whether reported by men (2.5% vs 0.7%, respectively) or by women (4.4 % vs 2.0%, respectively). Rates of moderate violence committed by military men and women were higher (10.8% and 13.1%, respectively) than by civilian men and women (9.9% and 10.0%, respectively).
The committee reviewed ten papers related to the associations between deployment and intimate partner violence. Two studies (Jordan et al. 1992; McCarroll et al. 2000) met the criteria for a primary study as described in Chapter 2. The Jordan et al. (1992) study focused on the effects of PTSD on intimate partner violence using data from the NVVRS. Studies by Orcutt et al. (2003) and by Taft et al. (2005) provided additional analyses of the Jordan data. The primary studies for intimate partner violence are summarized in Table 7-1.
McCarroll et al. (2000) conducted a cross-sectional survey in 1990-1994 of a randomly chosen sample of about 15% of the married, active-duty Army men and women at each of 47 Army installations; civilian spouses were not included in the survey. A modified version of the CTS, a widely used self-report instrument, was used to assess aggression in the past year by the military spouse toward the nonmilitary spouse. Responses were categorized as no, moderate, or severe physical aggression. Of the 37,514 surveys received, 26,835 were deemed eligible: 25,520 from men and 1315 from women. Dual military couples were excluded from the survey, and an analysis of responders and nonresponders found no pattern of bias. During the year prior to the survey, 11,540 soldiers had been deployed: 6195 of them for less than 3 months, 3944 for between 3 and 6 months, and 1402 for between 6 and 12 months; 15,294 soldiers had not been deployed. The nature and location of the deployments was not provided. Using multinomial logistic modeling, the researchers found that the predicted probability of moderate and severe aggression increased with increasing length of deployment. For moderate aggression the probability was 0.1762 for nondeployed, 0.1776 for less than 3 month deployments, 0.1793 for 3-6 month deployments, and 0.1850 for 6-12 month deployments; the probabilities for severe aggression were 0.0367, 0.0425, 0.0464, and 0.0495, respectively. The model was controlled for deployment, age, race, sex, rank, and children living with the respondent. A comparison of the predicted probabilities for moderate and severe aggression with no aggression found that the ratios increased with deployment length, although only the difference in ratios for severe aggression were significant (p < 0.05). For severe aggression, ratios of no deployment vs deployment were: 1.1580 (deployed less than 3 months, 95% CI 1.1370-1.1791), 1.2643 (deployed 3-6 months, 95% CI 1.2415-1.2872); and 1.3488 (deployed 6-12 months, 95% CI 1.3245-1.3731). For moderate aggression the ratios were: 1.0079 (95% CI 0.8431-1.1728), 1.0176 (5% CI 0.8518-1.1834), and 1.0499 (95% CI 0.8808-1.2190), respectively. The researchers note that although the increases were significant for severe aggression the changes represented small absolute values of probability. Although there are limitations inherent in a cross-sectional design, this study demonstrates a clear association between deployment and increased spousal aggression.
Intimate partner violence was studied by three research groups using information from the NVVRS discussed above. Based on responses to the NSVG, Jordan et al. (1992) selected all households of theater veterans who appeared to have PTSD and the households of a subset of veterans who did not appear to have PTSD to determine the effects of PTSD on family adjustment and marital conflict, including intimate partner violence. Of those without PTSD, the researchers oversampled for veterans who indicated they had had high combat exposure or high