Primary Submission Category: Difference in Differences
Impacts of Gun Violence on Health: A Risk Set-Matched Difference in Differences Study
Authors: Eric Cohn, Zirui Song, Jose Zubizarreta,
Presenting Author: Eric Cohn*
Injury due to firearms is a leading cause of morbidity in the United States, though research into its health and economic effects, both on survivors and those in their communities, is limited.
In this paper, we use a large longitudinal data set of health care claims to estimate the effects of nonfatal gun violence on the health and health care spending of those injured and their family members.
We obtain causal estimates from these observational data by profile matching each exposed unit to a set of unexposed units with similar covariate values prior to the exposure, that is, by risk set matching.
A difference-in-differences identification strategy allows for the presence of unobserved, time-invariant confounding, which is likely in this application, and we perform inference using design-based (i.e., randomization-based) procedures.
Additionally, we explore the presence of treatment effect heterogeneity by both prespecified subgroups and those discovered from the data, also within the difference-in-differences and design-based frameworks.
Finally, all inferences are subject to sensitivity analyses to assess the robustness of any statistically significant findings to unobserved confounding.