Legal cynicism and protective gun ownership among active offenders in Chicago

Abstract Most American gun owners report having their firearms for protection. However, these national estimates are likely to undersample residents of marginalized urban communities where rates of violent victimization, and presumably the need for personal protection, are more pronounced. Further, this undersampling limits our understanding of motivations for gun ownership within the “hidden” group of active criminal offenders that are more likely to be both victims and offenders of street crime. Drawing on past work linking neighborhood violence to legal cynicism, and using data gathered by the Chicago Gun Project (CGP), I employ measures of police legitimacy to explore the effect of distrust of legal agents on protective gun ownership among active offenders in Chicago. These data confirm that lower levels of police legitimacy are significantly related to a higher probability of acquiring a firearm for protection. I consider the ways that gang membership, legal changes in Chicago, and gun behaviors are related to protective gun ownership, as well as how community policing and procedural justice can improve perceptions of police and enhance their legitimacy, potentially reducing the incentives to engage in violent, extralegal “self-help” with a firearm.

ABOUT THE AUTHOR Michael Sierra-Arévalo is a PhD candidate in the department of sociology at Yale University, as well as an affiliate fellow at the Institution for Social and Policy Studies (ISPS) and a graduate student affiliate of the Center for Research on Inequalities and the Life Course (CIQLE). His dissertation research focuses on how police officers' perceptions of danger affect their behavior in multiple cities in the U.S. His research interests also include gangs, gun violence, social networks, and criminal justice policy.

PUBLIC INTEREST STATEMENT
As mass shootings and gun violence persist in the United States, it should come as no surprise that most American gun owners own their weapons for a simple reason: protection. However, these estimates are likely to undersample residents of marginalized urban communities where rates of violent victimization, and presumably the need for personal protection, are much higher than the country as a whole. Understanding the motivations for gun ownership in these highviolence communities is important for helping address the persistent issue of gun violence in these areas. Using a survey of active offenders in high-violence neighborhoods in Chicago, this research shows that more negative perceptions of police are significantly related to a higher probability of acquiring a firearm for protection, suggesting that improving community perceptions of police and enhancing police legitimacy in marginalized communities might reduce the perceived need to have a firearm for protection.

Introduction
As mass shootings continue to occur across the breadth of the United States-from Newtown, CT to Aurora, CO, Roseburg, OR to Orlando, FL-concerns for safety are becoming increasingly important for those choosing to arm themselves: in 2013, 48% of American gun owners reported owning their weapon(s) for protection, a 22 percentage point increase from 1999 (Cook & Goss, 2014;Dimock, Doherty, & Christian, 2013). With media headlines trumpeting the death tolls at schools, movie theaters, and churches, it is understandable that a public faced with the specter of unpredictable tragedy may turn to firearms as a way to provide safety for themselves and their loved ones.
But while protection is an increasingly common reason for owning a firearm, protective gun ownership is not distributed equally among the American public. In particular, protective gun ownership varies significantly by race, with 45% of white Americans citing protection as the primary reason for owning a firearm, compared to 71% of blacks (Dimock et al., 2013). Clearly, while protection is a common reason for gun ownership overall, it is a more salient motivation for black gun owners.
This racial difference in the prevalence of protective gun ownership is potentially related to disparities in the risk of victimization faced by different communities. With racial minorities being differentially exposed to neighborhoods characterized by structural conditions strongly related to violence (Massey & Denton, 1993;Peterson & Krivo, 2005;Sampson & Wilson, 1995), it is unsurprising that blacks, particularly young black males, find themselves at a significantly higher risk of violent victimization than whites (Fox & Zawitz, 2007;Harper, Lynch, Burris, & Davey Smith, 2007;Heron, 2012). 1 The unique context of poor, violent, and highly-policed neighborhoods has profound implications for the processes that motivate gun ownership. In neighborhoods characterized by high violence rates, many residents exhibit a unique, culturally-defined orientation known as legal cynicism, a frame in which the criminal justice system and its agents (e.g. police and the courts) are viewed as "illegitimate, unresponsive, and ill equipped" to ensure the safety of community residents (Kirk & Matsuda, 2011, p. 444;Kirk & Papachristos, 2011). Where police responses do little to resolve crime and violence (provided the police respond to calls for assistance at all), residents are often left with no legal recourse. The lack of legal avenues for addressing criminal victimization presents a likely motivation for extralegal remedies such as acquiring a protective firearm. As a form of "self-help" to prevent victimization (Black, 1983;Watkins, Huebner, & Decker, 2008), residents of marginalized inner-city communities may own firearms to ensure the protection they feel police cannot or will not provide.
Unfortunately, the so called high-risk gun owners (Lizotte, Tesoriero, Thornberry, & Krohn, 1994) of these marginalized neighborhoods that own their guns illegally make up the kind of "hidden population" that is unlikely to be captured in national-level surveys (Pettit, 2012;Watters & Biernacki, 1989). This exclusion is all the more likely when considering the small number of active criminal offenders within these communities. This exclusion is all the more likely when considering the small number of criminally active individuals within these communities that use firearms in the commission of crimes. While previous work has attempted to overcome this methodological hurdle by sampling from populations of felons (Wright & Rossi, 1986), juvenile offenders (Sheley & Wright, 1995), institutionalized youth (Watkins et al., 2008), and high-risk urban youth (Bjerregaard & Lizotte, 1995;Curry, Decker, & Egley, 2002;Lizotte, Krohn, Howell, Tobin, & Howard, 2000;Lizotte et al., 1994), these studies do not shed much light on the motivations for gun ownership among active criminal offenders who are on the street, continuing to be exposed to criminogenic neighborhood conditions.
To address the likely exclusion of this hidden population from samples used in previous work, I use survey data on 141 active offenders in two high-crime, high-violence Chicago neighborhoods to examine how legal cynicism influences the probability that the gun they most recently possessed was acquired for protection. The sample is drawn randomly from individuals aged 17 years or older, many of whom were past or current gang members, who had at least one prior arrest for a violent crime (e.g. robbery, assault, battery, etc.), and were within the first six months of release to either probation or parole at the time of data collection. As such, the sample is targeted at those who were very recently engaged in violent crime, the "active" nature of these individuals' offending highlighted by the fact that, since the time of data collection, approximately half of the sample has recidivated and returned to prison. 2 Thus, this study represents an advancement over previous work because it concentrates on a criminally-active subset of the population likely to be victims and perpetrators of gun violence and, presumably, to have all the more reason to own a firearm for protection (Braga, Hureau, & Winship, 2008;Kennedy, 1997;Papachristos, Braga, & Hureau, 2012).
Additionally, this analysis focuses on the relationship between protective gun ownership and a specific component of legal cynicism: perceptions of police. While legal cynicism encompasses a collection of attitudes and perceptions, including those towards judges, prosecutors, and the law more generally, I posit that understanding how individuals perceive the most visible and most contacted agents of the legal system-the police-is of particular import when considering behaviors that might be motivated by concerns for personal safety at the street-level. Though legal cynicism includes attitudes towards other sections of the legal system, it is not judges or prosecutors that are tasked with providing public safety; as such, using perceptions of police as an indicator of the broader construct of legal cynicism makes sense for examining legal cynicism's relationship to a behavior like protective gun ownership.
Leveraging the highly focused sample collected by the Chicago Gun Project (CGP), my analysis shows that among respondents that report having ever possessed a firearm, over 80 percent protection as the reason for acquiring their most recent firearm. Further, results from a series of logistic regression models show that criminal offenders who have more negative perceptions of police are significantly more likely to have ever possessed a firearm, and among criminal offenders who have ever possessed a firearm, those with more negative perceptions of police are significantly more likely to have acquired their most recent firearm for protection. These effects are present even when accounting for the plausible counter-explanation that protective gun ownership is explained by prior victimization. In light of these findings, I consider the role strategies like community policing and procedural justice techniques stand to play in violence reduction efforts. Beyond improving policecommunity relations and enhancing police legitimacy, I argue that these strategies can ameliorate legal cynicism, and enhance the damaged legitimacy of police that this analysis demonstrates is strongly associated with protective firearm ownership.

Gun ownership
Gun ownership varies significantly across the population. For instance, men are approximately three times more likely (37 vs. 12%) to personally own a firearm than women. There are also differences in gun ownership by age, with 16% of adults under 30 reporting personal gun ownership, compared 27% of those over 30 (Dimock et al., 2013). One of the more dramatic variations in gun ownership is by race: data from the 2012 General Social Survey (GSS) indicates that 85% of those who personally own a firearm are white. A 2013 survey by the Pew Research Center also finds a racial difference in gun ownership, with 25% of whites vs. 16% of blacks reporting personally owning a firearm.
Just as gun ownership is differentially distributed across the population, so too are the reasons for gun ownership. Since 1999 there has been a marked increase in the percentage of gun owners who say they own their firearm(s) primarily for protection. Forty-eight percent of gun owners cite protection as the reason for owning a gun, a 22 percentage point increase from the late 90s (Dimock et al., 2013). The racial disparities seen in general gun ownership can also be seen when looking at protective gun ownership: 45% of white gun owners own their weapon for protection, compared to 71% of black gun owners. Importantly, gun owners and their guns concentrate in particular places. Knowing that violent crime and gun violence concentrate on city streets and not suburban cul-de-sacs or rural towns (Braga, Papachristos, & Hureau, 2009;Branas, Nance, Elliott, Richmond, & Schwab, 2004;Weisburd, Bushway, Lum, & Yang, 2004;Weisburd, Groff, & Yang, 2012), it is unsurprising that disparities in general gun ownership and protective ownership persist when examining urban areas, specifically. Figure 1 compares the percentage of urban whites and blacks who are gun owners, as well as what percentage of those gun owners have their firearm for protection, specifically.
As shown in Figure 1, 25% of urban whites and 16% of blacks report personal firearm ownership. Additionally, trends in protective ownership are reflected in urban areas, with 49% of white gun owners vs. 68% of black gun owners reporting they have their firearm for protection. 3 While blacks are less likely to be gun owners than whites at both the national level and within urban contexts, protection is more often a concern that warrants acquiring a firearm among black gun owners than their white counterparts.
Clues for what drives the disparity in the need for owning a protective firearm can be found in the different social contexts in which urban blacks and whites find themselves. More specifically, the perceived need for protection is undoubtedly influenced by differences in neighborhood structure, and the resulting cultural understandings of violence, the police, and what constitutes a viable option for addressing the often violent reality of the inner-city. One of these guiding cultural frameworks often observed among residents of marginalized urban communities is legal cynicism, particularly as it relates to the most visible and contacted agents of the legal system: the police.

Structural and interactional causes of legal cynicism
Early conceptions of legal cynicism define it as a component of anomie, or normlessness, in which the rules of the dominant society and its institutions (e.g. courts, the police) no longer dictate proper behavior (Sampson & Bartusch, 1998;Sampson, Morenoff, & Raudenbush, 2005). More recent discussions of legal cynicism by Kirk and colleagues (Kirk & Matsuda, 2011, p. 447;Kirk & Papachristos,  Other Protection 2011) use a more focused conception of legal cynicism, and instead concentrate on the causes and consequences of seeing the law and police as "illegitimate, unresponsive, and ill equipped to ensure public safety." Interestingly, research has demonstrated that residents of disadvantaged neighborhoods disapprove of crime and violence more than residents of more advantaged neighborhoods, and refute the simplistic notion of a racially-linked "subculture of violence." Instead, violence and toleration of deviance is linked to normative orientations structured by neighborhood characteristics (Sampson & Bartusch, 1998). As described in ethnographies of the inner-city, residents of disadvantaged neighborhoods do not necessarily value deviant or criminal activity, but instead engage in these behaviors in response to particular stimuli and situations in their neighborhoods (Sampson & Bartusch, 1998). These residents may believe in dominant ideas about law and order, but because of the structural realities of their environment and the associated cultural understandings of how to navigate that environment, behaviors such as putting one's trust in police or calling on them for assistance are not viable solutions to solving street-level dilemmas (Horowitz, 1987;Sampson & Bartusch, 1998;Suttles, 1968).
Legal cynicism, then, is in part determined by the set of structural conditions of a neighborhood. Neighborhood characteristics such as poverty, unemployment, and racial segregation can inhibit social mobility through social isolation and restricted (or non-existent) opportunities. This inhibited mobility can engender cynicism towards institutions (and institutional agents) that community residents do not see as serving their interests, or as treating them unfairly. In his ethnographic study of an inner-city Philadelphia neighborhood, Anderson describes how the social context of unemployment, racial segregation, and drug use contributes to disenfranchised and cynical attitudes towards established social institutions: The hard reality of the world of the street can be traced to the profound sense of alienation from mainstream society and its institutions felt by many poor inner-city black people, particularly the young. The code of the street is actually a cultural adaptation to a profound lack of faith in the police and the judicial system-and in others who would champion one's personal security. (Anderson, 1999, p. 34) Thus, structural realities of the disadvantaged inner-city lead to feelings of isolation from and skepticism towards traditional social institutions like the legal system.
In particular, the distrust of legal institutions and their agents is deeply rooted in community interactions with police officers, the most visible and contacted agents of the criminal justice system (Banton, 1964;Hough, Jackson, Bradford, Myhill, & Quinton, 2010). In many cases, how police interact with residents is biased by simple geography: police act differently in different neighborhoods (Fagan & Davies, 2000;Smith, 1986;Weitzer, 2000). The salience of the neighborhood as a categorization tool for police to demarcate their jurisdiction into "good" and "bad" neighborhoods can lead to "ecological contamination", whereby all residents of a so-called bad neighborhood are yoked with "moral liability" (Silver, 1967;Terrill & Reisig, 2003, p. 295). When residents become inherently bad by merit of the block they live on, this can have considerable effect on interactions between police and community members.
It is precisely in the most structurally disadvantaged neighborhoods where police are most likely to engage in occupational misconduct or abuse their authority (Fagan & Davies, 2000;Kane, 2002Kane, , 2005. For example, previous work has found that suspects are three times as likely to be arrested in disadvantaged areas as compared to more advantaged ones, and recent work on New York City's controversial Stop-and-Frisk policy shows that stops concentrate in disadvantaged neighborhoods (Fagan & Davies, 2000;Smith, 1986). The use of physical force with suspects is also inequitably distributed between neighborhoods, this neighborhood-level effect existing independent of of situational factors (e.g. suspect resisting or being combative), officer characteristics (e.g. age, training), and the suspect's race; it is neighborhood-level characteristics, not individual ones, that most strongly predict use of police force (Terrill & Reisig, 2003).
As outlined at length by Kirk and Matsuda (2011), inequitable police behaviors have implications for how residents of disadvantaged communities perceive of the police and the law more generally. Police harassment and victimization lead juveniles to view police as ineffective, crooked, and in a generally negative way (Carr, Napolitano, & Keating, 2007), and these predatory police practices need not be experienced directly-youth have negative perceptions of the police even if their experience with police brutality, harassment, or misconduct is vicarious (Brunson, 2007).
These negative interactions are precisely the kind likely to damage the legitimacy of police, in large part because they are seen and experienced as unfair. Scholarship on procedural justice and legitimacy finds that fair interactions have three components: (1) neutrality and consistency; (2) citizens being treated with dignity and respect; and (3) the belief that the law and its agents are acting benevolently and with a sincere desire to be fair (Tyler, 2000;Tyler & Huo, 2002;Tyler & Wakslak, 2004). When police interactions with community residents lack one or more of these components, the legitimacy of the law and the police is called into question. When legitimacy is called into question, not only are police less likely to be cooperated with or deferred to, but residents (law-abiding and criminal, alike) are less likely to obey the law in general (Papachristos, Meares, & Fagan, 2012;Sunshine & Tyler, 2003;Tyler, 1990Tyler, , 2000Tyler, , 2006Tyler & Huo, 2002;Tyler & Wakslak, 2004). 4 These factors are particularly salient when considering active offenders who are unlikely to involve police in their affairs, especially if they involve illegal activity. Though perhaps victimized physically or by way of the theft of money or drugs, contacting police will not result in the return of the stolen (illegal) items, nor exact retribution needed to maintain status on the street (Jacobs, 2000). What's more, interacting with police could result in their arrest for real or fabricated crimes, not to mention expose them to harassment or brutality-unsurprisingly, active offenders largely avoid the police at all costs, even if they have legitimate reason to contact them (Rosenfeld, Jacobs, & Wright, 2003). In any case, whether active offenders do not call the police because they do not want to be detected committing crimes, or because they are fearful of the same legitimacy-damaging harassment that non-offenders face, the end result is the same: police are not viewed as a viable option for solving problems on the street.

Social adaptations to legal cynicism
In neighborhoods where police are seen as ineffective, unfair, and as potential victimizers, it makes sense that some residents would come to distrust police and the legal system they represent. What does one have to gain by calling the police if they may be hassled, or if they face retribution from other neighborhood residents who take exception to the police being called? Taking into account the potential ramifications for contacting police, individuals must look to other avenues to address interpersonal conflicts. Kirk and Papachristos (2011) argue that one alternative conflict resolution strategy is the use of violence. This use of violence as "self-help" occurs in social contexts where (1) formal institutions (e.g. the legal system) are weak or absent, and (2) violence is informally sanctioned as an acceptable way to solve problems and control social behavior (Black, 1983).
Interpersonal violence is but one type of self-help that community residents can engage in. Similar to the how legal cynicism can create a context where violence is a reasonable option, Reiss and Bordua (1967) argue that when citizens perceive the police to be ineffective, they may take measures to provide for their own protection. One such measure is the acquisition of firearms (Lavrakas, 1985;Smith & Uchida, 1988). Just as Black's (1980Black's ( , 1983 model of self-help is premised on weakened institutional controls, McDowall and Loftin (1983) find that handgun purchases increase when police strength declines and violence rates increase, and theorize that handgun acquisition is a product of a self-interested desire for protection. One way or another, people will take steps to protect themselves. In legally cynical neighborhoods with illegitimate law enforcement and high rates of violent crime, one way to do so is to acquire a firearm.
Despite their contribution to our understanding of gun ownership for self-protection, McDowall and Loftin's (1983) study examines legal gun ownership as a reaction to the perceived inability of police to protect citizens. Whether residents of disadvantaged neighborhoods who own their guns illegally acquire firearms for protection, and how perceptions of police affects the choice to own firearms, remains largely unexplored. Given this gap in our knowledge, the current study aims to explore how legal cynicism is related to firearm ownership among active criminal offenders. I hypothesize that active offenders with more negative perceptions of police are significantly more likely to have acquired their most recent firearm for protection.
My analysis proceeds in two stages. In the first stage, I use descriptive statistics to give a broad overview of patterns of protective gun ownership among active offenders in Chicago. In the second stage, I employ logistic regression to explore the relationship between legal cynicism (as measured by perceptions of the police) and protective gun ownership, finishing with the inclusion of statistical controls for the potential counter-explanation of protective gun ownership as a consequence of prior victimization.

Sample
The data used in this study were collected in 2006-07 by the CGP using a cross-sectional survey of active offenders as part of a violence-reduction field experiment. 5 The survey used by the CGP derived its questions from prior research projects, including work by Tom Tyler (1990), Sampson, Raudenbush, and Earls (1997), and Wright and Rossi (1986). The survey included items related to (1) perceptions of the law, various legal agents (including police, prosecutors), and legal authority, (2) guns, gun crime, and gun use, and (3) experience with gangs and associated criminal behavior.
While the past work from which the CGP survey instrument is derived has addressed matters concerning legal legitimacy, guns, and criminal behavior, almost all of this work addresses the effect of legitimacy on behavior among "normal" (i.e. non-criminal) citizens. As noted by Papachristos, Meares, and Fagan (2012), the real strength of the CGP is the particular population sampled: active offenders in some of Chicago's highest crime communities. The CGP provides unique insight into how legitimacy influences behaviors among the high-risk population of active criminal offenders, who are disproportionately likely to be perpetrators and victims of violent crime (Braga et al., 2008). Simply put, the high risk of victimization faced by active criminal offenders represented in the CGP makes it a particularly appropriate sample for exploring behavior designed to minimize victimization.
The CGP consisted of in-person surveys with 141 active offenders in and around high-violence neighborhoods in Chicago (see Figure 2). All survey respondents were at least 17 years of age, had at least one prior arrest for a violent crime, and had been released within the previous six months to some form of state supervision. Though not part of the selection criteria for the sample, many were current or previous members of street gangs. Seventy-one percent of the sample was drawn from high-violence neighborhoods on the south and west sides of the city, with an additional 41 cases drawn randomly from other neighborhoods in the city. 6 Sample participants were recruited in three ways. First, respondents were mailed a letter soliciting their participation in the study. The letter also made it clear that participation was not an additional condition of their probation or parole. Second, the letter was followed up with a phone call to explain the survey and its goals. The addresses and phone numbers for these two recruitment steps were acquired with the help of probation and parole officers. Lastly, participants were recruited from local service programs that provided services to the target population. There was an overall response rate of approximately 60 percent once respondents were contacted, and interviews took about an hour to complete. Respondents were compensated with twenty dollars and a bus pass upon completion of the interview.
Basic sample descriptive statistics are reported in Table 1. Almost 93% of the sample was male, and nearly 86% was African American. The average age of the sample was near 33 years of age, and less than half the sample (42%) was currently employed at the time of data collection. Finally, almost 30% of the respondents reported having ever joined a "gang" or "street organization." 7 As shown in Table 1, there is little difference in these demographic characteristics between the total sample and the subsample of those who had ever owned a firearm.

Measures
The analysis proceeds in several parts. First, I use OLS regression to see how control variables are related to perceptions of police. The objective of this stage is to explore which traditional individuallevel "risk factors" predict perceptions of police. The second stage of analysis includes these individual-level characteristics in two logistic regressions to predict how legal cynicism, as measured by perceptions of police, affects both gun ownership in general (i.e. for any reason) and protective gun ownership, specifically. Before jumping into the analysis, the following section describes the construction of the index used to measure legal cynicism, as well as the measure of protective gun ownership used in subsequent analysis.

Legal cynicism
Borrowing from recent work by Kirk and colleagues (Kirk & Matsuda, 2011;Kirk & Papachristos, 2011), I attempt to examine the police-specific facet of legal cynicism, the cultural frame characterized by the belief that law and police are "illegitimate, unresponsive, or ill-equipped" to ensure public safety.
Knowing that police are the most visible and contacted agents of the criminal justice system (Banton, 1964;Hough et al., 2010), as well as the fact the fact that police are intimately tied to citizens' perceptions of the legitimacy of law more generally (Sunshine & Tyler, 2003), it is reasonable to assume that individuals' feeling about police are more salient for street-level behavior such as protective gun ownership than feelings about faceless judges or prosecutors. To operationalize legal cynicism, I use items in the CGP survey that queried respondents about their feelings towards police to create an index. This index, hereafter referred to simply as the perceptions of police, is constructed using four items, coded from 1 to 4 for "strongly disagree" to "strongly agree". These survey items ask survey participants to indicate their level of agreement or disagreement with statements specifically about the police, such as "Most police treat people with respect." 8 The perceptions of police index is generated by predicting individual responses on survey items using a latent variable model. Emulating the application of the method used by Papachristos, Meares, et al. (2012) and originally proposed by Raudenbush, Johnson, and Sampson (2003), I employ a Rasch model with random effects to model the latent construct of legal cynicism, specifically the component of legal cynicism built on perceptions of police. 9 The log odds of a given response on each survey item depends on the respondent's propensity towards particular answers on other survey items related to perceptions of police. In other words, the log odds of a particular response to a survey item are dependent on the propensity for a respondent to have particular views of the police as captured in their response to other survey items included in the index. This method has two key assumptions: (1) the severity of the response on each item, as well as individual propensity for a given response, are additive in their effects; and (2) responses to survey items are conditionally independent. In the case of both of these assumptions holding, the resulting index is easily interpretable: those with a higher perceptions of police index score have more favorable views of the police and, by extension, are less legally cynical.
It bears reiterating that perceptions of police as a measure of legal cynicism speaks to a specific component of a broader cultural frame. In its totality, legal cynicism encompasses attitudes and perceptions of not just police, but courts, judges, and prosecutors, as well (Papachristos, Meares, and Fagan, 2007). However, unlike prosecutors and judges, police frequently encounter residents on the street and the corner as they go about their daily lives, acting simultaneously as agents of the law and members of the community (Banton, 1964). As put by Black (1980, p. 1), "Police work is arguably the most visible species of legal life, it touches the most people, and it is probably the most controversial." The focus on the particular facet of legal cynicism concerned with police hopes to narrow the analytic focus of this study, and provide a more precise exploration of the relationship between active offenders and police that can give rise to particular adaptive behaviors.

Protective gun ownership
The main dependent variable of interest in this paper is a binary variable indicating if the respondent reported having acquired their most recently possessed firearm "for protection." This measure is operationalized by leveraging two survey items, one asking respondents if they had ever possessed a firearm, the other asking those who reported previous possession, "What was the primary reason you got the gun?" Of the 141 total respondents in the survey, 74.47% (N = 105) reported having ever possessed a firearm. For those reporting firearm possession, a dummy variable was constructed where responses of "for protection" = 1, and all other responses = 0. 10 Table 2 presents simple descriptive statistics for both the protective gun ownership measure and the perceptions of police index, for the total CGP sample and CGP gun owners, respectively. http://dx.doi.org/10.1080/23311886.2016.1227293

Control variables
All regression models include controls for the basic individual level variables commonly associated with crime and violence: continuous age (in years), gender (1 = male, 0 = female), the race of the respondent (1 = black, 0 = non-black), employment status (1 = currently employed, 0 = unemployed), and gang membership (1 = ever in a gang, 0 = never in a gang).

Perceptions of police
The results of the OLS regression of perceptions of police on selected control variables is displayed in Table 3. Age is the only control variable significantly related to perceptions of police, with older individuals being more likely to have more positive perceptions of police (β = 0.008; p < 0.01). This relationship is in line with previous work showing that positive perceptions of police increase with age. (Preiss & Howard, 1958;Worrall, 1999) The lack of significance of the other controls points to the homogeneity of the sample. With the sample being composed largely of black males, it is likely this lack of variation across controls that is responsible for said controls not reaching significance when regressing on perceptions of police. Interestingly, while one might expect gang members to have highly negative perceptions of police, gang membership has no significant effect on perceptions of police. Plausible reasons for this result are described in the discussion and conclusion.

Protective gun ownership
Having ever possessed a firearm is common among active offenders, with over 74% (N = 105) of the sample reporting they have possessed a firearm at some point. This percentage is markedly higher than the percentage of the overall population that personally owns a weapon (22%), as well as the percentage of urban whites (25%) and urban blacks (16%) as reported in the GSS. Thus, even when compared to other urban residents, gun ownership is much more common for active offenders.  Most striking is the percentage of gun owners in the CGP sample that report protection as their reason for acquiring a firearm. Eighty-three percent of active criminal offenders who had ever possessed a firearm acquired their most recent gun for protection-23 percent more than what recent nationally representative surveys report for urban blacks, and over 69 percent more than urban whites (Dimock et al., 2013). As shown in Figure 3, protection is far and away the most common reason to have a firearm among CGP respondents; less than 3 percent of active offenders reporting that they acquired their most recent firearm to commit crimes.
Furthermore, like gun ownership more generally, protective gun ownership among active offenders is more common than it is among the average white (49%) or black (68%) urbanite. Not only do active offenders own guns more often than the average American or the average city dweller, but if they have a gun it is even more likely to be for protection. Table 4 presents the unstandarized coefficents and odds ratios of a logistic regressions to more deeply explore the determinants of general and protective gun ownership. The only control that is significantly related to the probability of having ever posessed a firearm is employment, with those who are employed being significantly more likely to have ever posessed a firearm. The main predictor variable, perceptions of police, is a moderately significant predictor of active offenders having ever posessed a firearm. 11 These results indicate that active offenders with less positive perceptions of police are more likely to have ever posessed a firearm.
This initial result, however, does not elucidate the particular reasons for obtaining those firearms in the first place, only that general ownership of firearms (for whatever reason) is more likely as perceptions of police decreases. To ascertain whether the chances of offenders owning a firearm for protection is significantly related to perceptions of police, I use another logistic regression model, this time restricting the sample to those who reported having ever posessed a weapon. Restricting the sample in this way allows me to predict the probability of protective gun ownership in comparison to all other reasons for acquiring a firearm. Table 5 presents the unstandarized regression coefficients and odds ratios for the logistic regression model predicting protective gun ownership among active offenders who have ever posessed a firearm.
The results of the logistic regression model indicate that perceptions of police is a negative and significant predictor of the probability of having acquired one's most recent gun for protection (β = −3.018; p < 0.05). Among respondents who report having ever possessed a firearm, those with more negative perceptions of the police are significantly more likely to report having acquired their most recent firearm for protection. As indicated by their negative perceptions of police, active offenders who are more legally cynical are significantly more likely to have acquired their most recent firearm for protection.

Percentage
Of the variables in Table 5 predicting protective gun ownership, gender (β = 4.045; p < 0.01), employment status (β = 1.043; p < 0.01), and having been in a gang (β = −2.603; p < 0.001) are significant predictors of protective gun ownership. Males and those who are currently employed are more likely to have acquired their most recent firearm for protection relative to other reasons, while those who  have ever been in a gang are significantly less likely to have gotten their most recent gun for protection. The direction of the gang membership coefficient is counterintuitive since gang members are frequently involved with the drug trade and inter-gang violence, which presumably increases their need for self-protection.

Alternative hypothesis
One potential confound of the relationship between legal cynicism and protective gun ownership stems from a likely determinant of the perceived need for protection: victimization. With gang members and criminal offenders embedded in social contexts and networks that are likely to expose them to high rates of criminal victimization (Braga et al., 2008;Kennedy, 1997;Papachristos, Braga & Hureau, 2012), and with prior victimization being linked to behaviors such as firearm acquisition (Kleck, 1988;Spano & Bolland, 2013), the question remains whether prior victimization can account for the effect of legal cynicism on protective gun ownership shown thus far.
To address this potential confound, I use two more logistic regressions: controlling for prior violent victimization, the first model predicts general gun ownership among the entire CGP sample, and the second predicts protective gun ownership among those in the CGP sample that report having ever possessed a firearm. Table 6 presents the unstandarized regression coefficients and odds ratios for the first model, and Table 7 presents the results for the second.
As shown in Table 6, the results of the logistic regression predicting general ownership among all active offenders is robust to the inclusion of previous victimization as a control. Perceptions of police (β = −1.221; p < 0.1) and employment (β = −1.155; p < 0.01) are still predictors of general ownership, while prior violent victimization is not significantly related (β = 0.239; p > 0.05). Turning to Table 7, prior victimization is not a significant predictor of protective ownership among CGP respondents that reported having ever possessed a firearm (β = 1.437; p > 0.05), and the relationship between perceptions of police and protective ownership shown in previous results holds when controlling for prior victimization: among active offenders who had ever owned a firearm, more negative perceptions of police are associated with a significantly higher probability of their most recent firearm being acquired for protection (β = −2.778; p < 0.05). Gender (β = 4.331; p < 0.01) and employment (β = 1.155; p < 0.01) are also significant predictors of protective ownership, while gang membership is strongly and negatively related (β = −2.740; p < 0.001).
These results suggest, then, that even when accounting for prior criminal victimization, negative perceptions of police are related to a significantly higher probability of active offenders (1) having ever possessed a firearm, and (2) among those reporting having ever possessed a firearm, having acquired their most recent firearm for protection. Thus, while prior victimization has been found in past research to be related to protective gun ownership, it appears that, at least among active offenders in Chicago, the effect of legal cynicism on general and protective gun ownership operates above and beyond the effect of having been a victim of violent crime in the past.

Conclusion and discussion
This study represents several advances in research on trends in gun ownership, as well as how legal cynicism is related to behaviors like protective firearm ownership. First, my findings highlight the fact that nationally-representative estimates of the motivations for gun ownership are unlikely to capture the motivations and gun ownership behavior of high-risk gun owners, especially hidden populations such as ex-felons and active criminal offenders. Additionally, my descriptive findings indicate that respondents in the CGP sample own a protective firearm at rates higher than urban blacks and urban whites, with 83 percent of respondents reporting their most recently possessed firearm was acquired for protection. To be sure, differences in sample composition between highly-selective samples (like the CGP) and nationally-representative samples make it prudent to use caution in making direct comparisons between these samples. Future research would do well to consider methods of conducting surveys on gun ownership to allow direct comparison between criminal and non-criminal populations.
Using perceptions of police as an indicator of legal cynicism, my analysis provides evidence that negative perceptions of police are a significantly related to protective gun ownership among respondents in the CGP sample. In neighborhoods characterized by high rates of crime and violence, interactions between community residents and the law are often strained. These strained relations can give rise to a widely held belief that agents of the law are unable or unwilling to provide for the safety of the neighborhood. This cultural frame of legal cynicism makes calling the authorities an untenable, even dangerous, strategy, and thus individuals engage in "self-help" to provide for their protection. This study suggests that the acquisition of a protective firearm, particularly among those residing in violent, high-crime neighborhoods, is strongly related to negative perceptions of police.
Some of my findings run counter to what might otherwise be expected, particularly the results pertaining to gang membership's relationship to perceptions of police and protective gun ownership. While one might expect gang members to have highly negative perceptions of police, gang membership shows no effect on perceptions of police in this analysis. One hypothesis is that gang membership is superseded by membership in a marginalized community. In other words, negative perceptions of the police may be products of neighborhood-level phenomena that do not discriminate between "decent" community residents and "street"-oriented individuals like gang members (Anderson, 1999).
Finally, knowing that gang members regularly own firearms (Decker & van Winkle, 1996;Hagedorn & Macon, 1988;Miller, 1982) at rates higher than non-members (Sheley & Wright, 1995), and that many gang members choose to arm themselves out of fear of being outgunned in exchanges of reciprocal gang violence (Block & Block, 1993;Horowitz, 1983;Papachristos, 2009), it is surprising that gang member are not more likely than non-members to have acquired a protective firearm.
One potential gang-specific explanation lies in the perception of what it means to "possess" a firearm. Recent ethnographic work in the South Side of Chicago finds that gangs frequently harness their social ties within the gang to borrow and lend guns, and that gang membership is often preserved by individuals who want to maintain the ability to access a weapon should they need it. In a thin, illegal gun market, gang leaders dictate which members are allowed to carry a firearm, and under what circumstances. In essence, the gang controls the supply of firearms, often to prevent wanton violence by younger members that inevitably draws police attention (Cook, Ludwig, Venkatesh, & Braga, 2007). This street-level gang dynamic could explain why gang members do not report having protective firearms-they may view their gang, not themselves, as the possessors of firearms. Even more simply, gang members may not actually have protective firearms because (1) they can access one if need be, or (2) being in a firearm-owning gang is protection enough.
Another potential explanation for a lack of reported protective gun ownership among gang members can be traced to recent legal-political changes in the city of Chicago. Chicago first enacted its handgun ban in 1982, which operated in large part by prohibiting the registration of most handguns. 12 However, in addition to this ban, a 2005 decision by the state legislature created a new ordinance whereby those charged with gun offenses could be tried as armed habitual criminals if they had two prior felony convictions. 13 The additional penalty ranges from an additional 6 to 30 years of imprisonment with no possibility of probation, to a potential extended term of 30 to 60 years. 14 Since gang members are often serial offenders with existing criminal records (Kennedy, 1997), getting caught with a firearm (even if it is one that they have for their personal protection) could potentially cost them the rest of their life in prison. Knowing these severe penalties, gang members may find that the cost of going unarmed is outweighed by the specter of a life behind bars, and choose not to acquire a handgun for protection. 15 While this study advances the knowledge of legal cynicism and gun ownership, it does have its weaknesses. Chief among these weaknesses is also one of the study's strengths: the specificity of the sample. With the purpose of the CGP being to explore the perceptions of the law and police held by active offenders, the very design of the study excludes non-offenders, and non-violent offenders such as drug users. More broadly, this study is unable to generalize to active offenders outside of the highcrime, predominantly African American neighborhoods that the sample was drawn from. Future research will hopefully extend the methods used in this study to other cities, neighborhoods, and types of offenders, as well as hopefully increase its sample size to allow for more intensive analysis.
Additionally, though I address the potential for prior victimization to be driving protective gun ownership instead of legal cynicism, there are other potential explanations for active offenders to own a protective firearm that are not rooted in the belief that police cannot provide for their safety. Though the descriptive results outlined earlier do not suggest it was the case in the CGP sample, the possibility exists that "for protection" is, in practice, protection while engaging in high-risk criminal activities, regardless of gang membership. Indeed, "The qualities of a gun that make it effective in fending off assailants and intruders are the same as those that make it an effective tool in personal crimes of violence. A gun greatly enhances most people's capacity to intimidate or incapacitate an assailant" (Cook, 1991, p. 53). In the case of robbery, not only does a firearm reduce the probability that a victim will resist, but also helps ensure that any such resistance can be quickly and effectively dealt with (Cook, 1980(Cook, , 1987(Cook, , 2011Cook & Ludwig, 2011). The same logic would apply to gang members who want protection from rival gangs they themselves violently victimize (Decker, 1996;Papachristos, 2009;Vigil, 2003).
Another alternative explanation for protective firearm ownership is rooted in the connection between status and the ability to provide for one's own safety on the street. As described by Anderson (1999), inner city residents (particularly young men) exist in a context where the ability to defend oneself and dominate others garners respect. In such an environment, firearms serve to enhance one's ability to maintain and enhance their social standing. Combined with the presence of violent street gangs, drugs, and the proliferation of firearms, guns themselves can become symbols of status, physical embodiments of power that can be employed to defend one's respect or take away another's (Fagan & Wilkinson, 1998;Wilkinson & Fagan, 1996). Within the context of such marginalized communities, guns for protection and guns for status are likely one and the same, the value of the gun deeply tied to the salience of that weapon as a means to preserve one's safety and, by extension, one's respect. Given that the CGP sampled within two such high-crime, high-violence communities, it is difficult to disentangle how much of "for protection" is tied to related concerns surrounding status.
On a more methodological note, an important limitation of this analysis centers on the crosssectional nature of the survey data generated by the CGP. Without longitudinal measures of perceptions of police or knowledge of the timing of when a protective firearm was acquired, claims to the tune of legal cynicism causing acquisition of a protective firearm cannot be made reliably. That said, the results of this analysis are, at the very least, strongly suggestive. As such, I urge that this study's inability to make concrete causal claims should not lead us to throw out the baby with the analytical bathwater. If nothing else, these results are evidence that there is a strong connection between legal cynicism and protective firearm acquisition, and serve as a promising jumping off point for further investigation into the determinants of firearm acquisition in the neighborhoods most direly affected by the fallout of persistent crime and gun violence.
Further, by linking perceptions of police with (illegal) protective firearm ownership, this research makes a compelling case for policing strategies that, through the enhancement of police legitimacy and the building of trust, might reduce such behaviors. If community residents, instead of distrusting or even fearing police, were to view them as a viable option for seeking redress to interpersonal conflicts or for the provision of personal safety, they may no longer feel the need to acquire a firearm. Knowing that having negative perceptions of police is related to an increased probability of protective gun ownership, it follows that policing strategies that engender trust in police and ameliorate legal cynicism might help reduce the probability of protective gun ownership.
One policing strategy with the potential to encourage more positive, trusting relationships between police and the community is community-oriented policing. A key tenet of community policing is the emphasis on partnerships, specifically those between police and the individuals and organization they serve. In this policing paradigm, community-members are seen as valuable resources for identifying and helping police solve problems in the community, and police actively seek to include the community in their efforts to reduce crime and disorder.
As a part of community policing, walking-beats are frequently employed to provide non-confrontational interactions between officers and community members. These walking beats not only enhance feelings of security among residents (Kappeler & Gaines, 2012), but also show the community that police exist for reasons other than to arrest their friends and family. Because of their daily onfoot presence on neighborhood streets, police become more accessible to the community, these non-punitive interactions fomenting familiarity that can ultimately result in enhanced trust between police and community members.
Importantly, the type of interactions that occur between the police and the public are crucial determinants of perceptions of police and legal legitimacy. It is not enough for police to simply exist on neighborhood streets or have many contacts with the public; in fact, order maintenance policing (e.g. New York City's infamous Stop-and-Frisk) that employs frequent stops of pedestrians and vehicles can contribute to negative perceptions of police and damage their legitimacy (Fine et al., 2003;Gau & Brunson, 2010). What's more, the resulting distrust and distaste for law enforcement produced by these practices can have dire consequences for police and communities alike.
Such adverse consequences are exemplified by scholars' descriptions of the pervasive moratorium in some inner-city communities against "snitching." Though originally concerned with an apprehended criminal's cooperation with police in exchange for leniency or some monetary reward (Rosenfeld et al., 2003), anti-snitching (or "stop snitching") extends beyond criminals, becoming embedded in a street code that also demands the silence of non-offenders and victims (Anderson, 1999;Downing & Copeland, 2014). Perpetuated by threats and violence against those who cooperate with police (Rosenfeld et al., 2003;Venkatesh, 2008;Whitman & Davis, 2007), the stop snitching code contributes to a milieu characterized by fear, distrust, and non-cooperation among community members (Woldoff, 2006;Woldoff & Weiss, 2010). As the ability for neighborhood residents to trust one another and cooperate is diminished, so too is their ability to collectively address neighborhood problems of crime and disorder (Sampson & Bartusch, 1998).
What's more, the lack of cooperation with police expected of those who believe in the stop snitching message reduces the ability of police to effectively address serious crime. As the Police Executive Research Forum (2009) describes, distrust in police and a lack of cooperation makes it difficult to secure witnesses to assist with prosecution of crimes, exacerbating worsening clearance rates that have been declining since the early 2000s. Naturally, as more crimes go unsolved and more criminals remain free on city streets, the beliefs of community members of police as uncaring and ineffectual are solidified, further entrenching their reticence to cooperate with law enforcement and confirming that they must solve their own problems and provide for safety themselves. As neatly summarized by Woldoff and Weiss (2010, p. 207), "an anti-snitching narrative constructs the police as the enemy, increases the possibility that crimes remain unsolved, and makes communities vulnerable to vigilantism." In order for repeated interactions with police to not worsen these problems, police-citizen encounters-be they through community policing or police activities more generally-must align with tenants of procedural justice that emphasize fairness, respect, and the preservation of the dignity of those contacted by police. When they do not, a lack of procedural justice damages the legitimacy of police, exacerbates legal cynicism, and decreases the chances of cooperation between police and the public (Gau & Brunson, 2010;Sunshine & Tyler, 2003;Tyler, 2003;Tyler & Huo, 2002). Encouragingly, a growing body of research shows that the distrust between police and marginalized communities is not immutable; when the public is treated in a procedurally just way, they are less likely to believe they have been profiled, trust in police and the law is enhanced, satisfaction with police increases, and citizens are more likely to cooperate with police (Mazerolle, Bennett, Davis, Sargeant, & Manning, 2013;Tyler, 2003;Tyler & Wakslak, 2004).
If members of legally cynical communities can come to view police not as enemies and victimizers, but as viable protectors and partners in reducing crime and violence, they might also come to see 911 as a better option than a pistol. It should go without saying that community policing and procedural justice are not cure-alls for the tangle of pathologies that affect inner-city communities and contribute to street violence (Moynihan, 1965). Enhancing police-community interactions is unlikely to fix failing schools, reduce unemployment, or resolve broader systemic inequalities that disproportionately affect poor, minority neighborhoods. Still, with urban gun violence continuing to afflict communities across the country, the possibility that protective gun ownership among active offenders might be reduced through legitimacy-and trust-enhancing policing strategies deserves serious consideration.

Notes
1. For a review of the large body of work on structural determinants of violence, see Sampson and Lauritsen (1994). 2. It is worth noting that 50% likely underestimates the proportion of the sample that continued to engage in criminal activity. Those that recidivated are only those detected engaging in criminal activity or violating the terms of their release. As such, it is entirely possible that others within the CGP sample were not detected while engaging in continued criminal activity upon release from prison. This is all the more likely considering past work that finds a link between imprisonment and subsequent criminal activity (Cullen, Jonson, & Nagin, 2011), as well as an increased likelihood of violent recidivism for those originally incarcerated for a violent offense (Schwaner, 1998). 3. These statistics and Figure 1 are derived from survey data presented by Dimock et al. (2013). 4. With few exceptions (e.g. Papachristos, Meares, et al., 2012), research on procedural justice and perceptions of general and police legitimacy do not concentrate on particular criminal acts, but instead on general perceptions and opinions of the law. 5. See Papachristos, Meares, and Fagan (2007) for a thorough analysis of Project Safe Neighborood (PSN) and the field experiment methodology. 6. Though the selected randomly, the additional 41 cases did cluster around the high-crime, high-violence neighborhoods where the bulk of the sample was draw from. 7. This is a term for gang frequently used in Chicago. 8. For the complete set of items used in the construction of the index, please see Appendix A. It should be noted that the questions used to construct this index are the same questions employed in prior research on legal cynicism (e.g. Papachristos, Meares & Fagan, 2012;Tyler, 1990). 9. The scale responses for perceptions of police were recoded to binary items to allow for index construction. 10. Option included: "for committing crimes", "as a gift", "for a job (military/law enforcement/security)", "for protection", and "other". 11. While only significant at the < 0.1 level, it bears keeping in mind that the sample used in the CGP is relatively small. The magnitude of the coefficient suggests that the statistical power lent by a larger sample size would likely produce results showing a higher level of statistical significance for the perceptions of police index's effect on the probability of gun ownership. 12. See Chicago Municipal Code §8-20-050 before July of 2010. 13. http://www.ilga.gov/legislation/ilcs/fulltext. asp?DocName=072000050K24-1.7 14. http://www.ilga.gov/commission/lru/2010PFC.pdf 15. 50% of gang members who had ever possessed a gun reported that their most recent firearm was acquired for protection. However, the next most common category was a non-response (N = 9, 21%), making more in-depth analysis difficult.