Campus Carry Attitude, Intention, Behavior, and Impact: A Multilevel Meta-analysis

Abstract Numerous studies have examined the correlates and predictors of campus carry among various campus communities. There were, however, too many risk factors included, raw effect sizes were too small, and the differences in effect sizes were negligible, making comparisons prohibitively difficult or even worthless. To address this issue, a multilevel meta-analysis was conducted to investigate thirteen domains of risk factors as potential predictors of four campus carry-related outcomes including attitudes, intentions, behaviors, and impacts. Thirty-seven studies were included, with 703 effect sizes in a unique population of 55,115 campus members. A series of three-level meta-analyses revealed that different risk domains contribute to different campus carry outcomes. Generally, social life- and campus life-related risk factors were strongly related to campus carry intentions and behaviors, while the political ideology domain influenced both attitudes and behaviors. Statistically significant effects were observed in multiple moderator analyses for the number of universities from which data were collected and sample affiliations, however, these effects were relatively small. An in-depth review of the literature, synthesis of empirical evidence, and discussion of implications provided a clear path for future research and policymaking.


Introduction
As reported by the FBI Uniform Crime Report for 2020, the national murder rate was 6.5 murders per 100,000 persons.The murder rate on college campuses was 0.2 per 100,000 persons (U.S.Department of Education, 2023).Overall, college campuses are much safer than the rest of the nation (Gius, 2019;Hassett & Kim, 2020).In Silva and Capellan (2019) examining media coverage of mass public shootings, findings indicated that school shootings (i.e.armed attacks at educational institutions) in the United States decreased substantially between 1966 and 2016, but the coverage of school shootings and the number of articles/words written about this particular type of shooting increased during the same time period.Shootings on college campuses are rare but high impact events attracting widespread media attention and causing space-based fear and campus safety concerns (Arrigo & Acheson, 2016;Burton et al., 2021;Hayes et al., 2021;Tanner & Kelemen, 2023).
After several school shootings occurred and the media's extensive coverage of such events, there has been an intense debate about how to make campuses safer.Among the many campus security strategies proposed, the reform of campus bans on concealed carry practices has proved to be one of the most controversial (Arrigo & Acheson, 2016;Hassett & Kim, 2020, 2021;Nodeland & Saber, 2019).Commonly known as campus carry-allowing people to carry a concealed firearm on a university campus-has long been deemed illegal for everyone but law enforcement (Hassett & Kim, 2020, 2021;Hayes et al., 2021).
The decisions on campus carry have been left to individual states resulting in considerable variation in policies across the country (Arrigo & Acheson, 2016).States can generally be categorized into those that permit campus carry, those that prohibit people besides law enforcement from carrying a firearm on campus, and those that allow individual institutions to make the decision (Hassett et al., 2020).In 2020, 10 states had provisions allowing the carrying of a concealed firearm on a university campus, one state (Tennessee) only permitted faculty members with a license to carry weapons on campus, 16 states banned carrying a concealed weapon on a college campus, while 23 states allowed individual institutions to enact their own policies on the matter (Burnett, 2020; National Conference of State Legislatures [NCSL], 2019).New campus carry proposals are introduced each year at state legislative sessions causing the policy landscape to continually change (Hassett et al., 2020;Tanner & Kelemen, 2023).Six states introduced campus carry bills in 2022, including Arizona, Florida, Kentucky, Missouri, New Hampshire, and Oklahoma for either implementing campus carry or repealing bans on guns on campuses, but Minnesota introduced a bill to further restrict guns on campus by civilians (Burnett, 2022).
Recent school shootings and violence on college campuses have sparked much discussion and debate on whether to allow concealed carry on campus (Nodeland & Saber, 2019).The idea of expanding gun rights after a campus shooting event may seem antithetical to some, but others see expanding concealed carry permissions as the only means of deterring potential mass shooters-the theory being that the knowledge of other armed (albeit concealed) individuals on campus might deter potential shooters (Hassett et al., 2020;Hayes et al., 2021).
There is no federal law governing the issuance of permits for carrying concealed firearms.Rather, each state has passed laws that allow qualified individuals to carry specific firearms in public, subject to certain restrictions (World Population Review, 2023).Concealed carry permits are divided into four categories, including unrestricted, shall-issue (i.e. the states must issue licenses if applicants meet certain requirements), may issue (i.e.licensing officials have the discretion to scrutinize and deny a license regardless of whether an applicant meets the statutory criteria), and no issue (Mayes, 2022).Several states have simplified the process of obtaining a permit to carry a concealed firearm for their residents and passed laws making it easier for citizens to carry concealed firearms in various locations (Hemenway et al., 2001a(Hemenway et al., , 2001b)).
Using county-level data from 1992 to 1997, Lott and Mustard (1997) found that states with shall-issue laws had lower rates of violent crime, murder, rape, and assault.In the wake of their "more guns, less crime" conclusion, numerous subsequent studies examined how relaxed concealed carry laws affected violent crime (Smart, 2023).While some studies suggested that relaxed concealed carry laws led to a decrease in violent crime, others found no significant effect, and some even reported an increase in violent crime (Hemenway et al., 2001a;Smart, 2023).
The conflicting findings have led to several attempts to synthesize primary studies on gun policy effects.In a systematic review of the early literature on shall-issue laws, Hahn et al. (2005) concluded there was insufficient evidence to assess their impact on violent crime.Similarly, the National Research Council (2004) reviewed much of the same literature as Hahn et al. (2005) and did not find a clear causal relationship between right-to-carry laws and crime rates.In 2023, an economist at the RAND Corporation, Smart, conducted the most comprehensive review.The conclusions of her review were based on 22 studies, which did not raise any major methodological concerns, although she incorporated all studies published between 1997 and 2020.In Smart's (2020) conclusion, shall-issue laws increase total homicides and firearm homicides, but there is insufficient evidence to show that they affect assaults and rapes, or whether they increase violent crimes.
Besides the actual effects of gun carrying laws on crime and violence, the psychological effects of firearm prevalence on safety and social capital have also drawn scholarly attention.Miller et al. (2000) examined how perceptions of safety are affected by an increasing number of people in a community who have firearms by using a national survey of 2,500 American adults in 1999.Only 14% of respondents reported feeling safer if others in their community acquired guns, while 50% reported feeling less safe.Hemenway et al. (2001b) examined the relationship between firearm availability and social capital across different states in the United States using the US General Social Survey data in conjunction with state-level proxy measures of firearm ownership.They found that states with higher levels of firearm ownership had lower levels of social capital, as measured by mutual trust and civic engagement.Hemenway et al. (2001a), which directly relates to the main research question of this study, explored public attitudes toward gun carrying in various locations, including college campuses, restaurants, sports stadiums, hospitals, and government buildings.Results revealed the majority of Americans surveyed thought concealed firearms should not be allowed in public spaces.Specifically, 94% of respondents believed that regular citizens should not be allowed to bring guns to college campuses.
A campus carry policy affects students, faculty, staff, and anyone who frequents a university campus including university/college police and nearby residents (Hassett et al., 2020).Over the course of the past decade, there has been a growing effort among scholars to understand the attitudes of those most affected by campus carry legislation.Recently, Hassett et al. (2020) conducted a systematic review of 17 existing studies on campus carry attitudes.Taking a look at the findings of previous studies of campus carry attitudes as a whole, the majority of campus community members reported some degree of opposition to campus carry.In particular, faculty and staff have consistently expressed opposition to campus carry, yet students have consistently reported being less opposed.Empirical evidence suggests, however, that college students are more uncomfortable with concealed firearms on a university campus than the general public.In their systematic review, Hassett et al. (2020) concluded the campus community primarily opposes the policy.
By surveying students and employees at universities where carrying guns is forbidden on campus, past research has also explored the potential impacts of campus carry policies.In these studies, researchers examined faculty, staff, and/or student perceptions of how the presence of guns on campus would impact personal safety and/or campus security (McMahon-Howard et al., 2021).Researchers have found that campus members perceive guns on campus as a negative threat to campus safety, and the majority of respondents indicated they would not feel safer on campus if they or others carried concealed guns (McMahon-Howard et al., 2021).
There have been a few studies that examined the actual impact of guns on campus at public colleges and universities that allow them.In recent studies using qualitative methods, several faculty and staff in Texas reported negative effects associated with having guns on campus (Jones & Horan, 2019;Somers et al., 2017Somers et al., , 2020)).Using survey data from institutions that allow concealed carry, quantitative studies also found that campus carry laws directly impacted faculty/administrators, staff, and students by causing increased fear of crime on campus, a lack of confidence in police, a perception that campus was unsafe, as well as concerns about mishandling or accidental discharge of guns and potential brandishing of handguns on campus (McMahon-Howard et al., 2021;Reimal et al., 2019;Scherer et al., 2022).As a result of campus carry, a significant number of respondents have decided not to enroll in or teach night courses, refused to discuss controversial topics in class, avoided challenging the majority opinion, and changed pedagogy and faculty-student interactions.Moreover, the comparisons revealed no change in the number of violent victimization on campus before and after the implementation of the campus carry law.However, following the implementation of the law, the number of campus members who reported experiencing fearful conflict on campus and carrying guns on campus for self-defense significantly increased (McMahon-Howard et al., 2021).
A limited number of empirical studies have investigated the impact of campus carry policies impacts on crime on campus.In comparing official crime rates from 2000 to 2009 for institutions in five states in the Western U.S.-two of which permit concealed firearms- Hayter et al. (2014) found no evidence that campus carry laws significantly affect crime rates on campuses.Gius (2019) also found no evidence that the law had a significant impact on state-level campus crime rates from 2005 to 2014.Input from both studies indicates that campus carry laws have not increased or decreased crime rates on campus significantly (McMahon-Howard et al., 2021).
As states continue to debate over legalizing guns in higher education institutions, it is imperative to recognize that such policy may negatively impact the day-to-day activities and feelings of safety of students and employees on university campuses (Hayes et al., 2021, McMahon-Howard et al., 2021).In the absence of clear evidence that campus carry policies prevent crime on campus, it may, however, be the case that campus carry policies remain linked to political discussions despite campus members' dissatisfaction with these policies (Carter et al., 2022).The empirical examination of this divisive policy is crucial.In particular, the correlates and predictors of campus carry attitudes, intention, behavior, and impact could be of great use to policymakers and university administration when designing and implementing college/ university shooting prevention policies.Hassett et al. (2020) systematic review provides the only existing synthesis of campus carry by reviewing 17 existing empirical studies examining the attitudes of students, faculty, and staff as well as the factors contributing to their attitudes.While the majority of individuals in the university community opposed campus carry, the studies showed considerable variations in campus carry attitudes depending on demographic factors, gun ownership, fear of campus crime, and past crime victim experiences.Hassett et al. (2020) also noted the possibility of evolving campus carry attitudes over time, with more recent publications showing more support for campus carry, but no definitive conclusion could be drawn due to a lack of recent studies.Since Hassett et al. (2020) synthesized the existing studies, the research literature on campus carry has grown.As Kelling et al. (2021) pointed out, the assumption that campus carry is universally opposed by the campus community is no longer true.There has been evidence that more diverse campus carry opinions have been reported in recent studies, although the majority of campus community members are still opposed to concealed carry on campus (Kelling et al., 2021;Satterfield & Wallace, 2020).Nodeland and Saber (2019), for example, found that 46% of student respondents and 27% of faculty/staff respondents supported campus carry.Furthermore, Hayes et al. (2021) found that over half of the participants favored campus carry.Hassett et al. (2020) noted that previous research on the attitudes toward campus carry had explored only a limited number of variables.However, there has been more recent research on the factors that predict support for the policy, ranging from demographic variables to variables specific to the campus (Hassett & Kim, 2021).In their systematic review, Hassett et al. (2020) focused only on campus carry attitudes.Recent studies have found that there are differences between factors that predict attitudes toward campus carry and those that predict engagement in campus carry behaviors (Hayes et al., 2021).The findings of Hassett et al.'s systematic review of campus carry attitude may not be generalizable to campus carry intention, behavior, or perceptions of campus carry effects.

Current Meta-analysis
Gun control is a highly controversial and divisive social issue (Hassett et al., 2020).As campus carry legislation and policies have been introduced over the last few years, this issue has been thrust into the academic arena (Hassett et al., 2020).The recent study found that, on average, campus safety concerns increased after the campus carry policy was passed, supporting the arguments made by opponents of campus carry that guns on campuses could negatively impact campus safety perceptions (McMahon-Howard et al., 2021).Since campus carry policies are perceived as having negative impacts on campus community members and have little or no crime deterrence effect, empirical studies are crucial to acquiring integrated knowledge about the factors that influence attitudes, intentions, behaviors, or perceived impacts of campus carry.It is particularly important to gain insight into dynamic (i.e.changeable) factors for intervention purposes, while a comprehensive assessment of both dynamic and static contributing factors is essential (Emmelkamp et al., 2020).
There has been a growing body of primary research examining a range of correlates and predictors of campus carry, including attitude, intention, behavior, or impact.At the time of writing, there has been only one systematic review that focuses on campus carry attitudes.The synthesis research of Hassett et al. (2020) provides a comprehensive overview of research on attitudes toward campus carry policies.The systematic review, however, cannot provide clarification of the relative importance of different factors influencing campus carry intention, behavior, and impact.For the campus community as a whole and university administrators and policymakers responsible for assessing and intervening with campus safety issues, the relative composite effect sizes for attitudes, intentions, behaviors, and impacts for each domain of risk factors must be of considerable interest.
Using a meta-analytic approach, the present study examined how risk factors at distinct domains affect campus carry attitudes, intentions, behaviors, and impacts among students, faculty, staff, and administrators.Our first step was to estimate the composite effect sizes of risk domains for each campus carry outcome (i.e.attitude, intention, behavior, and impacts).We also examined if and how study, sample, and risk factor characteristics moderate the effects of risk domains.

Inclusion Criteria
Several inclusion criteria were established before conducting the literature search.First, the focus of the study must match with the goals of the current research, which was to examine the correlates and predictors of campus carry attitudes, intentions, behaviors, or impacts.Second, the research sample had to be drawn from university-affiliated individuals-students, faculty, administrators, and/or staff.Next, studies must report the results of bivariate or multivariate analyses of the association between contributing factors and campus carry, or provide sufficient details to compute a bivariate test statistic.No restrictions were placed on the year of publication.The current research has some limitations in terms of inclusion criteria, which should be acknowledged.The meta-analysis solely included published and peer-reviewed studies to avoid selection bias and maintain the quality of the research (Kim, 2022b).As high-quality dissertations are often published in peer-reviewed journals, a systematic search in the current study was likely to have found them.Nonetheless, it cannot be entirely ruled out that some relevant studies may not have been located.

Search Strategies
The present study employed a systematic review approach to identify all relevant studies by searching existing previous research of interest according to the pre-determined inclusion criteria (Cooper, 2017).The first search was conducted in September 2020 and an update was carried out in August 2022.Two researchers searched 10 electronic databases (Criminology: A Sage Full-Text Collection, Criminal Justice Abstracts, Social Service Abstracts, SocINDEX with full text, Sociological Abstracts, Sociological Collection, Health and Safety Science Abstracts, JSTOR, Psych INFO, and Psychological & Behavioral Science Collection).The following search terms were used to search the selected database: (college, school, campus, or university) AND (guns, firearm, weapon, shooting, possession, control, carry, concealed, safety, or policy).This set of search terms is similar to those used in Hassett et al. (2020) previous campus carry systematic review.Additionally, citation searches were conducted on all papers that met the inclusion criteria.To further detect works by active researchers cited, Google Scholar was searched with the authors' names after obtaining initial sample studies through online databases and reference lists (Cooper, 2017).
A database search returned 4616 hits, and a reference tree search identified four sources.
Papers were then excluded from the sample in the following manner.As a first step, all duplicate references were removed from the original sample of 4620 articles (n = 1995).Second, all titles and abstracts of the remaining articles (n = 2625) were screened.Those that did not meet the inclusionary criteria were removed.These included studies where topics are irrelevant to campus carry (n = 1713); qualitative studies or legal review papers (n = 409); and studies without specific statistics regarding campus carry and correlates/predictors (n = 451).In the third step, we obtained full-text copies for all 52 remaining citations.Each paper was evaluated according to the inclusionary criteria.The full-text review excluded seventeen articles due to the following reasons: the same data was used in another study (n = 3), and there was insufficient data on effect size or information necessary to convert effect size estimates (n = 14).Upon the reviewer's recommendation, two new studies were added in February 2023.A total of 37 empirical studies were included in the final sample.

Data Extraction and Coding of Studies
Two researchers coded data from each article using an a priori data extraction form.A key variable in this meta-analysis was the risk domains in which risk factors were classified.A risk domain is a group of homogeneous risk factors that are classified together based on their similarity (Emmelkamp et al., 2020).By clustering individual factors into domains, the composite effect size of each domain within each campus carry outcome could be estimated and compared (Eisenberg et al., 2019;Kim, 2022a).All risk factors identified in the included studies were clustered into "risk domains." After the classification of risk factors into domains, the following 13 risk domains were distinguished: demographics, personal characteristics, ecology, political ideology, gun socialization, media exposure, social life, campus life, non-gun self-protection measures, safety-personnel/ measures, perceived safety on campus, fear/risk of crime/victimization, and experience of crime/victimization.The factors specific to each domain are listed in Table 1.
A measure of campus carry outcome (i.e.attitudes, intentions, behaviors, and impacts) as well as corresponding descriptive and inferential statistics were obtained from the primary studies.As potential moderators in the relationships between risk domains and campus carry outcomes, general study characteristics (i.e.data year, number of universities where data was collected, sample size, survey type: in-person, online, or both) and sample characteristics (i.e.campus carry laws in the state where the data were collected: prohibited, permitted, up to the individual institution, sample affiliation: students, faculty/staff, or whole campus community) were extracted.Kappa statistics of .78 to .99 indicated excellent inter-coder agreement.Coding discrepancies were resolved by reviewing the paper in question and referring to the established methodology in the meta-analysis (Cooper, 2017).

Effect Size Estimation
All effect sizes (ES) of risk factors and risk domains were converted to Pearson's product-moment correlation coefficients (r) to investigate their explanatory power Note: r = reverse coding of the effect direction.(Emmelkamp et al., 2020).The ES r is most commonly used in meta-analyses concerning two continuous variables and is easy to interpret.Bivariate effect sizes were used in the analysis, but there were some exceptions.In a few studies, the multivariate effect sizes (such as standardized coefficients) were only presented.For this case, we retrieved multivariate effect sizes.When the r was not given in the included paper, the Campbell Collaboration's Practical Meta-Analysis Effect Size Calculator (Lipsey & Wilson, 2001) was used along with formulas of Peterson and Brown (2005) to convert other test statistics into ES r (Kim et al., 2020).

Meta-analytic Method
Each ES r represented a relationship between a risk factor and a campus carry outcome.Primary studies included in this meta-analysis reported multiple effect sizes that could be categorized under the same risk domain.An important assumption of meta-analysis, namely that all reported effect sizes should be independent, is violated when more than one effect size from a single study is extracted and analyzed together.
The effect sizes within a study are likely to be more similar (in size) than those in different studies (Emmelkamp et al., 2020).Using a three-level random-effects meta-analytic model to account for the dependency of effect sizes, the combined effect sizes were estimated for each risk domain (Assink & Wibbelink, 2016).In a three-level meta-analysis, all relevant information reported in primary studies is considered by modeling three sources of variance: sampling variance of the observed effect sizes (Level 1); variance of effect sizes within studies (Level 2); and variance of effect sizes between studies (Level 3) (Eisenberg et al., 2019).In this way, composite effect sizes can be better estimated, and maximum statistical power can be achieved (Assink et al., 2015;Geerlings et al., 2020;Kim, 2022a).The statistical analyses were performed using the "rma.mv"function of the "metafor" package in the program R studio (version 1.4.1717)(Emmelkamp et al., 2020;Kim, 2022a).Data was compiled in Excel.Based on Assink and Wibbelink (2016) R syntax, a multilevel meta-analysis was conducted.Each risk domain in the present study contained homogeneous risk factors for campus carry outcomes; therefore, a separate meta-analysis was conducted for each domain (Assink et al., 2015;Emmelkamp et al., 2020;Kim, 2022a).All model parameters were estimated using the restricted maximum likelihood estimation method.Knapp and Hartung (2003) adjustment with the t-distribution was also used to test the individual regression coefficients of the meta-analytic models and calculate the corresponding confidence intervals (Assink et al., 2015;Geerlings et al., 2020).P-values less than .05were considered statistically significant in all analyses (Emmelkamp et al., 2020;Geerlings et al., 2020;Kim, 2022a).The resulting mean ES estimates (Mr) were interpreted based on Rice and Harris (2005) benchmark: a "small" effect when .10 ≤ Mr < .24, a "moderate" effect when .24 ≤ Mr < .37,and a "substantial" effect when Mr is .37 or above (Eisenberg et al., 2019).
To determine the significance of variance in effect sizes within (Level 2) and between studies (Level 3), two separate one-sided log-likelihood ratio tests were performed.These tests involved comparing the deviance of the full model with the deviance of a model excluding one of the variance parameters (Assink et al., 2015;Assink & Wibbelink, 2016;Eisenberg et al., 2019).The result of these tests indicated a significant decrease in model fit when level 2 or level 3 variance was excluded, suggesting that there was significant variance at those levels (Geerlings et al., 2020).Therefore, it can be concluded that effect sizes were heterogeneously distributed within or between studies (Assink & Wibbelink, 2016;Kim, 2022a).
The level 2 and/or level 3 variances within each campus carry outcome were statistically significant as a whole, so four multiple moderator analyses were performed to determine whether moderators could explain variation (Wibbelink et al., 2017).For moderator analyses, continuous variables were centered around their means whereas categorical variables with k categories were transformed into k-1 dummy variables through binary coding (Wibbelink et al., 2017).In the moderator analysis, all potential moderators were tested simultaneously using an omnibus test (F test).When the F test was statistically significant, the regression coefficient (slope) of each variable was tested via t-test (Eisenberg et al., 2019).

Publication Bias Tests
The largest threat to the credibility and validity of meta-analysis results is publication bias, "a phenomenon where studies with large and/or statistically significant effects, relative to studies with small or null effects, are more likely to be published" (Polanin et al., 2016, p. 207).To investigate whether publication bias led to the severely inflated mean effect size estimates in the meta-analysis and to determine the number of hypothetical unpublished studies required for a significant reduction in the mean effect size, various graphical and analytic techniques were performed using the "trimfill," "regtest," and "fsn" functions of the "metafor" package in the program R Studio (Geerlings et al., 2020;Kim, 2022aKim, , 2022b)).

Risk Domains for Campus Carry Attitudes
The composite effect size of the 13 risk domains for campus carry attitudes was statistically significant and positive (Table 2).Based on Rice and Harris (2005) criteria for interpreting effect sizes, however, the overall effect size Mr = .075is trivial.The variance results (i.e. the statistical significance of level 2 and level 3 variance) support the need for separate tests of risk domains within studies of campus carry attitudes.The strongest correlation was found between gun socialization and campus carry attitude, and the effect size was medium (Mr = .262).And, political ideology had a medium effect size (Mr = .225).There was a small effect in the domain of personal characteristics (Mr = .161).The remaining 10 risk domains had trivial effects.

Risk Domains for Campus Carry Intentions
Twelve out of the 13 risk domains were tested for campus carry intention (Table 3).The composite effect size of the 12 risk domains for campus carry intentions was not statistically significant.While its composite effect size (Mr = .088)was still considered trivial (Rice & Harris, 2005), it was larger than one for campus carry attitude (Mr = .075).Among the domains, we found a large effect for the social life domain (Mr = .351).The Note: # studies = number of studies; # es = number of effect sizes; se = standard error; ci = confidence interval; % Var = percentage of variance explained; level 2 variance = variance between effect sizes from the same study; level 3 variance = variance between studies.+ p < .10;* p < .05;** p < .01;*** p < .001.
following three domains had a small effect size, according to Rice and Harris (2005) criteria: ecology (Mr = .168),personal characteristics (Mr = .130),and perceived safety on campus (Mr = .125).A trivial effect was found for the remaining eight domains.

Risk Domains for Campus Carry Behavior
Table 4 presents the results of campus carry behavior.A composite effect of nine (9) domains was statistically and substantively non-significant (Mr = .014,p = .564),and the magnitude of each individual risk domain ranged from trivial (Mr = 0.027 for crime/ victimization experience) to small (Mr = 0.133 for political ideology).According to the criteria of Rice and Harris (2005), the effects of the three domains were small (i.e.political ideology, safety personnel/measures, and campus life), while the effects of the six remaining domains were trivial.Positive effect size was found for safety personnel/ measures, indicating that those with a high level of trust and confidence in safety personnel/measures on campus were likely to carry a concealed gun on campus.In a discussion section, the implications of this result were discussed in more detail.

Risk Domains for Campus Carry Impacts
In the study, 11 out of 13 risk domains were tested for perceived positive impacts of campus carry (Table 5).There was a significant and positive composite effect size Note: # studies = number of studies; # es = number of effect sizes; se = standard error; ci = confidence interval; % Var = percentage of variance explained; level 2 variance = variance between effect sizes from the same study; level 3 variance = variance between studies.+ p < .10;* p < .05;** p < .01;*** p < .001.Note: # studies = number of studies; # es = number of effect sizes; se = standard error; ci = confidence interval; % Var = percentage of variance explained; level 2 variance = variance between effect sizes from the same study; level 3 variance = variance between studies.+p < .10;* p < .05;** p < .01;*** p < .001.
for the 11 risk domains.This composite effect size (Mr = .104)is small, but larger than those of other campus carry outcomes.Political ideology showed the strongest correlation with the perceived positive impacts of campus carry with a medium effect size (Mr = .213).There was a small effect (Mr = .155)in the domain of gun socialization.In the remaining nine risk domains, the effect was trivial.

Sensitivity Analyses: Multiple Moderator Analyses and Publication Bias Analyses
It was found, from the likelihood-ratio tests, that there were significant variances in composite effect size estimates in attitude, intention, and perceived positive impact, both between effect sizes from the same study (i.e. level 2 variance) and between studies (i.e. level 3 variance).Only level 2 variance was statistically significant for campus carry behavior.The small number of studies may explain this non-significance of level 3 variance.In fact, all 36 effect sizes were derived from only three studies.Overall, these heterogeneity analyses indicated that effect sizes were more variable than expected based solely on sampling variance.It is likely that moderators, along with types of risk domains, influence campus carry outcomes (Eisenberg et al., 2019;Emmelkamp et al., 2020).
To examine the unique influence of each significant moderator, multiple moderator analyses were conducted.Multiple moderator models for campus carry attitudes, intentions, behaviors, and impacts are presented in Appendix A, B, C, and D, respectively.A number of universities were found to have statistically significant positive effects in our campus carry intention model; studies that collected data from a larger number of universities showed greater effect sizes for the relationship between risk domains and campus carry intentions.The campus carry impact model, however, found statistically significant negative effects for a number of universities.In other words, studies with data collected from a larger number of universities had smaller effect sizes of risk domains with perceived positive impacts of campus carry.Additionally, in the campus carry impact model, we found a statistically significant positive effect for sample affiliation; studies with data from faculty/staff samples had a larger effect size on risk domains with perceived campus carry positive impacts than studies with sample sizes containing all campus community members.A model of campus carry behaviors did not find a significant moderator.The results showed that although there were some sample and study characteristics that moderated the association between risk domains and campus carry outcomes, these effects were relatively weak.
Each composite effect size was subjected to a separate publication bias analysis.A visual inspection of the funnel plots revealed no significant publication bias for all four campus carry outcomes (Emmelkamp et al., 2020).The original composite effects sizes were not changed by the trim-and-fill analyses (Eisenberg et al., 2019).In four campus carry outcomes, all fail-safe Ns exceeded their thresholds.In sum, these findings indicate that publication bias was not an issue in this study, which supports the validity of the results.

Summary and Discussion
The central aim of this multilevel meta-analytic study was to determine the relative strength (expressed as effect size) of risk domains for predicting different campus carry outcomes (including attitudes, intentions, behaviors, and perceived impacts) among campus community members (students, faculty, staff, and administrators).All four campus carry outcomes had trivial to small composite effect sizes, but the current study found considerable variation in effect sizes across risk domains within and between campus carry outcomes.The unique risk domains were associated with specific outcomes within the context of campus carry, but only one domain was consistently linked to both attitudes and behaviors.It was the political ideology domain that influenced campus carry attitudes, behaviors, and impacts, but not intentions.Intentions to carry guns on campus were influenced by a domain of social life-related variables (such as drinking, spending time with friends, social acceptance, and deviant lifestyle).Its effect size (Mr = .351)was the largest among all effect sizes estimated in the current study.Aside from campus carry intentions, campus carry behavior would be the most important and interesting topic for campus community members and policymakers since attitudes might not always lead to actions, but behaviors can directly affect the campus community.Results from the current meta-analysis revealed that both perspectives on safety personnel/measures and variables related to campus life, as well as political ideology significantly influenced campus carry behaviors.
Political ideology-related variables, such as voting behavior and political affiliation, have consistently been associated with gun-related attitudes in previous studies.The current meta-analysis confirmed that political ideology significantly affects both attitudinal support for campus carry and concealed carry behaviors.This result is informative but does not offer any concrete implications for policies and practices because political ideology is relatively stable.
A better approach would be to focus on changeable environmental factors, which can yield important policy implications, rather than simply confirming the common demographic predictors or other established independent variables (Hassett et al., 2020).Our discussion here largely focused on dynamic (rather than static) risk domains (Eisenberg et al., 2019).

Social-Life Domain and Campus Carry Intentions
Interventions and prevention strategies that aim to reduce school shootings often target bullied and socially isolated students.These strategies include fostering inclusive and safe learning environments as well as bullying prevention programs (Burton et al., 2021).According to the current findings, however, more socially connected students intend to carry guns on campus.Wallace (2019) found that socially active students, especially those perceived to be leaders in their peer groups, viewed gun carrying as more beneficial to society.The importance of college shooting prevention programs, according to Wallace (2019), lies in helping students disrupt myths about gun carrying and popularity, seek out alternatives to peer pressure, and understand the consequences of carrying a gun or violating weapons laws.Among the specific recommendations was Teens on Target, a peer-based program that involves other teens meeting with youth sanctioned for weapons violations, which could be adapted for college campuses to address peer perceptions of guns and gun carrying (Conduent, 2017;Wallace, 2019).Carter et al. (2019) examined school counselors' acceptance of school shooting prevention strategies.A multi-tiered approach to school safety, which was rated significantly higher than all other procedures by this expert group, could also be adapted for college students.The multi-tiered approach to school safety includes providing social, academic, and emotional support strategies for all children and differential strategies for students who have engaged in destructive behaviors or are at risk of doing so (Carter et al., 2019).As it addresses social life as well as campus life for different types of students, whether active or passive, the Multi-tiered Safety Approach may be more effective than other prevention methods.Carter et al. (2019) advocated that school counselors select, adopt, and implement evidence-based counseling curricula within a multi-tiered process that addresses students' personal/social, academic, and career needs.This recommendation is also applicable to college settings.On college campuses, counselors who work in mental health programs are often the only persons familiar with the mental health, emotional, and behavioral needs of their students.Comprehensive school counseling programs that include effective school shooting prevention/intervention can help improve campus climate, assist students' personal/social development, and reduce violence.A university's mental health services can develop comprehensive counseling-related protocols that are an essential component of school shooting prevention and intervention (Carter et al., 2019).

Campus Safety Personnel/Security Domain, Campus-Life Domain, and Campus Carry Behaviors
According to a previous systematic review of campus carry attitudes, higher confidence in existing (non-gun related) campus safety measures was associated with a less favorable campus carry attitude within the total sample as well as within the faculty subgroup, while distrust of the police was associated with higher support for campus carry policies (Hassett et al., 2020).Taking this into account, Hassett et al. (2020) concluded that universities also need to enhance campus police trust and legitimacy in addition to improving campus security measures.This meta-analysis confirmed the previous systematic review result by showing the statistically significant negative association between security personnel/measures and campus carry attitudes, even though the effect size was small.In contrast, the current meta-analysis results found that campus carry behavior and intention were positively associated with campus security measures/personnel.It, therefore, appears that those who are confident in campus security and personnel also carry guns on campus.
The empirical evidence in this study is in accordance with Carter et al. (2022) study, which examined whether campus community members (i.e.students, faculty, and administrators) would accept 11 different strategies for preventing active shooters on college/university campuses.Other school shooting prevention measures were significantly correlated with each other in a positive direction, but not security personnel and campus carry.Campus carry was significantly related only to training, security technology, and security personnel procedures.Those who carry firearms on campus or intend to do so are apparently neither distrustful nor unconvinced of campus security measures/personnel.According to Kelling et al. (2021), those reporting gun ownership, concealed carry licenses, and/or plans to carry on campus were most likely to believe that both universities and individuals are responsible for protecting them from violence on campus.
Another key finding of the current study is that campus life-related variables (e.g.days on campus, residence on campus, full-time students, and taking in-person courses rather than online courses) are strongly and positively correlated with campus carry behavior.This study, however, cannot explain the mechanism behind this relationship.The results of this study indicated that both perceived fear of crime on campus and offending/victimization experiences had trivial effects on campus carry behavior.As discussed above, the current meta-analysis results also suggested that simply adding security measures and security personnel on campus is unlikely to reduce the carrying behavior.According to Nodeland et al. (2021), university programs, such as counseling or mental health services, can help campus community members consider options other than campus carry.Additionally, regular updates about campus safety are provided, as are potential security concerns and changes to campus carry policies, which may reduce the likelihood of carrying guns on campus.In the context of campus carry, Nodeland et al. (2021) highlighted the importance for university decision-makers or administrators to take a holistic approach that includes all stakeholders on campus, community members, as well as the parents of students residing on or near the campus.They all could influence campus carry behaviors, either directly or indirectly (Nodeland et al., 2021).

Experts' Opinion on the Acceptability of Active Shooter Prevention Strategies
Identifying campus shooting prevention strategies that are both effective and acceptable to stakeholders will become easier as more research and evidence become available (Carter et al., 2019;Hassett & Kim, 2020).However, as of now, there is little clear evidence for the effectiveness of active shooter prevention strategies including campus carry policies, and stakeholders on university and college campuses are unsure of their perspectives on various prevention measures (Carter et al., 2019).In a recent study, Carter et al. (2019) investigated the social validity of interventions designed to prevent shootings on college and university campuses by evaluating which commonly proposed prevention measures are most acceptable to stakeholders.It appeared that college and university stakeholders could find several different active shooter prevention strategies acceptable.Carter et al. (2019) found campus carry, peer evaluation and responses, zero tolerance, and profiling procedures to be the least acceptable.Experts such as criminologists, school psychologists, and security professionals would be valuable in rating various campus shooting prevention measures for acceptability.It would be possible to propose policy recommendations based on the results that are not only evidence-based but also probable (Hassett & Kim, 2020).

New Predictors and Risk Domains Specific to Each Campus Carry Outcome
According to the current meta-analysis, the composite effect size of the combined domains is trivial to small across all four campus carry outcomes.In most cases, the effects of the risk domains within each campus carry outcomes were trivial to small.Future primary studies should explore additional potential predictors of campus carry outcomes.Discovering changeable environmental factors can yield important policy results rather than simply confirming demographics or other established independent variables.Campus services and attitudes toward other campus shooting prevention strategies may warrant further examination.There is evidence in this study that different predictors are associated with different campus carry outcomes.It is therefore imperative that future primary research focuses on identifying unique predictors of campus carry outcomes.

Moderating Effect of Standing State Laws among Faculty Members
The results of previous studies suggest that faculty members, in comparison with students or staff in non-student-centered or nonsupervisory roles, experience an increased fear of crime and an increase in fear-producing interpersonal campus conflicts caused by guns on campus due to the nature of their roles (McMahon-Howard et al., 2021).It is important to note, however, that not all faculty members oppose campus carry policies (Hayes et al., 2021).According to Hassett et al. (2020), geographical regions and standing gun policies significantly influence faculty members' support for campus carry.Although it was among a sample of criminologists, results showed that faculty, staff, and administrators in states where campus carry has been implemented were more supportive of it compared to those in states where it is prohibited or up to individual universities (Hassett & Kim, 2020).It is possible that job seekers may take campus gun policies into account when selecting positions.As a result, criminologists who work in states where campus carry is legal may be more conservative from the start.The policy could have also led to those opposed to campus carry moving to another school if the policy had been implemented at their institution (Flaherty, 2015;Soboroff et al., 2019).
There is a lack of recent studies on faculty in states where campus carry is still prohibited.As a result, we could not test whether standing state law moderates faculty campus carry.Research has increasingly examined faculty attitudes toward campus carry in states that allow it (Scherer et al., 2022;Somer et al., 2020).Campus carry attitudes among faculty in states that prohibit campus carry have not been studied recently, preventing comparisons with their counterparts in states that allow campus carry.If we can collect data from studies conducted in states that allow campus carry, from institutions that allow campus carry, or from national data, then we will be able to evaluate any potential effects that the policy may have on faculty members.In light of state campus laws, policymakers and university administrators can determine strategies for hiring and retaining faculty members based on this information.
There are differences in the restrictions on where guns are permitted between states with this law, which means that campus carry laws may not have the same impact across states.Each state law limits the places on campus where guns are prohibited.For example, Georgia's law ( § 16-11-127.1)prohibits guns in faculty, staff, and administrative offices.Consequently, most campus members are unable to legally carry their guns on campus since they spend a considerable amount of their time in their office spaces (McMahon-Howard et al., 2021).
It is necessary to conduct more primary research in states where campus carry is permitted as well as those where it is prohibited in order to test the moderating effect of state laws.A greater focus should be placed on campus carry-related outcomes other than attitudes.It was found in this study that while prior research has documented attitudes toward campus carry among faculty and staff, less is known about their intentions to engage in campus carry behavior if/when permitted.

The Impacts of the Bruen Decision
The Supreme Court's decision in New york State Rifle & Pistol Association v. Bruen expanded the right of Americans to bear arms under the Second Amendment to the Constitution.The Bruen decision significantly limited the ability of states to restrict citizens' rights to carry firearms for self-defense in public (Mayes, 2022).Despite the Court's recognition of the legitimacy of prohibitions against gun possession in "sensitive places, such as schools and government buildings," the definition of "sensitive" was unclear in the Bruen ruling.Several lawsuits have been filed against the state's designation of sensitive locations in New york, citing the Bruen decision (Finerty, 2022).Legal experts anticipate additional future lawsuits throughout the country challenging both existing gun laws and new laws passed to address the impact that the Bruen decision will have on the increase in the public carrying of firearms (Finerty, 2022;Mayes, 2022).Campus carry practices may be affected differently by the Bruen decision depending on the laws and regulations in each state.A better understanding of Bruen's implications requires further research.

A Comparison of Risk Domains by University Affiliations
In the current meta-analysis, moderating tests of affiliation status (i.e.students, faculty, staff, and administrators) were not conducted due to the lack of studies using non-student samples.Only campus carry attitudes and campus carry impacts were evaluated with regard to the moderating effects of affiliation status.The university affiliation did not moderate attitudes toward campus carry, but it did moderate the perceived impacts of campus carry.These results imply that there may be significant risk factors for perceived campus carry impacts that vary among students, faculty, and staff.Nonetheless, the current meta-analysis results cannot answer the important question of which risk domains for the campus carry impacts are unique to different members of a campus community, which is an important question to address.The unique risk domains related to campus carry outcomes, such as intention, behavior, and impact, should be identified for each subgroup of the campus community in a future meta-analysis.

Conclusion
A campus carry policy affects students, faculty, staff, university/college police, and residents living near campuses.The existing empirical evidence indicated that campus carry negatively affected campus life and did not result in a reduction in crime.Even so, a small but significant number of stakeholders reported favorable attitudes toward gun carry on campus, intended to carry a gun if allowed, and already carried a gun.The current study contributes to the campus carry literature by synthesizing empirical studies and meta-analyzing the effects of various risk domains.Three-level meta-analysis results revealed that different risk domains are associated with different campus carry outcomes.It appeared that social life-and campus life-related risk factors were strongly associated with campus carry intentions and behaviors, respectively, while the political ideology domain influenced both attitudes and behaviors.Evidence-based college shooting prevention measures and campus security procedures targeting risk factors in the social life and campus life domains are likely to reduce campus carry and improve the educational atmosphere.The current comprehensive systematic review and meta-analysis of campus carry are also of value to the field in terms of guiding further studies in this area.Considering the lack of criminological literature on campus carry intention and behaviors, it becomes more important than ever for scholars to continue to advance knowledge in this area.

Table 1 .
Domains and factors classified in each domain.

Table 2 .
campus carry attitude: Overall mean effect sizes (correlations) for the domains, with corresponding level 2 and level 3 variance estimates.

Table 3 .
campus carry intention: Overall mean effect sizes (correlations) for the domains, with corresponding level 2 and level 3 variance estimates.

Table 4 .
campus carry behavior: Overall mean effect sizes (correlations) for the domains, with corresponding level 2 and level 3 variance estimates.

Table 5 .
campus carry (positive) impact: Overall mean effect sizes (correlations) for the domains, with corresponding level 2 and level 3 variance estimates.