Cross-cultural measurement equivalence of the Japanese version of Revised Conflict Tactics Scales Short Form among Japanese men and women

Authors

  • Maki Umeda PhD, MPH,

    Corresponding author
    1. Department of Mental Health, Graduate School of Medicine, the University of Tokyo, Tokyo, Japan
    • Correspondence: Maki Umeda, PhD, MPH, Department of Mental Health, Graduate School of Medicine, the University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan. Email: makiumeda-tky@umin.ac.jp

    Search for more papers by this author
  • Norito Kawakami MD, DMSc

    1. Department of Mental Health, Graduate School of Medicine, the University of Tokyo, Tokyo, Japan
    Search for more papers by this author

Abstract

Aim

The Revised Conflict Tactics Scales Short Form (CTS2SF) is an instrument used to measure intimate partner violence (IPV) perpetration and victimization over the past 12 months.

Methods

The CTS2SF was translated into Japanese, and the reliability (internal consistency and 4-week test–retest reliability) and the concurrent and factor-based validity were examined using two waves of Internet surveys over an interval of 4 weeks. Participants of the survey were 393 Japanese men and women who were registrants of an Internet survey company.

Results

Cronbach's α was greater than 0.5 for most scales, while it was low (α = 0.18) for sexual coercion by partner. The test–retest reliability of the binary variable for the presence or absence of IPV was high (Yule's Q, 0.79–1.00), and moderate between the scores (Spearman's rank correlation, 0.38:0.70). Concordance with the Buss–Perry Aggression Questionnaire, Violence Against Women Screen, and Kessler 6 generally indicated good concurrent validity. The results of the exploratory factor analysis confirmed the three-factor structure of the Japanese version of the CTS2SF.

Conclusion

Although the internal consistency reliability was limited for some sub-scales, its moderate internal consistency and test–retest reliability and good factor-based validity highlighted the benefit of using the Japanese version of the CTS2SF in a large-scale community survey where a shorter scale is required to assess IPV.

The prevalence of intimate partner violence (IPV) varies across countries: the lifetime prevalence of IPV against women ranged from 15% to 71% in a multinational study across 10 countries,[1] which implies the complex nature of causes, where individual and interpersonal properties interact with social norms and structures to increase the risk of IPV.[2, 3] In Japan, 32.9% of women and 18.3% of men experienced physical, psychological, and/or sexual violence repeatedly by their spouses or ex-spouses in their lifetime.[4] IPV can be the cause and consequences of mental health problems, being associated with depressive disorders, anxiety disorders, post-traumatic stress disorder, substance abuse, suicide ideation, and child maltreatment among victims and perpetrators.[5-7] In Japan, the Act on the Prevention of Spousal Violence and the Protection of Victims was enforced in 2001, and national and local governments are obligated to implement the policies that enhance the prevention of IPV and the support for victims. However, in Japan, an epidemiological community survey on the mental health impact of IPV is still limited in number, and existing studies were conducted only among women with the focus on female victimization.[8, 9] To promote the development and evaluation of IPV policies in Japan, a standardized measurement of IPV perpetration and victimization needs to be administered both among men and women.

The Revised Conflict Tactics Scales Short Form (CTS2SF) is a 20-item scale that measures IPV victimization and perpetration among men and women during the past 12 months.[10] The CTS2SF is a short version of the Revised Conflict Tactics Scales (CTS2), which is a dominant research instrument validated in over 20 languages,[11-15] including the Japanese.[16] The CTS2SF has five sub-scales, each of which consists of two items that corresponds to a different type/level of act: negotiation (cognitive/emotional), psychological aggression (minor/severe), physical assault (minor/severe), injury (minor/severe) and sexual coercion (minor/severe). Based on the respondent's self-report, 10 items measure the respondent's act and perpetration of violence (self/perpetration), while the other 10 items measure their partner's act and the respondent's victimization of violence (partner/victimization).

The CTS2SF has the following advantages over other research scales for use in community surveys. First, it is a standardized scale that can reduce the burden of responding to the 78 items on the full CTS2, many of which are related to issues that may not have been experienced by the respondents participating in the community surveys.[10] Second, the CTS2SF can reduce refusal and negative reactions from respondents, due to its conceptual framework, which defines IPV as couple conflicts, rather than abuse.[15] Third, the CTS2SF can assess both perpetration and victimization in male-to-female and female-to-male violence, while other scales focus on the victimization of women[17-19] or are only validated among women.[20, 21]

The aim of the current study was to develop the Japanese version of CTS2SF (J-CTS2SF) and examine its reliability and validity. The original English version was translated by the authors. Then, an Internet survey was conducted to examine the reliability and validity of the J-CTS2SF among Japanese men and women. To the best of our knowledge, this is the first study that examined the test–retest reliability of the CTS2SF in any language. In addition, this study would contribute to cross-cultural research on IPV by examining the cultural adaptability of the scale developed in the USA.

Methods

Translation of the CTS2SF

The authors translated the original English version of the CTS2SF into Japanese. The Japanese version was back-translated by an independent translator, and sent to the original author to assess the equivalence with the original English scale in terms of literal and general meaning. Accordingly, two items of the Japanese version were revised.

Internet survey on the reliability and validity of the J-CTS2SF

Participants and data collection

Two waves of Internet surveys were conducted in October and in November 2011. Invitations were sent via emails to the registrants of an Internet company who were over the age of 18 years. The survey was kept open until 100 respondents were obtained for each of the marital status categories by sex (married and not-married by male and female). As a result, 412 respondents were included in the first survey (103 respondents for each category of each sex). Four weeks after the first survey, the second survey was conducted and data were obtained from 393 respondents (retention rate = 95.4%). Analysis was conducted on the data of those who responded to the first and the second surveys (n = 393). The Research Ethics Committee of the University of Tokyo Graduate School of Medicine approved the research procedure for this study (No. 3403).

Measurements

J-CTS2SF

The full-set of the J-CTS2SF was administered, and scored on an 8-point scale (1 = once, 2 = twice, 3 = 3–5 times, 4 = 6–10 times, 5 = 11–20 times, 6 = more than 20 times, 7 = not in the past year, but it did happen before, and 8 = never happened).[10] The acts conducted by the respondents and resulting injury of the partner were defined as ‘self /perpetration’ and the acts conducted by their partners and resulting injury of the respondents were defined as ‘partner/victimization’. In the second survey, the J-CTS2SF was assessed by looking back at the past 12 months, from the time of the first survey, to make the duration of assessment comparative between the two surveys.

Buss–Perry Aggression Questionnaire

The Japanese version of the Buss–Perry Aggression Questionnaire (BAQ) measures personal traits of aggression with 24 items on a five-point Likert scale (1 = strongly disagree to 5 = strongly agree).[22, 23] The total scores of the sub-scales were added together to yield a general total score from 24 to 120. A higher score means a higher degree of aggression.

Violence Against Women screen

The Violence Against Women (VAW) is a Japanese screening instrument to detect IPV victimization among women during the prenatal period.[24] While the VAW was validated in a female prenatal population, its items were assumed to be applicable to the general female and male population without the items associated with pregnancy-related experiences or gender-laden incidences as in other validated screening instruments. The VAW consists of seven items on physical violence (two items), psychological violence (six items), and sexual violence (one item) with three response options on the frequency of incidents (1 = never, 2 = sometimes, and 3 = often). Each score was added together to yield a total score and a total score of 9 or higher was defined as indicating IPV victimization. We measured the VAW only at the second wave. To examine the association with J-CTS2SF at the same second wave of the survey, respondents were asked to look back at the past 12 months from the time of the first survey.

Kessler 6

Kessler 6 (K6) assesses psychological distress that is not specific to certain types of mental disorders.[25] It asks about the frequency of listed psychological states in the past 30 days with six items on a five-point scale (0 = none of the time, 1 = a little of the time, 2 = some of the time, 3 = most of the time, and 4 = all of the time). The total score of 13 and above is defined as having a serious mental disorder.[26, 27] K6 was validated in a Japanese population and showed good concordance with the diagnosis by Composite International Diagnostic Interview on mood and anxiety disorders.[28]

Demographic variables

Demographic variables included in the analysis were sex, marital status (married and not married), age (18–29 years old, 30–39 years old, 40–49 years old, and 50 years old or older), educational attainment (high school, college, and university), employment status (employed and not employed), and having one or more children (none, and one or more). Household income was measured by 10 income categories, and divided into quartile (low, low–average, average–high, and high).

Statistical analysis

Reliability was assessed on the basis of internal consistency and test–retest reliability. Standardized Cronbach's α coefficients were computed using the data from the first survey for each type/level of the subscale by self/perpetration and partner/victimization status. Test–retest reliability was assessed by Spearman's correlation coefficients for each type/level of the subscale by self/perpetration and partner/victimization status. The J-CTS2SF scores were re-ordered as a rank (0 = never happened, 1 = not in the past year, but it did happen before, 2 = once, 3 = twice, 4 = 3–5 times, 5 = 6–10 times, 6 = 11–20 times, 7 = more than 20 times). Yule's Q was also calculated for a bivariate variable of the presence/absence of an IPV-related act (0 = no incidence, and 1 = one or more incidences), considering the frequent use of this binary definition in community surveys. Intra Class Correlation Coefficients were not calculated because of the highly skewed distribution of items (Appendix S1).

The criterion-related validity was assessed by the bivariate association of each J-CTS2SF subscale with aggression (BAQ) and IPV victimization (VAW). The negotiation items related to self were expected to have an inverse association with aggression, as were those related to the partner with IPV victimization.[11, 14] The self/perpetration items of psychological, physical, sexual violence and injury were hypothesized to have positive associations with aggression, and the partner/victimization items were thought to have positive associations with IPV victimization. Logistic regression analyses were conducted, having dichotomous variables of CTS2SF subscales (0 = never, 1 = one or more instances) in either type/level as the dependent variables. The BAQ, for which no cut-off point was available to define the level of aggression, was entered in the logistic models as a continuous variable. VAW was dichotomized at the cut-off points of 9 as a criterion of victimization.[24]

Criterion-related validity was also tested on the association between the J-CTS2SF at the first survey and psychological distress measured by K6 at the second survey. Dichotomous variables on ‘IPV perpetration and/or victimization’ were created from the J-CTS2SF items. The responses to each item were dichotomized into ‘0’ and ‘once or more times’, and having experienced any instances in the subscale was coded as having experienced IPV (0 = neither by respondents nor their partners and 1 = one or more instances by respondents and/or their partners). K6 was dichotomized at the cut-off point of 13 for having psychological distress at the level of severe mental disorders.[26, 27] Greater psychological distress at the second survey was hypothesized to have positive associations with psychological, physical, and sexual violence and injury at the first survey, while negotiation was assumed to have a negative association. Generalized estimating equations models were applied to the longitudinal data from the first and the second surveys as to control psychological distress at the first survey.

All models were adjusted for sex, age, marital status, having one or more children, household income, and employment status. The construct validity was examined by exploratory factor analysis based on the maximum likelihood method with promax rotation.

Results

Reliability of the J-CTS2SF

Standardized α was generally higher for items that measured the respondents' acts (0.52–0.85) than for those of partners (0.18–0.83) (Table 1). The items on physical assault had the highest internal consistency (0.85 for respondents; 0.83 for partners), and those on sexual coercion had the lowest (0.52 for respondents; 0.18 for partners). Test–retest reliability assessed by Spearman's correlation coefficients was low, especially among sexual coercion items (0.38–0.68) and injury items (0.48–0.65), while Yule's Q coefficients were considerably high (0.99–0.78) (Table 2).

Table 1. Standardized Cronbach's α coefficients for the subscales of the J-CTS2SF calculated from the first wave of Internet surveys (two items for each subscale by self/perpetration and partner/victimization)
 Self/perpetrationPartner/victimization
  1. J-CTS2SF, Japanese version of the Revised Conflict Tactics Scales Short Form.
Negotiation0.670.69
Psychological0.530.47
Physical0.850.83
Injury0.850.53
Sexual0.520.18
Table 2. Test–retest reliability of the J-CTS2SF for the Internet survey conducted over the interval of 4 weeks
  Spearman correlation coefficient (SE)Yule's Qa (SE)
Self/perpetrationPartner/victimizationSelf/perpetrationPartner/victimization
  1. aConcordance of binary variables for the J-CTS2SF between the two surveys (with or without an intimate partner violence incident).
  2. J-CTS2SF, Japanese version of the Revised Conflict Tactics Scales Short Form.
NegotiationCognitive0.56 (0.04)0.52 (0.04)0.83 (0.04)0.81 (0.04)
Emotional0.57 (0.04)0.51 (0.04)0.83 (0.04)0.79 (0.04)
PsychologicalMinor0.68 (0.04)0.70 (0.04)0.90 (0.03)0.89 (0.03)
Severe0.55 (0.09)0.52 (0.08)0.99 (0.01)0.93 (0.04)
PhysicalMinor0.66 (0.06)0.69 (0.05)0.99 (0.01)0.96 (0.02)
Severe0.67 (0.08)0.53 (0.09)0.99 (0.01)0.97 (0.02)
InjuryMinor0.48 (0.08)0.65 (0.06)0.98 (0.01)0.98 (0.01)
Severe0.63 (0.14)0.61 (0.12)1.00 (0.00)0.97 (0.03)
SexualMinor0.38 (0.12)0.63 (0.07)0.98 (0.02)0.96 (0.02)
Severe0.47 (0.20)0.42 (0.15)1.00 (0.00)0.98 (0.02)

Validity of the J-CTS2SF

Higher BAQ was positively associated with psychological aggression (odds ratio [OR] = 1.06; P < 0.001), physical assault (OR, 1.07; P < 0.001), and injury perpetration (OR, 1.07; P = 0.001) (Table 3). IPV victimization assessed was positively associated with psychological aggression (OR, 21.77; P < 0.001), physical assault (OR, 14.46; P < 0.001), sexual coercion (OR, 24.15; P = 0.002), and negotiation (OR, 2.26; P < 0.001) (Table 3). The iteration of the maximum likelihood estimation did not converge for the association of injury with the VAW because of the complete separation, where the VAW was always negative in the absence of injury victimization. Severe psychological distress measured by K6 at the second survey was positively associated with psychological aggression (OR, 2.41), physical assault (OR, 3.09), injury (OR, 2.88), and sexual coercion (OR, 3.46) at the first survey (Table 3).

Table 3. Association between any incidences of conflict tactics in the past 12 months by BAQ, VAW, and K6a using the data from Internet surveys
 BAQb and self (perpetration)VAW 9+c and partner (victimization)K6 13+d and either self or partner (victimization/perpetration)
OR95% CIP valueOR95% CIP valueOR95% CIP value
  1. aAll models were adjusted for sex, age, marital status, having any children, equivalized household income, and employment status.
  2. bEntered in the model as a continuous variable based on the first-wave data.
  3. cEntered in the model as a dichotomous variable (VAW score 8 or lower as reference) based on the second-wave data.
  4. dEntered in the model as a dichotomous variable (K6 score 12 or lower as reference) based on the first- and second-wave data.
  5. BAQ, Buss–Perry Aggression Questionnaire; K6, Kessler 6; VAW, Violence against Woman Screen.
Negotiation1.000.99–1.020.7822.261.46–3.50<0.0011.400.91–2.140.125
Psychological1.061.04–1.09<0.00121.779.72–48.72<0.0012.411.56–3.71<0.001
Physical1.071.03–1.10<0.00114.463.32–63.01<0.0013.091.77–5.39<0.001
Injury1.071.03–1.120.001NANA–NANA2.881.41–5.910.004
Sexual1.010.91–1.110.92024.153.10–187.930.0023.461.60–7.490.002

Construct validity was examined by exploratory factor analysis. A scree plot suggested three factors for the J-CTS2SF, accounting for 85% of the variation among the items. The factor pattern after promax rotation is presented in Table 4. Factor 1 concerned physical and psychological aggression and injury, Factor 2 concerned sexual coercion and severe injury of partners, and Factor 3 concerned negotiation tactics.

Table 4. Rotated factor pattern of the J-CTS2SF based on the data from the first wave of Internet surveysThumbnail image of

Discussion

Internal consistency of the J-CTS2SF was relatively low, and differed by types of subscales and perpetration/victimization status. Test–retest reliability was high when the binary variables for the presence or absence of IPV-related acts were measured, but low when it measured the frequency. Concordance with the related scales generally indicated good concurrent validity, and the construct of the J-CTS2SF was generally consistent with the original CTS2.

Lower internal consistency was expected because the subscales of the J-CTS2SF included only two items, each of which measures different levels/types of acts that were not necessarily correlated.[10] The test–retest reliability of the frequency scales was low, possibly because the frequency of incidence was most likely subject to recall bias.[15] The binary variable for the presence or absence of IPV would be more precise if it were assessed based on the recall over a period of 4 weeks.

The results of logistic regression showed that the increase in the BAQ total score by one unit increased the odds of psychological, physical, and sexual violence by about 6–7%. On the other hand, the negotiation and sexual coercion items were not significantly associated with BAQ. This result indicated that the negotiation and sexual coercion measured by the J-CTS2SF may not be strongly related to one's personal traits of aggression among the Japanese. The IPV victimization was positively associated with negotiation by partners and victimization in psychological, physical, and sexual violence against our hypothesis.

Psychological distress had positive associations with psychological, physical, and sexual violence and injury regardless of one's victimization/perpetration status. Negotiation did not have a significant negative association with psychological distress, although it had positive association with IPV victimization. In the current Japanese sample, the ways negotiation tactics were used and its consequences could have been diverse, either preventing, resolving or aggravating marital conflict.[29]

The exploratory factor analysis found a smaller number of factors in the J-CTS2SF than in the original CTS2, even though the CTS2SF was constructed based on the factor structure of the CTS2.[10] The overall factor structure found in this study was considered to be in agreement with that of the full CTS2 theoretically. Factor 1 included items on aggression and its consequence, Factor 2 included sexual coercion, and Factor 3 included negotiation tactics. The severe injury of partners was categorized in Factor 2 together with sexual coercion. These results indicate that the severe injury of partners may be highly correlated with sexual coercion in the current Japanese sample. Another possibility is that sexual coercion may have been largely underreported because of the justification of sexual violence and stigmatization of the victims in a Japanese society.[30, 31] The factor pattern we found may be a reflection of reporting patterns, where those who reported their partners' severe injury were more likely to report socially unfavorable behaviors, including sexual coercion.

This study has the following limitations. First, the survey was conducted among the registrants of an Internet survey company. These participants were users of the Internet, and they tended to have a higher level of educational attainment and lower household income compared to the general population used in other studies (Appendix S2).[32] As it is unknown how these traits affected the results, generalizing the results to the larger Japanese population cannot be done without further investigation. Second, we examined test–retest reliability for an interval of 4 weeks. How precisely the scale would measure IPV for an interval of 12 months has yet to be examined.

The J-CTS2SF was found to have fair reliability and good validity in Japanese men and women. Although the reliability was limited due to the smaller number of items, its good validity indicates that the J-CTS2SF has the capacity to accurately capture IPV, especially when the scores were dichotomized as the presence or absence of IPV. The use of J-CTS2SF enables a large-scale community survey to assess IPV with a limited number of internationally standardized items. Negotiation items in this short version may need further scrutiny to see whether they would capture IPV-related behaviors and incidences accurately in a Japanese population.

Acknowledgments

This study was supported by a Grant-in-Aid for Scientific Research (A) 2009−2013 (No. 20240062) from the Ministry of Education, Culture, Sports, Science and Technology, Japan. There is no conflict of interest. Material from the CTS2 ‘short form’ © 2003 by Western Psychological Services. Translated and adapted by M. Umeda, the University of Tokyo, for specific, limited research use under license of the publisher, WPS, 625 Alaska Avenue, Torrance, California 90503, U.S.A. (rights@wpspublish.com). No additional reproduction, in whole or in part, by any medium or for any purpose, may be made without prior, written authorization of WPS. All rights reserved.

Ancillary