generalized estimating equation estimates
Translation (even with the statistical word salad definition) it still comes to: There’s actually no way to figure this out, so I’ll make my SWAG look good on paper

John Lott (more guns/less crime) was right, but they couldn’t let that be confirmed, so they kept torturing the statistics hoping for something different, but the best they could come up with was that gun sales don’t have any effect on crime rates.

Legal Firearm Sales at State Level and Rates of Violent Crime, Property Crime, and Homicides

Journal of Surgical Research, Volume 281, January 2023, Pages 143-154



The effects of firearm sales and legislation on crime and violence are intensely debated, with multiple studies yielding differing results. We hypothesized that increased lawful firearm sales would not be associated with the rates of crime and homicide when studied using a robust statistical method.


National and state rates of crime and homicide during 1999-2015 were obtained from the United States Department of Justice and the Centers for Disease Control and Prevention. National Instant Criminal Background Check System background checks were used as a surrogate for lawful firearm sales. A general multiple linear regression model using log event rates was used to assess the effect of firearm sales on crime and homicide rates. Additional modeling was then performed on a state basis using an autoregressive correlation structure with generalized estimating equation estimates for standard errors to adjust for the interdependence of variables year to year within a particular state.


Nationally, all crime rates except the Centers for Disease Control and Prevention–designated firearm homicides decreased as firearm sales increased over the study period. Using a naive national model, increases in firearm sales were associated with significant decreases in multiple crime categories. However, a more robust analysis using generalized estimating equation estimates on state-level data demonstrated increases in firearms sales were not associated with changes in any crime variables examined.


Robust analysis does not identify an association between increased lawful firearm sales and rates of crime or homicide.

Based on this, it is unclear if efforts to limit lawful firearm sales would have any effect on rates of crime, homicide, or injuries from violence committed with firearms.

The effects of firearm ownership are widely debated in both the social science and medical literature. Given the lack of universal firearms registration in the United States, no method exists to precisely quantify legal firearm ownership, although a 2015 study estimated the number to be 265 million weapons.1 Important questions exist regarding any potential effects of legal firearm ownership and purchases on crime. Prior studies concerning the potential risks and benefits of firearms ownership have yielded conflicting results. Some suggest an increased risk of death and injury,2, 3, 4, 5, 6, 7, 8 whereas others suggest a significant protective effect.9, 10, 11, 12 Further studies suggest no significant changes in violent crime,11,13, 14, 15, 16 whereas others suggest a mixed effect.17, 18, 19 These inconsistent results could hinder the creation of informed policy decisions. Because of the lack of specific data on firearm sales, multiple prior studies have suggested the use of firearm background checks as a surrogate marker for legal firearm sales.3,13,20,21 The analysis of this surrogate marker allows for the investigation of trends, which might be affected by firearm purchases.

Given the inconsistency of prior studies, our group investigated whether legal firearms sales, using the surrogate marker of firearm background checks, are associated with changes in publicly reported crime or homicide rates. This is an important consideration, as some are concerned that increased firearm sales might both directly and indirectly lead to increases in crime and homicide. We hypothesized that increased lawful firearm sales would not be associated with rates of crime and homicide. However, as always, it is important to remember that association does not necessarily equate with causation. Robust conclusions in this area may help better inform policy decisions on firearms ownership as well as injury and crime prevention.


Our work was a retrospective analysis of national- and state-level publicly available data on crime and firearm sales during the study period of 1999-2015. The aim was to assess any relationship between legal firearm sales and rates of violent and nonviolent crime on both national and state levels. This study was reviewed by the Institutional Review Board of the Mayo Clinic and was determined not to constitute human subjects research, as it uses only publicly available deidentified data. This


Over the study period, complete yearly data were available for the United States in aggregate, as well as for each individual state and District of Columbia for all UCR, CDC, and NICS datapoints. A total of 12 datapoints encompassing the rates for UCR Total Violent Crime, UCR Murder, UCR Robbery, UCR Rape, UCR Aggravated Assault, UCR Total Property Crime, UCR Burglary, UCR Larceny, UCR Vehicle Theft, CDC Homicide, CDC Firearm Homicide, and NICS Checks. This resulted in 204 national-level and


The Second Amendment of the United States Constitution holds that “A well-regulated Militia, being necessary to the security of a free State, the right of the people to keep and bear Arms, shall not be infringed.” Although the intent of this amendment has been the subject of much historical debate, the recent landmark United States Supreme Court Cases District of Columbia et al. v. Heller and McDonald v. Chicago have held that the right to own firearms is an individual right, which applies to

Author Contributions

M.E.H., M.C.H., K.R.B., and H.J.S. were involved in initial study design. M.E.H., M.C.H., and K.R.B. performed statistical analysis. M.E.H., M.C.H., K.R.B., M.D.Z., and H.J.S. performed statistical interpretation. M.E.H., M.C.H., and K.R.B. wrote the initial article draft. All authors were involved in the critical revision of the article.


None declared.


The authors report no external sources of funding for the research.

Availability of Data

This article has not been published elsewhere and is not currently under consideration for publication elsewhere.

Meeting Presentation

A preliminary version of this article was presented at the 2020 meeting of the Western Trauma Association, February 25, 2020, in Sun Valley, Idaho.