This paper asks whether the threat of adverse policy change—specifically the prospect of Democratic control of the White House or salient mass shootings—shifts the behavior of gun owners, measured through background checks for firearm purchases. The authors contribute by merging records on mass-shooting incidents with gun-purchase background-check counts from 2006–2020 and using quasi-experimental timing variation to compare responses across different presidential outcomes and administration periods.
Key data merger and timing variation Regression discontinuity designs Comparative election windows (Obama vs. Trump)
Data and methods
The authors compile location- and timing-specific data on mass shootings (2006–2020) together with counts of firearm-purchase background checks over the same period. They exploit the exogenous timing of closely contested presidential elections and use regression discontinuity designs to estimate changes in background-check activity around election and re-election events, comparing periods under Democratic versus Republican administrations.
Main findings
Using the regression discontinuity framework, the study reports increases in background-check–measured gun purchases following the election and re-election of Barack Obama, while the election of Donald Trump was not associated with similar increases. The authors also report that salient mass shootings were associated with larger increases in gun sales during periods when a Democratic president held office.
- Reported magnitudes for post-election increases are described in the paper as in the range of 70–210 additional background checks per 100,000 following Obama’s election and re-election.
Qualifications and scope
All descriptions above follow the authors’ abstract: results are conditional on the study’s measurement (background checks) and identification strategy (timing-based regression discontinuity). The analysis covers 2006–2020 and contrasts responses under Democratic and Republican presidential administrations as observed in those data.
