Differences in beliefs about COVID-19 by gun ownership: a cross-sectional survey of Texas adults.

OBJECTIVES: We investigated the association between gun ownership and perceptions about COVID-19 among Texas adults as the pandemic emerged. We considered perceived likelihood that the pandemic would lead to civil unrest, perceived importance of taking precautions to prevent transmission and perceptions that the threat of COVID-19 has been exaggerated.
METHODS: Data were collected from 5 to 12 April 2020, shortly after Texas' stay-at-home declaration. We generated a sample using random digit dial methods for a telephone survey (n=77, response rate=8%) and by randomly selecting adults from an ongoing panel to complete the survey online (n=1120, non-probability sample). We conducted a logistic regression to estimate differences in perceptions by gun ownership. To account for bias associated with use of a non-probability sample, we used Bayesian data integration and ran linear regression models to produce more accurate measures of association.
RESULTS: Among the 60% of Texas adults who reported gun ownership, estimates of past 7-day gun purchases, ammunition purchases and gun carrying were 15% (n=78), 20% (n=100) and 24% (n=130), respectively. We found no evidence of an association between gun ownership with perceived importance of taking precautions to prevent transmission or with perceived likelihood of civil unrest. Results from the logistic regression (OR 1.27, 95% CI 0.99 to 1.63) and the linear regression (β=0.18, 95% CI 0.07 to 0.29) suggest that gun owners may be more likely to believe the threat of COVID-19 was exaggerated.
CONCLUSIONS: Compared with those without guns, gun owners may have been inclined to downplay the threat of COVID-19 early in the pandemic. © Author(s) (or their employer(s)) 2021. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ.

We used a mixed mode approach to sampling Texas adults that included random digit dial sampling and non-probability sampling from an ongoing panel of Texas adults.

Use of non-probability sampling methods allowed us to rapidly gauge how Texas adults’ perceptions about the COVID-19 pandemic differed by gun ownership.

The probability sample had a low response rate (8%), increasing the likelihood of misestimation of associations because of sampling error.

The mixed mode sampling design enabled us to use information from the probability sample in a sensitivity analysis. We conducted Bayesian data integration to generate more accurate estimates of association.

Findings cannot be generalised beyond the state of Texas.

Introduction

In the USA, SARS-CoV-2 first occurred in Washington State in March 2020. COVID-19, the disease caused by SARS-CoV-2, has since emerged as a public health catastrophe with more than a half-million deaths in the USA.1 Given the highly infectious nature of SARS-CoV-2, states began to enact stay-at-home orders to slow the spread in mid-March.2 The public health response quickly became enmeshed in ‘culture wars’, with groups staging protests in opposition to those orders.3 Protesters expressed their view that COVID-19 was not a serious disease, no worse than the influenza, and characterised stay-at-home orders and other measures to prevent transmission (eg, closures of schools and restaurants) as federal over-reach and a threat to freedom and liberty. News and social media have given the impression that there was substantial overlap in groups demonstrating against stay-at-home orders and those supporting individualistic interpretations of the Second Amendment. Some protestors openly carried firearms at demonstrations, further suggesting links between those who favour unrestricted rights to have and carry guns and those who oppose public health measures to control COVID-19.4

In this study, we sought to explore the overlap in perceptions about COVID-19 and gun ownership in Texas, a state where there was strong opposition to stay-at-home orders.5 6 Texas has the second-largest population size of all US states, and it is a conservative-leaning state with a culture supportive of gun ownership and a history of independence from the federal government. The state established a stay-at-home order in April 2020, after many states across the USA and nearly one-third of counties in Texas had already done so.2 5 Shortly thereafter, there were protests at the Texas State Capitol with demands for reopening businesses and schools, and calls for ‘freedom from tyranny’.6 7

Connections between support for individualistic interpretations of the Second Amendment and opposition to the public health response to COVID-19 may have roots in politics and industry practices. The National Rifle Association and the firearm industry more broadly capitalised on fear and uncertainty around COVID-19 to promote guns as necessary during the pandemic8; these efforts may have ramped up beliefs that there would be civil unrest. After several states classified gun dealers as non-essential businesses, the president of the USA ordered the firearm industry be classified as essential at the federal level, forcing states that had closed gun shops to allow them to operate.9 This action bolstered support for the president and others in his political party from gun rights activists in an election year. There were dramatic increases in firearm sales as the COVID-19 pandemic emerged.10 The number of monthly background checks conducted by the Federal Bureau of Investigation is an indicator of gun purchases; the number of monthly background checks in June 2020 was 70% higher than in June 2019.11

In the USA, and in Texas specifically, the COVID-19 pandemic was highly politicised and became intertwined with conservative political ideologies, including ideas around individual gun rights. There are many potential adverse outcomes related to politicising the COVID-19 pandemic; people may downplay the severity of the disease and become less willing to take the recommended public health precautions or support public health mandates, people may buy guns and ammunition and people may be more inclined to carry their guns around. Failure to take recommended precautions could lead to increased spread of SARS-CoV-2, whereas gun acquisitions and increased gun carrying may increase risk for firearm suicide, lethal assaults and unintentional injuries.10 Apparent connections between beliefs about gun rights and about perceptions of COVID-19 raise the possibility that people with guns may be less supportive of public health strategies to respond to the pandemic. To enhance what is known on this topic, we investigate: (1) differences in perceptions about COVID-19 among Texas adults with versus without guns, and (2) recent gun acquisitions, gun carrying and purchases of ammunition among those with guns. We explore perceptions that the COVID-19 pandemic will lead to civil unrest, perceived importance of taking precautions to prevent transmission and perceptions that the threat of COVID-19 has been exaggerated.

Methods

Data for this cross-sectional study come from the Texas Mental Health Survey, which was a state-wide sample of adult residents conducted from 5 to 12 April 2020. Data collection began shortly after the state-wide stay-at-home order went into effect. SARS-CoV-2 infections and COVID-19 deaths in the state nearly doubled over the data collection period; reported infections increased from 7276 to 14 624, and fatalities increased from 140 to 318.1 2 Eligible respondents were Texas residents, fluent in English or Spanish and aged 18 or older. The mixed mode sample included 77 residents who were contacted by telephone using random digit dial (RDD) sampling and 1120 residents who were randomly selected from a panel of adults in the state who opted in to take online surveys through Dynata, a survey research company. Members of the Dynata panel conducted an informed consent process on enrolling in the panel. For respondents in the RDD sample, interviewers conducted informed consent prior to beginning the survey. The online and telephone surveys were conducted in both English and Spanish. The response rate for RDD sample was 8%.12 We restricted analysis to the 1183 respondents who answered the item about household guns.

The main exposure variable was household gun ownership, assessed with the following question: ‘Do you happen to have any guns or revolvers in your home, garage, or car?’ All respondents were asked about plans to acquire guns: ‘Are you or is anyone in your household considering getting a gun for your home in the next 2 weeks?’ Respondents with a household gun were asked about: (a) the number of guns (ie, ‘How many guns are there in your home, garage, or car?’; options were 1, 2, 3+ and ‘not sure’); (b) personal gun ownership (‘Do any of the guns belong to you, personally?’); (c) recent gun acquisitions (‘Were any of the guns in your home purchased or obtained within the last 7 days?’); and (d) recent ammunition purchases (‘In the last 7 days, have you purchased bullets and ammunition?’). Response options for the latter three questions were yes, no and not sure.

Outcome variables included three perceptions about the COVID-19 pandemic, including: likelihood of civil unrest, importance of taking precautions and exaggerations of its danger. Respondents were asked how much they agree or disagree with the following statements: ‘Coronavirus and the COVID-19 pandemic will probably lead to civil unrest’; ‘It is important to take precautions to avoid potentially infecting other people, even for people who don’t have symptoms’; and ‘The threat of coronavirus and COVID-19 has been blown out of proportion’. We created binary versions of these variables to compare those who agree or strongly agree versus those who indicated that they disagree, strongly disagree, or neither agree nor disagree.

Additional study variables included age category (18–25, 26–44 or 45+ years), sex (male, female), presence of children younger than 18 years of age in the home, residence in a rural county, whether the respondent was living with a spouse or partner and race/ethnicity. The race/ethnicity categories were Hispanic/Latino of any race, non-Hispanic White, non-Hispanic Black and ‘all other’, which included respondents who were Asian, American Indian/Alaska Native, Native Hawaiian/Pacific Islander, more than one race or in another race category. To maximise power, we used binary measures of age (ie, <45 years vs ≥45 years) and race/ethnicity (ie, non-Hispanic White vs all other groups) in the final analyses. Rural is specified by matching county of residence to US Department of Agriculture rural-urban continuum codes; rural counties were those with an urban population of less than 2500 as of the 2010 census.13

To adjust for non-response and non-coverage, the data were weighted based on the known population characteristics of the Texas adult population derived from the 2018 Current Population Survey and the 2017 American Community Survey. The sample was balanced to match parameters for sex, age, race/ethnicity and educational attainment using raking ratio estimation, an iterative proportional fitting method. The use of sample weights in analysis ensures that the characteristics of the sample reflect the characteristics of the Texas population. First, we summarised the sample based on demographic factors, perceptions about the COVID-19 pandemic and gun ownership. Then we conducted multiple logistic regression models (with listwise deletion) to assess perceptions in relation to household gun ownership, adjusting for demographic factors. In those models, we used binary measures of perceptions about the COVID-19 pandemic. Analyses were conducted in Stata V.14.2.14

Given the emerging nature of COVID-19, online samples represent an opportunity to rapidly gather information to inform health promotion and policy development. However, the probability of inclusion in the sample is unknown for online surveys. To address this bias, we conducted additional analyses to further assess the representativeness of estimates.15 Specifically, to ensure the smaller variance of our estimates is not biased by the parameters of the large non-probability sample, we ‘borrow’ information from the non-probability sample to produce estimates of the probability sample that have more variance. In a sensitivity analysis, we conducted linear regression modelling using Bayesian data integration with responses from the RDD and online samples.16 17 We retained the five-level response options for each of the three variables measuring perceptions about COVID-19 for these analyses. The Bayesian framework is well suited for integrating multiple data sources of varying quality, such as probability and non-probability samples. We treated the probability-based RDD sample as having higher quality (ie, less selection bias) relative to the online sample, an assumption consistent with the survey literature.18 We constructed informative prior distributions based on data from the online sample to increase the efficiency of the coefficient estimates derived from the smaller RDD sample. We considered four prior specifications that inform the resulting posterior estimates. In this article, we report conjugate difference specification, as it has been shown to have superior properties in simulation studies even in the presence of large selection biases in non-probability samples and in other real-world applications.16 We used a linear regression model to estimate the association between having a household gun (vs not) with perceptions about COVID-19. To ensure comparability, linear regression models controlled for the same set of demographic factors used in the logistic regression models described above. The analysis was conducted in R V.3.6.0.19 Additional details on sensitivity analyses as well as results for the other three prior specifications are available online (see online supplemental material).

Results

There were 1183 respondents in the sample, and 40.8% reported having a household gun (table 1). Nearly one-fifth of the respondents were aged 18–25, 37.5% were aged 26–44 and 44.1% were 45 or older. Thirty per cent had a child in the home, and 46.4% were living with a spouse or partner. The sample was sex balanced, and 45.3% of respondents were non-Hispanic White. Respondents who were White or who were living with a partner were more likely to report a household gun. Thirteen per cent of the respondents indicated plans to acquire a gun in the next 2 weeks, two-thirds of that group indicated there was already a gun in the home (111 out of 164).

Table 1

Respondent and household characteristics, Texas adults (n=1183)

	Full sample(n=1183)	Gun in household
		Yes(40.8%, n=483)	No(59.2%, n=700)	Χ2 (P value)
Age (years)				12.65 (0.055)
18–25	18.3% (217)	15.2% (81)	20.5% (133)
26–44	37.5% (444)	34.8% (186)	39.5% (255)
45+	44.1% (522)	50.1% (268)	40.1% (259)
Race/ethnicity				59.96 (<0.001)
Hispanic/Latino, any race	35.1% (415)	28.3% (152)	39.9% (257)
Black, non-Hispanic	11.3% (134)	07.1% (38)	14.3% (92)
White, non-Hispanic	45.3% (534)	58.5% (314)	36.1% (233)
All other	8.3% (98)	06.1% (33)	09.7% (63)
Sex				4.77 (0.220)
Male	48.6% (575)	52.4% (281)	46.1% (298)
Female	51.1% (604)	47.3% (253)	53.7% (347)
Married or living with a partner				25.81 (<0.001)
Yes	46.4% (549)	55.3% (296)	40.3% (260)
No	53.6% (634)	44.7% (240)	59.7% (386)
Any children <18 years in home				0.19 (0.747)
Yes	30.0% (355)	30.5% (197)	29.3% (157)
No	70.0% (828)	69.5% (450)	70.7% (379)
Live in rural area				0.14 (0.808)
Yes	14.1% (166)	14.5% (78)	13.8% (89)
No	85.9% (1017)	85.5% (458)	86.2% (558)
Plan to purchase a gun				34.72 (<0.001)
Yes	13.9% (164)	21.0% (111)	9.1% (59)
No	85.6% (1011)	79.0% (418)	90.9% (587)

Values are weighted percentage (unweighted n); values may not sum to total due to missing data. Percentages sum to 100% by column, except for the header row (ie, percentage with and without household guns), which sums to 100% by row. Respondents in the ‘all other’ race/ethnicity group were Asian, American Indian/Alaska Native, Native Hawaiian/Pacific Islander, more than one race or were not in any of the race groups listed on the survey.

Among the 483 respondents with household guns, 65% had two or more guns (n=315) and 71.9% (n=348) indicated personal ownership of a gun (table 2). When asked about the past 7 days, 15.4% reported a gun purchase (n=74), 19.6% reported a purchase of ammunition or bullets (n=95) and 25.4% said they carried a gun most or all of the time when away from home (n=123). Twenty-eight per cent of the respondents who reported a past 7-day gun purchase indicated there was just one gun in their household.

Table 2

Prevalence estimates (95% CI) for number of guns, recent purchases and gun carrying among people with household guns, Texas (unweighted n=483)

Characteristic	Prevalence (95% CI)
Two or more household guns.	65.2% (59.2% to 70.6%)
Respondent is personal owner of a household gun.	71.9% (65.9% to 77.2%)
At least one member of the household purchased a gun in the past 7 days.	15.4% (10.8% to 21.5%)
At least one member of the household purchased bullets or ammunition in the past 7 days.	19.6% (14.8% to 25.6%)
Respondent carried a gun when away from home, all or most of the time, in the past 7 days.	25.4% (20.4% to 31.1%)

Prevalence estimates and CIs are weighted.

A large majority (87.9%) agreed that it was important to take precautions to prevent transmission of the virus (table 3); agreement was high among those with and without guns in the home (89.6% vs 86.3%). Forty-two per cent agreed that COVID-19 would lead to civil unrest and 37.6% agreed that the pandemic has been ‘blown out of proportion’. Differences in agreement with these statements did not vary substantially by household gun ownership in bivariate analyses.

Table 3

Percentage of respondents who agree or strongly agree with statements about COVID-19 and SARS-CoV-2, by household gun ownership (n=1183)

	Total(n=1183)	Gun in household
		Yes (n=483)	No (n=700)	χ2 (P value)
‘Coronavirus and the COVID-19 pandemic will probably lead to civil unrest.’	42.2% (499)	43.4% (210)	40.2% (280)	1.20 (0.42)
‘It is important to take precautions to avoid potentially infecting other people, even for people who don’t have symptoms.’	87.9% (1039)	89.6% (433)	86.3% (604)	2.72 (0.28)
‘The threat of coronavirus and COVID-19 has been blown out of proportion.’	37.6% (444)	39.8% (192)	34.3% (240)	3.66 (0.17)

Reported numbers of subjects (n) are unweighted.

Table 4 shows associations between perceptions about COVID-19 and the set of seven binary predictor variables (ie, household gun, White race, male, partner, children in home, 45+ and rural). Estimates in the first column are from the logistic regression models, and estimates in the second column are from the Bayesian linear regression models that integrated the probability and non-probability samples. Both sets of models applied type III sum of squares, that is, every term in the model is tested in light of every other term in the model.

Table 4

Association between perceptions about the COVID-19 pandemic among those with (vs without) guns in the household

	Logistic regressionOR (95% CI)	Linear regressionCoefficient estimates (95% credible interval)
‘Coronavirus and the COVID-19 pandemic will probably lead to civil unrest.’
Household gun (vs none)	1.38 (1.07 to 1.78)	0.28 (−0.14 to 0.71)
White (vs not)	1.07 (0.78 to 1.46)	−0.09 (−0.29 to 0.11)
Male (vs female)	0.80 (0.62 to 1.02)	−0.26 (−0.42 to −0.10)
Live with partner (vs not)	0.89 (0.68 to 1.17)	−0.10 (−0.38 to 0.17)
Children in home (vs not)	1.39 (1.03 to 1.86)	0.22 (0.05 to 0.40)
Age ≥45 (vs <45)	0.43 (0.32 to 0.58)	−0.40 (−0.80 to 0.01)
Rural (vs not)	1.42 (0.95 to 2.11)	0.23 (0.00 to 0.47)
‘It is important to take precautions to avoid potentially infecting other people, even for people who don’t have symptoms.’ *
Household gun (vs none)	0.90 (0.57 to 1.40)	–
White (vs not)	0.81 (0.47 to 1.39)	–
Male (vs female)	0.71 (0.45 to 1.12)	–
Live with partner (vs not)	1.77 (1.06 to 2.94)	–
Children in home (vs not)	1.00 (0.58 to 1.72)	–
Age ≥45 (vs <45)	2.27 (1.27 to 4.04)	–
Rural (vs not)	0.82 (0.45 to 1.50)	–
‘The threat of coronavirus and COVID−19 has been blown out of proportion.’
Household gun (vs none)	1.27 (0.99 to 1.63)	0.18 (0.07 to 0.29)
White (vs not)	0.85 (0.62 to 1.16)	0.10 (−0.02 to 0.22)
Male (vs female)	1.49 (1.15 to 1.92)	0.42 (0.02 to 0.83)
Live with partner (vs not)	1.02 (0.78 to 1.30)	−0.08 (−0.35 to 0.20)
Children in home (vs not)	1.21 (0.89 to 1.64)	0.11 (−0.03 to 0.25)
Age ≥45 (vs <45)	0.65 (0.49 to 0.87)	−0.40 (−0.71 to −0.08)
Rural (vs not)	1.32 (0.91 to 1.91)	0.04 (−0.53 to 0.60)

Bold values indicate statistical significant estimates.

*Insufficient variability in responses did not allow applying a linear regression model.

Analyses do not provide evidence of an association between gun ownership with perceived likelihood of civil unrest or with perceived importance of taking precautions. Although logistic models indicated that those with household guns were 1.38 times more likely to agree that COVID-19 would lead to civil unrest (95% CI 1.07 to 1.78), this finding was not observed in the linear regression model, indicating the possibility of sampling bias. Respondents with children were significantly more likely to agree on the possibility of unrest, whereas those aged 45 or older were significantly less likely to. The logistic regression model did not indicate an association between having a household gun and agreement on the importance of taking precautions to prevent transmission, although respondents aged 45 or older and with a partner in the home were more likely to agree. (Because of limited variation in responses about perceived importance of taking precautions, we were unable to conduct a sensitivity analysis for this variable.)

In the final pair of models, the logistic regression model showed that those with a household gun were 1.27 times more likely to agree that the threat of COVID-19 has been ‘blown out of proportion’, although the interval estimate included the null (95% CI 0.99 to 1.63). Men were more likely to agree, whereas respondents aged 45 or older were less likely to agree. Results from the Bayesian linear regression are consistent with the conclusion that people with household guns were more likely to agree the pandemic has been exaggerated. The parameter estimate of 0.18 was statistically significant, with a 95% credible interval of 0.07–0.29. Thus, the evidence suggests that there is a modest association between perceptions that the pandemic has been overblown and gun ownership, with gun owners being more likely to agree that it has been overblown.

Discussion

In the USA, the COVID-19 pandemic has become intertwined with advocacy for gun rights, leading us to consider that adults with household guns may have different ideas about the pandemic than those without guns. We investigated whether there were differences in perceptions about COVID-19 among Texas adults with versus without guns, and also assessed changes in gun ownership in the early stage of the COVID-19 pandemic in the USA. Data collection took place relatively early in the pandemic, before protests against the coronavirus response at the Texas state capitol and prior to the large public protests for racial justice following the death of George Floyd.

Nearly 90% agreed that it was important to take precautions to prevent transmission of the virus, and less than one-half agreed that the pandemic would lead to civil unrest. Surprisingly, results did not offer evidence of differences in the perceived likelihood of civil unrest or the perceived importance of taking precautions to prevent transmission among those with versus without household guns. It may be that some of the most vocal people out protesting COVID-19 response measures represent a minority of gun owners. Our work suggests that gun owners may be open to public health messaging around ways to prevent the spread of COVID-19. On the other hand, this study offers evidence that Texas adults with household guns may be slightly more likely to downplay the threat of COVID-19 than those without guns. The polarised discourse about COVID-19 may have led to confusion about the severity of COVID-19, suggesting a need for messaging that offers concrete facts about the effects of COVID-19 on individuals and communities in a manner that does stir up political identities.20

Two-fifths of the respondents in our sample of Texas adults had household guns, a figure consistent with previous estimates for the state.21 22 Although there has been a secular trend of declining household gun ownership in the past half century, gun sales in the USA have steadily increased over the past 15 years.11 21 Recent reports demonstrate a substantial spike in gun sales that coincides with the emergence of the COVID-19 pandemic.23 Our findings offer preliminary evidence that the pandemic may have prompted people to buy or consider buying guns and ammunition. Among those with guns, 15% reported a past 7-day gun purchase and nearly one-fifth reported a recent ammunition purchase. Twenty-eight per cent of respondents who had purchased a gun in the prior 7 days had only a single gun in their home, suggesting they may have been first-time gun purchasers. Of those without household guns, 9% indicated plans to buy one within the next week. Increases in firearm ownership, particularly during a stressful time for people across the USA, could pose risks for public health such as intimate partner violence, suicide and access to unsecured guns by children or teens.

Results of this investigation should be considered in the context of limitations with regard to sampling and generalisability. This sample was drawn from Texas and findings should not be assumed to reflect the US population. We used RDD sampling and non-probability sampling, which allowed us to rapidly gather information about this emerging health issue. However, both modes of samples are subject to bias. As with most general population survey efforts these days, we had a low response rate for the probability sample, which increases the likelihood of non-response error. We do not have information about those who did not participate, although the literature suggests that they are likely to be younger than survey respondents. Older adults are more likely to have household guns, but are also more inclined to believe that COVID-19 is a serious illness and that precautions should be taken. It is therefore likely that our results are conservatively biased, that is, biased toward the null. To address the limitations of non-probability sampling, we applied innovative strategies for sensitivity analyses to strengthen conclusions.16 17 Additionally, we adjusted for non-response in both samples by using sample weights based on several sociodemographic characteristics (ie, sex, race/ethnicity, age and educational attainment), a standard procedure for addressing non-response in surveys.

Given the rapid pace of the pandemic and changes in mandates and norms about preventive behaviours, findings provide insight in early stage of COVID-19 in the USA but cannot be generalised beyond that period. Unfortunately, the COVID-19 pandemic continues to devastate the USA and Texas despite the availability of a vaccine.1 2 Because effective control of COVID-19 in the USA depends on the ability to persuade citizens to comply with public health guidelines, knowledge about COVID-related health beliefs can inform more effective communication strategies.24–26 People with guns have been portrayed as opposed to COVID prevention strategies under the guise it is inconsistent with values of freedom and liberty. Consistent with that portrayal, our findings suggest that gun owners in Texas were more likely to think that the threat of COVID-19 has been exaggerated. However, we did not observe differences in the perceived importance of taking precautions to prevent COVID-19 among those with versus without household guns. This is positive news that indicates the potential for effective health communication with gun owners about home safety, gun safety and COVID prevention, even within the USA’s hyperpartisan social environment.