Method

SAMPLE

Data are drawn from two cross sectional telephone surveys of registered voters in the US (n = 4138). Because the data were originally collected as part of a larger study to evaluate a public education campaign about gun violence, surveys were administered at two different times: June 1994 and February 1995. Survey questions asked about awareness of the campaign, experiences with violence, general beliefs about guns, and support for violence prevention interventions. Participants in the two surveys did not differ significantly on any of the study variables of interest and the two samples were therefore combined. The sampling frame consisted of households with telephones in metropolitan statistical areas that included a city with a population of at least 500 000. Respondents were selected using random digit dialing methods. To facilitate subgroup analyses, African Americans were oversampled. This oversampling was achieved by creating a separate sampling frame of phone exchanges that were known to have at least a 30% African American population. One out of three individuals contacted agreed to participate and 98% of those who participated completed the entire survey.

MEASURES

The outcome variable reflects responses to a question asking respondents whether they believed keeping a gun in the home makes the people who live there more or less safe. While “it depends” was not explicitly offered as a response option, it is included in the analysis since nearly one quarter of study participants (23%) gave this as an unsolicited response. We also hypothesized that the group responding “it depends” might be substantially different than either the “more safe” or “less safe” groups. If this is true, having three respondent groups provides an opportunity to better understand a group that might be more susceptible to attitudinal change.

Factors hypothesized to influence beliefs about the risks of guns in the home that were examined included: (1) exposure to violence (violent victimization, gun related crime in the neighborhood), (2) fear of violent victimization, (3) trust in the police to keep the neighborhood safe, (4) personal characteristics (sex, age, ethnicity/race, education, household income, and children living at home), and (5) political ideology as measured by self reported major US political party affiliation (Democrafts, Independent, and Republican).

Two measures of exposure to violence were used: (1) previous victimization of the respondent, the respondent's family members, or a close friend, within the past five years and (2) perceived levels of gun related crime in the respondent's neighborhood. To determine potential differences due to type of weapon used by perpetrators, violent victimizations were analyzed in two ways: one variable differentiated between victimizations that involved a gun while the other did not.

The inclusion of variables measuring fear of victimization were used to differentiate those respondents whose fear of victimization is based on actual exposure to violence, by comparison with those who are fearful but have not experienced significant exposure to violence. For example, certain segments of the population such as the elderly and women have higher levels of perceived risk of victimization relative to their actual rates of victimization.^18,29–32

Previous studies have found that as trust in the police decreases, the level of gun ownership increases.^33–35 As a result, we hypothesized that individuals would be more likely to view guns as protective if they did not trust the police to provide protection.

Respondents were grouped into three age categories: 18–34 years (22%), 35–65 years (58%), and over 65 years (18%). In the US, persons 18–34 years experience the highest rates of gun related homicides and assaults; persons over 65 years of age experience the highest rates of gun related suicides and reported levels of fear of violent crime.^36,37

Other demographic factors may also be associated with beliefs about guns in the home. For example, African Americans who are disproportionately affected by gun violence,^38–40 and respondents with children, may be more concerned about the safety of having guns in the home. Those with higher levels of education and income are more likely to take measures to protect their health (for example, wear seat belts, exercise).^41,42 Beliefs about guns and gun policy are also strongly associated with self reported political party affiliation.^43–45 Therefore, we hypothesized that Republicans would be more likely than Independents or Democrats to believe that guns make a home more safe. Because handgun owners commonly cite protection as one of the primary reasons for owning a handgun,^18,19 we hypothesized that handgun owners would be more likely to believe in the protective value of guns in the home. Since the ownership of long guns (rifles and shotguns) for protection in urban areas within the US is uncommon, we did not focus on this type of ownership.

DATA ANALYSIS

For all analyses, sampling weights were applied to adjust for oversampling of African Americans. The outcome variable, beliefs about whether a gun in the home makes a home more safe or less safe, was cross tabulated with each independent variable. The statistical significance of these bivariate relationships was evaluated using the likelihood ratio χ² test when the independent variable was categorical and one way analysis of variance when the independent variable was continuous. Multinomial logistic regression analysis was used to examine the relationship between each independent variable and the outcome variable while simultaneously controlling for other variables in the model.⁴⁶

Although all respondents are included in our univariate and bivariate analyses, several factors required exclusion of some respondents for the multivariate analysis. Multinomial logistic regression analysis does not reliably identify the best model fit when variables with small categorical percentages are included in the model. Therefore, ethnic/racial groups that comprised fewer than 5% of the sample population were excluded. Similarly, respondents who stated that they did not know whether a gun in the home was more protective or perilous (9%) were excluded. Twenty five per cent of the cases (n = 1034) were also excluded in the final model due to the exclusions mentioned above or to missing information concerning one or more variables of interest. Refusal and “don't know” responses comprised most of the missing information, and ranged from less than 1% to 3.8% for individual variables in the final model.

Handgun ownership is more likely to be a consequence of an individual's belief about the risks and benefits of keeping a gun in the home rather than a cause of those beliefs. Including handgun ownership in the model could, therefore, account for variance due to precursors of beliefs about guns and safety, and add little to our understanding of the determinants of those beliefs. For this reason, we chose not to include handgun ownership in the final model. Because 17% of respondents (n = 626) eligible for inclusion refused to disclose their income, income was also excluded from multivariate analysis. Other variables were excluded from the final multivariate model if they were not significantly associated with the outcome and their exclusion did not substantially alter the parameter estimates for the other covariates. This enabled us to identify the most parsimonious model that included all relevant confounders for which data were available.

Results

Twenty nine per cent of the respondents believed keeping a gun in the home makes the home more safe, 40% said it makes it less safe, 23% said “it depends”, and 9% said “don't know”. Respondent characteristics are described in table 1.

Table 2 shows that groups more likely to believe that a gun in the home increases safety include males, Republicans, handgun owners, and those who had very little or no trust in the police to keep the neighborhood safe. Having 12 years of education or fewer, or having no children living at home, were each associated with the belief that guns in the home generally enhance the safety of its occupants.

The multinomial logistic regression analysis controlled for the effects of other variables, but, with few exceptions, produced results (table 3) similar to the bivariate analyses (table 2). Compared with respondents with “a lot of trust” in the police, those with “very little” or “no trust” in the police were much more likely to believe that keeping a gun in the home increased household safety (odds ratio (OR) 4.5; 95% CI 2.1 to 9.2). Those with “some trust” in the police were slightly more likely than those with “a lot of trust” in the police to believe that a gun would make a home safer (OR 1.28; 95% CI 1.25 to 1.3).

Previous violent victimization and perceived level of gun related crime in the neighborhood were not associated with beliefs about a gun in the home. Findings did not differ by whether or not the violent victimization was committed with a gun, therefore these variables were excluded from the final analysis.

The multinomial logistic regression analysis also showed that being male, having 12 years or fewer of education, and no children living at home was associated with the belief that keeping guns in the home makes it safer. African Americans were more likely to respond that the risks and benefits depend on other factors as compared with whites (OR 1.5; 95% CI 1.1 to 2.0).

Political party affiliation was also strongly and independently associated with beliefs about gun risks and benefits. Compared with Democrats, Republicans (OR 16.6; 95% CI 9.2 to 29.7) and Independents (OR 11.1; 95% CI 6.1 to 20.1) were much more likely to believe they were more safe with a gun in the home.

For most of the independent variables, the odds ratios for survey participants who responded “it depends” were similar in direction, but of lesser magnitude, to those responding “more safe”. For instance, men were more likely than women to respond more safe (OR 2.4; 95% CI 2.1 to 2.9), more likely to respond it depends (OR, 1.6; 95% CI 1.3 to 1.9), and less likely to respond less safe (OR 0.26; 95% CI 0.18 to 0.38).

A sensitivity analysis demonstrated that the findings were substantially unchanged when income was included in the final model. A separate sensitivity analysis showed that the inclusion of handgun ownership in the final model did not affect other estimates. The one exception was the association between age and beliefs about gun risks. With handgun ownership in the model, the youngest and middle aged respondents were 2.0 (95% CI, 1.6 to 2.3) and 1.7 times (95% CI, 1.2 to 2.1) more likely, respectively, to believe that they were more safe with a gun in the home as compared with the oldest respondents.

Discussion

To our knowledge, this is the first study of individuals' beliefs about the effects of keeping a gun in the home on household safety. The large sample size and oversampling of African Americans provided an opportunity to assess the views of a group disproportionately affected by violence with a relatively high level of precision. Although the large US sample enhances the generalizability of our study, the focus on registered voters in large metropolitan areas precludes us from generalizing the findings to persons living in small cities or rural areas, or to individuals who are not registered to vote. Different factors may be associated with these groups' beliefs about the protective nature of firearms in the home. For example, those living in rural areas are more likely to own guns for other reasons besides protection (for example, hunting) and individuals who are not registered to vote are more likely to be poorer and disproportionately affected by crime.

Our inability to include racial groups other than whites and African Americans in multivariate analysis may also limit the generalizability of the findings. The bivariate analyses indicated that, as a group, individuals falling into the “other” racial and ethnic group held views about guns and safety that were similar to those of whites. Nevertheless, there may be important differences among racial and ethnic groups represented in this heterogeneous category.

While the response rate was relatively low, other than somewhat higher reported levels of education, the sample population was similar to the US Bureau of Census information on the national electorate in regard to sex, ethnicity/race, age, and income.⁴⁷ Prevalence of gun ownership among respondents was also similar to estimates from other national surveys.¹⁸

As exposure to violence was limited to two questions that did not distinguish personal victimization from that of family or friends, additional questions about the type of violent victimization, its frequency and location, and the relationship of the perpetrator to the victim may provide a more comprehensive picture of how exposure to violence affects beliefs about the risks associated with keeping a gun in the home.

Only 40% of our study population believes that keeping a gun in the home makes it “less safe”. Those most likely to believe they are safer with a gun in the home include: males, those with at most a high school education, individuals with no children living at home, Republicans, and those with very little or no trust in the police.

Although beliefs about the relative protection and risks provided by guns may indeed reflect risk awareness gaps, the associations may also reflect real differences in risks among the groups being compared. For example, risks associated with keeping guns in the home may be greater for households with children and adolescents. Children's future risk of death may also be affected by parents' attitudes about keeping a gun in the home since gun ownership as an adult is associated with growing up in a household with a firearm.²³ Although epidemiologic studies indicate that, on average, keeping a gun in the home is more perilous than protective,^9,14 this relationship (or at least the strength of it) is likely to vary depending on the characteristics of the household and its members.

The two direct measures for exposure to violence did not predict the belief that it is safer to keep a gun in the home. However, groups at highest risk for violence, young adults and males,^36,37 were more likely than groups at lower risk—older adults and females—to believe that keeping a gun in the home increases safety.

Republican party affiliation and lack of trust in the police to provide protection were the variables most strongly associated with the belief that keeping a gun in the home enhances household safety. As a group, Republicans may be more skeptical of some collective efforts to address social problems, such as crime (particularly if those efforts involve government institutions), and more likely to believe that individuals should be responsible for their own welfare. Therefore, our findings suggest that skepticism about the effectiveness of collective measures to address crime may be closely tied to the belief that one's household is safer with a gun than without one.

Implications for prevention

A variety of methods are currently used to educate people about the potential risks associated with gun ownership including media campaigns, community health education campaigns, and anticipatory guidance by health care providers. Our findings can assist the development and implementation of these efforts by providing insight into the cluster of characteristics associated with the belief that keeping a gun in the home makes it more safe rather than less safe.

Although these findings enhance our understanding of individuals' beliefs about the risks and benefits of keeping guns in the home, additional qualitative research is needed to assess what educational strategies would actually change those attitudes as well as gun acquisition. For example, it is unclear how easy it would be to enhance public confidence in the ability of the police to provide protection against home intrusion. But people who lack confidence in the police may find alternative means of home protection more acceptable if the risks of keeping guns in the home are underscored.

It may be especially advantageous to focus intervention efforts on the group who responded “it depends”. This group may be more receptive to gun safety interventions since they already believe that situations exist in which guns in the home make household members less safe.

Although the patterns of gun ownership⁴⁸ and reasons for gun ownership in other countries (for example, cultural, political) may differ, individuals throughout the world who keep guns in the home are at risk of gun injuries. While the investigation of unique factors associated with beliefs about the protective nature of guns in the home in other countries is needed, this study may provide a basis for additional studies and interventions outside of the US.