Abstract
Previous research has paid little attention to legal firearm demand, instead often focusing on illegal firearm demand. This study expands sociological research on firearms by theoretically identifying and empirically examining a new type of legal firearm demand, status anxiety demand, while also examining recreational and security firearm demand. We use unique background check data from the National Instant Criminal Background Check System to measure firearm demand and test our hypotheses using random effects pooled time-series generalized least squares (GLS) regressions. Findings indicate that our measures for recreational demand and status anxiety demand both affect overall firearm sales, but that actual crime risk and the number of police in a state do not. We find that both increases in National Rifle Association (NRA) membership and Democratic Party strength in the federal government increase firearm demand, suggesting that changes in legal firearm demand are associated with political factors and not just recreational or self-defense motivations.
Introduction
Among Western nations, the United States has the most guns owned by civilians (Cook and Goss 2014). Social scientists have explored how such high firearm prevalence may relate to firearm violence, producing a broad and sometimes conflicting literature (Wellford, Pepper, and Petrie 2004). One concern within this literature is firearm demand. Criminologists have explored why criminals seek out black market firearms or use firearms for illicit purposes (Braga et al. 2002; Braga et al. 2012). Similarly, public health scholars have examined how firearm dealers are able to supply firearms to proscribed buyers (Wintemute 2009, 2010). While this literature has focused on the problem of illegal firearms in the United States, little work has examined the determinants of legal firearm demand.
This discrepancy is important. Demand for legal firearms affects legal firearm prevalence, which in turn is also a source for illegal firearms obtained via theft, straw purchases, or illegal private sales, or borrowed from acquaintances (Braga et al. 2002). In addition, firearms are durable goods that can exist for decades after they are purchased, indeed firearms made 100 years ago may operate as well as newly made firearms. Such longevity means that previous firearm demand can impact overall firearm prevalence well into the future. Understanding contemporary or past changes in firearm demand may illuminate changes in firearm prevalence over time.
To assess firearm demand, we use state-level data from the Federal Bureau of Investigation’s (FBI) National Instant Criminal Background Check System (NICS) that includes the purpose codes for NICS checks conducted at the state level from 1999 to 2010. In doing so, we assess two well-established trends in firearms demand and introduce an argument for a third that we examine here. First, we consider firearm demand as it relates to recreational uses. Second, we examine firearm demand as a security response to crime rates and weakening police strength. Finally, we argue that firearm demand may be an outcome of status anxieties around gun rights. Our findings suggest that increases in recreational demand are associated with increases in firearm demand, but we find no support for a relation between actual crime rates or police strength with firearm demand. Finally, we find that indicators of status anxiety around gun rights are associated with firearm demand. We conclude with policy implications of our findings and suggestions for future research on macro-level firearm ownership and demand.
Macro-level Firearm Demand
This study is concerned with macro-level firearm demand, the causes of which are not mutually exclusive and can be thought of as components of a broader phenomenon of firearm demand. Previous research has identified two forms of macro-level firearm demand: recreational demand and security demand. Here, we also present a new type of macro-level firearm demand: status anxiety demand. Before moving to introduce our arguments for status anxiety demand, we review both recreational demand and security demand. Our discussion of firearm demand only refers to the acquisition of firearms purchased from licensed firearm dealers; it does not encompass why people may buy guns for the first time or how many firearms a person may want or need. By focusing on firearm demand, our study seeks to advance research on the determinants of firearm sales and the firearms market in the United States.
Recreational Demand
The United States has a long history of public access to land and firearms, helping make shooting sports a common recreational activity (Hillyard and Burridge 2012; Kohn 2005; Lizotte and Bordua 1980). Thus, it is unsurprising that growing up and living in rural areas is linked with hunting participation as rural areas offer more opportunities for shooting participation by allowing access to open spaces (Stedman and Heberlein 2001; Young 1986). Likewise, areas with large urban populations reduce these opportunities for shooting sports as access to land and game is much more limited (Kleck 1997; Kohn 2005; Stedman and Heberlein 2001).
Previous research also repeatedly finds that hunting and recreational shooting are strongly associated with firearm ownership and support for gun rights (Brennan, Lizotte, and McDowall 1993; Celinska 2007; Kleck 1996; Kleck, Gertz, and Bratton 2009). People who participate in recreational shooting often buy new firearms to add to their collections and to trade with other collectors and shooters (Taylor 2009). Also, because recreational shooters use different types of firearms for different types of sport shooting, firearm owners participating in sport shooting have larger collections of firearms than those who own guns exclusively for protection (Wyant and Taylor 2007). Thus, greater hunting and recreational shooting participation should increase demand for firearms because having access to a firearm is crucial to participation. The recreational demand explanation leads to the following hypotheses about firearm demand:
Security Demand
The second explanation in the literature suggests that firearm demand increases when the risk of criminal victimization increases, as well as when trust in collective security falters. Perhaps the most well-established discussion about firearms in society within criminology is that exploring the use of firearms for self-defense. 1 Lizotte and Bordua (1980) and Clotfelter (1981) found that higher crime rates were associated with increases in handgun ownership and demand. Kleck and colleagues (2011) also found that both perceptions of risk and actual victimization have positive associations on individual plans to purchase a firearm. Most recently, Carlson’s (2015) work on concealed weapon carriers in Detroit illustrates that many who carry firearms for protection do so because they feel at risk of victimization and seek personal responsibility for their safety. In short, increases in perceived and actual risks of criminal victimization can increase firearm demand.
A related concept to criminal victimization that can affect firearm demand is trust in collective security mechanisms like the police. McDowall and Loftin (1983) found that demand for handgun licenses in Detroit, 1955–1977, increased when rising neighborhood crime rates were combined with a declining police force. Using the Police Services Study, Smith and Uchida (1988) found that respondents’ perceptions of increasing crime and decreasing police effectiveness increase the odds of owning a firearm for self-defense. Similarly, Kleck and Kovandzic (2009) found that handgun ownership is associated with higher rates of homicide in large American cities, but that larger police forces were associated with lower levels of handgun ownership. Finally, Gau (2008) used data on concealed pistol permits in Spokane to demonstrate that fear of crime and distrust in police increase demand for such permits. Thus, previous research makes clear that both weak perceptions of police efficacy to stop crime and lower levels of police strength are associated with increased demand for firearms. This literature leads us to the following hypotheses about macro-level firearm demand:
This study focuses on the concept of actual crime risk as a determinant of firearm demand, as opposed to perceptions of crime risk. We are likewise concerned with actual state capacities for law enforcement, as opposed to perceptions of such capacities. It is possible that heightened perceptions of criminal victimization, as well as weakened views on law enforcement, may increase firearm demand as perceptions of victimization risk are often greater than their actual likelihood. Indeed, advocates of gun rights often promote firearm ownership as a key means to prevent such victimization despite the fact that crime rates have been declining in recent decades (Carlson 2015; Melzer 2009). However, such perceptions of risk vary across race, gender, and even views on politics and race relations (Blair and Hyatt 1995; Ferraro 1996; Kleck et al. 2011; Young 1985). Even though perceived crime risk and perceptions of law enforcement efficacy may affect firearm demand, this study can only test the effect of actual crime and the actual capacity of law enforcement in a state. This is because measures on perceptions of crime, disorder, and law enforcement are not available at the state level. Our measures of actual crime risk and law enforcement capacities thus come from the FBI’s Uniform Crime Reports (UCR), which we discuss in detail below. However, we address in our conclusions below the implications that perceptions of crime and law enforcement have for this study.
Status Anxiety Demand
Finally, we present a novel form of macro-level firearm demand: status anxieties around gun rights. Status anxieties arise when a group (be it a race, class, social movement, or a political identity) fears that its values, power, and/or status in society are declining at the same time as competing groups’ values, power, and/or status are on the rise (Gusfield 1963; Zurcher and Kirkpatrick 1976). Previous research demonstrates that such culture conflicts and status anxieties exist between gun rights advocates and gun control advocates (Kleck 1996; Kleck et al. 2009; Melzer 2009). The ideological, cultural, and political divisions between these groups foster status anxieties as one group gains political ground at the expense of the other.
Many scholars recognize that gun ownership, use, and demand are related to locality, culture, and ideology (Brennan et al. 1993; Burbick 2006; Celinska 2007; Kleck 1996; Kleck et al. 2009; Kohn 2005; Melzer 2009; O’Connor and Lizotte 1978; Wright and Marston 1975). The academic consensus is that gun ownership is most often associated with southern, white, protestant, heterosexual, males living in rural and suburban areas. Wright and Marston (1975) were the first to empirically demonstrate that gun ownership is associated with what they called traditional “American Way” values of rugged individualism, traditional masculinity, and frontier myths (p. 99). More recent work has confirmed that supporters of gun rights and ownership broadly hold conservative ideologies on issues, including religion, sexuality, criminal justice, and the welfare state (DeZee 1983; Gahman 2015; Melzer 2009; Micklethwait and Wooldridge 2005; Stroud 2012; Young and Thompson 1995). Many gun rights advocates view themselves as the guardians of conservative values on marriage, small government, heterosexuality, and traditional masculinities, while also being skeptical of the federal government (Jiobu and Curry 2001; Kohn 2005; Lio, Melzer, and Reese 2008; Melzer 2009; Stroud 2012). While not all gun rights advocates or owners are staunch conservatives, conservative movements in the United States have often relied on them as a vanguard for their causes (Berlet and Lyons 2000; Blee and Creasap 2010; Micklethwait and Wooldridge 2005).
In contrast, supporters of gun control are broadly associated with liberal political ideologies, urban residence, and nonwhite populations (Carter 1997; Goss 2008). Kaplan (1979) argued gun control advocates frame gun control as a symbolic crusade, similar to prohibition and the war on drugs. Kaplan (1979) also stated that gun control advocates are often hostile to the values of gun rights supporters and promote gun control laws as a means to “stigmatize” gun ownership (p. 6). Supporting Kaplan’s arguments, Kleck and associates (Kleck 1996; Kleck et al. 2009) found support for gun control is strongest within groups hostile to the concept of gun ownership itself and tend to hold negative stereotypes of gun owners as “racist, violent, politically reactionary, ignorant ‘hicks’”(Kleck et al. 2009:498). Conversely, gun rights advocates also hold negative views of gun control advocates. Melzer (2009) found that gun rights supporters view gun control advocates as “ ‘anti-gunners,’ revisionists, socialists, and feminists” who are attacking “gun rights, individual rights, heritage, [and] freedom” (p. 168). This conflict of values and symbolism around firearms catalyzes and heightens status anxieties. 2
Status anxieties can mobilize social movements by way of a threat response (Meyer and Staggenborg 1996; Van Dyke and Soule 2002). Threat occurs when changes in a population’s political, economic, demographic, or cultural characteristics provoke status anxieties among a group within the population, which in turn mobilizes the threatened group to minimize their status loss. For example, Van Dyke and Soule (2002) found that patriot-movement militias mobilized where both increases in minority group political power and majority group economic declines occurred. McVeigh and Maria-Elena (2009) found that the decline of traditional gender roles and relationships influenced people opposed to same-sex marriage to vote for bans against same-sex marriage. Similarly, Crockett and Kane (2012) found the proliferation of pro-LGBT religious congregations and increasing pro-LGBT public opinion fostered the creation of ex-gay religious “therapy centers.” In all of these studies, groups experiencing status loss responded by mobilizing to reduce further losses. Likewise, we anticipate that the mobilization of gun rights advocates may be a product of status anxiety, which may increase firearm demand.
Indeed, the National Rifle Association (NRA), the largest gun rights social movement organization in the United States, regularly uses rhetorical strategies in its publications and media communications that portray conservative values as under siege from progressive liberals. Many people join the NRA because they take this status struggle seriously (Melzer 2009). The ability to legally buy firearms is a fundamental right for gun rights advocates, as the NRA (2015) stated, it “is committed to preserving the right of all law-abiding individuals to purchase, possess and use firearms for legitimate purposes” (emphasis added). Thus, we would anticipate that the mobilization of NRA members may increase firearm sales as people seek to reaffirm and protect their status as gun rights advocates.
Partisan politics may also foster status anxieties around gun rights. Since 1972, the Republican Party platform has included a gun rights plank stating the need to “safeguard the right of responsible citizens to collect, own and use firearms for legitimate purposes, including hunting, target shooting and self-defense” (The American Presidency Project 2015b). The Democratic Party platform has had a gun control plank since 1968, stating the need to “promote the passage and enforcement of effective federal, state and local gun control legislation” (The American Presidency Project 2015a). It is not surprising then that Democratic voters are more likely to support gun control and not own guns (Gewurz 2013). With this history of partisanship, gains by the Democratic Party are often framed by gun rights advocates as declines to the status of gun rights and gun ownership (Goss 2008; Melzer 2009). This relationship between partisan politics and status anxieties around gun rights leads us to make the following hypothesis:
Data
Dependent Variable: Firearm Demand
Our dependent variable is a measure of firearm demand constructed as the natural log of the annual number of background checks conducted for firearm purchases per 100,000 residents in a state (1999–2010). We use the natural log to improve the linearity of the variable. Our background check data come from a Freedom of Information Act (FOIA) request to the FBI requesting state-level data on the number of background checks from the National Instant Background Check System (NICS) for 1999–2010. The primary purpose of the NICS is to enforce the Brady Law, which mandates that Federal Firearms Licensees (FFLs) conduct a background check pursuant to a firearm sale (Public Law 103-159). Many states also use the NICS for regular checks on concealed firearm permit holders, which is why we obtained the purpose codes with our FOIA request to differentiate between NICS checks used for firearm purchases and other uses.
To illustrate our data, consider the following example. Imagine an individual decides to purchase a firearm. They could legally obtain a firearm in two ways. First, they may seek a private sale, meaning they find another person who already owns a gun to sell them. In order for such a transaction to qualify as a private sale, the private seller of the firearm must not sell firearms for a profitable business (this often means they are hobbyists). Such private sales typically do not require an NICS background check under federal law, but there are exceptions. 3 Second, the person wishing to buy a firearm may visit a FFL’s store, where the buyer would undergo an NICS check (FBI 2015). An important consideration in going to an FFL store is that only FFLs can sell newly manufactured firearms, as firearm manufacturers are only allowed to distribute their products to FFLs (Braga et al. 2002).
The total proportion of private and FFL sales is unknown. Despite reports and statements from politicians claiming that “40 percent of all gun purchases are conducted without a background check “ (Kessler 2013a; The White House 2013), this means that a majority of all firearm sales would still be captured in the NICS system. Yet, the 40 percent statistic itself is problematic. Only one study has examined this question and the circumstances around its timing may be problematic (Cook and Ludwig 1996). First, the 40 percent statistic uses the top of the confidence interval for private gun sales; Cook and Ludwig stated the interval to be 30 and 40 percent (Cook and Ludwig 1996:27; Kessler 2013b). Second, the study was conducted in November and December of 1994, a cathartic moment when the Federal Assault Weapons Ban and the Brady Bill were being implemented and firearms sales were spiking (Koper 2013; Koper and Roth 2002). As background checks in 1994 could take days, gun buyers may have sought private sales to buy firearms quickly before prices increased or because stock had sold out. When the NICS system was implemented, it effectively ended long waiting periods for background checks, potentially reducing the attractiveness of private sales for gun buyers since then. Thus, the 30 to 40 percent estimate of private sales for firearms is suspect and likely lower for the time period of our study.
There are caveats to our use of the NICS data. Ideally, assessing firearm demand would occur at the individual or household level. However, NICS data are only available as state-level aggregated counts. This limits our analyses to state-level trends and precludes us from assessing city or neighborhood dynamics of firearm demand as well as individual perceptions of crime risk. In addition, though NICS checks capture a majority of firearm sales, they do not capture private sales of firearms in the United States. It is also likely that the motivations to buy firearms from a private seller, rather than an FFL, are systematically different. 4 Unfortunately, an assessment of individual motivations is not feasible as private sales data do not exist. Finally, we use NICS data to measure changes in demand for all firearms rather than assessing demand for specific types of firearms (i.e., handguns or long guns). Previous research suggests that our proposed theories of firearm demand may be more applicable to handguns in the case of security demand (Gau 2008; McDowall and Loftin 1983) and to long guns in the case of recreational demand (Wright and Marston 1975; Wyant and Taylor 2007). We opt for a broad definition of firearm demand to include all firearm types because previous research looking at personal and household firearm ownership finds that stated motivations to own a firearm (either for recreation or protection) do not always discretely determine the type of firearms one owns or buys (Lizotte, Bordua, and White 1981; Wyant and Taylor 2007). Given our aggregated level of analysis, we opt to conceptualize firearm demand as conservatively as possible, defining it as NICS checks for all types of firearms.
Independent Variables
To examine our hypotheses, we use a number of independent variables to capture the concepts theoretically linked to firearm demand. All measures described below cover the years 1999–2010. All of our demographic data were obtained using U.S. Census Bridged Race Data and use state populations over the age of 15.
First, to account for factors theoretically associated with recreational firearm demand, we use measures of hunting participation and urbanization within a state. Hunting participation is argued to capture changes in recreation demand because of the use of firearms in hunting. To construct this measure, we use the percentage of a state’s population issued a hunting license. Data on hunting licenses were obtained from the U.S. Fish and Wildlife Service, Historical License Data Web page. Urbanization is argued to capture recreational demand because it indicates a lack of opportunity for recreational firearm use and is negatively associated with hunting and sport shooting participation (Kleck 1997; Kohn 2005; Stedman and Heberlein 2001). To construct our measure of urbanization, we used the percentage of a state’s population living in urban areas as defined by the U.S. Decennial Census in 1990, 2000, and 2010, and interpolated the years between these time points.
The next set of variables measures factors associated with security demand for firearms. Crime is measured as the annual homicide rate per 100,000 people in a state. To construct our measure of crime, we use data on murders and nonnegligent manslaughters from the FBI’s UCR. Police strength is measured as the ratio of sworn law enforcement officers per 100,000 residents in a state. Data on sworn officers were obtained from the UCR Law Enforcement Officers Killed and Assaulted reports. Finally, although we make no hypothesis in this regard, the legality of firearms for self-defense may impact firearm demand. So, we include a dichotomous indicator for concealed carry weapons (CCW) laws. CCW laws specify the criteria for the issuance of CCW licenses in a state and fall into four categories. First, states can prohibit concealed firearms altogether. Second, states may allow local authorities discretion to issue CCW licenses as they see fit, often referred to as “may-issue” laws. Third, states may have “shall-issue” laws that mandate local authorities to issue CCW licenses to any applicants qualified to have one. Fourth, some states may even allow CCW without a license at all. State-years with a “shall-issue” or license-free CCW law in effect were coded as “1” while state-years with more restricted CCW laws were coded as “0.” Data on CCW laws came from the authors’ own legal research (Steidley, forthcoming).
Finally, we include variables to indicate changes theoretically linked to status anxieties around firearms. First, following the argument that gun rights mobilization might affect firearm demand, we use a measure of gun rights mobilization as the percent change in state-level NRA membership from the previous year. Data for NRA membership come from a proxy of NRA membership: the number of subscriptions to NRA magazines as reported by the Alliance for Audited Media. These magazines come with an NRA membership and include the titles American Rifleman, American Hunter, and America’s 1st Freedom. To construct our measure of gun rights mobilization, we first calculated the number of NRA members per 100,000 residents and then calculated the percent change in NRA membership from the previous year. We argue that this measure accurately captures rising or falling status anxieties around firearms because it shows the relative growth and decline of gun rights mobilization in a state. 5
Next, we use three variables to capture partisan political factors that may influence status anxieties around gun rights in the form of Democratic Party power. The first is the percentage of state residents who identify as Democratic Party voters. Our data on Democratic voters come from Enns and Koch’s (2013a, 2013b) work on state-level public opinion. Enns and Koch constructed this measure by using multiple national surveys asking respondents about their political views, which were weighted using demographic data (e.g., race, age, education) for respondents’ states to calculate the percentage of a state that self-identified as Democrat voters (for details on their method, see Enns and Koch 2013a). Second, we use Enns’s (2014) measure of Democratic Party strength in state government, a continuous measure of Democratic strength in a state’s legislative chambers and gubernatorial office. This measure is coded on an additive scale where Democratic governors are coded as 1 (or 0 if the governor is not a Democrat), with the proportion of Democratic seats in the upper and lower state legislative chambers added to this measure. The scale has a maximum of 3 (Democratic governor and both state legislative chambers held entirely by the Democratic Party) and a minimum of 0 (Democrats are completely absent from elected state government). To construct the scale, we used Klarner’s State Partisan Balance Data (Klarner 2013). Klarner’s measures do not include data on Nebraska’s unicameral nonpartisan senate, so we used the Nebraska Blue Book to identify party affiliations of Nebraska legislators and multiplied the proportion of Democrats in the unicameral senate by two (Nebraska Library Commission 2017). Third, we use a measure of Democratic Party strength in Federal Government. This measure is identical to the state Democratic Party strength variable but captures Democratic Party strength in both houses of Congress and the Presidency. Finally, given that some states regulate firearm ownership more rigorously than others, it is possible that status anxieties may be tempered by existing regulations. Although we make no formal hypothesis, we feel it necessary to include such a measure. So we use a dichotomous indicator for state-years where a firearms registry for any or all types of firearms is enforced. We use registration laws because firearm registration is a common grievance for gun rights advocates (Goss 2008; Melzer 2009). Data on state firearm registration laws were collected in the same manner as the CCW laws.
We also include a number of control variables that previous research suggests are related to firearm demand. First, we include a proxy of firearm prevalence in a state by using the commonly utilized measure of the percentage of firearm suicides out of all suicides in a state (firearm suicides/all suicides). This measure is often employed in research on firearm prevalence, proving to be a well-validated proxy of gun ownership in a state (Cook and Ludwig 2006b; Siegel, Ross, and King 2013). Our data on state mortality come from the National Center for Health Statistics Compressed Mortality Files, which use death certificate data to establish causes of mortality in a state. Second, we use a measure of the percentage of a state’s residents who are white non-Hispanic using data from the U.S. Census Bridged Race Estimates. We also included Protestant adherents, measured as the ratio of Protestant faith adherents per 100,000 residents in a state. Religion data were obtained from the 2000 and 2010 Religious Congregations and Membership Studies from the Association of Religion Data Archives. As religion data are only available for years 2000 and 2010, we linearly interpolated observations for the years between and extrapolated for the year 1999. Next, we include controls for the states’ regions with dichotomous indicators for states in the South and West (as defined by U.S. Census with “1” meaning yes and “0” meaning no for each region); we use the Northeast and the Midwest States as a reference category. Finally, we included a temporal control by including the year of each observation in our data. 6
Analytic Strategy
Our data are comprised of 600 state-year observations for all 50 states from 1999 to 2010. As our data are multiple time series, we use a pooled time-series generalized least squares (GLS) regression analysis with random effects using the xtreg command in Stata 14. A random effects term used is to exploit serial correlation in the composite error term inherent in multiple time series. Because observations are clustered on states, we use adjusted standard errors using the Huber/White formula provided in Stata’s cluster option (Rogers 1993; Williams 2000). This assumes that observations are independent across states, but not necessarily within states, and produce more conservative estimates of standard errors (Wooldridge 2010). Given that we are conducting multiple hypothesis tests with a limited sample size, we also use Holm-Bonferroni sequential corrections for conservative alpha levels which reduce the likelihood of committing Type I error (Holm 1979).We use three models to test our main hypotheses regarding firearm demand. Our first model examines our hypotheses for recreational firearm demand. The second model examines our hypotheses for security firearm demand. The third model examines our hypotheses for status anxiety firearm demand. We then use a final “full” model to test the robustness of significant findings from the first three models net of each other.
Findings
Table 1 reports summary statistics and measurement details for the variables in our dataset. As the strength of our data is that it exploits longitudinal variance in firearm demand, we present a timeline of NICS checks per 100,000 people for the entire United States in Figure 1. To illustrate the potential argument for a status anxiety firearm demand, Figure 1 depicts time periods where Democratic politicians controlled the presidency and/or a federal legislative chamber. Overall, firearm demand declined from 1999 to 2003. However, 1999 and 2000 were still years where firearm demand was relatively high. Those were also years when the Clinton administration and Al Gore’s candidacy for president were portrayed by the NRA as a threat to gun rights. It may also be the case that popular concern about the Y2K bug and the Columbine shooting increased firearm demand in 1999, but this is difficult to verify without data from before 1999. Firearm demand appears to have stayed the same from 2000 to 2001, as the September 11th terror attacks kept firearm demand high (Aisch and Keller 2015). From 2002 to 2006, firearm demand was relatively low when Republicans controlled the presidency and both legislative chambers in the federal government. Once the Federal Assault Weapons Ban expired in 2004, we see that firearm demand begins to pick up again. However, it is not until 2007, when Democrats take control of the House of Representatives, that firearm demand begins to make a sharp increase and again return to relatively high levels. These findings offer some preliminary support for our arguments of status anxiety firearm demand and partisan politics in the United States, but to account for the influence of recreational and security demand, we must now turn to our regression analyses.
Summary Statistics for Dataset.
Note. Data sources for all variables are described in text. NICS = National Instant Criminal Background Check System; CCW = concealed carry weapons; NRA = National Rifle Association.
Dichotomous indicator (“1” = yes, “0” = no).
National Firearm Demand, 1999–2010.
Results from the random effects GLS models with robust standard errors are reported in Table 2. A benefit of having a logged dependent variable for firearm demand is that we can interpret the effects of the reported coefficients in terms of percent change in firearm demand. 7 Model 1 examines recreational firearm demand, and we find that both hunting participation and urbanization are significantly associated with firearm demand. Findings in Model 1 indicate that a 1 percent increase in the population issued hunting licenses is associated with a 4 percent increase in firearm demand. Similarly, a 1 percent increase in the percentage of a state’s population living in urban areas is associated with about a 2 percent decline in firearm demand.
Results from Pooled Time-Series Generlized Least Squares Regressions with Random Effects.
Note. Robust standard errors are reported in parentheses; all hypotheses are tested using Holm-Bonferroni adjusted p values. CCW = concealed carry weapons.
p < .05. **p < .01. ***p < .001.
Model 2 examines the association between actual crime risks and firearm demand by incorporating variables for police strength, crime rates, and the CCW law status of a state. We find that there is no significant association between police strength, crime rates, or CCW laws in a state and firearm demand. Thus, our results indicate that security demand is not a reliable predictor of state-level firearm demand.
Model 3 examines indicators of status anxiety firearm demand by introducing variables for Democratic Party power, gun rights mobilization, and firearm registration. Model 3 finds that state-level Democratic Party strength in government and the percentage of Democratic voters are not significantly associated with firearm demand. We find a sizable, positive, and highly significant association between Democratic Party presence in the Federal government and firearm demand. A one unit increase in Democratic Party power in the federal government (tantamount to a Democratic president or a completely Democratic House of Congress) is associated with a nearly 8 percent increase in firearm demand. We also find that gun rights mobilization is positively associated with firearm demand, with a 1 standard deviation increase in annual NRA membership corresponding with a 2 percent increase in firearm demand. Finally, we find that it was prudent to account for state regulations of firearms, as our indicator of state firearm registries shows that states with registries had significantly lower rates of firearm demand, about a 53 percent reduction, compared with states without such registration.
Model 4 offers a robustness test for findings in Model 1 and Model 3. We excluded insignificant variables from Model 2 and Model 3 as Wald tests indicated that these variables did not significantly improve fit in a “full” model. Model 4 confirms the findings for associations found in Model 1 and Model 3. Hunting participation and Democratic Party power in the federal government are both positively and significantly associated with firearm demand. Urbanization and state firearm registries are both negatively and significantly associated with firearm demand. Model 4 finds that gun rights mobilization is statistically significant and positively associated with firearm demand. While we conducted exploratory analyses (not reported here) examining the potential for interactions between state NRA membership and Democratic Party strength in state or federal government, no such interaction was found to be statistically significant.
We now briefly touch on our control variables. The firearm prevalence proxy is positively associated with firearm demand but is only significant when testing recreational and security firearm demand. This finding suggests that state firearm prevalence does not affect firearm demand once status anxiety is considered and that the percentage of whites in a state has a significant association with firearm demand only when examining security demand. Protestant adherence in a state is significantly associated with firearms demand in Model 2 and Model 3. States in the Southern and Western regions all have positive associations with firearm demand compared with the Northeast, but this association is only statistically significant with regard to the Western region. According to Model 4, firearm demand in Western states is 75 percent higher than in Northeastern and Midwestern States.
Discussion and Conclusion
This research makes two contributions to the research literature on firearms in the United States. First, we drew upon previous research to examine how recreational demand and security demand affect overall firearm demand. Second, we present an argument for what we call status anxiety demand, and then present analyses that demonstrate how status anxiety demand can predict the influence of partisan politics and gun rights mobilization on firearm demand. This work also makes a contribution by examining dynamics of legal firearm demand, a relatively understudied area of firearms research when compared with illegal firearm demand.
How do our hypotheses fare? We find support for our hypotheses that firearm demand is associated with hunting participation and urbanization. This finding speaks to the fact that the United States is a nation with a large number of hunters and sport shooters; thus, recreational firearm use is important for legal firearm demand. We do want to note though that recreational demand is likely broader than our measure of hunting participation, new shooting sports are emerging (e.g., cowboy action shooting and three-gun competitions; Kohn 2005), and gun collecting can be a hobby without actually shooting a gun (Taylor 2009). It remains to be seen how such activities change macro-level firearm demand, as these newer shooting sports attract shooters outside the white, male, and rural demographics. If recent news stories are a reliable indication that more women are beginning to participate in handgun shooting sports (e.g. Goode 2013), then future research on recreational firearm demand would do well to explore recreational firearm ownership and demand outside the realm of hunting.
Our hypotheses for security demand did not hold. We found no association between police strength and firearm demand in a state. We likewise found no association between state-level crime rates and firearm demand. There was also no association between CCW law status in a state and firearm demand. This finding is contrary to several previous studies, perhaps because security demand is likely a city, neighborhood, and individual-level phenomenon that is not suitable to be tested with state-level data (Gau 2008; Kleck and Kovandzic 2009; McDowall and Loftin 1983; Smith and Uchida 1988). This study is also limited in its analysis of security demand by the use of UCR data to measure actual crime risk and police presence as we are unable to account for perceptions of crime risk and law enforcement efficacy, unlike previous studies that assessed individual perceptions of crime (Gau 2008; Kleck et al. 2011; Young 1985). We use actual crime and police strength data from the UCR out of necessity as no state-level data on perceptions of criminal risk and law enforcement efficacy exist. Such findings lead us to conclude that, at the state level, actual crime rates and actual levels of police strength do not change firearm demand. Any future research on state-level firearm demand needs to assess perceptions of crime and law enforcement efficacy in conjunction with actual crime rates and police strength to confirm our findings.
Our final hypotheses sought to test our arguments for status anxiety firearm demand. First, due to highly partisan political views on gun rights and gun control in the United States, we anticipated that the number of Democratic voters in a state and the presence of Democratic officials in both state and federal government would increase status anxieties among gun rights supporters. We also hypothesized that the mobilization of gun rights supporters, measured as a growing NRA membership, would reflect increases in status anxieties as well. Our results indicate that Democratic officials in federal government did increase demand for firearms at the state level, but state-level Democratic officials and Democratic voters did not. These findings suggest that federal Democratic officials are more likely to create status anxieties around gun rights than state-level Democratic officials. This makes sense, as gun rights advocates regularly frame national Democratic leaders (such as Barack Obama, Bill and Hillary Clinton, and Al Gore) as more threatening actors than state-level politicians (Melzer 2009). We also find that yearly increases in state NRA membership are positively associated with increases in firearm demand, suggesting that the mobilization of gun rights supporters increases firearm demand. These findings make a new contribution to firearms research, as we demonstrate that changes in political factors and gun rights mobilization can impact firearm demand in a state. Our findings for status anxiety demand also contribute to a growing literature in social movements on the concept of threat (Crockett and Kane 2012; McVeigh and Maria-Elena 2009; Van Dyke and Soule 2002), as we find that factors related to status anxiety may influence both movement mobilization and actions meant to address status anxieties.
As is always the case in social science research, limitations in the current study must be acknowledged. First, our measure of firearm demand is a necessarily conservative one. We define firearm demand as all NICS checks conducted pursuant to a firearm sale, which does not include all private sales. Also, our measure of firearm demand does not capture changes in overall firearm ownership, and we cannot speak to who is buying firearms in the United States with these data. However, our indicator of firearm prevalence was not significant for models testing status anxiety arguments, suggesting that there is not an association between existing levels of gun ownership and firearm demand when status anxiety is taken into account. Finally, we want to make clear that this study is not meant to assess individual motivations to buy a firearm. As the NICS data are limited to the state level, our study is limited to considering changes in firearm demand at an aggregated level.
Some limitations of this study also highlight new directions for firearms research. While we demonstrate that partisan politics and gun rights mobilization impact firearm demand at the state level, more research is needed to understand how these factors might be related to perceptions of disorder and crime (Beckett 1999; Carlson 2015). If minorities are perceived as criminogenic, or minorities are perceived to be political and/or economic threats, then firearms might be purchased for “security.” However, such demand may also be thought of as a form of status anxiety. Our data cannot speak to these nuanced possibilities, although we demonstrate that status anxiety firearm demand can exist separate from actual crime risks. Specifically, we feel there is fruitful work in exploring the theoretical linkages between status anxieties and perceptions of crime risk and their impact on firearm demand.
Another implication for research is that firearm demand may have long-term implications for firearm prevalence by virtue of firearms being durable goods. Currently, it is unknown how a one-time spike in legal firearm sales may influence firearm prevalence or the rate at which people become first-time firearm owners. Future work should find fertile ground exploring such questions. Namely, we think is it crucial to understand how rates of firearm demand translate into rates of first-time firearm owners, and how periodic surges of new firearms into the civilian firearm stockpile may affect overal firearm prevalence over time.
This research also has an important policy implication. While it is no surprise that gun rights and gun control are politically and ideologically divisive, such divisions have not been considered as a source for firearm demand itself. Our findings suggest that policymakers seeking to implement gun control policies must consider the purpose of such policies and how supporters of gun rights will view them, lest status anxieties create (unintended) new demand. Public health researchers have argued that to reduce firearms violence, new gun control laws and policies in the United States should be implemented to reduce overall firearm prevalence (Hemenway 2006). However, proposing and implementing such policies create political tensions and culture conflicts that exacerbate status anxieties around firearms. Indeed, gun rights advocates often capitalize on modest gun control proposals as a slippery slope to total gun confiscation (Goss 2004). If the goal of a new gun control law or policy is to reduce firearms prevalence, then policymakers and public health advocates must recognize that firearm sales will increase if gun rights supporters feel threatened by the policy. Thus, a policy meant to reduce firearm prevalence which also provokes gun rights opposition will create a spike in firearm demand that may render subsequent reductions in firearm prevalence moot. Policymakers and public health advocates concerned about gun control must recognize that status anxiety firearm demand has the potential to have firearm prevalence reduction efforts backfire into increased firearm demand.
Footnotes
Acknowledgements
The authors would like to thank our faculty and graduate student colleagues in the Department of Sociology at Ohio State University (OSU) for feedback on earlier drafts, the Criminal Justice Research Center at OSU for funding to present this work at the 2012 American Society of Criminology Annual Meeting, and our undergraduate research assistants—Thomas Bailey, Rachel DeLucia, Kevin Knipe, and Rachel Thompson—for their help during data collection.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.