Between 2.5 and nearly 5.0 million physical assaults per year are perpetrated against women by their intimate partners (Rand and Rennison 2005).  Card and Dahl (2011) describe two theoretical explanations for why men might commit acts of violence against their intimate partners or other family members: (i) men use violence as a mechanism to control their partners (or children), and (ii) violence arises unintentionally via the interaction of verbal arguments with stress and emotional cues.  This second explanation raises concerns that stress may be an important contributing factor to incidents of domestic violence.

Because of the substantial occupational stress that accompanies military service, the families of servicemembers have been a vulnerable population thought to be in need of protection from such violence.  The Department of Defense (DOD) has taken a strong position against domestic violence since the 1981 implementation of DOD Directive 6400.1, when it set out administrative procedures through which suspected incidents of abuse would be dealt.  Specifically, this directive:

"…required each branch of military service to establish (a) a Family Advocacy Program to prevent child maltreatment and spouse abuse; and (b) a confidential registry to collect and analyze Family Advocacy Program data. Suspected incidents of child maltreatment and/or spouse abuse in military families are referred to Family Advocacy Programs where a case review committee, composed of a multidisciplinary team of designated individuals working at the military installation level, is tasked with the evaluation and determination of abuse and/or neglect and the development and coordination of treatment and disposition recommendations." (Rentz et al., 2006; p. 94)

Prior research that has studied intimate partner violence or child abuse in families of military servicemembers have either (i) focused on military populations and studied the prevalence of domestic violence among these individuals or (ii) compared domestic violence rates of military service members to civilians to assess the “effect” of military-related stress.  Studies comparing military servicemen to civilians have found that spousal abuse is more common among military than civilian families; findings on child mistreatment are more mixed.  While descriptively informative, civilians may be a poor control group for military servicemen because those who select into military service may come from backgrounds or have personality traits that are associated with different propensities for later violence than civilians.

To overcome this problem, we exploit plausibly exogenous variation in deployment assignment among active duty deployed servicemen in the Iraq and Afghanistan wars to estimate the causal effect of combat exposure on the likelihood that servicemen commit domestic violence against their wives, girlfriends, or children.  Using data drawn from the Department of Defense Health and Related Behaviors Survey and the National Longitudinal Study of Adolescent Health, the findings show that those assigned to combat zones are 5 to 10 percentage-points more likely to abuse their wives and children than their counterparts assigned to non-combat zones overseas.  These findings are robust to controls for individual observables—including military occupation and family background—as well as pre-deployment measures of propensity for violence, including own abuse victimization.