Marital violence is a public policy concern that is crucial for the health and human rights of citizens of international society, including Japan. Health care providers are one of the most frequently contacted professionals for those experiencing marital violence, and can play significant roles in the secondary and tertiary prevention of marital violence. Although previous research found that disadvantageous socioeconomic conditions can be barriers to health care, little is known about how socioeconomic conditions interact with marital violence and affect one's care-seeking behaviors. This cross-sectional study exploratorily examined the moderating effect of educational attainment, household income, and employment status on the association between marital violence and health care utilization. The mechanisms underlying the moderating effect of socioeconomic status were further investigated by examining indirect effect of the moderation through psychosocial resources, such as mastery, health literacy, instrumental support, and informational support. The current study also explored the gender differences in the effect of socioeconomic conditions on the association between marital violence and health care utilization. The thesis consists of two sections. The first section is on the development of scales that measure intimate partner violence (IPV) in a community survey, the Japanese Version of Revised Conflict Tactics Scales Short Form (J-CTS2SF). The second section is on the secondary analysis of the Japanese Study of Stratification, Health, Income, and Neighborhood (J-SHINE), in which the J-CTS2SF was used to measure marital violence.

As a part of this study, the Revised Conflict Tactics Scales Short Form (CTS2SF) was translated into Japanese. The CTS2SF is a short form of the Revised Conflict Tactics Scales (CTS2) and measures incidences of IPV with 20 items. The reliability and validity of J-CTS2SF were tested using the data from the two waves of internet survey (N = 393). The reliability was examined by internal consistency and test-retest reliability. Two types of validity were examined: criterion-related validity with the scales on psychological aggression, IPV victimization screening, and psychological distress, and construct validity by factor analysis. Internal consistency of the J-CTS2SF was generally low, and differed by types of subscales and perpetration/victimization status. Compared to the high internal consistency of the original CTS2, low internal consistency of the J-CTS2SF was expected, because the subscales of J-CTS2SF included only two items, each of which measures different levels/types of acts/events. Test-retest reliability examined by Yule's Q was high for the interval of four weeks. Concordance with related scales generally indicated good concurrent validity; expect the negotiation subscale that had a positive association with IPV victimization against the hypothesis. The negotiation subscale did not have significant associations with psychological aggression and psychological distress either. The overall factor structure of J-CTS2SF generally corresponded to that of the full CTS2 theoretically. Because of these findings, the use of J-CTS2SF can be justified especially in a large-scale community survey where the full CTS2 cannot be administered for the limited amount of time. Negotiation items in this short version may need further scrutiny on its accuracy to assess IPV related behaviors and incidences in a Japanese population.

Secondary analysis was conducted on the data from the J-SHINE. The J-SHINE survey collected data from October 2010 to February 2011 in four municipalities in and around the Tokyo metropolitan area. Respondents were community residents aged 25 to 50 years, and had spouses at the time of survey (N = 2,984). Marital violence in the past 12 months was assessed by the J-CTS2SF, and dichotomized into 1) having any perpetration and/or victimization experience of physical and/or psychological violence in marital relationship, and 2) having none of such experience. Another variable, any victimization, was created to explore possible differences in the associations by the presence of victimization. Those who experienced any victimization, regardless of the presence or absence of perpetration, were coded as having any victimization. A dichotomous variable of health care utilization was coded as positive if the respondents received outpatient treatment at clinics or hospitals about their illness or injury in the past 12 months. Socioeconomic conditions included educational attainment (junior high school and high school/two years' college and vocational college/university and higher), equivalized household income (low/average/high), and employment status (employed/not employed). Psychosocial resources examined in this study were mastery, health literacy, instrumental support, and informal support (three-level ordered variables). Control variables were gender, age, number of children, any difficulty in accessing to health care, and survey area.

Bivariate associations of marital violence and socioeconomic conditions with health care utilization were estimated using logistic regression (0.05 level two-sided test). Independent association between marital violence and health care utilization was also examined by multiple logistic regression analysis. Moderating effects of socioeconomic conditions on the association between marital violence and health care utilization were examined by interaction terms added in the multiple logistic regression model. Statistical significance for the interaction terms were set at 0.10 level in this exploratory study. Underlying mechanisms that explained differential effects of socioeconomic conditions on the association between marital violence and health care utilization was further explored by hierarchical logistic regression analysis and mediation analysis. The magnitude and statistical significance of indirect effect through psychosocial resources examined by mediation analysis were estimated using boot-strapping method (0.05 level two-sided test). Gender difference was explored in this series of analysis by interaction terms with gender and by gender subgroup analyses.

The presence of marital violence was associated with increased use of health care independent of socioeconomic conditions among Japanese community residents. The likelihood of using health care in the past 12 months was 1.36 times higher among those experiencing marital violence than those not experiencing marital violence after adjusting the effect of socioeconomic conditions (95% CI = 1.12-1.66, p = 0.002). Educational attainment and household income were independently associated with health care utilization, but the adjustment for these socioeconomic conditions did not weaken the strength of the association between marital violence and healthcare utilization.

The association between marital violence and health care utilization significantly differed by employment status in the total sample. The odds ratio of health care utilization in the presence of marital violence to the absence of marital violence were 1.52 times higher among those who were employed compared to those who were not employed (95% CI = 0.94-2.44, p = 0.086). The association between marital violence and health care utilization did not significantly differ by the level of educational attainment (p = 0.544) and household income (p = 0.661). None of the mastery, health literacy, instrumental support, and informational support explained the differential association by socioeconomic conditions in the total, male, and female samples.

The association between victimization in marital violence and health care utilization differed by gender on the basis of statistical interaction at the 0.10 level. Gender-subgroup analyses found that the presence of marital violence had positive association with increased health care use in men (OR = 1.69, 95% CI = 1.25-2.27, p < 0.001), but not in women (OR = 1.18, 95% CI = 0.90-1.55, p = 0.222). The moderating effect of employment on the association between any victimization and health care utilization differed by gender at the 0.10 level; the likelihood of health care use in the presence to the absence of victimization was 1.67 times higher among men out of employment than among women out of employment (95% CI = 0.97-2.86, P = 0.067).

The internet survey demonstrated that the fair reliability and sound validity of J-CTS2SF. However, the interpretations of results require careful consideration to make conclusion about the association of perpetration and victimization with outcome measures and its gender differences. The J-CTS2SF lacked the assessment on the heterogeneous experience in mutual violence (respondents experienced both perpetration and victimization), which accounted for the majority of positive responses to having marital violence. It was also suggested that more detailed assessment of marital violence should be incorporated with the J-CTS2SF to reflect possible differentials in the patterns and impacts of marital violence between men and women. As a result of secondary analysis on the J-SHINE, it was found that marital violence was associated with increased use of health care independent of the socioeconomic conditions in a Japanese population. Employment status moderated the association, especially in women who had victimization experience. None of mastery, health literacy, instrumental support or informational support accounted for the moderating effect of socioeconomic conditions. Further studies are needed to explore the mechanisms in which a particular social condition encourages or discourages the use of social resources in marital violence and its gender difference.

This study exploratorily examined the socioeconomic conditions that affect the health care utilization in marital violence. The thesis consists of two sections. The first section is on the development of scales that measure intimate partner violence (IPV) in a community survey, the Japanese Version of Revised Conflict Tactics Scales Short Form (J-CTS2SF). The second section is on the secondary analysis of the Japanese Study of Stratification, Health, Income, and Neighborhood (J-SHINE), in which the J-CTS2SF was used to measure marital violence.

1.The Revised Conflict Tactics Scales Short Form was translated into Japanese with the approval of the original author and the publisher. The internal consistency of the J-CTS2SF was generally low because of the small number of items that measure different levels/types of IPV. The test-retest reliability was high. Concordance with related scales generally indicated good concurrent validity; expect the negotiation subscale. Negotiation items in this short version may need further scrutiny on its accuracy to assess IPV related behaviors and incidences in a Japanese population.

2.Secondary analysis was conducted on the data derived from the Japanese Study of Stratification, Health, Income, and Neighborhood (J-SHINE). The likelihood of using health care among those experiencing marital violence was 1.36 higher than that among those without marital violence (p = 0.002).

3.The association between marital violence and health care utilization significantly differed by employment status in the total sample at the 0.01 level. The odds ratio of health care utilization in the presence of marital violence to the absence of marital violence were 1.52 times higher among those who were employed compared to those who were not employed (95% CI = 0.94-2.44, p = 0.086). The association between marital violence and health care utilization did not significantly differ by the level of educational attainment (p = 0.544) and household income (p = 0.661).

4.None of the mastery, health literacy, instrumental support, and informational support explained the differential association by socioeconomic conditions in the total, male, and female samples.

5.The association between victimization in marital violence and health care utilization differed by gender on the basis of statistical interaction at the 0.10 level. Gender-subgroup analyses found that the presence of marital violence had positive association with increased health care use in men (OR = 1.69, 95% CI = 1.25-2.27, p < 0.001), but not in women (OR = 1.18, 95% CI = 0.90-1.55, p = 0.222). The moderating effect of employment on the association between any victimization and health care utilization differed by gender at the 0.10 level; the likelihood of health care use in the presence to the absence of victimization was 1.67 times higher among men out of employment than among women out of employment (95% CI = 0.97-2.86, P = 0.067).

6.The internet survey demonstrated that the fair reliability and sound validity of J-CTS2SF. However, the interpretations of results require careful consideration to make conclusion about the association of perpetration and victimization with outcome measures and its gender differences. The J-CTS2SF lacked the assessment on the heterogeneous experience in mutual violence (respondents experienced both perpetration and victimization), which accounted for the majority of positive responses to having marital violence. It was also suggested that more detailed assessment of marital violence should be incorporated with the J-CTS2SF to reflect possible differentials in the patterns and impacts of marital violence between men and women.

In summary, this study found that the presence of marital violence was associated with the increased use of health care and that employment status moderated the association among community residents in Japan. This study addressed the effect of socioeconomic condition on the care-seeking behaviors in marital violence, which had been scarcely examined in previous studies. Thus, this study has been approved as a PhD dissertation.