![[ Visit Client Website ]](images/banner.gif)
Schedule:
Sunday, January 19, 2025
Medina, Level 3 (Sheraton Grand Seattle)
* noted as presenting author
Background: While the vast majority of the war-related literature has focused on the perpetration of gender-based violence (GBV) in the public sphere, there are far fewer population-based studies linking armed conflict with GBV perpetrated in the private sphere (Ostby et al., 2019). While the group-level socialization theory provides a cogent framework for understanding why certain armed groups engage in rape as part of armed conflict (Cohen & Green, 2012; Cohen & Nordås, 2015; Wood, 2009, 2018), this theory has yet to be applied to understanding the IPV women experience in times of war. Grounded in group-level socialization theory to examine the effects of armed groups on the GBV continuum, this analysis hypothesized that geographic areas controlled by armed groups known to perpetrate rape (exposed) will have a higher prevalence of sexual IPV and severe physical IPV than non-exposed settings. This analysis drew from data collected in Colombia, where 60 years of armed conflict has had deleterious impacts on the civilian population.
Methods: A semi-panel dataset was created using two waves of Colombia’s Demographic and Health Surveys (DHS) in 2005 and 2016 (n= 34,256), a nationally representative survey of women aged 13 to 49. Two dichotomous variables from the DHS dataset, past-year severe physical IPV and sexual IPV were used as outcome variables in the current analysis. The time variable captured the years before and after an area was exposed to armed groups. Geographic areas exposed to armed groups who have perpetrated rape were coded as yes=1, no=0. Covariates were women’s age, education, job, residence in urban/rural setting, wealth, married under the age of 18, number of children, and exposure to IPV during childhood. Using matching techniques, the sample was balanced to ensure comparability between exposed and non-exposed settings. The estimation of the treatment effect proceeded in two steps. Firstly, the Average Treatment Effect (ATE) was calculated. Employing the matched dataset, the differences-in-differences (DD) analysis then gauged the interaction between time points and geographic area exposed to armed groups perpetrating rape.
Results: The one-one-matching on propensity score was the best approach for matching. After leveraging the optimal match, the treatment effect was 0.012. Furthermore, the DD analysis suggest that both models examining severe physical IPV and sexual IPV exhibited a significant interaction between time and geographic area exposed to armed groups perpetrating GBV, with results statistically significant at p >.000. As hypothesized, geographic areas under the control of armed groups known to perpetrate rape exhibited higher prevalence of severe IPV and sexual IPV over time. Additionally, settings with the presence of armed groups showed an increase in severe IPV compared to settings without such exposure.
Conclusions: These results add to the growing evidence-base indicating that armed conflict exacerbates the violence women experience in the home by specifically highlighting the use of rape by armed actors as a salient factor. Future research should further explore mechanisms underlying the differential impact of armed groups by considering not only the territories in dispute but the territories that are under their control.
Methods: A semi-panel dataset was created using two waves of Colombia’s Demographic and Health Surveys (DHS) in 2005 and 2016 (n= 34,256), a nationally representative survey of women aged 13 to 49. Two dichotomous variables from the DHS dataset, past-year severe physical IPV and sexual IPV were used as outcome variables in the current analysis. The time variable captured the years before and after an area was exposed to armed groups. Geographic areas exposed to armed groups who have perpetrated rape were coded as yes=1, no=0. Covariates were women’s age, education, job, residence in urban/rural setting, wealth, married under the age of 18, number of children, and exposure to IPV during childhood. Using matching techniques, the sample was balanced to ensure comparability between exposed and non-exposed settings. The estimation of the treatment effect proceeded in two steps. Firstly, the Average Treatment Effect (ATE) was calculated. Employing the matched dataset, the differences-in-differences (DD) analysis then gauged the interaction between time points and geographic area exposed to armed groups perpetrating rape.
Results: The one-one-matching on propensity score was the best approach for matching. After leveraging the optimal match, the treatment effect was 0.012. Furthermore, the DD analysis suggest that both models examining severe physical IPV and sexual IPV exhibited a significant interaction between time and geographic area exposed to armed groups perpetrating GBV, with results statistically significant at p >.000. As hypothesized, geographic areas under the control of armed groups known to perpetrate rape exhibited higher prevalence of severe IPV and sexual IPV over time. Additionally, settings with the presence of armed groups showed an increase in severe IPV compared to settings without such exposure.
Conclusions: These results add to the growing evidence-base indicating that armed conflict exacerbates the violence women experience in the home by specifically highlighting the use of rape by armed actors as a salient factor. Future research should further explore mechanisms underlying the differential impact of armed groups by considering not only the territories in dispute but the territories that are under their control.