultural context as culture is not static or rigid but dynamic and everchanging and reacting to as well as producing changes in individuals and society. Fourth, the cultural aspects of perception of abuse and abuse intolerance as well as those factors that may influence women’s propensities to leave their abusive husbands need to be further studied. Fifth, socio-structural power in relation to age, education, and income needs to be further studied, as younger women’s propensities to leave their abusive husbands were different from those of older women. In addition, re-conceptualization of socio-economic status in Korean culture is needed because education and income did not support the hypotheses.

175

Sixth, the model testing of Korean women’s responses to domestic violence needs to be further studied based on socio-structural and psychological relational power.

Implications for Nursing

The purpose of the study was to test the psychometric properties of two newly developed measurements and a hypothetical model of Korean women’s responses to domestic violence. There were significant results of relevance to nursing. First, abuse significantly influenced the development of women’s Hwabyung. Women with Hwabyung have received little attention even though many Korean women suffer from the perception of an obstructive mass in the epigastrium associated with shortness of breath, palpitation, and fear of impending death. Abused women can develop complex psychological problems in the context of their husbands’ ongoing abusive behaviors. This is a significant public health problem due to high prevalence among married women. Surprisingly, little research has been conducted about the relationship of Hwabyung and domestic violence. Furthermore, many married women with Hwabyung do not seek medical assistance until they feel fear of impending death. Family support, medical assistance, and universal screening for abused women and women with Hwabyung are needed in community settings. Second, cultural norms, values, and beliefs are powerful forces in shaping women’s perceptions of domestic violence and dealing with psychological illness. Women with collectivistic values sacrifice their personal desires for the sake of family

176

harmony because they have believed this value is as a virtue as wives and mothers. As a result, women must suppress their anger and rage, and this leads women to develop mental illness. Cultural norms have much more influence on women’s perceptions than the influx of Western culture. Regardless of resource availability for abused women, some abused women tend to be concerned with family rather than taking steps to leave their abusive husbands. The cultural value that influenced women’s perceptions of domestic violence and their propensities to leave abusive husbands is important in nursing research. Cultural values are important variables that influence women’s perceptions of domestic violence. The values that affirm women’s cultural identities not only influence women’s mental illness but also impact on women’s vulnerability. Many women with mental illness avoid getting medical help due to a stigmatization of mental illness. These cultural beliefs are different for each individual as younger women had higher levels of psychological relational power in the model. However, little research has been conducted about women’s perceptions of domestic violence and mental illness in power dynamics. Understanding cultural assumptions that guide Korean women’s behaviors would facilitate the understanding of their propensities to leave their abusive husbands. This would help to clarify the individual variations in dynamic cultural power.

177

Conclusion

This study is important in Korean cultural context for several reasons. First, in Korea, prior to this study there did not exist a culturally appropriate women’s abuse screening tool or a tool measuring women’s propensities to leave their abusive husbands. This is important because many women are victims of domestic violence in the family system and the incidence of domestic violence has increasingly been recognized. The high Cronbach’s alphas of two newly developed measurement instruments indicated homogeneity of variances among items that support an excellent value of reliability. Construct validity assessment using CFA supported a hypothesis that abuse predicts abuse intolerance. Second, the findings of the theoretical model testing are significant in the Korean cultural context. Many married Korean women do not perceive domestic violence in their marriages because of cultural acceptance of men’s abusive behaviors and traditional wives’ values. Results of the model testing suggest that theories that explain women’s responses to abuse in Western culture have explanatory power in Korean culture. Third, a theoretical model of women’s responses to domestic violence has not been developed or tested in Korean society. The model tested in this study is an important first step in understanding domestic violence in a Korean context. In addition, this study was designed to aid in understanding women’s dynamic power (sociostructural and psychological-relational power) in the Korean cultural context. Therefore, significant findings including supporting hypotheses would be the basis for a future study.

178

Process of Theory Development

Interview of Korean married abused women Review literature Extract relevant

Theory Analysis

Review definition and meaning of each theory

Appendix A.

Examine strengths and weaknesses of each theory to determine logical adequacy

Extract major concepts of each theory

Generate hypotheses to determine testability of each theory

Extract and graph relational statements of each theory using directions and signs Integrate empirical research findings with extracted relational statements, definitions, and major concepts

Theory Synthesis Confirm purposes and testability by conducting literature review, placing relational statements into graphic form

Socio-structural theory Theory Derivation

Age Education Income

Cultural theory Traditionalism Collectivism

Generate four theories to explain married Korean women’s responses to abuse

Patriarchal theory Power ascription

Identify content and structure of each theory Redefine new concepts and statements

Social exchange theory Marital satisfaction

179

SUBJECT'S CONSENT FORM

Appendix B

Married Korean Women’s Responses to Domestic Violence Within The Framework of Socio-Cultural Context I AM BEING ASKED TO READ THE FOLLOWING MATERIAL TO ENSURE THAT I AM INFORMED OF THE NATURE OF THIS RESEARCH STUDY AND OF HOW I WILL PARTICIPATE IN IT, IF I CONSENT TO DO SO. SIGNING THIS FORM WILL INDICATE THAT I HAVE BEEN SO INFORMED AND THAT I GIVE MY CONSENT. FEDERAL REGULATIONS REQUIRE WRITTEN INFORMED CONSENT PRIOR TO PARTICIPATION IN THIS RESEARCH STUDY SO THAT I CAN KNOW THE NATURE AND RISKS OF MY PARTICIPATION AND CAN DECIDE TO PARTICIPATE OR NOT PARTICIPATE IN A FREE AND INFORMED MANNER.

PURPOSE I am being invited to participate voluntarily in the above-titled research project. The purpose of this project is two-fold. First, this study is to test a theoretical model of married Korean women’s responses to domestic violence within socio-cultural context. Second, new culturally appropriate measurements for screening abuse and assessing intolerance of abuse in Korea will be tested.

SELECTION CRITERIA I am being invited to participate in the study of Korean women’s responses to domestic violence because I am a married Korean woman self-identified as being physically, emotionally, sexually or financially abused. Approximately 200 subjects at four sites including the Yonsei-University Psychology Department, Daejeon Home Care Center, Daejeon Konyang Home Health Center, and Pusan Women Hot Line will be enrolled in this study.

STANDARD TREATMENT(S) If the subjects refuse to consent, this will not affect the care they receive.

180

PROCEDURE(S) If I agree to participate, I will be asked to consent to the following: to fill out the questionnaires at the Yonsei University, Department of Psychiatry, and the Daejeon Home Care Center. It will take approximately one hour to fill out the questionnaires. It will take about 5 to 10 minutes to receive instructions on how to fill out the questionnaires and how to return the questionnaires. RISKS There are no risks associated with participating in this research study. BENEFITS There are no benefits for participating in this research study.

CONFIDENTIALITY Principal investigator, Myunghan Choi will access to the data, and collaborators Drs. Phillips, Cromwell, Insel, Figueredo, and Guggenheim will access to the data if they need it for the future study. The data will be maintained in a lock drawer for five years and then will be shredded. No names are included on questionnaires and individual questionnaires will be kept in a private drawer PARTICIPATION COSTS AND SUBJECT COMPENSATION I will receive no compensation for participating in this research. The only cost to me will be my time approximately one hour. CONTACTS I can obtain further information from the principal investigator _Myunghan Choi, PhD (c), MPH, RN__at (520) 909_-_7654. If I have questions concerning my rights as a research subject, I may call the Human Subjects Committee office at (520) 626-6721. AUTHORIZATION BEFORE GIVING MY CONSENT BY SIGNING THIS FORM, THE METHODS, INCONVENIENCES, RISKS, AND BENEFITS HAVE BEEN EXPLAINED TO ME AND MY QUESTIONS HAVE BEEN ANSWERED. I MAY ASK QUESTIONS AT

181

ANY TIME AND I AM FREE TO WITHDRAW FROM THE PROJECT AT ANY TIME WITHOUT CAUSING BAD FEELINGS OR AFFECTING MY MEDICAL CARE. MY PARTICIPATION IN THIS PROJECT MAY BE ENDED BY THE INVESTIGATOR OR BY THE SPONSOR FOR REASONS THAT WOULD BE EXPLAINED. NEW INFORMATION DEVELOPED DURING THE COURSE OF THIS STUDY WHICH MAY AFFECT MY WILLINGNESS TO CONTINUE IN THIS RESEARCH PROJECT WILL BE GIVEN TO ME AS IT BECOMES AVAILABLE. THIS CONSENT FORM WILL BE FILED IN AN AREA DESIGNATED BY THE HUMAN SUBJECTS COMMITTEE WITH ACCESS RESTRICTED TO THE PRINCIPAL INVESTIGATOR, Myunghan Choi_ OR AUTHORIZED REPRESENTATIVE OF THE _College of Nursing__ DEPARTMENT. I DO NOT GIVE UP ANY OF MY LEGAL RIGHTS BY SIGNING THIS FORM. A COPY OF THIS SIGNED CONSENT FORM WILL BE GIVEN TO ME. __________________________________________________________________ Subject's SignatureDate __________________________________________________________________ Witness (if necessary)Date INVESTIGATOR'S AFFIDAVIT I have carefully explained to the subject the nature of the above project. I hereby certify that to the best of my knowledge the person who is signing this consent form understands clearly the nature, demands, benefits, and risks involved in his/her participation and his/her signature is legally valid. A medical problem or language or educational barrier has not precluded this understanding. __________________________________________________________________ Signature of InvestigatorDate 01/21/04

182

Collectivism; Marital Satisfaction, Traditional Family Ideology, Attitude Toward Power Ascription Appendix C Strongly agree

Strongly disagree

1. Family members should stick together no matter what sacrifices are required.

7

6

5

4

3

2

1

2. It is my duty to take care of my family even when I have to sacrifice what I want.

7

6

5

4

3

2

1

3. Parents and children must stay together as much as possible.

7

6

5

4

3

2

1

4. It is important to me that I respect the decisions made by my family.

7

6

5

4

3

2

1

5. Children should be taught to place duty before pleasure.

7

6

5

4

3

2

1

6. I usually sacrifice my self-interest for the benefit of my family.

7

6

5

4

3

2

1

7. How satisfied are you with your marriage?

7

6

5

4

3

2

1

8. How satisfied are you with your husband as a spouse?

7

6

5

4

3

2

1

9. How satisfied are you with your relationship with your husband?

7

6

5

4

3

2

1

10. Women have as many rights as men do.

7

6

5

4

3

2

1

11. It goes against nature to place women in positions of authority over men.

7

6

5

4

3

2

1

12. A woman should never be allowed to talk back to her husband or else he will lose respect for her.

7

6

5

4

3

2

1

13. Women who want to remove the word “obey” from the marriage ceremony don’t understand what it means to be a wife.

7

6

5

4

3

2

1

14. Assertive

7

6

5

4

3

2

1

15. Strong personality

7

6

5

4

3

2

1

16. Forceful

7

6

5

4

3

2

1

17. Dominant

7

6

5

4

3

2

1

18. Aggressive

7

6

5

4

3

2

1

19. Acts as leader

7

6

5

4

3

2

1

183

Appendix D HB Scale The following statements describe ways in which you might feel as a result of events that have occurred in marriage. For each statement, please respond to each item by indicating “yes” or “no.” Causes of Hwabyung 1. I have been repeatedly treated unfairly by significant people in my life.

Yes No

2. I have lived in poverty or had little opportunity for education or been abused, experienced sudden death of spouse or child, or forced to act illegally etc.

Yes No

3. I have felt angry when the situation in # 1 or # 2 happened.

Yes No

4. I usually repress my feelings or am depressed.

Yes No

5. My current physical and psychological symptoms are due to above situations and these symptoms are related to Hwabyung.

Yes No

Following statements describe ways in which you might feel as a result of events that have occurred in marriage. For each statement, please respond to each of the statements by indicating the extent to which that statement describes your symptoms and feelings: 0: Never; 1: Rarely; 2: Sometimes; 3: Often; 4: Usually; 5: Always.

Symptoms of Hwabyung 1. I have been feeling angry.

5

4

3

2

1

0

2. I have been feeling enraged.

5

4

3

2

1

0

3. I have been feeling persecuted.

5

4

3

2

1

0

4. I have wanted to get away from it all.

5

4

3

2

1

0

5. I have wanted to express my feelings to somebody.

5

4

3

2

1

0

6. I have been depressed.

5

4

3

2

1

0

7. I have been agitated.

5

4

3

2

1

0

8. I have been anxious and have repeated the same action again and again.

5

4

3

2

1

0

9. I have felt that I have a disease.

5

4

3

2

1

0

10. I have felt regret.

5

4

3

2

1

0

11. I have been embarrassed.

5

4

3

2

1

0

12. I have felt guilty.

5

4

3

2

1

0

13. I have felt useless.

5

4

3

2

1

0

184

14. I have been subservient.

5

4

3

2

1

0

15. I have been easily startled.

5

4

3

2

1

0

16. I have been unable to concentrate.

5

4

3

2

1

0

17. I have been irritable.

5

4

3

2

1

0

18. I have had pain.

5

4

3

2

1

0

19. I have felt that I hid my feelings.

5

4

3

2

1

0

20. I have had hot flashes.

5

4

3

2

1

0

21. I often sigh.

5

4

3

2

1

0

22. I have felt something is pushing up (a tightening) in my chest

5

4

3

2

1

0

23. I have had trouble sleeping.

5

4

3

2

1

0

24. I have had headaches.

5

4

3

2

1

0

25. I have felt my heart pound.

5

4

3

2

1

0

26. I have felt dizzy.

5

4

3

2

1

0

27. I have felt tired.

5

4

3

2

1

0

28. I have been very thirsty.

5

4

3

2

1

0

29. I have less or lost interest in sex.

5

4

3

2

1

0

30. I have had stomach problems.

5

4

3

2

1

0

31. I have been unable to tolerate a hot bath or hot room temperature.

5

4

3

2

1

0

185

KWAST

Appendix E

Demographic Information 1.Age:___________Yrs

2.Education:______ Yrs

3. Income: ___________won/month

4. Religion:_________

5. How would you rate your own current health status?: 6. Marital status:

Married

7. Who do you live with?:

1.

Divorced

Separated

Excellent Widow

Good

Fair

Single

Poor

Other: _____________

With husband & no children

With husband & children

With his & my parents/family

With my parents/family

With husband’s parents/family Alone

Other:__________

In general, how would you describe your relationship with your husband? A lot of tension

No tension

5-----------------------4---------------------3------------------------2-------------------------1------------------------0 2.

How do you and your husband work out arguments? With great difficulty

With no difficulty

5-----------------------4---------------------3------------------------2-------------------------1------------------------0 3.

Do arguments ever result in you feeling down or bad about yourself? Always

Never

5-----------------------4---------------------3------------------------2-------------------------1------------------------0 4.

Do arguments ever result in hitting, kicking, or pushing? Always

Never

5-----------------------4---------------------3------------------------2-------------------------1------------------------0 5.

Do you ever feel frightened by what your husband says or does? Always

Never

5-----------------------4---------------------3------------------------2-------------------------1------------------------0

Have you ever been physically abused by your husband? ( Yes

No). “If yes, please mark below.”

Everyday 1-2 times a week 1-2 times a month Several times this past year 1-2 times this past year 5-------------------------4----------------------------3---------------------------------2-----------------------------1

186

Please read the following definitions and then answer the question following each definition. In answering each question, please answer only for the past year. Physical abuse: can include hitting, punching, slapping, pushing, pinching, kicking, burning, stabbing or cutting. Physical abuse is legally known as assault. Your husband intentionally uses force or tries to use force against you without consent. 6. Have you ever been physically abused by your husband? ( Yes below.”

No). “If yes, please mark

Everyday 1-2 times a week 1-2 times a month Several times this past year 1-2 times this past year 5-------------------------4----------------------------3---------------------------------2-----------------------------1 Psychological/Emotional abuse: can include threats, constant criticism and put downs, control of activities, humiliation, insults unjust blaming, name calling, ignoring you, false accusations about loyalties or actions, damaging property, reading another person's mail, harassment at work, and destruction of possessions. 7. Have you ever been psychologically/emotionally abused by your husband? ( Yes please mark below.”

No). “If yes,

Everyday 1-2 times a week 1-2 times a month Several times this past year 1-2 times this past year 5-------------------------4----------------------------3---------------------------------2-----------------------------1

Sexual abuse: can include forcing someone to participate in unwanted, unsafe or degrading sexual activity, or using ridicule or other tactics to try to denigrate, control or limit their sexuality or reproductive choices. 8. Have you ever been sexually abused by your husband? ( Yes below.”

No). “If yes, please mark

Everyday 1-2 times a week 1-2 times a month Several times this past year 1-2 times this past year 5-------------------------4----------------------------3---------------------------------2-----------------------------1 Financial abuse: can include taking your pay check, withholding money from you so that you cannot buy essential items, withholding access to family money, withholding control over what is spent or saved, withholding control over any purchasing decisions, and withholding money for personal use. 9. Have you ever been financially abused by your husband?( Yes below.”

No). “If yes, please mark

Everyday 1-2 times a week 1-2 times a month Several times this past year 1-2 times this past year 5-------------------------4----------------------------3---------------------------------2-----------------------------1 If you answered “yes” to any of the types of abuse above, then please fill out the next questionnaire. If you answered “no” to ALL of the questions, do not complete the next questionnaire.

187

Appendix F KWAIS The following statements describe ways in which you might feel as a result of abuse. For each statement, please indicate the extent to which you feel that way by circling a number from 5 (indicating you strongly agree with the statement) to 1 (indicating that you disagree with the statement). Strongly Agree

Strongly Disagree

1. I feel like I want to run away because of my abusive husband.

5

4

3

2

1

2. I would be happy if I could leave my abusive husband.

5

4

3

2

1

3. I have never thought about leaving my abusive husband.

5

4

3

2

1

4. It is difficult for me to be happy living with my abusive husband.

5

4

3

2

1

5. I am accustomed to my husband’s abuse and it doesn’t bother me.

5

4

3

2

1

6. I think being abused is normal for a wife.

5

4

3

2

1

7. I often imagine leaving my abusive husband.

5

4

3

2

1

8. If there were a way, I would leave my abusive husband.

5

4

3

2

1

9. I have been thought to tolerate my husband’s abuse.

5

4

3

2

1

10. Sacrificing myself for my family harmony is more important than leaving my abusive husband.

5

4

3

2

1

11. I think having a child makes it impossible for a wife to leave an abusive husband.

5

4

3

2

1

12. I would leave my abusive husband if my family, friends, and neighbors understood and supported me.

5

4

3

2

1

13. I would leave my abusive husband if I knew of social support systems (e.g., shelters, job training, etc.) to support me.

5

4

3

2

1

188

REFERENCE Abrams, L. S. (2003). Contextual variations in young women’s gender identity negotiations. Psychology of Women Quarterly, 27, 64-74. Ahn, M. E., Ahn, H. C., Choi, J. T., Choi, Y.M., You, K. C., Cho, Y. J., Hwang, J. H., Song, J. H., Shin, D. H., & Song, K. J. (2000). The patient who presented to the Emergency Department because of domestic violence. Journal of the Korean Society of Emergency Medicine, 11(1), 54-65. Allison, P. D. (2003). Missing data techniques for structural equation modeling. Journal of Abnormal Psychology, 112(4), 545-557. Anastasi, A. (1988). Psychological Testing (6th Ed.). New York: Macmillan. Anderson, N. B., & Armstead, C. A. (1995). Toward understanding the association of socioeconomic status and health: A new challenge for the biopsychosocial approach. Psychosomatic Medicine, 57(3), 213-225. Arbukle, J. L., & Wothke, W. (1999). AMOS 4.0 user’s guide [Computer software manual]. Chicago: Smallwaters. Baron, L., & Straus, M. A. (1988). Cultural and Economic Sources of Homicide in the United States. Sociological Quarterly 29, 371-392. Baron, L., & Straus, M. A. (1989). Four theories of rape in American society: A statelevel analysis. New Haven, CT: Yale University Press. Barriball K. L., Christian S. L., While A. E., & Bergen A. (1996). The telephone survey method: A discussion paper. Journal of Advanced Nursing, 24(1), 115-121.

189

Baxter, J. (1993). Work at home: The domestic division of labour. St. Lucia: University of Queensland Press. Becker, G. S. (1981). A treatise on the family. Cambridge, MA: Harvard University Press. Bellah, R. N., Madsen, R., Sullivan, W. M., Swidler, A., & Tipton, S. M. (1985). Habits of the heart: Individualism and commitment in American Life. San Francisco: Harper & Row. Bentler, P. M. (1990). Comparative fit indexes in structural models. Psychological Bulletin, 107, 238-246. Bentler, P. M. (1992). On the fit of models to covariances and methodology to the bulletin. Psychological Bulletin, 112(3), 400-404. Bernard, T. (1990). Angry aggression among the “truly disadvantaged” Criminology, 28, 73-96. Berry, J. W., Segall, M. H., & Kagitcibasi, C. (1997). Handbook of cross-cultural psychology. In J. W. Berry, M. H. Segall, C. & Kagitcibasi (Eds.), Sex, gender, and culture (pp. 163-212). Boston: Allyn and Bacon. Blau, P. M. (1964). Exchange and power in social life. New York: Wiley and Sons. Blumberg, R. L., & Coleman, M. T. (1989). A theoretical look at the gender balance of power in the American couple. Journal of Family Issues, 10, 225-250. Bollen, K. A. (1986). Sample size and Bentler and Bonett’s nonnormed fit index. Psychometrika, 51, 375-377. Bollen, K. A. (1989). Structural equations with latent variables. New York: Wiley.

190

Bourque, L. B. & Clark, V. A. (1992). Processing data: The survey example. Series: Quantitative applications in the social sciences. London: SAGE Publications. Bramlett, M. D., & Mosher, W. D. (2001). First marriage dissolution, divorce, and remarriage: United States. Advance data from vital and health statistics no. 323. Braun, D. (1995). Negative consequences to the rise of income inequality. Res. Pol. Soc, 5, 3-31. Byrne, B. M. (2001). Structural equation modeling with AMOS: Basic concepts, applications, and programming. New Jersey: Lawrence Erlbaum Associates Publishers. Byrne, C. A., Resnick, H. S., Kilpatrick, D. G., Best, C. L., & Saunders, B. E. (1999). The socioeconomic impact of interpersonal violence on women. Journal of Consulting and Clinical Psychology, 67(3), 362-366. Campbell, J. C. (1989). A test of two explanatory models of women’s responses to battering. Nursing Research, 38, 18-24. Campbell, J. C., Rose, L., Kub, J., & Nedd, D. (1998). Voices of strength and resistance: A contextual and longitudinal analysis of women’s responses to battering. Journal of Interpersonal Violence, 13(6), 743-762. Campbell, J. C., & Soeken, K. L. (1999). Women’s response to battering: A test of the model. Research in Nursing and Health, 22(1), 49-58. Carmines, E. G., & Zeller, R. A. (1979). Reliability and validity assessment. London: Sage Publications. Cassani S. H. D. B., Zanetti M. L., & Rotter N. T. P. (1992). The telephone survey: A

191

methodological strategy for obtaining information. Journal of Advanced Nursing, 17, 576-581. Chadwick-Jones, J. K. (1976). Social exchange theory: Its structure and influence in social psychology. New York: European Association of Experimental Social Psychology by Academic Press. Chang, H. (1995). The effects of employment status, occupation, and employment sector on conjugal power relations and marital satisfaction among Korean immigrant married women. Unpublished doctoral dissertation, UCLA. Chang, H. K., & Moon, A. (1998). Work status, conjugal power relations, and marital satisfaction among Korean immigrant married women. In. Y. I. Song, & A. Moon (Eds.), Korean American women: From tradition to modern feminism (pp. 75-87). London: Praeger. Check, J. V. P., & Malamuth, N. M. (1983). Sex role stereotyping and reactions to depictions of stranger versus acquaintance rape. Journal of Personality and Social Psychology, 45, 344-356. Cho, H. (1998). Male dominance and mother power: The two sides of Confucian patriarchy in Korea. In W.H. Slote, & G.A. DeVos (Eds.), Confucianism and the family (pp. 187-207). Albany, New York: State University of New York Press. Choi, M., & Harwood, J. (2004). A hypothesized model of Korean women’s responses to abuse. Journal of Transcultural Nursing, 15(3), 207-216. Chon, K. K., Whang, W. W., Kim, J. W., & Park, H. K. (1997). Emotional stress and Hwa-Byung. Korean Journal of Health Psychology, 2(1), 168-185.

192

Chou, C. P., & Bentler, P. M. (1995). Structural equation modeling: Concepts, issues, and applications. In Rick H. Hoyle, (Ed.). Thousand Oaks, CA: Sage Publications. Cohen, J, & Cohen, P. (1983). Applied multiple regression/correlation for the behavioral sciences (2nd Ed.). London: Lawrence Erlbaum Associates, Publishers. Coontz, S. (1992). The way we never were: American families and the nostalgia trap. New York: Basic Books. Daly, M., & Wilson, M. (1996). Sex, power, conflict: Feminist and evolutionary perspectives. In D. M. Buss, & N. Malamuth (Eds.), Evolutionary psychology and marital conflict: The relevance of stepchildren (pp. 9-28). New York: Oxford University Press. Daly, J. L. (2001). Gender equality rights versus traditional practices: struggles for control and change in Swaziland. Development Southern Africa, 18(1), 45-56. Danielian, J. (1969). Development of construct-relevant and culturally non-biased criteria for measuring judging accuracy. International Journal of Psychology, 4(2), 129-134. Davey, S. G., Carroll, D., Rankin, S., & Rowan, D. (1992). Socioeconomic differentials in mortality: Evidence from Glasgow graveyards. BMJ, 305, 15541557. Dawson, M. M. (1942). The basic teachings of Confucius. New York: The New Home Library. Dempster, A., Laird, N., & Rubin, D. (1977). Maximum likelihood from incomplete data via the EM algorithm. Journal of the Royal Statistical Society, 39(1), 1-38. DeVellis, R. F. (2003). Scale development : Theory and applications (2nd Ed.). SAGE,

193

Applied Social Research Methods Series, Vol 26. Devine, J. A., Sheley, J. F., & Smith, M. D. (1988). Macroeconomic and social-control policy influences on crime rate changes, 1948-1985. American Sociological Review, 53, 407-420. Dietz, T., Stern, P. C., & Guagnano, G. A. (1998). Social structural and social psychological bases of environmental concern. Environment and Behavior, 30(4), 450-472. Digital Korea Herald (2002). Korea Now. Retrieved April 11, 2003, from http://kn.koreaherald.co.kr/site/data/html_dir/2002/04/11/200204110031.asp Draper, N, & Smith, H (1981). Applied Regression Analysis, 2nd ed., John Wiley and Sons. Duncan, J. (1998). The Korean adoption of Neo-Confucianism: The social context. In W. H. Slote, & G. A. DeVos (Eds.), Confucianism and the family (pp. 37-51). Albany, New York: State University of New York Press. Enders, C. K. (2001a). The relative performance of full information maximum likelihood estimation for missing data in structural models. Structural Equation Modeling: A Multidisciplinary Journal, 8(3), 430-457. Enders, C. K. (2001b). The impact of non-normality on full information maximumlikelihood estimation for structural equation models with missing data. Psychological Methods, 6(4), 352-370. Enders, C.K. & Bandalos, D.L. (1999). The effects of heterogeneous item distributions on reliability. Applied Measurement in Education, 12, 133-150.

194

Figueredo, A. J., Ferketich, S. L., & Knapp, T. R. (1991). Focus on psychometrics more on MTMM: The role of confirmatory factor analysis. Research in Nursing & Health, 14, 387-391. Figueredo, A. J. (1999). The entity of emergent variables II: Applicable quantitative methods. Paper presented at the meeting of the American Evaluation Association Annual meeting, November, Orlando, Florida. Fitzpatrick, M. A. (1993). Communication in family relationships. In P. Noller, & M.A. Fitzpatrick (Eds.), Communication and marital satisfaction (pp. 163-179). New Jersey: Prentice Hall, Englewood Cliffs. Fitzpatrick, M. A., & Ritchie, L. D. (1992). Communication theory and the family. In P. G. Boss, W. J. Doherty, R. LaRossa, W. R. Schumm, & S. K. Steinmetz (Eds.), Sourcebook of family theories and methods: A contextual approach (pp. 565-585). New York: Plenum Press. Forbes, D. A., King, K. M., Kushner, K. E., Letourneau, N. L., Myrick, A. F., & ProfettoMcGrath, J. (1999). Warrantable evidence in nursing science. Journal of Advanced Nursing, 29(2), 373-379. Furstenberg, F. F., & Cherlin, A. (1992). Divided families: What happens to children when parents part. Cambridge, MA: Harvard University Press. Gelles, R. J. (1972). The violent home: A study of physical aggression between husbands and wives. London: Sage Publications. Gelles, R. J. (1976). Abused wives: Why do they stay? Journal of Marriage and the Family, 38, 659-668.

195

Gilligan, C. (1982). In a different voice: Psychological theory and women’s development. Cambridge, Mass: Harvard University Press. Glassman, J. (2003). Rethinking overdetermination, structural power, and social change: A critique of Gibson-Graham, Resnick, and Wolff (pp. 678-698). Malden, MA: Blackwell Publishing. Graham, J. W., Hofer, S. M., & MacKinnon, D. P. (1996). Maximizing the usefulness of data obtained with planned missing value patterns: An approach of maximum likelihood value procedures. Multivariate Behavioral Research, 31, 197-218. Grana, S. J. (2001). Sociostructural considerations of domestic violence. Journal of Family Violence, 16(4), 421-434. Green, L. W. (1970). Manual for scoring socioeconomic status for research on health behavior. Publich Health Reports, 85(9), 815-827. Hafner, R. J. (1984). The marital repercussions of behavior therapy for agoraphobia. Psychotherapy, 21, 530-542. Hansen, G. L. (1987). Reward level and marital adjustment: The effect of weighting rewards. The journal of Social Psychology, 127(5), 549-551. Hash, V., Donlea J., & Wallasper, D. (1985). The telephone survey: a procedure for assessing educational needs of nurses. Nursing Research 34, 126-127. Helton, A. S., McFarlane, J., & Anderson, E. T. (1987). Battered and pregnant: A prevalence study. Am J Public Health, 77, 1337-1339. Hoelter, J. W. (1983). The effects of role evaluation and commitment on identify salience. Social Psychology Quarterly, 46, 140-147.

196

Hofstede, G. (1980). Culture’s consequences. Beverly Hills, CA: Sage. Hofstede, G. (1994). Cultures and organizations: Software of the mind. London: McGraw Hill. Hofstede, G. (2001). Individualism and collectivism. Culture’s consequences (pp. 209278). CA: Sage Publication, Inc. Homans, G. C. (1964). Contemporary theory in sociology. In Faris, R. E. L. (Ed.), Handbook of Modern Sociology, Chicago: Rand McNally. Hox, J. J., & Bechger, T. M. (1998). An introduction to structural equation modeling. Family Science Review, 11, 354-373. Hoyle, R. (1995). Structural equation modeling: concepts, issues and applications. Thousand Oaks, CA: Sage Publications. Hu, L., & Bentler, P. M. (1995). Evaluating model fit. In R. H. Hoyle (Ed.), Structural equation modeling: Concepts, issues, and applications (pp. 76-99). Thousand Oaks, CA: Sage Publications. Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6(1), 1-55. James, S., & Dallacosta, M. (1973). The Power of Women and the Subversion of the Community. Falling Wall Press. Jones, E. (1986). Translation of quantitative measures for use in cross-cultural research. Nursing Research, 36(5), 324-327.

197

Jones, P. S., Lee, J. W., Phillips, L. R., Zhang, X. E., & Jaceldo, K. B. (2001). An adaptation of Brislin’s translation model for cross-cultural research. Nursing Research, 50(5), 300-304. Joreskog, K. G. (1993). Testing structural equation models. In K. A. Bollen & J. S. Long (Eds.), Testing structural equation models (pp. 294-316). Newbury Park, CA: Sage. Jorgensen, S. R., & Johnson, A, C. (1980). Correlates of divorce liberality. Journal of Marriage and The Family, August, 617-626. Kaplan, D. (1995). The impact of BIB spiraling-induced missing data patterns on goodness-of-fit tests in factor analysis. Journal of Educational and Behavioral Statistics, 20, 69-82. Kerlinger, F. N. (1986). Foundations of behavioral research (3rd ed.). New York: Holt, Rinehart and Winston. Kim, E. (1998a). The social reality of Korean American women: Toward crashing with the Confucian ideology. In Y. I., Song & A. Moon (Eds.), Korean American women: From tradition to modern feminism (pp. 23-36). London: Praeger. Kim-Goh, M. (1998). Korean women’s Hwa-Byung: Clinical issues and implications for treatment. In Y. I., Song & A. Moon (Eds.), Korean American women: From tradition to modern feminism (pp. 225-233). London: Praeger. Kim, J. (1998b). A study of Korean domestic violence and social factors. Journal of Korean Social Welfare, 35.

198

Kim, L. I. (1998c). The mental health of Korean American women. In Y. I., Song & A. Moon (Eds.), Korean American women: From tradition to modern feminism (pp. 209-223). London: Praeger. Kim, J. Y. (1998d). Study of relations of SES variables and domestic violence in Korea. Journal of Korean Social Welfare, 35, 133. Kim, J. W. (1997e). Hwa-Byung. Seoul, Korea: Yousung Sinmoon Sa. Kim, J. W., Lee, J. H., Lee, S. G., Eom, H. H., & Whang, W. W. (1996). A clinical study on Hwa-Byung with Hwa-Byung model of oriental medicine. Korean Journal of Stress Research, 5(1), 23-32. King, I. (1981). A theory of nursing: Systems, concepts, process. New York: Wiley. Klein, D. M., & White, J. M. (1996). Family theories: An introduction. London: SAGE Publications. Korea Focus (2002). Third divorce rate and first in tutoring expenses. Korea Focus, 10(6). Korea Institute for Health and Social Affairs (1999). A study of domestic violence and reality in Korea. KIHASA, RIA 09-9904059. Retrieved April 2, 2003, from http://old.welfare.net/library/9901/060.txt Korea National Statistical Office (2000). Retrieved April 8, 2003, from http://www.nso.go.kr/cgibin/SWS_1021.cgi?KorEng=2&A_UNFOLD=1&TableID=MT_ETITLE&TitleID =E3&FPub=4&UserID

199

Langer, E. J., & Rodin, J. (1976). The effects of choice and enhanced personal responsibility for the aged: A field experiment in an institutional setting. Journal of Personality and Social Psychology, 34, 191-198. Lee, J. R. (1999). 한국가정 폭력의 실태와 처벌: 한국에서의 적응과 대책을 위하여 [ The reality of Korean domestic violence and legal regulation: The adaptation and prevention]. Seoul, Korea: Sookmyung Women’s University, College of law. Levinson, D. J. & Huffman, P. E. (1955). Traditional family ideology and its relation to personality. Journal of Personality, 23, 251-278. Lin, K. M. (1983). Hwa-Byung: A Korean culture-bound syndrome? American Journal of Psychiatry. 140, 105-107. Lin, K. M., & Cheung, F. (1999). Mental health issues for Asian Americans. Psychiatric Services, 50(6), 774-780. Little, R. J. A., & Rubin, D. B. (1987). Statistical analysis with missing data. New York: John Wiley. Little, R. J. A., & Rubin, D. B. (2002). Statistical analysis with missing data (2nd ed.). Hoboken, New Jersey: John Wiley & Sons, Inc. Lynn, M. R. (1986). Determination and quantification of content validity. Nursing Research, 35(6), 382-385. MacCallum, R. C., Browne, M. W., & Sugawara, H. M. (1996). Power analysis and determination of sample size for covariance structure modeling. Psychological Methods, 1, 130-149.

200

MacCallum, R. C., & Austin, J. T. (2000). Applications of structural equation modeling in psychological research. Annual Review of Psychology, 51, 201-226. MacDonald, G. W. (1981). Structural exchange and marital interaction. Journal of Marriage and The Family, Nov, 825-839 Mansour, W. (2003). Feminist literary theory. Retrieved April, 3, 2003, from http://www.geocities.com/Athens/Academy/4573/Lectures/feminism.html Markowitz, F.E. (2001). Attitudes and family violence: Linking intergenerational and cultural theories. Journal of Family Violence, 16(2), 205-218. Marsh, H. W., Balla, J. R., & McDonald, R. P. (1988). Goodness-of-fit indexes in confirmatory factor analysis: The effects of sample size. Psychological Bulletin, 103(3), 391-410. Marsh, H. W., & Balla, J. R. (1994). Goodness of fit indices in confirmatory factor analysis: The effect of sample size and model complexity. Quality and Quantity, 28, 185-217. Matsumoto, D. (1996). Culture and psychology. Pacific Grove, CA: Brooks Min, S. K., & Lee, H. Y. (1989). A clinical study of Hwa-Byung. Transcultural Psychiatric Research Review, 26, 140-144. Min, S. K., Park, C. S., & Han, J. O. (1993). Defense mechanism and coping strategies in Hwa-Byung. Journal of Korean Neuropsychiatry Association, 32(4), 506-516. Moon, C. K., Choi, S. H., Jeun, J. M., Lee, S. W., & Hong, Y. S. (1998). Clinical analysis of domestic violence in emergency department. Journal of the Korean Society of Emergency Medicine, 9(2), 311-316.

201

Morash, M., Bui, H., & Santiago, A. (2000). Culture-specific gender ideology and wife abuse in Mexican-descent families. International Review of Victimology, 7, 67-91. Noller, P., & Feeney, J. A. (2002). Understanding marriage: Development in the study of couple interaction. New York: Cambridge University Press. Noller, P & Fitzpatrick, M. A. (1993). Communication in family relationships. New York: Prentice-Hall. Nunnally, J. (1978). Psychometric theory. New York: McGraw Hill. O’Hagan, K. (1999). Culture, cultural identity, and cultural sensitivity in child and family social work. Child and Family Social Work, 4, 269-281. Osmond, M.W., & Thorne, B. (1993). Feminist theories: The social construction of gender in families and society. In. P. G. Boss, W. J. Doherty, R. LaRossa, W. R. Schumm, & S. K. Steinmetz (Eds.), Sourcebook of family theories and methods: A contextual approach (pp. 591-623). New York: Plenum Press. Park, Y. J., Kim, H. S., Kang, H. C., & Kim, J. W. (2001). A survey of Hwa-Byung in middle-age Korean women. Journal of Transcultural Nursing, 12(2), 115-122. Park, Y., Kim, H. S., Schwartz-Barcott, D., & Kim, J. (2002). The conceptual structure of Hwayung in middle-aged Korean women. Health Care for Women International, 23 (4), 389-397. Phillips, L. R., de Hernandez, L., de Ardon, T. (1994). Strategies for achieving cultural equivalence. Res Nurs Health, 17(2), 149-154. Pinkley, R. L., & Northcraft, G. B. (1994). Conflict frames of reference: Implications for dispute resolution and outcomes. Academy of Management Journal, 37, 193-205.

202

Polit, D. F. & Beck, C. T. (2004). Nursing research: Principles and methods (7th ed.). Philadelphia: Lippincott Polit, D. F., Beck, C. T., & Hungler, B. P. (2001). Essentials of nursing research: Methods, appraisal, and utilization (5th ed.). Philadelphia: Lippincott. Raaijmakers, Q. A. (1999). Effectiveness of different missing data treatments in surveys with Likert type data: Introducing the relative mean substitution approach. Educational and Psychological Measurement, 59, 725-748. Raudenbush, S. W., & Sampson, R. (1999). Assessing direct and indirect effects in multilevel designs with latent variables. Sociological Methods & Research, 28(2), 123-153. Raven, B. H. (1992). A power/Interaction model of interpersonal influence: French and Raven thirty years later. Journal of Social Behavior and Personality, 7, 217-244. Reykowski, J., & Smolenska, Z. (1993). Collectivism, individualism and interpretation of social change: Limitations of a simplistic model. Polish Psychological Bulletin, 24, 89-107. Roberts, D. M. (2000). Face validity: Is there a place for this in measurement? SHIKEN: The JALT Testing & Evaluation SIG Newsletter, 4(2), 5. Robinson, J. P., Shaver, P. R., Wrightsman, L. S. (1991). Measures of personality and social psychological attitudes. San Diego, CA: Academic Press. Rodin, J. & Langer, E. J. (1977). Long-term effects of a control-relevant intervention with the institutionalized aged. Journal of Personality and Social Psychology, 35, 897-902.

203

Rogers, S. J. (2003). Wives’ income and marital quality: Are there reciprocal effects? Journal of Marriage and the Family, 61(1), 123-132. Rowland, R. & Klein, R. (1990). Radical feminism: Critique and construct. In S. Gunew (Ed.), Feminist knowledge: Critique and construct. London: Routledge. Rubin, D. B. (1987). Multiple imputation for nonresponse in surveys. New York: John Wiley Sons. Sarantakos, S. (2002). Beyond domestic patriarchy: Marital power in Australia. Nuance, 4, 12-34. Schafer, J. L. (1997). Analysis of incomplete multivariate data. London: Chapman & Hall. Schafer, J. L., & Graham, J. W. (2002). Missing data: Our view of the state of the art. Psychological Methods, 7(2), 147-177. Schumm, W. R., Paff-Bergen, L. A. Hatch, R. C., Obiorah, F. C., Copeland, J. M., Meens, L. D., & Bugaighis, M. A. (1986). Concurrent and discriminant validity of the Kansas marital satisfaction scale. Journal of Marriage and the Family, 48, 381-387. SEMNET archieves (2003 & 2004). Retrieved from http://bama.ua.edu/cgi-bin/wa?A1=ind0312&L=semnet Shapiro, S. S., & Wilk, M. B. (1965). An analysis of variance test for normality: complete samples. Biometrika, 52, 591-611. Shaw, M. E. & Wright, J. M. (1967). Scales for the measurement of attitudes. NY: McGraw Hill. Simpson, P. L. P. (1997). Aristotle: Politics. Chapel Hill: University of North Carolina Press, 1997.

204

Singelis, T. M. (1994). The measurement of independent and interdependent selfconstruals. Personality and Social Psychology Bulletin, 20(5), 580-591. Smallwaters

(2004).

Data

retrieved

Aug

9,

2004

from

http://www.smallwaters.com/amos/faq/faua-missdat.html Snyder, M.G. (1995). Feminist theory and planning theory: Lessons from feminist epistemologies. Berkeley Planning Journal, 10, 91-106. Somers, S. L. (1998). Examining anger in culture-bound syndromes. Psychiatric Times, 15(1), Retrieved April 20, 2003, from http://www.psychiatrictimes.com/p980145.html Song, Y. I. (1998). Life satisfaction of the Korean American Elderly. In Y. I. Song, & A. Moon (Eds.), Korean American women: From tradition to modern feminism (pp. 193-207), London: Praeger. Song, Y. I. & Moon, A. (1998). Korean American women: From tradition to modern feminism. In Y. I. Song & A. Moon (Eds.), The domestic violence against women in Korean immigrant families: cultural, psychological, and socioeconomic perspectives (pp. 161-173). London: Praeger. Spanier, G. B. (1976). Measuring dyadic adjustment: New scales for assessing the quality of marriage and similar dyads. Journal of Marriage and the Family, 38, 15-28. Spector, P. E. (1992). Summated rating scale construction: An introduction. London: Sage Publications.

205

Straus, M. A. (1994). State-to-state difference in social inequality and social bonds in relation to assaults on wives in the United States. Journal of Comparative Family Studies, 25(1), 7-24. Taylor, E. B. (1871). Primitive Culture. London: John Murray. Thibaut, J. W. & Kelley, H. H. (1959). The social psychology of groups. New York: John Wiley. Tonnies, F. (1957). Community and society (C.P. Loomis, trans). East Lansing: Michigan State Press. Triandis, H. C., Betancourt, H., Bond, M., Leung, K., Brenes, A., Georgas, J., Hui, C. H., Marin, G., Setiadi, B., Sinha, J. B. P., Verma, J., Spangenberg, J., & de Montmollin, G. (1986). The measurement of the etic aspects of individualism and collectivism across cultures. Austrian Journal of Psychology, 38(3), 257-267. Triandis, H. C. (1995). Individualism and collectivism. USA: Westview Press. Triandis, H. C., & Gelfand, M. (1998). Converging measurement of horizontal and vertical individualism and collectivism. Journal of Personality and Social Psychology, 74, 118-128. Triandis, H. C. (1988). Cross-cultural studies of personality, attitudes and cognitions. In G. K. Verma & C. Bagley (Eds.), Collectivism and individualism: A reconceptulaization of a basic concept in cross-cultural psychology (pp. 60-95). London: MacMillan.

206

Triandis, H. C. (1999). Cross-cultural psychology. Asian Journal of Social Psychology, 2, 127-143. Vor der Bruegge, E. (1995). Credit with education: A self-financing way to empower women. Convergence, 28(3), 26-35. Vor der Bruegge, E. (1995). Credit with education: A self-financing way to empower women. Convergence, 28(3), 26-35. Walker, L. O. & Avant, K. C. (1983). Strategies for theory construction in nursing. London: A Publishing Division of Prentice-Hall, Inc. Waltz, C. F., Strickland, O. L., & Lenz, E. R. (1986). Reliability and validity of normreferenced

measures.

Measurement

in

nursing

research

(pp.

133-158).

Philadelphia: F. A. Davis Company. Wayman, J. C. (2003). Multiple imputation for missing data: What is it and how can I use it? Paper presented at the meeting of the American Educational Research Association, Chicago, IL. WHO-WPRO (1992). Data provided by WHO-WPRO and Korean National Statistical Office (Social Statistics Survey). Retrieved April 25, 2003, from http://www.blackbox.com.ph/ncd/tables/_RepublicofKorea.html Williams, D. R. (1990). Socioeconomic differentials in health: A review and redirection. Soc Psychol Q, 32, 81-99. Wothke, W. (2000). Longitudinal and multi-group modeling with missing data: Practical issues, applied approaches, and specific examples. In T. D. Little, K. U. Schnabel & J. Baumert (Eds.), Mahwah, Jew Jersey: Lawrence Erlbaum Associates.

207

Yamaguchi, S. (1994). Collectivism among the Japanese: A perspective from the self. In U. Kim, H. C. Triandis, C. Kagitcibasi, & G. Yoon (Eds.), Individualism and collectivism: Theoretical and methodological issues (pp. 175-188). Thousand Oaks, CA: Sage. Zamarripa, M. X., Wampold, B. E., & Gregory, E. (2003). Male gender role conflict, depression and anxiety: Clarification and generalizability to women. Journal of Counseling Psychology, 50, 167-174.

208