Theses Doctoral

Estimating the negative and racialized consequences of the police-centric response to intimate partner violence

Kajeepeta, Sandhya

Intimate partner violence (IPV) is estimated to impact about one in four U.S. women in their lifetime and represents 15% of all violent crime. Total violent crime rates have steadily declined across the country, but rates of IPV victimization have fallen at far slower rates and the incidence of intimate partner homicide has been increasing in recent years. These alarming trends suggest that current strategies for IPV prevention are insufficient and may even be counterproductive. Since the 1970s, the U.S. has developed and maintained a police-centric response to IPV—a response that relies on arrest as its primary tool through practices and policies including mandatory arrest laws and other pro-arrest policing practices.

This police-centric response to IPV persists despite increased recognition of the harms of mass criminalization and incarceration and growing calls for criminal legal reform, and despite a lack of empirical evidence that policing and arrest in fact prevent or reduce IPV.In addition, there are strong theoretical reasons to believe, and emerging empirical evidence to suggest, that there are negative consequences of the police-centric response to IPV that extend beyond subsequent IPV victimization, including the increased risk of all-cause mortality among survivors of IPV and the increased risk of child protective services involvement for families. However, there is very limited quantitative research estimating these negative consequences of IPV policing. There are also likely profound racialized disparities in the consequences of IPV policing because of the ways in which policing, and the criminal legal system more broadly, disproportionately harms Black communities and other communities of color. In this dissertation, I assessed the state- and county-level consequences of the police-centric response to IPV for the health and safety of IPV survivors and estimated the extent to which those consequences have differential impacts across racialized groups.

The first chapter presents a systematic scoping review synthesizing the existing evidence concerning the negative and racialized consequences of the police-centric response to IPV. The review included all empirical studies (quantitative and qualitative) focused on a U.S. population that assessed consequences of IPV policing. A total of 34 articles were included in the review. I found that survivor criminalization was the most studied negative consequence of IPV policing and the existing evidence suggests that IPV policing has increased the risk of survivor arrest. I also found that there have been numerous rigorous studies on the effects of mandatory arrest laws on population-level measures of IPV victimization, including IPV homicide rates. The evidence to date generally suggests there is no association between mandatory arrest rates and population-level IPV victimization rates. The review also identified gaps in the evidence base: specifically, there is a need for research on additional potential consequences of IPV policing such as police violence against survivors, child protective services involvement, and measures of the psychosocial and physical health of survivors.

The second chapter presents a difference-in-differences analysis estimating the effect of state-level changes in warrantless arrest legislation for IPV on the all-cause mortality of IPV survivors from 1980-2019 in the U.S. I analyzed panel data measured at the state-year level and included data for all U.S. states and the District of Columbia. I used women’s all-cause mortality age 20-54 as a proxy for all-cause mortality among IPV survivors. I used quasi-Poisson regression models with a population offset term with robust standard errors to model the association between state-year changes in warrantless arrest legislation and all-cause mortality. I also fit two models with race-specific rates of women’s all-cause mortality (20-54 years) to assess if there are differences by racialized group. Overall, the findings suggested that there is no detectable effect of mandatory arrest laws on women’s mortality (20-54 years) at the population level, however, there appears to be a harmful effect of preferred arrest laws. Despite no documentation of a harmful effect of mandatory arrest laws on women’s mortality (20-54 years), this null effect should be weighed against known, documented harmful effects of mandatory arrest such as its role as a driver of female arrests and arrests of IPV survivors.

The third chapter presents a county-level analysis of the intersection between IPV arrest practice and family surveillance from 2000-2019 in large U.S. counties. I hypothesized that family surveillance and subsequent intervention by child protective services agencies would be a negative consequence of the police-centric response to IPV because of direct coordination between police and child welfare systems and the subsequent increased risk of child protective services involvement that may follow from the arrest of a parent. I evaluated family surveillance and child protective services involvement as a negative consequence because of the harmful and racialized ways in which the U.S. child welfare apparatus targets, punishes, and breaks apart Black families. The criminal legal system and child welfare system are both rooted in structural racism and, thus, their coordination is expected to cause generational harm for Black families. The outcome of interest was the rate of child maltreatment reports that received a response from child welfare agencies and the exposure of interest was the percentage of IPV incidents that were reported to police and resulted in arrest, measured at the county-year level. I used Poisson regression and modeled between- and within-county effects using three types of models: 1) a multilevel county random intercept model, 2) a multilevel county random intercept model with state fixed effects, and 3) a county fixed effects model. I also fit two additional models with race-specific outcome data and conducted an interaction analysis by the percentage of Black residents in the county to assess if there were differences by racialized groups and within different racialized contexts.
The findings presented in the third chapter demonstrated no overall association between the percentage of police-reported IPV incidents that resulted in arrest and child maltreatment report rate at the county level. In addition, there was no county-level evidence to suggest that the association differed for the Black vs. non-Hispanic white child maltreatment report rates. These county-level findings are inconsistent with existing individual-level survey research and qualitative research that provide evidence that IPV policing is associated with child protective services involvement through direct coordination between police and the child welfare system. However, the interaction analysis demonstrated that the percentage of police-reported IPV incidents that resulted in arrest was positively associated with the Black child maltreatment report rate within counties with a low percentage of Black residents. Therefore, the association between IPV policing and family surveillance may vary by individual and contextual factors. This was the first study to examine the intersection between IPV policing and family surveillance at the county level, so additional research is warranted to assess whether the findings can be replicated. Individual family-level data would be useful to further interrogate the relationship between IPV policing and family surveillance.

Geographic Areas


  • thumnail for Kajeepeta_columbia_0054D_17525.pdf Kajeepeta_columbia_0054D_17525.pdf application/pdf 1.54 MB Download File

More About This Work

Academic Units
Thesis Advisors
Prins, Seth Jacob
Ph.D., Columbia University
Published Here
October 12, 2022