THE CODE OF THE STREET AND INMATE VIOLENCE: INVESTIGATING THE SALIENCE OF IMPORTED BELIEF SYSTEMS

Abstract

Scholars have long argued that inmate behaviors stem in part from cultural belief systems that they “import” with them into incarcerative settings. Even so, few empirical assessments have tested this argument directly. Drawing on theoretical accounts of one such set of beliefs—the code of the street—and on importation theory, we hypothesize that individuals who adhere more strongly to the street code will be more likely, once incarcerated, to engage in violent behavior and that this effect will be amplified by such incarceration experiences as disciplinary sanctions and gang involvement, as well as the lack of educational programming, religious programming, and family support. We test these hypotheses using unique data that include measures of the street code belief system and incarceration experiences. The results support the argument that the code of the street belief system affects inmate violence and that the effect is more pronounced among inmates who lack family support, experience disciplinary sanctions, and are gang involved. Implications of these findings are discussed.

Scholars have long been interested in understanding the conditions of social order among incarcerated populations (e.g., Irwin, 2005; Sykes, 1958; see, generally, Sparks, Bottoms, and Hay, 1996). The emergence of historically unprecedented levels of incarceration during the past three decades (Clear, 2007; Visher and Travis, 2011) has contributed to a resurgence of interest in undertaking studies that illuminate how and why inmates behave as they do (e.g., Bottoms, 1999; Liebling and Arnold, 2012; Useem and Piehl, 2008). One of the central theoretical arguments advanced by researchers has been that inmates “import” cultural belief systems conducive to violence (Irwin and Cressey, 1962; Thomas, 1977; Wright, 1991). Even so, few rigorous assessments of the salience of imported belief systems have been undertaken; instead, prior research has primarily examined such dimensions as age, sex, race, prior record, educational background, and other such factors.

This gap is striking because importation theory anticipates that it is cultural belief systems that individuals carry with them into incarcerative settings and is not, per se, the social and demographic characteristics of these individuals that contributes to violence. Necessarily, given the limited research that has examined directly the salience of imported belief systems for inmate violence, little is known about the related possibility that the effects of these beliefs may be altered by other social forces, such as exposure to specific conditions, experiences, or deprivations (Sykes, 1958), which restrain or reinforce violence-promoting beliefs. The potential for such interactions derives from theoretical accounts of prison life, which link the insights offered by importation and deprivation perspectives to suggest that inmate violence results from a confluence of factors (Bottoms, 1999; Dhami, Ayton, and Loewenstein, 2007; Tasca, Griffin, and Rodriguez, 2010; Toch and Adams, 1989). A finding that imported beliefs and in-prison deprivations interact in predicting prison violence would highlight the importance of integrating these theoretical models rather than pit them against each other. Notably, however, empirical tests of this interactive argument are rare, and none, to our knowledge, examine directly how the influence, if any, of cultural belief systems may be conditioned by incarceration experiences.

Against that backdrop, the goal of this study is to contribute to scholarship aimed at understanding conditions of order among inmates and, in particular, at examining the role that imported inmate cultural belief systems may have on violence. Specifically, we draw on prior theory and research to develop hypotheses about the influence of one type of belief system—the code of the street (Anderson, 1999)—on inmate behavior and how its effect may be conditioned by experiences inmates have while incarcerated. To this end, we first discuss importation theory and its utility for understanding violence among incarcerated individuals. We then discuss Anderson’s (1999) account of the code of the street belief system and its relevance for examining inmate misconduct. In so doing, we build on his arguments and related scholarship to develop two interrelated hypotheses, the first arguing that inmates who adhere strongly to the street code will engage in more violence when incarcerated and the second arguing that this effect will be amplified by specific incarceration experiences. We then turn to a description of the data and methods and present the findings, which indicate that a code of the street belief system increases the likelihood of inmate violence and that some types of incarceration experiences, such as gang involvement, amplify this effect. We conclude by discussing the implications of the study for theory and research, and in particular, we highlight the importance of integrated theoretical accounts that focus on ways in which inmate characteristics and cultural belief systems combine with incarceration experiences to influence inmate behavior and social order.

BACKGROUND

IMPORTATION THEORY AND INMATE VIOLENCE

A large literature exists that investigates the conditions of order and violence in correctional systems (see, generally, Bottoms, 1999; DiIulio, 1987; Useem and Piehl, 2008). One of the most prominent theoretical accounts is importation theory, which Irwin and Cressey (1962) developed in part as a response to Clemmer’s (1940) work on prisonization and to Sykes’s (1958) argument that prison life entails deprivations conducive to violence. A central starting point for Irwin and Cressey’s argument was the recognition implicit in much prior work that inmates have past lives that they do not necessarily leave behind when incarcerated. Indeed, in developing their theory, they emphasized that scholars, such as Erving Goffman, acknowledged “that inmates bring a culture with them to the institution” (Irwin and Cressey, 1962: 143), but then they dismissed, wrongly in Irwin and Cressey’s (1962) view, the potential for such culture and associated beliefs to affect the behavior of these individuals while institutionalized. Irwin and Cressey (1962) then extended their argument to describe individuals who fit different cultural (thief, convict, and legitimate) belief systems and how each might contribute to prison violence.

The underlying theoretical observations—1) that individuals adhere to cultural belief systems prior to incarceration, 2) that the beliefs may affect individuals’ behavior once incarcerated, and 3) that the content of the belief system directs or channels individuals toward behaviors, such as violence—have garnered the most attention from scholars (e.g., Cao, Zhao, and van Dine, 1997; Jacobs, 1974, 1977; Porporino and Zamble, 1984; Tasca, Griffin, and Rodriguez, 2010). Notably, however, extant studies by and large have not examined the salience of preincarceration belief systems on inmate violence. Rather, typically they have examined the more general corollary that flows from the logic of importation theory, namely, that individuals bring into correctional settings a set of characteristics and experiences that may affect their behavior. Based on this logic, a considerable body of work has emerged that examines whether such factors as age, sex, race, ethnicity, education, marital status, prior record, and the like are associated with inmate behavior, including misconduct and violence. Empirical studies and reviews of importation research have identified that such factors predict these behaviors and, more generally, inmate adjustment (e.g., Adams, 1992; Arbach-Lucioni, Martinez-García, and Andrés-Pueyo, 2012; Dhami, Ayton, and Loewenstein, 2007).

The summary assessment thus is that importation theory is empirically well supported. Yet, and without suggesting that this assessment is incorrect, it remains the case that these studies have not examined imported cultural belief systems directly (e.g., Harer and Steffensmeier, 1996; Paterline and Peterson, 1999; Thomas and Foster, 1973). Indeed, as Lahm (2008: 121) has emphasized, the characteristics, such as race, typically included in tests of importation theory “are presumed to serve as proxies for adherence to a criminal value system. However, it would appear that these proxy measures do not really indicate anything directly about adherence to subcultural values, attitudes, or beliefs.” Put differently, studies to date have supported the view that inmates bring with them characteristics that affect their behavior while incarcerated, but it remains largely unknown how specific imported belief systems influence such behavior. To our knowledge, Lahm’s (2008) study of Kentucky, Ohio, and Tennessee inmates stands as the only exception. She drew on self-reported data from inmates about the extent to which they viewed specific activities (e.g., selling illegal drugs and stealing) as wrong. The analyses identified no effect of the belief measure on inmate violence. Lahm (2008: 134) called for further studies of the salience of inmate beliefs and, at the same time, identified the need for research that relied on preincarceration measures of beliefs, noting that inmate recall “can be highly problematic and could explain the null effect of the belief variable.”

In this investigation, we echo that assessment but also emphasize the importance of examining specific cultural beliefs systems. It is reasonable to anticipate that a belief system that views criminal acts as acceptable holds the potential to influence behavior. Yet, as Irwin and Cressey’s (1962) account and the broader importation literature suggest, specific cultural belief systems, and the content of them, should lead to predictions about the direction of effects on inmate violence. Cultural belief systems, of course, can and do vary greatly. As we argue, one belief system, the code of the street, provides one example of how certain beliefs may be of particular salience when individuals enter prison.

CODE OF THE STREET

Cultural belief systems vary greatly, but one, the code of the street, is especially relevant for investigating the influence of imported beliefs on inmate violence. Anderson (1999) has described the street code as an informal set of beliefs that are learned during childhood and govern how some individuals, especially young African American males, act. According to this belief system, maintaining a tough identity is critical for securing the respect of others. Respect constitutes a central guiding focus. When an individual is disrespected, the code calls for individuals to respond forcefully and, if need be, with violence. Maintaining one’s self-respect dictates such a response. It creates or reinforces an identity that lets others know that “you are ‘about serious business’ and not to be trifled with” (p. 130). The code calls on an individual to display a tough or violent demeanor both to communicate to others a street culture identity and to deter others from attempting to disrespect or victimize oneself (Baron, Kennedy, and Forde, 2001; Jacobs, 2004). Indeed, as Anderson (1999: 92) has emphasized: “For those who are invested in the code, the clear object of their demeanor is to discourage strangers from even thinking about testing their manhood.” The commitment to the code—and the attendant commitment to “respect, toughness, and retribution” (Stewart, Schreck, and Simons, 2006: 431)—thus is motivated not only by a desire to fit in, to belong, with a group of peers but also by self-preservation.

Prior research has lent support to the idea that the code of the street governs conduct and leads to violent threats or acts as a means of creating a self-identity and of protecting oneself. Jacobs (2004), for example, interviewed street offenders and found that they used threats and violence to convey a reputation of toughness that requires respect and that, if challenged, would result in violent retaliation. A similar finding holds in other studies, including Baron, Kennedy, and Forde’s (2001) study of homeless youth and Rich and Grey’s (2005) interviews with young Black male victims (see, generally, Courtwright, 1996; see also Brezina et al., 2004; Brookman et al., 2011; Jacobs and Wright, 2006; Stewart, Schreck, and Simons, 2006). Most extant work, however, has involved qualitative accounts that have not empirically linked prior beliefs to subsequent behavior or, of particular importance for this study, examined the potential effect of the street code on inmate violence.

The nature of prison settings provides a unique context for investigating the code of the street and simultaneously for examining whether imported beliefs influence how some inmates act. The risk of violence, the ubiquity of threats, the imbalance in power among inmates and between inmates and officers, and the deprivations that can affect inmates’ sense of identity—these and other aspects of the prison experience create conditions that increase the likelihood of violence (see, e.g., Irwin, 2005; Rhodes, 2004; Sparks, Bottoms, and Hay, 1996). They also constitute conditions that the code of the street is, as in areas of disadvantage, designed to address (Anderson, 1999; Kubrin and Weitzer, 2003; Stewart, Schreck, and Simons, 2006).

Notably, however, the possibility that an imported street code belief system may contribute to inmate violence remains to be investigated. To our knowledge, the sole exception is Brookman, Copes, and Hochstetler’s (2011) qualitative study of 118 violent inmates in the United Kingdom; they found that the inmates used narratives that drew on beliefs consistent with the code of the street to make sense of and to justify the behaviors that led to their incarceration. However, inmates did so in ways that suggested an ad hoc use of the street code, wherein they “adhere to images from a code as part of a ‘line’ or front that they provide to researchers” (p. 418). The analyses thus raised questions about the extent to which inmates indeed adhered to the code of the street and, by extension, whether they did so while incarcerated and, more generally, whether it affected their conduct before and during incarceration.

In short, although it seems reasonable to anticipate that an imported set of beliefs such as the street code may affect inmate behavior, direct tests of this idea remain to be undertaken. In addition, recent scholarship on the code of the street points to another avenue of investigation that has yet to be examined. A central feature of Anderson’s (1999) argument is the claim that recourse to the street code derives in part from contexts in which, as Stewart, Schreck, and Simons (2006: 434) have emphasized, “street code values and violence can flourish.” These contexts may include neighborhoods in which individuals reside. More generally, they may include contexts that may serve either to inhibit or to fuel or amplify the street code logic or the felt necessity of adhering to it. The potential for belief systems to interact with social context is suggested by many accounts of social life (Gould, 1987), including studies of prison order (see, generally, Clemmer, 1940; DiIulio, 1987; Sparks, Bottoms, and Hay, 1996; Useem and Piehl, 2008). Accordingly, we argue next that aspects of the prison experience may inhibit or amplify the influence of street code beliefs on inmate violence. Evidence of such interactions would underscore the importance of extending scholarship on the code of the street and importation theory to consider how beliefs and social conditions together contribute to violence.

THEORETICAL HYPOTHESES

Two related hypotheses emerge from the preceding observations. The first is that individuals who enter an incarcerative setting with a stronger adherence to the street code belief system will be more likely to engage in violence while incarcerated. The focus on violence stems from the very nature of the street code, which calls for a violent response to provocation (Anderson, 1999: 73). The directionality of this effect bears emphasizing—the code of the street is a belief system that has special salience within correctional settings because it specifically directs individuals to respond with violence when insulted, threatened, or attacked and because, as a large literature has attested to (see, e.g., Adams, 1992; Bottoms, 1999; Schenk and Fremouw, 2012), the potential in these settings for such provocations is great. Accordingly, adherence to a belief system that essentially demands retaliation should increase the likelihood that individuals who strongly adhere to this set of beliefs will engage in more violence.

The second hypothesis is that the street code effect will be amplified by incarceration experiences that either 1) fail to impede or counter the logic of this belief system or 2) serve to reinforce it. Specifically, the effect will be greater among inmates who do not participate in educational or religious programming, who lack family support, who experience disciplinary sanctions, and who are involved with gangs. This expectation derives directly from integrated theoretical accounts that have suggested that inmate preincarceration experiences and characteristics interact with the conditions and deprivations associated with incarceration (see, e.g., Blevins et al., 2010; Bottoms, 1999; DeLisi et al., 2011; Hochstetler and DeLisi, 2005).

We focus on these dimensions because collectively they capture a diverse range of experiences that occurs within correctional settings and because they each capture different ways in which beliefs might be restrained or reinforced. Some prison experiences may provide or promote alternative ways of dealing with situations where the street code would dictate violence. For example, consistent with a social learning theory perspective (Akers, 2009), educational and religious programming can teach individuals about nonviolent ways of responding to provocation. Inmates who do not participate in such programming thus can be anticipated to be more susceptible to the dictates of the street code. Furthermore, consistent with social bond theory (Hirschi, 1969), we can anticipate that inmates who lack family support may lack access to resources that would promote or give greater support to the importance of beliefs that promote prosocial behavior (Jiang and Winfree, 2006). They also may have less reason to care about the potentially harmful consequences to them and their families of engaging in violence. Here, again, the result would be a greater tendency for a street code ethic to govern inmate conduct.

Other prison experiences may elicit or encourage behavior that adheres to the street code. First, scholars have drawn attention to the considerable importance inmates place on perceptions of correctional system legitimacy (see, e.g., Reisig and Mesko, 2009; Sparks, Bottoms, and Hay, 1996). Similarly, Sherman’s (1993) theory of defiance underscores the centrality that individuals give to the perceived legitimacy of authority. These lines of work have suggested that criminal and correctional system sanctions can worsen behavior by creating or reinforcing perceptions of unfairness and the view that the exercise of legal authority is illegitimate (see, e.g., Bottoms, 1999; Maruna, 2001, 2012; Poole and Regoli, 1980; Tyler, 2003). This reasoning suggests that inmates who experience sanctions while incarcerated may be more likely to view correctional system authority as illegitimate, an effect that in turn would free adherence to a street code belief system to direct an inmate’s behavior toward violence.

Finally, gangs have featured in many accounts of inmate social order and violence, including those that have elaborated on Irwin and Cressey’s (1962) importation theory (see, e.g., Thomas, 1977). Gangs clearly constitute a social group within which inmates may feel that they belong and in which they may seek safety. Regardless of the motivation to participate in a gang, the end result is exposure to a group that may endorse or encourage beliefs, such as the street code, which places a premium on the use of violence to resolve conflict (see, e.g., Matsuda et al., 2012). Here, as with the other types of inmate experiences, the effect would be to amplify the influence of adherence to the code of the street, resulting in more violence than otherwise would arise.

Research demonstrating that an imported belief system, such as the street code, influences inmate violence would provide direct evidence that supports Irwin and Cressey’s (1962) theory. If this effect is moderated by the experiences that inmates have while incarcerated, then there is support for the argument that importation and aspects of the prison experience affect behavior and do so interactively (Bottoms, 1999). Put differently, to the extent that evidence of an interaction between imported beliefs and incarceration experiences surfaces, it suggests support for a theoretical account of inmate behavior that does not pit “importation,” “deprivation,” or other arguments against one another and that instead views them as complementary perspectives that together provide a richer, more nuanced account of prison order and violence. At the same time, it would underscore the importance of extending scholarship on the code of the street by investigating conditions under which the code may exert a weaker or stronger influence (Anderson, 1999; Stewart, Schreck, and Simons, 2006; Stewart and Simons, 2010).

DATA AND METHODS

Data for this study were drawn from the Family and Community Health Study (FACHS), a multisite investigation of neighborhood and family effects on health and development. These data provide a unique opportunity to investigate the salience of the street code belief system for inmate violence and whether its effect is amplified by the incarceration experience. A full description of FACHS is available from the Institute for Social and Behavioral Research (http://www.isbr.iastate.edu/FACHS) and in studies that have employed these data (e.g., Murry et al., 2008; Simons et al., 2002; Stewart et al., 2009; Su, Simons, and Simons, 2011). Briefly, FACHS was designed to identify the neighborhood and family processes that contribute to school-age African American children’s development in families living in a wide variety of community settings. The data were collected in Georgia and Iowa using similar research procedures. Families were recruited from neighborhoods that varied on demographic characteristics, specifically racial composition (i.e., percent African American) and economic conditions (i.e., percent families with children living below the poverty line). Using 1990 census data, block group areas (BGAs) were identified in Iowa and Georgia in which the percentage of African American families was high enough to make recruitment economically practical (10 percent or higher) and in which the percentage of families with children living below the poverty line ranged from 10 percent to 100 percent. Using these criteria, 259 BGAs were identified (115 in Georgia and 144 in Iowa). Families were recruited from these BGAs. The final sample of families and neighborhoods recruited involved participants who ranged from extremely poor to middle class.

Among the identified families, interviews were conducted with the target child who was in fifth grade at the time of recruitment, his or her primary caregiver, and a secondary caregiver when one was present in the home. The first wave of data was collected in 1998, the second in 2001, the third in 2004, the fourth in 2007, the fifth in 2009, and the sixth in 2011. In wave 1, the participants were 867 African American children (400 boys, 467 girls; 462 in Iowa and 405 in Georgia) and their primary caregivers. In wave 6, 661 participants (the youth themselves, not their parents) were interviewed again. Attrition has not been problematic for the FACHS study. For example, Simons et al. (2011) recently provided a systematic assessment of attrition and found that across the different waves of data collection, no differences emerged in the characteristics of the respondents, their family structure, or the neighborhoods from which they come; the sole exception was that at wave 5, proportionally fewer respondents were male. We repeated the attrition analyses with wave 6 and identified the same pattern.

We use data from wave 6 for the 219 individuals who became incarcerated after wave 4 and before wave 6, and who were no longer incarcerated at wave 6. These respondents experienced a term of incarceration between these two points in time. We excluded two respondents who were incarcerated twice during this span; the results were not changed by their exclusion. We include a measure of the code of the street from wave 4; we also include controls from wave 4 and, for neighborhood disadvantage, from wave 1. This approach is strategic—it enables us to examine whether adherence to the code of the street prior to incarceration is associated with violence during incarceration. It also means that our sample is likely underrepresentative of more serious offenders, who may have been more likely to have been still incarcerated at the time that the wave 6 data were collected. Thus, the results may not generalize to such individuals, although there is little a priori basis to anticipate that an association between adherence to the code of the street and inmate violence would substantially differ among less serious and more serious offenders. Regardless, this limitation dictates caution in generalizing the results to other inmate populations.

MEASURES

An advantage of the current study is the longitudinal nature of the data. One of the primary questions that the current study investigates is whether endorsement of a street culture affects violence while incarcerated. To model this process, we use a measure of street code beliefs and controls from wave 4 to predict incarceration violence, as reported at wave 6. Furthermore, we predict incarceration violence while controlling for offending at wave 4. As a result, the analyses provide a conservative test of the proposed research hypotheses. Although a longitudinal design does not rival an experiment when making causal inferences, measuring predictor variables in a time order consistent with the proposed theoretical effects provides greater confidence in the results by reducing a central threat to internal validity. Another strength of the data is the wide range of measures that prior research has identified as potential confounders (see, e.g., Adams, 1992; Arbach-Lucioni, Martinez-García, and Andrés-Pueyom, 2012; Bottoms, 1999; Gendreau, Goggin, and Law, 1997; Schenk and Fremouw, 2012) and that can be included as controls.

Dependent Variable

The main outcome consists of the number of times respondents reported using violence while incarcerated between waves 4 and 6. At wave 6, respondents used a life calendar to indicate the number of times they had engaged in various violent acts in recent years. Self-reported data are advantageous because administrative records data tend to suffer from underreporting of inmate misconduct (Reisig, 2002; Steiner and Wooldredge, 2012). In addition, the use of a life-event calendar approach in the FACHS study helped to ensure that reports of acts occurring in recent years were accurate (see, generally, Caspi et al., 1996; Sutton et al., 2011). We would have included analyses of administrative data, but these were not collected as part of the study. Even so, Steiner and Wooldredge’s (2012) analyses indicated that study results tend to be similar regardless of which data source is used.

Of all the inmates in the sample, 32 percent reported engaging in one or more violent acts while incarcerated. The prevalence of committing specific types of acts varied. For example, among all inmates, 24 percent reported threatening other inmates, 16 percent reported physical fights with other inmates, and 6 percent reported physical fights with correctional officers. (Among the inmates who reported engaging in any type of violent acts, 44 percent committed two or more acts of violence; thus, the offense-specific percentages do not add up to 32 percent.) The correlations among these type-specific offense counts ranged from .72 to .88. Accordingly, we combined the items to create a more global measure of violence. This approach is consistent with the fact that Anderson’s (1999) discussion did not differentiate between types of violence, such as threats or assaults, that result from adherence to the street code. (Ancillary analyses examining each type of violence separately identified results that were statistically and substantively similar.) The range for the measure is between 0 and 12 (mean = 2.07, standard deviation = 3.94) (see table 1).

Table 1.

Descriptive Statistics

Variable	Mean	(SD)	Minimum	Maximum
Dependent Variable
Incarceration violence W6	2.07	(3.94)	0	12
Independent Variable
Street code beliefs W4	20.12	(3.77)	7.00	28.00
Inmate Background Controls
Neighborhood disadvantage W1	17.68	(4.07)	1.01	25.62
Family SES W4	16.19	(2.58)	.08	24.05
Family structure W4	.42	(.50)	0	1
Male W4	.53	(.49)	0	1
Prior violent offending W4	.33	(.42)	0	1
Urban W4	.61	(.49)	0	1
South W4	.43	(.49)	0	1
Criminal Justice Controls
W4–6 incarceration—violent W6	.07	(.13)	0	1
W4–6 incarceration—property W6	.31	(.45)	0	1
W4–6 incarceration—drug W6	.25	(.43)	0	1
W4–6 incarceration—supervision violation W6	.23	(.38)	0	1
W4–6 incarceration—other W6	.12	(.17)	0	1
W4–6 incarceration—length (months) W6	7.33	(8.11)	1	18
Prior incarceration W4	.28	(.44)	0	1
Incarceration Experiences
Lack of educational training W6	5.68	(1.04)	2	8
Lack of religious participation W6	4.25	(1.27)	1	6
Poor family social support W6	7.28	(2.97)	3	12
Disciplinary actions W6	.18	(.35)	0	1
Gang involvement W6	.14	(.26)	0	1

NOTE: N = 219.

ABBREVIATIONS: SD = standard deviation; SES = socioeconomic status; W4 = wave 4; W6 = wave 6.

Independent Variable

The main independent variable is a measure of street code beliefs at wave 4. It captures the extent to which African American respondents viewed violence as justifiable or advantageous to earning respect. Although race has been implicated in discussions of the role of culture in prison and jail settings (Harer and Steffensmeier, 1996; Steiner and Wooldredge, 2009), research typically has employed race as a proxy for adherence to beliefs that may be more conducive to violence. Here, then, a focus on an African American sample of individuals and the use of a measure of a specific cultural belief system provides an opportunity to investigate more directly this argument. Specifically, we use a measure that consists of a seven-item, self-report scale from responses at wave 4. Analyses using a wave 1 measure of street code beliefs produced similar results to those reported here and, thus, suggest support for the causal order discussed subsequently, with street code beliefs both preceding and contributing to inmate violence. Here, we use the wave 4 street code beliefs measure because it directly gauges the extent of street code adherence among individuals in the period of time shortly before they were incarcerated. (For more detailed discussions about modeling the street code acquisition process, see Matsueda, Drakulich, and Kubrin, 2006; Stewart and Simons, 2010.)

The scale here has been used in prior studies and accords with Anderson’s (1999) conceptualization of the street code (Stewart, Schreck, and Simons, 2006; Stewart and Simons, 2010). Respondents answered the following questions (1 = strongly disagree to 4 = strongly agree):

1. When someone disrespects you, it is important that you use physical force or aggression to teach him or her not to disrespect you.

2. If someone uses violence against you, it is important that you use violence against him or her to get even.

3. It is important not to back down from a fight or challenge because people will not respect you.

4. People do not respect a person who is afraid to fight physically for his or her rights.

5. Being viewed as tough and aggressive is important for gaining respect.

6. It is important to show courage and heart and not be a coward in a fight or challenge in order to gain or maintain respect.

7. People tend to respect a person who is tough and aggressive.

The responses were summed to obtain a score concerning the extent to which the respondent held beliefs that were consistent with adopting a street code. The alpha coefficient was .89. To ensure that the estimated effects of the code of the street on inmate violence are not spurious, the regression models incorporate a wide range of controls that includes inmate background measures; criminal justice involvement, including type of offense and length of incarceration; and incarceration experiences. These are discussed in the following sections.

Inmate Background Controls

We include controls that reflect not only the characteristics of the individual in the sample but also the characteristics of the areas in which the individuals resided and of their families. We include, for example, neighborhood disadvantage as measured at wave 1. Although this was the only time at which such information was collected, few individuals in the study moved from one neighborhood to another, and any movement likely entailed transitions to neighborhoods that, on average, would not have differed appreciably from those areas from which the subjects moved. Accordingly, the neighborhood measure likely serves as a useful proxy for neighborhood conditions to which the subjects were exposed not only as of wave 1 but also in later waves. The measure was created using the following census items: the proportion of households that were female headed, proportion of persons on public assistance, proportion of households below the poverty level, proportion of persons unemployed, and proportion of persons who are African American. Previous studies have used combinations of these variables to assess neighborhood socioeconomic status (SES) (Sampson, Raudenbush, and Earls, 1997). The items were standardized and combined to form a measure of disadvantage. A constant of 10 was added to eliminate negative values. The alpha coefficient was .89.

Family SES at wave 4 was measured by primary caregiver education level and family income. These two items were standardized and summed to form a composite measure of family SES. Family structure is a dichotomous variable denoting households in which there were two caregivers in the home, in comparison with single-caregiver homes (1 = two-caregiver family, 0 = one-caregiver family).

Male is a dichotomous variable. Males were assigned a value of 1, and females were assigned a value of 0. Prior violent offending is a dichotomous variable, drawn from the wave 4 interview, that captures whether respondents had a history of violent offending, such as threats, assaults, robbery, and use of a weapon, prior to wave 4. Those who had a history of offending were coded as 1, and those who did not were coded as 0. The focus on violent offending stems from the logic of the street code, which calls for violent responses to insults, threats, and aggression. A global measure of prior violent offending was used because responses to the items were highly correlated and because the results were similar using offense-specific measures of prior violence.

Urban is a binary variable identifying those who lived in urban areas at wave 4 (1) with nonurban areas (0) as the reference group. South is a binary variable indicating respondents who lived in the southern United States at wave 4 (1) with midwestern (0) as the reference group.

Criminal Justice Controls

We include several measures that capture the type of offense for which the individual was convicted for the incarceration that occurred between waves 4 and 6, the length of time served for this incarceration, and incarceration events that occurred prior to wave 4. The offense type measure consists of five categories, ones typically used in sentencing and recidivism research, for which individuals were convicted: violent offenses, property offenses, drug offenses, supervision (probation or parole) violations, or other offenses. Each category was converted into a dummy variable, with property offense serving as the omitted category. (Analyses using other offending type classifications did not affect the estimated effects of the street code on inmate violence.) Incarceration length measures the number of months respondents spent incarcerated for the incarceration that occurred after wave 4 and prior to wave 6; it thus serves as a control for the duration of incarceration for the period of time to which the incarceration violence measure applies. Prior incarceration is a dichotomous variable that captures whether respondents had a history of being incarcerated prior to wave 4; inmates with prior incarceration experiences were coded as 1, and those who had none were coded as 0.

Incarceration Experiences

The data for this study are unique in providing measures that tap into five distinct dimensions of the incarceration experience for incarceration that occurred between waves 4 and 6. In each instance, the coding is such that a higher value indicates what is expected to contribute to a greater likelihood of violence while incarcerated. This coding also facilitates interpretation of the interaction effects. The first measure, lack of educational participation, captures whether respondents had limited participation in educational or vocational training while incarcerated. Respondents answered the following questions:

1. While incarcerated, how many programs did you participate in that provided educational advancement (Adult Basic Education [ABE], General Education Development Test [GED], or college courses)?

2. While incarcerated, how many programs did you participate in that provided vocational training while in jail or prison (auto mechanic, welding, etc.)?

The response format ranged from 1 (three or more programs) to 4 (none of the programs). The zero-order correlation between the items was .61. Responses were summed, with higher values on this construct representing lower levels of program participation.

The lack of religious participation is a single item that measures whether respondents had limited participation in religious services while incarcerated between waves 4 and 6. Respondents answered the following question: While incarcerated, how often did you attend religious services, functions, or gatherings? The response format ranged from 1 (everyday participation) to 6 (never participated). Higher values represent lower levels of religious participation.

Educational and religious programming opportunities may vary across the facilities where the individuals were incarcerated. However, as a general matter, educational and religious programming is widely available in U.S. prisons and jails. Harlow (2003: 4), for example, examined educational programming nationally in both prisons and jails, and she identified that 1) in 2000, 91 percent of state prisons offered education programs and 2) in 1999, more than 60 percent of jails offered such programs. Similarly, religious programming, including volunteer efforts from local ministries, has been a mainstay of the correctional system, jails and prisons inclusive, throughout the history of U.S. incarceration (Johnson, Larson, and Pitts, 1997; Mears et al., 2006). For example, according to Cornelius (2012), more than 70 percent of jails offer religious programming. Accordingly, we believe it is appropriate to include the measures of educational and religious programming. However, because we lack a direct measure of available programming, the results should be interpreted with caution.

Poor family social support is measured by three items and captures whether respondents had limited support from family members while incarcerated between waves 4 and 6. Respondents answered the following questions about events while they were incarcerated:

1. How often did you receive visits from family members on the outside?

2. How often did you speak by phone to family members on the outside?

3. How often did you receive letters from family members on the outside?

The response format ranged from 1 (more than once a week) to 4 (once a month or less). The alpha coefficient for the construct was .77. Responses were summed, with higher values on this construct representing lower levels of family social support during the period in which respondents were incarcerated.

Disciplinary actions is a dichotomous variable that measures whether respondents were disciplined (i.e., written up, lost privileges, or placed in segregation) for rule violations while incarcerated between waves 4 and 6. Respondents who indicated that they had received a disciplinary action were coded as 1, whereas those who did not were coded as 0. Several caveats about this measure bear emphasis. First, it captures formal sanctioning of inmates in response to any of a range of acts; the data do not permit, however, identification of which acts contributed to sanctioning. Second, it also does not allow for identifying how inmates perceived the sanctioning and, for example, whether it was perceived as fair; here, we assume that sanctioning is likely to be viewed as unfair. Third, it is possible that some violent acts—the dependent variable in this study—contributed to disciplinary actions. However, as with offending, not all such acts come to the attention of authorities. Indeed, a strength of these data is the use of self-reported misconduct and thus the ability to capture more accurately the amount of inmate violence, not just those that come to the attention of officials. The issue is not, on the face of it, problematic given that the vast bulk of acts that result in disciplinary actions involve nonviolent misconduct (see, e.g., Cochran et al., 2012). The focus here is on whether, separate from the commission of misconduct, disciplinary actions constitute an experience that may condition the effect of the street code. It is a limitation of the study, however, that we cannot isolate which acts contributed to disciplinary actions and that we cannot identify whether any identified effect stems from perceptions about the perceived legitimacy of sanctioning.

Gang involvement is a dichotomous variable that measures whether respondents self-identified as a gang member during the incarceration that occurred between waves 4 and 6. (There was no measure of gang involvement for incarceration events prior to wave 4.) Respondents who indicated that they were involved in a gang were coded as 1, whereas those who did not were coded as 0.

ANALYTIC STRATEGY

Because our dependent variable, incarceration violence, has a distribution that is skewed and discrete, it is appropriate to treat incarceration violence as event count data. In this situation, ordinary least-squares (OLS) regression is not appropriate because count distributions are likely to violate OLS assumptions of normality and homoscedasticity for the disturbance term (Long, 1997). One approach used to analyze count data is Poisson regression. Poisson regression often is used to analyze discrete events that take place randomly and independently in either time or space (Long, 1997). An important assumption of the Poisson model is that the conditional mean equals its variance. The violation of this assumption results in standard errors that are biased downward and leads to inflated significance levels (Long, 1997; Osgood, 2000). This assumption often is too strong and, as in this study, is unmet. Specifically, the likelihood ratio test for the Poisson model revealed that the data were overdispersed, evidence of which can be found in table 1, and that the negative binominal model was the appropriate modeling technique.

Negative binomial regression can be perceived as an extension of the Poisson regression that relaxes the assumption that the variance is equal to the mean. To account for overdispersion, the negative binominal model introduces an additional parameter that estimates the extent of overdispersion (Long, 1997). This technique assumes a Poisson process for all observations but does not assume that all observations have the same mean (Gardner, Mulvey, and Shaw, 1995; Liao, 1994; Long, 1997). The observations are gamma distributed, and the model includes this added source of variance. To estimate our theoretical models, we used the negative binominal commands in the STATA 10 program (StataCorp, College Station, TX). Negative binomial models do not solve model specification concerns (Berk and MacDonald, 2008). In this study, we include controls that are richer than those typically used in inmate violence studies; in the conclusion, we note that further studies using additional sets of potential confounding measures are needed. Ancillary analyses using binary versions of inmate violence identified similar results to those using the negative binomial models, indicating that the results are robust across modeling approaches. A power analysis for the full model indicated that, given our sample size and the number of predictors, we have a 99.9 percent chance of detecting moderate effect sizes at an alpha of .05.

FINDINGS

Our main focus is, first, on whether the street code belief system increases inmate violence and, second, whether this effect is amplified by certain experiences while incarcerated. We begin by examining whether greater adherence to the street code is, as hypothesized, associated with a greater likelihood of engaging in violence while incarcerated. An initial bivariate analysis, shown in model 1 of table 2, indicates that it is, indeed, a statistically significant predictor of inmate violence (b = .229, p < .05), with an estimated pseudo R-squared of 6 percent. Thus, for every one-unit increase in the street code, the log count of the amount of violence increases by .229. Expressed as a relative risk ratio, each unit increase in adherence to the street code—which in this study ranges from 7 to 28—is associated with an estimated 26 percent increase in violent acts while incarcerated [exp(.229) = 1.25734]. This effect is multiplicative. Accordingly, a two-unit increase is associated with a 58 percent increase in violent acts (1.25734²), a three-unit increase is associated with a 98 percent increase in such acts (1.25734³), and so on.

Table 2.

Negative Binomial Regression of Incarceration Violence _W6 on Code of the Street _W4

Variable	Model 1			Model 2			Model 3			Model 4
	b	(SE)	Exp(B)	b	(SE)	Exp(B)	b	(SE)	Exp(B)	b	(SE)	Exp(B)
Code of the Street
Street code beliefs W4	.229*	(.067)	1.26	.178*	(.062)	1.19	.139*	(.062)	1.15	.122*	(.059)	1.13
Inmate Background Controls
Neighborhood disadvantage W1	—	—	—	.161*	(.055)	1.17	.129*	(.055)	1.14	.117*	(.054)	1.12
Family SES W4	—	—	—	−.014	(.041)	.99	−.014	(.041)	.99	−.014	(.041)	.99
Family structure W4	—	—	—	.035	(.192)	1.03	.035	(.192)	1.03	.021	(.192)	1.02
Male W4	—	—	—	.277*	(.082)	1.32	.259*	(.082)	1.30	.198*	(.079)	1.22
Prior violent offendingW4	—	—	—	.888*	(.179)	2.43	.888*	(.179)	2.43	.864*	(.179)	2.37
Urban W4	—	—	—	−.038	(.267)	.96	−.038	(.267)	.96	−.038	(.267)	.96
South W4	—	—	—	.031	(.255)	1.03	.031	(.255)	1.03	.031	(.255)	1.03
Criminal Justice Controls
Violent offense W6	—	—	—	—	—	—	.363*	(.131)	1.44	.345*	(.131)	1.41
Drug offense W6	—	—	—	—	—	—	−.035	(.122)	.96	−.035	(.122)	.96
Supervision violation W6	—	—	—	—	—	—	−.041	(.172)	.96	−.039	(.192)	.96
Other offense W6	—	—	—	—	—	—	−.038	(.181)	.96	−.038	(.122)	.96
Incarceration length (months) W6	—	—	—	—	—	—	1.313*	(.179)	3.72	1.283*	(.179)	3.61
Prior incarceration W4	—	—	—	—	—	—	.762*	(.181)	2.14	.719*	(.181)	2.08
Incarceration Experiences
Lack of educational training W6	—	—	—	—	—	—	—	—	—	−.047	(.242)	.95
Lack of religious participation W6	—	—	—	—	—	—	—	—	—	−.044	(.218)	.96
Poor family support W6	—	—	—	—	—	—	—	—	—	.331*	(.142)	1.39
Disciplinary actions W6	—	—	—	—	—	—	—	—	—	.293*	(.129)	1.34
Gang involvement W6	—	—	—	—	—	—	—	—	—	.515*	(.187)	1.67
Overdispersion Parameter	−.989*	(.229)		−.751*	(.215)		−.699*	(.210)		−.693*	(.212)
Model Chi-Square	25.8*			86.6*			91.4*			94.7*
Pseudo R2	6%			22%			29%			32%

NOTES: N = 219. The — (em dash) indicates entries that are not applicable.

ABBREVIATIONS: SE = standard error; SES = socioeconomic status; _W4 = wave 4; _W6 = wave 6.

^*p ≤ .05 (two-tailed tests).

A key question, however, is whether the effect holds once controls are included. As shown in table 2, model 2, the street code effect was only slightly muted when we included inmate background controls. Here, again, the code of the street exerts a statistically significant effect in the expected direction—those individuals who more strongly abide by the street code are more likely to engage in violence while incarcerated (b = .178, p < .05). That is, for every one-unit increase in the street code, the log count of the amount of violence increases by .178, holding all other factors constant. That translates into an estimated 20 percent increase in violent acts while incarcerated [relative risk ratio = exp(.178) = 1.19483]. In short, after controlling for such background factors as conditions in the neighborhood context from which inmates came, family socioeconomic status, gender, and prior violent offending, preincarceration adherence to the street code is still associated with an increased likelihood of engaging in inmate violence.

Nevertheless, inmates endorsing street code beliefs 1) could simply be different types of offenders or 2) could have had past experiences with the justice system, which in turn could explain their behavior while incarcerated between waves 4 and 6. If so, then we would expect there to be no street code effect once such factors are held constant. Model 3 investigates this issue. An inspection of the model indicates that having been incarcerated prior to wave 4, committing a violent offense, and the time served between waves 4 and 6 are all associated with inmate violence. More relevant to our purposes is that the significant association between street culture and incarceration violence remains, although it is somewhat reduced after inclusion of the controls (b = .139, p < .05). Thus, a unit increase in the street code translates into an estimated 15 percent increase in violent acts while incarcerated [relative risk ratio = exp(.139) = 1.14912].

Model 4 further investigates how robust the street code and violence association is by including measures that tap into different dimensions associated with the incarceration that occurred between waves 4 and 6. The logic is that street code adherence may be related to whether inmates participate in or receive educational or religious programming, are supported by family while incarcerated, receive disciplinary actions related to rule violations, or are involved with gangs. The question is whether the beliefs represented by the street code contribute to inmate violence net of such factors. As can be observed in model 4, it does. Poor family support, disciplinary actions, and gang involvement all contribute to inmate violence. Over and above such effects and those associated with the other controls, the street code belief system remains statistically significant, although the effect is reduced (b = .122, p < .05). Put differently, in the full model, a unit increase in the code of the street scale is associated with a 13 percent increase in violent acts [exp(b) = 1.12975], a two-unit increase is associated with a 28 percent increase in such acts [exp(b) = 1.12975²], and a three-unit increase is associated with a 44 percent increase in them [exp(b) = 1.12975³]. (Multicollinearity diagnostics identified no problems; for example, no variance inflation factors [VIFs] were above 2. In addition, analyses of potential gender interactions revealed no significant differences in the effect of the street code on violence among males vs. females.)

The analyses to this point thus indicate that cultural belief systems, specifically the street code, imported into incarcerative settings may contribute to violent behavior. We turn now to table 3, which presents the results of five interactional models aimed at testing the hypothesis that the street code’s effect will be moderated by each of the five incarceration experience dimensions. Recall that theoretical accounts of inmate behavior, especially those that have emphasized the role of culture, point to the idea not only that belief systems affect behavior but also that this effect can be channeled or amplified by other social factors.

Table 3.

Negative Binomial Regression Coefficients of the Interactive Effects of the Code of the Street _W4 on Incarceration Violence _W6

	Model 1		Model 2		Model 3		Model 4		Model 5
	b	(SE)	b	(SE)	b	(SE)	b	(SE)	b	(SE)
Code of the Street
Street code beliefs W4	.138*	(.059)	.131*	(.059)	.126*	(.059)	.117*	(.059)	.119*	(.050)
Incarceration Experiences
Lack of educational participation W6	−.029	(.242)	−.041	(.242)	−.047	(.242)	−.047	(.242)	−.047	(.242)
Lack of religious participation W6	−.041	(.218)	−.027	(.218)	−.044	(.218)	−.044	(.218)	−.044	(.218)
Poor family social support W6	.328*	(.142)	.328*	(.142)	.351*	(.142)	.333*	(.142)	.328*	(.142)
Disciplinary actions W6	.289*	(.129)	.289*	(.129)	.294*	(.129)	.289*	(.129)	.261*	(.129)
Gang involvement W6	.515*	(.187)	.515*	(.187)	.515*	(.187)	.515*	(.187)	.588*	(.187)
Interactions
SC beliefs × lack of educational training	−.005	(.135)	—	—	—	—	—	—	—	—
SC beliefs × lack of religious participation	—	—	−.003	(.119)	—	—	—	—	—	—
SC beliefs × poor family support	—	—	—	—	.046*	(.022)	—	—	—	—
SC beliefs × disciplinary actions	—	—	—	—	—	—	.045*	(.022)	—	—
SC beliefs × gang involvement	—	—	—	—	—	—	—	—	.048*	(.019)
Overdispersion Parameter	−.699*	(.212)	−.699*	(.212)	−.711*	(.212)	−.709*	(.212)	−.693*	(.212)
Model Chi-Square	91.7*		91.8*		97.2*		96.9*		95.5*
Pseudo R2	29%		29%		34%		35%		35%

NOTES: N = 219. Models include all additional control variables reported in table 2. The — (em dash) indicates entries that are not applicable.

ABBREVIATIONS: SC = street code; SE = standard error; _W4 = wave 4; _W6 = wave 6.

^*p ≤ .05 (two-tailed tests).

Drawing on this work, we hypothesized that the violence-inducing effect of the street code would be greater among inmates who are exposed to incarceration experiences that either fail to impede the street code’s influence or that, conversely, enable the street code to assume greater prominence in how individuals act. Specifically, we anticipated that the street code would exert a greater effect among individuals who, while incarcerated, 1) failed to participate in educational programming, 2) did not participate in religious programming, 3) lacked family support, 4) were subjected to disciplinary actions, or 5) were gang involved. To test this idea, we created interaction terms between the street code and centered versions of each of these dimensions, and we included them in separate models: street code × lack of educational programming (model 1); street code × lack of religious programming (model 2); street code × lack of family support (model 3); street code × disciplinary actions (model 4); and street code × gang involvement (model 5). The models include all controls included in table 2, but to simplify the presentation of the results, table 3 presents only the effects associated with code of the street, incarceration experiences, and the interaction of the code of the street with each of the five types of experiences.

As an inspection of the models reveals, we find partial support for the hypothesis that the street code effect is moderated by incarceration experiences. On the one hand, a lack of participation in educational programming or religious programming did not amplify the effect of adhering to a code of the street belief system. On the other hand, lacking family support, receiving disciplinary sanctions, and being involved in a gang all enhanced the street code effect. In models 3, 4, and 5, for example, we can see that adherence to the street code is associated with an increased likelihood of engaging in violence while incarcerated. We can see as well that each type of prison experience is associated with incarceration violence. And in particular, as evidenced in the statistically significant interaction terms, we can see that the combination of 1) greater adherence to the street code and 2) having poor family support, disciplinary actions, or gang involvement contributes to an even greater likelihood of violence over and above the direct effects of the street code and these dimensions.

CONCLUSION

Scholars have long been interested in understanding the conditions of order and violence in prisons. A central theoretical argument, expressed most forcefully by Irwin and Cressey (1962) but since elaborated on at length in numerous other accounts (e.g., Bottoms, 1999; DeLisi et al., 2011; Dhami, Ayton, and Loewenstein, 2007; Thomas, 1977; Wright, 1991), has been that inmates import belief systems that influence their conduct while incarcerated. This line of work has argued that inmate violence stems from many factors, including belief systems, and that the effects of such factors may be moderated by dimensions of the prison experience. It is not only scholarly interest that has driven such work. The large-scale increase in incarceration in recent decades has engendered renewed attention to understanding the causes of prison violence.

Despite the longstanding view that belief systems matter for understanding how inmates act, few empirical studies exist that directly examine the beliefs held by inmates prior to incarceration and whether such beliefs indeed are associated with prison violence (Lahm, 2008). Building on prior research, which has established that dimensions such as age, sex, race, prior record, educational background, and the like may influence inmate conduct, this article aimed to test directly whether one prominent belief system—the code of the street (Anderson, 1999)—is related to violence among African American inmates and whether incarceration experiences moderate this association. One advantage of this approach is that it does not treat race as a proxy for certain beliefs and, furthermore, that it recognizes that, among African Americans, there will be varying degrees of endorsement of any given belief system.

The results can be summarized briefly. First, adherence to the street code belief system is associated with a greater likelihood of inmate violence, net of individual characteristics, prior record, conditions in the areas from which inmates come, and participation in programming or gangs. Second, the effect of the code of the street is greater among inmates who do not enjoy family support while incarcerated, who receive disciplinary actions, and who are involved in gangs. It is not moderated by a lack of participation in educational or religious programming. In short, consistent with importation theory, prior belief systems, especially those that promote, endorse, or require a violent response when provoked, seem to contribute to violence in prison. At the same time, consistent with calls for integrated theoretical accounts of inmate violence, the effect of such belief systems seems to be amplified in the absence of factors, such as family support, which otherwise might constrain them and in the presence of such factors, such as gangs, that might facilitate or support their logic or expression.

Several limitations of the study warrant mention. We focus only on individuals who were incarcerated after wave 4 and then released prior to data collection at wave 6. Accordingly, we examine inmates who may be less serious offenders than those who are incarcerated for longer periods of time. Although there is little reason to anticipate that the identified associations would differ for individuals serving longer sentences, we cannot be sure that they would be the same either. In addition, it would be ideal to have a wider array of measures of inmate violence and of how inmates perceived their treatment. We used disciplinary actions to gauge experiences with formal sanctioning and assumed that such experiences would be viewed by inmates as unjust; direct measures of the perceived legitimacy of prison authority would be preferable. It also would be helpful to have data that more directly allowed for identifying the extent to which the code of the street is imported into incarcerative settings and the extent to which such settings create or amplify adherence to the street code. In this study, the results were robust even when we included a measure of street code adherence from wave 1, when the respondents were in fifth grade, which reinforces the notion that the street code is imported into jails and prisons. Even so, future research ideally will directly examine changes in the street code prior to first incarceration events and after subsequent incarceration events. Also, studies are needed that examine the extent to which a street code effect on inmate violence is robust across different samples and that include additional sets of potential confounding measures. Not least, future research ideally would have more measures of a wider variety of inmate experiences and would include direct measures of opportunities for such experiences to occur. With such information, researchers will be positioned more accurately to identify the specific factors that may amplify or inhibit the influence of the code of the street on inmate violence.

Several implications flow from the study’s findings. First, and echoing a conclusion that flows from a long line of scholarship on incarceration, theories of prison order and violence are likely to provide a more complete account of inmate behavior if they consider the role of cultural belief systems. Many scholarly accounts assert that beliefs affect how inmates act, but few empirical studies examine directly the beliefs that inmates hold prior to incarceration or how these beliefs influence inmates’ behavior while incarcerated. Tests of importation theory illustrate this point—many empirical studies that purport to test the theory show that dimensions such as age or race are related to inmate violence. As a general matter, we can view an age–violence or race–violence relationship as evidence that individuals bring with them characteristics, beliefs, resources, social capital, or some other dimension that affects their behavior. However, such tests do not identify directly whether belief systems are imported, much less which types of beliefs are imported or affect inmate violence. The current study supports directly a central tenet of importation theory—namely, inmates do carry with them preexisting beliefs and these can contribute to violence. That finding does not imply that deprivation, administrative approaches, or prison conditions are inconsequential. To the contrary, prior research has clearly implicated these different conditions as factors that contribute to inmate behavior. Rather, the finding simply suggests that in fact some beliefs, the code of the street in this instance, may be imported into institutional settings and contribute to how individuals act while incarcerated. To the extent that this claim is correct, there is a need for research that extends scholarship on importation theory by examining the potentially differential salience of cultural belief systems among inmates who vary with respect to age, gender, race, ethnicity, country, and other such dimensions (see, e.g., Slotbloom et al., 2011).

Second, extending the previous observation, both theories and tests of them are needed that show how inmate characteristics, backgrounds, and beliefs intersect with prison experiences and settings to contribute to violence. We investigated this idea in this study by examining whether the effect of the code of the street is moderated by such experiences as educational or religious programming, family support while incarcerated, disciplinary actions, and gang involvement. Programming seemed to exert no conditioning effect, whereas the other dimensions did. It seems likely, then, that imported belief systems not only affect inmate behavior but also indicate that their influence can be enhanced or impeded by other facets of the incarceration experience. Disciplinary actions that are perceived to be unjust, for example, may reinforce beliefs about the legitimacy of responding to perceived or actual mistreatment with violence. Similarly, gangs may provide a social structural support that further legitimizes a belief system that calls for violence when instigated. At the broadest level, such possibilities, and the findings in this study, provide warrant for theories of inmate violence that identify the implications of specific belief systems for particular institutional contexts and the conditions in these contexts that may channel the influence of such systems on inmate conduct. These conditions may include deprivations and the general conditions of confinement (Rhodes, 2004; Sykes, 1958). However, they may include also administrative philosophies and approaches (DiIulio, 1987; Huebner, 2003; Irwin, 2005), prison culture and perceptions about the legitimacy of prison authority (Bottoms, 1999; Mears, 2008; Reisig and Mesko, 2009; Useem and Piehl, 2008), and the attitudes and characteristics of inmates and corrections staff (Adams, 1992; Liebling and Arnold, 2012; Sparks, Bottoms, and Hay, 1996; Steiner and Wooldredge, 2008, 2009).

Third, this study adds to a growing body of research that has been documenting the salience of the street code for social behavior (see, e.g., Jacobs and Wright, 2006; Piquero et al, 2012; Rich and Grey, 2005) and is the first, to our knowledge, to demonstrate that it is associated with inmate violence. An important line of inquiry is to examine how the street code belief system is reinforced, impeded, or modified by the experience of incarceration and the conditions of confinement. Clearly, beliefs may remain stable over time, but just as clearly they may be modified. A central question that relates to the code of the street is the extent to which adherence to the code is stable or variable over time (Anderson, 1999) and whether particular experiences, such as incarceration, that may occur in prison settings reinforce it. Brookman, Copes, and Hochstetler (2011) showed how inmates may use a street code type of philosophy to make sense of and defend their crimes and, more generally, to create an identity for themselves. A prison setting thus may cement adherence to the street code. Whether it does so by providing reinforcement of these beliefs or through some other mechanisms, such as sustained exposure to perceived injustice, remains to be investigated.

We close with a possible implication of the study for policy. As identified in this study, the street code belief system may influence inmate violence and, by extension, conditions of order in incarcerative settings. The code of the street is not per se criminogenic—it does not call for criminal activity. It does, however, call for proactive responses to perceived insults and provocations. Such responses, especially in a jail or prison setting, normatively would entail violence as a way of commanding respect.

Juxtaposed against such considerations is the fact that corrections programs that target criminogenic beliefs are more effective at reducing misconduct and offending (Cullen and Gendreau, 2000; French and Gendreau, 2006; see also Andrews, Bonta, and Wormith, 2011; Gendreau, Goggin, and Law, 1997; Goetting and Howsen, 1986). It might be reasonably inferred, then, that augmenting risk prediction instruments (Gottfredson and Moriarty, 2006) through inclusion of measures that tap into beliefs similar to the code of the street could increase the ability to identify inmates who may be, or are signaling that they will be, more likely to engage in violence. This idea fits squarely with recent arguments for finding new measures and approaches for predicting inmate misconduct and recidivism (Bushway and Apel, 2012; DeLisi, 2003).

A central question, however, for future research is how much better might be the risk prediction efforts if they incorporated measures of criminal and cultural belief systems systematically. Another question is what could be done to respond to the effects of the street code. One avenue correctional systems could pursue would be to attempt to change inmate adherence to the code of the streets. That said, changing culturally rooted beliefs may be difficult. Another possibility would be to target factors, such as increasing family support and reducing gang involvement, that might inhibit the code of the street from influencing an inmate’s behavior.

When considering such possibilities, it is important to consider the fact that adherence to the code of the street falls primarily along racial lines. Accordingly, any actuarial-based approach that relies on information about adherence to the code of the street would run the risk of targeting Black inmates more so than other groups. That concern might be alleviated if research identifies that analogous belief systems exist among Whites and other racial and ethnic groups (see Brookman, Copes, and Hochstetler, 2011). If so, such beliefs might be used safely with little risk of creating a form of empirically based racial profiling within correctional systems.

Biographies

Daniel P. Mears is the Mark C. Stafford Professor of Criminology in the College of Criminology and Criminal Justice at Florida State University. He conducts basic and applied research on a range of crime and justice topics. His work has appeared in Criminology, the Journal of Research in Crime and Delinquency, Justice Quarterly, and other crime and policy journals and in a recent book, American Criminal Justice Policy (Cambridge University Press, 2010).

Eric A. Stewart is a professor in the College of Criminology and Criminal Justice at Florida State University. He is a member of the National Consortium on Violence Research, a W.E.B. DuBois Fellow with the National Institute of Justice, and a member of the Racial Democracy, Crime and Justice Network. His research interests include racial inequality and criminal outcomes, crime over the life course, and contextual processes and microprocesses that affect adolescent development. His work has appeared in Criminology, the Journal of Research in Crime and Delinquency, and Justice Quarterly.

Sonja E. Siennick is an assistant professor in the College of Criminology and Criminal Justice at Florida State University, Gainseville. Her research examines the interpersonal causes and consequences of crime over the life course, including, most recently, the family dynamics of adult offenders. Her work has appeared in Criminology, the Journal of Research on Adolescence, the Journal of Research in Crime and Delinquency, and other journals.

Ronald L. Simons is a Foundation Professor in the School of Criminology and Criminal Justice at Arizona State University and research fellow in the Institute for Behavioral Research at the University of Georgia. His research has focused upon onset, amplification, and desistance from various externalizing and internalizing problems. This work has investigated the various avenues whereby community conditions, parents, schools, and peers during adolescence, and romantic partners, routine activities, incarceration, and employment during adulthood influence trajectories of maladjustment. Recently, he has expanded his research program to include genes and other biological variables. His work has appeared in such journals as Criminology, the Journal of Health and Social Behavior, and the American Sociological Review.