Public Opinion Perceptions, Private Support, and Public Actions of US Adults Regarding Gun Safety Policy

Key Points Question Is correcting misperceptions on gun safety public opinion associated with an increase in private support for and public actions on gun safety in the US? Findings In this 3-part survey study with a total of 1262 respondents, a nationally representative survey found that a majority of gun owners and non–gun owners support key gun safety policies, yet both groups significantly underestimate gun owners’ support for these policies. The 2 additional survey experiments, including one of university students and one with a representative sample of gun owners, found that corrective information about gun owners’ support for gun safety policies was associated with a significant increase in participants’ support for the policies both privately and publicly. Meaning These findings suggest that increasing awareness of the high levels of support for gun safety policies among gun owners may lead individuals to be more supportive of the policies both in private and in public.

that 63% to 91% of gun owners and 83% to 93% of non-gun owners supported key gun safety policies, yet both groups significantly underestimated gun owners' support for these policies by between 12% and 31%. Studies 2 and 3 found that exposure to corrective information was associated with a small increase in support for 2 gun safety policies of between 4% and 15%, both in terms of participants' privately held beliefs and the beliefs they would be willing to share publicly.

Introduction
Gun violence is a public health crisis in the US, with over 255 000 deaths recorded between 2012 and 2018. 1 Policies aimed at curbing gun violence are widely supported by the American public. Polls show that an overwhelming majority of US adults-gun owners and non-gun owners alike-support expanded background checks, mandatory waiting periods, and requirements for safe storage of firearms. [2][3][4][5] Despite this support, many gun owners express apprehension about enacting new gun safety laws, and they tend to favor political candidates who reject such policies. 5,6 Furthermore, supporters of gun safety measures show less willingness than gun rights advocates to express their views, sign petitions, and contact elected officials. 7 Misperceptions of the gun safety opinion climate may help to explain the disconnect between policy support and policy action. Individuals who perceive themselves to be in the minority are less likely to express their views publicly, regardless of whether their perception is accurate. 8 Although individuals often overestimate public support for their positions, that is unlikely in this case. 9 In the US, news media regularly misrepresent public views on gun policy, inaccurately suggesting a deep divide between gun owners and non-gun owners. 2 This misrepresentation could give rise to pluralistic ignorance, whereby individuals whose attitudes and judgments are shared by a majority erroneously believe that they are in the minority, 10,11 which could lead people to overestimate polarization about gun safety policy more generally. 12 Pluralistic ignorance and false polarization can have a profound effect on how people evaluate and voice their private views. Pluralistic ignorance can lead people to adjust their attitudes and behaviors toward a misperceived norm and to reduce their willingness to disclose their private views on the issue in public. 10,11,[13][14][15][16][17] Social conformity also plays an important role in this process, because individuals often display an aversion to expressing beliefs that contradict a perceived norm among in-group members. 18,19 Individuals also self-censor to avoid anticipated negative social consequences associated with expressing a viewpoint that deviates from the in-group's position, such as a loss of respect or lower perceived competence. 11,20 Such self-censorship sometimes takes the form of political nonparticipation. 20 Communicating accurate information about gun owners' views toward gun policies could help correct opinion-climate misperceptions, which could make it more likely that US adults-especially gun owners-will publicly express their support for gun safety policies. 11,15 For gun owners, emphasizing congruence between their private views and the views of others could reduce the threat of pluralistic ignorance resulting from in-group misperceptions. For non-gun owners, it could reduce the risk of false polarization resulting from outgroup misperceptions.
Conveying opinion-climate information is not without risk. If individuals perceive a message to be a persuasive appeal or a form of propaganda, then boomerang effects are more likely. 21 Individuals respond negatively to being told what to think or how to behave, and this can cause their beliefs and attitudes to become more entrenched. In this case, however, the risk appears modest. Learning that gun owners generally support gun safety policies will be congruent with most recipients' private viewpoints, whether they own guns or not. Motivations to push back against such messages should be limited.
The benefits of correcting misperceptions about the gun policy opinion climate may be contingent on other factors. Those who take a position endorsed by the in-group often avoid challenging the status quo if they think change is unlikely or impossible. [22][23][24] Conversely, individuals are more likely to challenge the status quo when they perceive that it can be or has been changed. 25 In the context of this study, we argue that correcting misperceptions of public opinion about gun safety will be uniquely effective among individuals primed to think that change is possible.
A history of federal inaction on gun safety in the wake of deadly mass shootings has left many US adults with the impression that changing gun laws will be slow and difficult, perhaps even impossible. 26 Still, there is some evidence that change is possible. For example, Florida's state legislature passed a law broadening firearm restrictions and implementing a mandatory waiting period following the Marjory Stoneman Douglas High School shooting in 2018. 27 Calling attention to this fact might augment the effect of providing accurate information about public opinion toward gun safety policies.
To test these predictions, we conducted a series of 3 studies that documents the presence of misperceived opinion climates and examines whether correcting these misperceptions, with or without emphasizing system change, influences attitudes and actions on gun safety policy.

Study 1 Methods
The marketing research company Ipsos conducted a 15-minute (median length) omnibus survey with a nationally representative sample of US adults from January 8 to 22, 2019. Questions for the current project were presented in the last of 3 survey blocks, each of which concerned a different topic. No participants skipped the whole block, and 2% or less skipped questions used in these analyses. There is no evidence that participants' demographic characteristics influenced these missing data. Ipsos We assessed participants' knowledge and attitudes about 3 gun safety policies: universal background checks, mandatory waiting periods, and safe storage laws. We described each topic with wording used in prior studies: "Requiring a background check for all gun sales to make sure a purchaser is not legally prohibited from having a gun" 2 ; "Imposing a federal mandatory waiting period on all gun purchases, so that everyone who purchases a gun must wait a certain number of days before taking the gun home" 4 ; and "Requiring by law that a person lock up the guns in their home when not in use to prevent handling by children or teenagers without adult supervision." 2 We asked 2 kinds of questions about each topic. First, we asked, "To what extent do you support or oppose the following policies?" (1 = strongly opposed to 6 = strongly supportive), which we dichotomized (support or oppose) in order to estimate the proportion of the population supporting each policy.
Given the high-profile topic, the absence of a neutral midpoint for these items reduces the risk of participant satisficing, and has been shown to improve validity and reliability. 28 Second, we asked, "What percent of American gun owners do you think support the following policies?" Participants entered their estimate into a textbox that accepted values from 0 to 100. All analyses were conducted from May 25 to October 16, 2020, using Stata/IC, version 14.2 (StataCorp LLC). All P values were 2 sided, and a P value of less than .05 was considered significant. comparable with the national average (mean [SD] = 47.7 [17.5] years). More generally, weighted participant demographic details were comparable with the US population (Table). Importantly, gun owners represented 30.2% of the sample (n = 150.2), mirroring Gallup's recent finding in a nationally representative poll. 29 Study 1 found that 63% to 91% of gun owners and 83% to 93% of non-gun owners supported key gun safety policies, including universal background checks, mandatory waiting periods, and safe storage laws, which is consistent with previous studies (Figure 1). 2 Support for universal background checks among gun owners and non-gun owners was comparable (F 1,492 = 0.63; P = .43; Cohen d = 0.14). Gun owners were less supportive than non-gun owners of mandatory waiting periods (F 1,492 = 9.47; P = .002; Cohen d = 0.36) and of safe storage laws (F 1,492 = 18.44; P < .001; Cohen d = 0.52). We noted that support for background checks and waiting periods was modestly higher (<5%) than in other published reports and speculate that this reflects normal variation in public opinion over time. 2 Despite high levels of support among gun owners for these policies, both gun owners and non-gun owners underestimated support among gun owners.  by gun owners and 12.3 points by non-gun owners. All differences between perceived and actual gun owner support were statistically significant (eTables 1 and 2 in the Supplement). There were no significant differences between gun owners and non-gun owners on any of these perceived support measures.

Study 2 Methods
Study 2, the student sample, was conducted from August 27 to October 17, 2019. Our second study was an experiment designed to test the hypothesis that correcting misperceptions about gun owners' support for gun safety policy would be associated with an increase in participants' policy support and would encourage them to disclose their support publicly. Participants were recruited from a participant pool composed of students enrolled in large introductory classes at the authors' university. A total of 354 participants consented to and then completed the study (a 100% completion rate). We conducted a 2 (corrective information vs no corrective information) × 2 (system changeable vs system not changeable) laboratory experiment in which college students, most of whom did not own guns, were randomly assigned to 1 of 4 conditions. The conditions were Following exposure to the manipulation, participants answered questions identical to those used in Study 1 for 2 gun safety issues. We excluded safe storage laws, because the system changeability manipulation referred to the Florida Public Safety Act, which did not include provisions related to safe gun storage. The survey also included a measure related to the strength of participants' public disclosure of their support for the policies: "If you were to publicly share your views on the following gun policies, how would you describe your position to others?" Response options mirrored those used when soliciting their private support for the policies (1 = strongly opposed to 6 = strongly supportive) (eAppendix 1 in the Supplement).
Last, the study included a pair of behavioral measures. We invited participants to sign gun safety petitions directed at the US president, members of Congress, and the governor of Ohio. We then created a dichotomous variable indicating whether participants chose to sign any of them. After debriefing at the end of the study, the researcher presented participants with a box filled with stickers that read, "I support universal background checks," and asked if they wanted any. We recorded the number of stickers each participant took and constructed a dichotomous measure indicating whether a participant took any stickers.
The study received approval from the Ohio State University institutional review board and was preregistered. 30 Before participation, students were informed that they would be presented with information about a public policy issue and would be asked questions about it. Participants were also invited to contact the principal investigator with questions about the study before consenting, but none did. All participants provided consent via an electronic form before completing the study.  (Table). As this was a convenience sample, we have no reason to believe that it was representative of the larger student body. Furthermore, very few participants owned a gun. Nonetheless, the study allows us to test our prediction among a group that is broadly supportive of gun safety policies.
Study 2 found that exposure to corrective information was associated with a small increase in support for 2 gun safety policies of between 4% and 7%, both in terms of participants' privately held beliefs and the beliefs they would be willing to share publicly. Manipulation checks demonstrated that the opinion-climate corrective information and system changeability manipulations were successful. An independent samples t test indicated that, compared with those not exposed to corrective information, participants exposed to corrective information tended to provide higher (that is, more accurate) estimates of gun owner support for universal background checks (t 352 = 9.49; P < .001; Cohen d = 1.01) and for mandatory waiting periods (t 352 = 12.03; P < .001; Cohen d = 1.28) (eFigure 1 in the Supplement). To assess the system changeability manipulation, participants were asked after exposure to the stimuli "To what extent do you believe that change is possible with gun policy in the United States?" (1 = not at all to 7 = very much so). Participants assigned to the system changeability conditions provided significantly higher scores than those assigned the system not changeable conditions (t 352 = 3.42; P < .001; Cohen d = 0.36) (eFigure 2 in the Supplement).
We used a series of factorial analyses of variance to test our hypotheses about gun attitudes.
Providing corrective information about the opinion climate had a significant association with participants' support for universal background checks (F 1, 351 = 9.64; P = .002; η p 2 = 0.03) and mandatory waiting periods (F 1, 351 = 7.13; P = .01; η p 2 = 0.02). Estimated marginal means were higher for participants exposed to corrective information than for participants not exposed to corrective information (Figure 2A). Corrective information also had a significant association on how supportive participants said they would be when sharing their views publicly about universal background checks (F 1, 351 = 7.96, P = .005, η p 2 = 0.02) and about mandatory waiting periods (F 1, 351 = 10.1, P = .002, η p 2 = 0.03). Estimated marginal means were higher for those exposed to corrective information than for those not exposed to corrective information (Figure 2A). Neither private nor public levels of support were conditioned on system changeability for either issue.

JAMA Network Open | Public Health
Opinion, Support, and Actions of US Adults Regarding Gun Safety Policy We estimated relative risk of the 2 behavioral outcomes using a series of Poisson regression models with robust error variance. Corrective information had no association on whether participants signed any petitions (relative risk = 0.98; 95% CI, 0.89-1.08; P = .72) or took any stickers (relative risk = 1.00; 95% CI, 0.92-1.09; P = .97).

Study 3 Methods
The third study sought to replicate the findings from Study 2 using a nationally representative sample All data collection was conducted online. Like Study 2, this was a 2 (corrective information vs no corrective information) × 2 (system changeable vs system not changeable) experiment in which participants were randomly assigned to 1 of 4 conditions. The study used the same stimuli and selfreport measures, but behavioral measures were different. First, we only invited participants to sign a single gun safety petition directed at Congress. Second, instead of offering stickers to participants, we provided a $20 cash incentive and invited participants to donate any portion of this payment to the Sandy Hook Promise, an organization devoted to gun safety. We provided a short description of the charity, including its mission to lobby for universal background checks (eAppendixes 2 and 3 in the Supplement).
This study received approval from the Ohio State University institutional review board and was preregistered. 32 Panelists were informed that they would be presented with information about a   Study 3 found that exposure to corrective information was associated with a small increase in support for 2 gun safety policies of between 6% and 15%, both in terms of participants' privately held beliefs and the beliefs they would be willing to share publicly. Manipulation checks indicated that all manipulations were successful. An analysis of variance indicated that, compared with those not exposed to corrective information, participants exposed to corrective information provided higher (that is, more accurate) estimates of gun owner support for universal background checks (F 1, 396 = 27.46; P < .001; η p 2 = 0.06) and for mandatory waiting periods (F 1, 396 = 50.72; P < .001; η p 2 = 0.11) (eFigure 3 in the Supplement). Similarly, participants assigned to the system changeable conditions were significantly more likely to describe political change as possible than those assigned the system not changeable conditions (F 1, 398 = 11.36; P < .001; η p 2 = 0.03) (eFigure 4 in the Supplement).
A series of factorial analyses of variance confirms that corrective information had a significant association with participants' support for both universal background checks (F 1, 397 = 6.31; P = .01; η p 2 = 0.02) and mandatory waiting periods (F 1, 396 = 8.44; P = .004; η p 2 = 0.02). Estimated marginal means were higher for participants exposed to corrective information than for participants not exposed to corrective information ( Figure 2B). Corrective information did not have a significant association with how supportive participants said they would be when sharing their views publicly for universal background checks (F 1, 397 = 3.33; P = .07; η p 2 = 0.01), but the association with mandatory waiting periods was significant (F 1, 397 = 5.86; P = .02; η p 2 = 0.02). Again, estimated marginal means are higher for those exposed to corrective information than those not exposed to corrective information ( Figure 2B). None of these relationships were conditioned on system changeability.
Corrective information did not significantly associate with participants' observed behaviors. We estimated relative risk of the 2 behavioral outcomes using a series of Poisson regression models with robust error variance (Figure 3). We found that the system changeability message was associated with the participants' likelihood of signing the petition, but corrective information had no relationship. Neither system changeability nor corrective information has a significant association with donating.

Discussion
The representative survey data from Study 1 suggest that gun owners and non-gun owners overwhelmingly support universal background checks, mandatory waiting periods, and safe storage laws. Furthermore, both groups underestimated the level of support for these policies among the

JAMA Network Open | Public Health
Opinion, Support, and Actions of US Adults Regarding Gun Safety Policy gun-owning public. The 2 follow-up experiments (Studies 2 and 3) found that providing accurate information about the opinion climate was associated with increases in private and public expression of support for gun safety policies among both gun owners and non-gun owners.
We found little evidence, however, that information about the opinion climate was associated with behavior. Individuals' willingness to take stickers expressing their policy support, to sign a petition advocating for gun safety policies, and to make a donation in support of gun safety were all unchanged following exposure to this information.

Limitations
One important limitation of the research is that our associations, though statistically significant, were small in magnitude. However, small associations can still be meaningful when considered in context. 34 Even modest increases in individuals' willingness to voice support for gun safety policies could have important consequences when multiplied across the population. Small associations can also be important at the local level, where policy change is often more politically feasible.
The nonrepresentative student sample used in Study 2 is another limitation, especially in contrast to the representative samples used in Studies 1 and 3. Study 2 participants were generally young, disproportionately women, and rarely owned guns. However, the consistency observed across the results of Studies 2 and 3 is striking, suggesting that the findings are robust in the face of significant demographic differences.
These experiments do not take into consideration participants' prior attitudes about gun safety policies. It is possible that individuals who oppose them would question the accuracy of polling data, undermining message effects. Alternatively, learning that they are in the minority might cause some participants to entrench their position. 21 This might be especially likely for those who hold extreme anti-gun policy attitudes and for those who hold those attitudes with high confidence. 35 Receptivity to corrective information could also be influenced by its source. Corrective information is often more effective when it comes from a credible source, especially one associated with the in-group. 36,37 Finally, drawing on the common in-group identity model, we speculate that accurate information about public opinion might elicit recategorization of gun owners by non-gun owners, thereby reducing intergroup distinctions. 38 This can, in turn, reduce intergroup bias, as group members' cognitive representations change from 2 groups to 1-that is, from an "us" vs "them" to an inclusive "we." 38 We cannot test this idea with these data, but future research should examine whether recategorization helps explain why correcting perceptions of public opinion is effective.