Research has identified relationships between interpersonal violence and a range of negative long-term health and other consequences.¹⁻⁵ Previous studies have also estimated the substantial cost of violence,⁶⁻¹² hence providing important information for understanding the cost-effectiveness of violence prevention strategies.¹³⁻¹⁵ Recent cost-of-violence studies have focused on the impact of violence on health using mathematical modeling (combined data from separate studies) of observational outcomes obtained from cross-sectional sample survey data.⁹⁻¹¹ For example, to estimate the long-term cost of intimate partner violence, researchers combined the lifetime cost of medical care for depression with the attributable prevalence of depression among the survivors of intimate partner violence.¹¹ Using observational cross-sectional data in this way is relevant given the data landscape, but it complicates the interpretation of violence as a cause for long-term health-related conditions and costs.

Furthermore, there is limited recent information about the impact of interpersonal violence on individuals’ non-health-related experiences and attainment, including criminal activity, education, employment, family (including marital) status, housing, income, quality of life, or wealth. For example, one frequently cited study compared arrest records of adults up to the age of 40 years with or without substantiated reports of child maltreatment processed by U.S. courts during the years 1967−1971; this study quantified the way in which childhood abuse and neglect were associated with worse educational performance, worse mental health, lower achievement, and higher incidence of arrests in adulthood.¹⁶ New original direct estimates of the socioeconomic impact of interpersonal violence using individual-level longitudinal data and econometric methods to strengthen the causal interpretation of the impact of violence would improve estimates of the economic burden of violence and the cost-effectiveness of the prevention strategies. This study aimed to identify opportunities in the current publicly available representative data sources to measure the impact of interpersonal violence on socioeconomic outcomes in the U.S.

A total of 16 data sources were identified that provide an opportunity to study the longitudinal socioeconomic impact of violence at the individual level (Table 2¹⁹⁻³⁴). Adverse childhood experiences, intimate partner violence, and sexual violence were the most common analyzable violence measures (Table 1). Income, education, and family status were the most common analyzable socioeconomic measures. Self-reported information on violence victimization and socioeconomic status from longitudinal surveys was the source of information in nearly all (14 of 16) of the analyzed data sources; however, 2 data sources were based on administrative records (e.g., substantiated child maltreatment or criminal justice records) (Table 2). The 2 administrative data sources were unique among the analyzed data sources in terms of being linkable at the record level to external longitudinal data sources reporting socioeconomic measures (Table 2). Nearly all (15 of 16) data sources had national geographic coverage, and most had extensive temporal coverage (the longest was from 1970 to present, and the shortest was from 2016 to 2018) (Table 2).

This study generated an inventory of publicly available representative data sources to classify current opportunities for expanding research evidence on the socioeconomic impact of experiencing interpersonal violence. Past research in this area has faced methodologic limitations that have inhibited our understanding of the causal link between experiencing interpersonal violence and its subsequent outcomes. This study’s primary contribution is a comprehensive and concise description of the current data landscape to investigate a complex and understudied question of critical public health importance. Most opportunities identified here to study the level of socioeconomic impact of violence on an individual were based on self-reported longitudinal survey data. A major benefit of this type of data is that it can capture a wider variety of violence exposure than administrative data sources based on, for example, law enforcement involvement. Another major benefit is that surveys typically include richer measures of various forms of violence than administrative records. This is important for studying the implications of experiencing different types of violence across the life course. However, longitudinal surveys are resource intensive and for sensitive topics such as violence exposure and socioeconomic status, survey respondents’ willingness to report and recall bias might reduce the value of the information that is gathered.

This study identified a limited number of administrative data sources that offer applicable analytic opportunities—primarily based on being linkable to other data sources. Recent advancements in data linkage infrastructure, data and network security, and research surrounding methodologies for preventing inadvertent disclosure of personal information have created new opportunities to conduct research using linked administrative records. For example, the FSRDC network facilitates data linkage across disparate data resources through the use of personal identification keys that are generated using probabilistic matching algorithms. These personal identification keys are an anonymized identifier which facilitate data linkage while maintaining individual anonymity.

It should be noted that administrative records of exposure to violence are necessarily a subset of all the violence that occurs, which limits the generalizability of related research findings. Making generalizations to populations based on administrative records carries the risk of perpetuating the biases inherent in the systems that generate these records. Therefore, caution should be taken when considering the external validity of findings from studies using administrative records. However, the major benefit of administrative records as highlighted here is that data linking infrastructure can enable expansive opportunities for longitudinal analysis of victims’ and perpetrators’ subsequent socioeconomic experiences. It should also be noted that surveys share similar limitations in terms of external validity. For example, a survey that aims to be nationally representative may miss certain hard-to-reach groups of the population. Victims of violence are a notable example of this because the consequences of violence exposure could make these individuals less likely to be represented in surveys that are not specifically designed to measure this group, more susceptible to social desirability bias, and more difficult to measure at multiple points in time.

Linked administrative data currently appears to offer the most comprehensive opportunities to examine the long-term impact of violence on individuals’ livelihoods. Despite the limitations of administrative data to comprehensively capture violence exposure, administrative data −linking infrastructure may provide cost-effective research opportunities to better understand elements of the economic burden of violence and improve the targeting of prevention strategies.