Establishments that were gun-prohibiting in Texas in 2021 and 2022 had fewer shootings in and around their premises than establishments that were gun-allowing, after adjusting for confounders, including alcohol sales revenue. This association was significant when we examined bars only but not restaurants. This is perhaps due to the ways in which these establishments are utilized by customers and the nature of interactions. For example, restaurants often close earlier than bars and are more family-friendly. However, the lower number of gun-prohibiting restaurants in our study may have also contributed to the lack of significance, as this reduced our statistical power to detect meaningful differences for this group. As a whole, these findings were consistent across sensitivity analyses and align with established research on the behavioral and spatial dynamics of alcohol-serving venues and violence [33, 34].

The difference in shootings was statistically significant when the buffer size around the bars and restaurants was 50 m but not 100 m. This may be because areas further from the bar or restaurant are less influenced by the gun-free zone status of that venue, and that the 100 m buffers introduce noise to our estimates. A greater proportion of individuals 50 to 100 m from the establishment may not intend to enter, nor be aware of the existence of the policy.

These results are in line with the limited research on the effect of gun-free zones. A study on the effect of gun-free school zones and crimes committed with a firearm in Saint Louis, MO found that gun-free zones in their entirety (1000 feet from the school) were not associated with a difference in crimes committed with a firearm compared to other areas immediately around the gun-free school zone; however, there did seem to be a protective association immediately around the premises of the school [10]. Another study of gun-free zones on college and university campuses, however, did not find an association with rates of violence [35]. Additionally, a nationwide case–control study on the association between gun-free zones and active shootings found a protective association with gun-free zones on the occurrence of active shootings [36]. These findings also concur with previous research indicating that increased firearm availability correlates with a rise in gun-related injuries and deaths [37, 38].

We were unable to crosswalk all of the restaurants and bars across data sources. We spoke to the administrators of both datasets to improve linking, but record keeping was inconsistent. We do not believe that misspelled names, slight variations in business names across datasets, or typos would be associated with either the exposure (gun-free status) or the outcome (shootings), and therefore these omissions, to the extent that they were the cause of failed linkages, should not have biased our results. Nevertheless, to understand how these omissions might affect our findings, we conducted a comparative analysis of characteristics between the bars and restaurants included in our study and those that were not. This comparison showed that in 2022, the establishments that we were able to link on average had higher alcohol sales revenue ($535,992 versus $358,256 for excluded establishments). Moreover, a smaller proportion of these linked establishments prohibited firearms (8.7% as opposed to 14.7% among the excluded), and a slightly higher percentage were classified as bars (22.9% against 20.7%). To address potential biases introduced by these discrepancies, we incorporated these variables (alcohol sales revenue, firearm prohibition status, and establishment type) into our analysis. This approach aimed to mitigate the impact of the observed limitations on our study’s conclusions. Our ability to generalize to all establishments selling alcohol in Texas is hindered by these omissions. However, we have no reason to believe a similar association would not hold among the excluded bars and restaurants.

Another limitation worth noting is that our measure of alcohol consumption was based on the total revenue for alcohol sold. Places that sell more high-end liquor likely sell a lower volume of alcohol per dollar of revenue than places with lower-quality products, resulting in some measurement error. Additionally, while we did obtain time-varying data on the volume of alcohol sold by each establishment, our dataset did not include a similar time-varying measure for red sign designation (indicating gun-free status) for the years 2021 and 2022. The dataset, provided by the TABC, covered both years but recorded the red sign status in only one column. This limitation may have introduced non-differential misclassification of our exposure variable.

Texas was chosen as the location for this cross-sectional study due to the availability of data on gun-prohibited status and alcohol revenue by establishment and due to their consistently applied criteria for a bar or restaurant being designated as gun-prohibited. The results of this study may only be generalizable to Texas. We restricted the analysis to 2021 and 2022, as 2020 consisted of abnormalities presented by the pandemic. For example, any establishment that received more than 50% of their revenue from alcohol in Texas was shut down during the height of the COVID-19 pandemic, and other establishments were limited to 50% occupancy, which would make 2020 data untenable for this study [39]. Complicating matters further, many establishments that sold above this threshold temporarily converted to selling below the 50% mark, blurring the distinction between gun-free and gun-allowing establishments [40]. By 2021, Texas had mostly returned to business as usual [31], as indicated by the similar median alcohol revenue between 2021 ($251,572) and 2022 ($295,305). Future studies should explore the effect of gun-free zones in places that sell alcohol in different time periods and jurisdictions to establish consistency.

Finally, there is the possibility of omitted variable bias. For example, it is also possible that areas with high rates of gun violence are also areas where people consume more alcohol relative to other non-alcoholic goods at bars and restaurants, so the difference we see may not be explained only by the policy, but also this difference in behavior.