The Social Context of Criminal Threat, Victim Race, and Punitive Black and Latino Sentiment

The Social Context of Criminal Threat, Victim Race, and Punitive Black and Latino Sentiment Abstract A well-established body of research focuses on the relationship between criminal threat and the exercise of formal social control, and a largely separate literature examines the effects of victim race in criminal punishment. Despite their close association, few attempts have been made to integrate these related lines of empirical inquiry in the sociology of punishment. In this article, we address this issue by examining relationships among criminal threat, victim race, and punitive sentiment toward black and Latino defendants. We analyze nationally representative survey data that include both subjective and objective measures of criminal threat, and we incorporate unique information on victim/offender dyads to test research questions about the that role victim race plays in the formation of anti-black and anti-Latino sentiment in the criminal justice system. The results indicate that both subjective perceptions of criminal threat and minority population growth are significantly related to punitiveness among whites, and that punitive sentiment is enhanced in situations that involve minority offenders and white victims. Moreover, we show that aggregate indicators of racial threat strongly condition the effect of victim race on punitive attitudes. Implications of these findings are discussed in relation to racial group threat theories and current perspectives on the exercise of state-sponsored social control. criminal threat, victim race, punitive sentiment, social control; victim-offender dyads The formal exercise of social control is a topic of enduring sociological interest, particularly as it relates to racial and ethnic inequality in society. Theoretical treatments of race and punishment are routinely cast in terms of racial group threat perspectives that suggest large and growing minority populations generate group prejudice and hostility on the part of the white majority, which is then translated into enhanced social control efforts in the criminal justice system (Blalock 1967; Liska 1992; Quinney 1970). At the same time, sociological formulations of the behavior of law argue that the capacity for government social control is directly related to the relative social status of affected parties in the justice system—larger quantities of law are expected when social distance is greater between the structural positions of victims and offenders (Black 1976:2). Because race is a key component of social stratification in American society, stronger legal responses are expected in situations involving white victims and racial minority offenders (Xie and Lauritsen 2012). Taken together, group threat theories suggest a key role for minority group size in the exercise of social control in society, whereas theoretical propositions about the behavior of law highlight the importance of victim and offender race. The current research unites these arguments, examining the complementary and interactive influences of aggregate population contexts, localized perceptions of criminal threat, and race of victim effects in public support for enhanced social control directed at minority offenders. This study contributes to prior work on group threat and punitiveness in several important ways. First, it integrates insights about the importance of victim race into contemporary formulations of racial group threat theory. Second, it expands the ken of prior work by examining group threat associated with both black and Latino populations in the United States. Third, it incorporates both objective and subjective measures of criminal threat and, finally, it focuses on a unique measure of social control that captures punitive sentiment targeting black and Latino defendants in the American criminal court system. The article proceeds by first considering the evidence linking racial threat to social control outcomes. It then introduces relevant scholarship on the role that victim race plays in social control processes, before outlining and testing specific hypotheses about group threat, victim race, and punitive sentiment. BACKGROUND Racial Threat and Social Control Race, crime, and punishment are intertwined in a complex historical legacy that involves past racial oppression and contemporary social stereotypes that link minority groups to enhanced punishment in the criminal justice system (Bonilla-Silva 2006; Brewer and Heitzeg 2008). Public opinions regarding race, crime, and public policy remain sharply divided along racial and ethnic lines in American society (Bobo and Johnson 2004). In particular, racial and ethnic minorities tend to be less supportive of punitive justice policies than majority whites (Hagan and Albonetti 1982; Hagan, Shedd, and Payne 2005). This racial divide is maintained across various domains of the criminal justice system, including policing (Brunson 2007; Russell 1999; Weitzer and Tuch 2005), criminal courts (Brooks and Jeon-Slaughter 2001), and corrections (Baumer, Messner, and Rosenfeld 2003; Unnever and Cullen 2007). James Unnever (2008), for instance, has recently demonstrated that black respondents are more than twice as likely as white respondents to identify police bias and unfairness in the courts as major reasons for the disproportionate incarceration of African American men. Although less research focuses on ethnic differences, Latinos also tend to be less punitive than whites, expressing higher levels of support for rehabilitation and lower levels of support for capital punishment (Baik 2012). Much of the empirical research examining the influence of racial group dynamics in the justice system finds evidence that perceptions of racial threat are spatially patterned. Communities with relatively large or increasing minority populations tend to be associated with higher levels of racial prejudice and increased support for harsh punishment (Bobo and Hutchings 1996; Quillian 1996; Stewart et al. 2015; Taylor 1998). Ryan King and Darren Wheelock (2007), for instance, have shown that spatial variations in racialized perceptions of economic and criminal threat are directly related to individual punitive attitudes. Related work reveals significant associations among minority population size, self-reported fear of crime, and subjective perceptions of the risk of criminal victimization (Quillian and Pager 2010; Taylor and Covington 1993). Over three decades ago, Allen Liska, Joseph Lawrence, and Andrew Sanchirico (1982) demonstrated that the relative size of the non-white population was associated with greater fear of crime across U.S. cities. Similarly, Jeannette Covington and Ralph Taylor (1991) found that when the racial composition of the neighborhood was a predominantly black neighborhood, residents were more fearful of crime. More recent work indicates that individual perceptions of racial demographics are often more salient than objective indicators of population size (Chiricos, Hogan, and Gertz 1997), and it demonstrates that similar findings occur with regard to proximity to Latino populations (Chiricos, McEntire, and Gertz 2001). Indicators of minority group size have also been linked explicitly to formal social control efforts in the criminal justice system. Policing research shows that as the size of the minority population grows so does the size and funding of local police departments (Chamlin 1989), as well as race-specific arrest rates across local communities (Harer and Steffensmeier 1992; Liska, Chamlin, and Reed 1985). Rates of police killings of minority suspects are also higher in cities with large or increasing African American populations (Jacobs and O’Brien 1998). Parallel work on American prison populations links racial threat indicators to spatial and temporal variations in incarceration rates (Greenberg and West 2001). George Bridges, Robert Crutchfield, and Edith Simpson (1987), for instance, found that counties with larger non-white populations also had higher rates of non-white incarceration, and David Greenberg and Valerie West (2001) reported that states with larger black populations not only experienced higher incarceration rates but also greater prison population growth in recent decades. Some research also suggests individual sentencing decisions are affected by broader racial dynamics in the community (Johnson 2006; Ulmer and Johnson 2004). For example, Chester Britt (2000) found that the proportion of black residents in a county increased the individual probability of incarceration, and Xia Wang and Daniel Mears (2010) reported evidence that custodial sentences were more likely when racial group threat was high and increasing in an area. The relative size of the black population has even been tied to the likelihood of receiving the death penalty (Jacobs and Carmichael 2002). Not all studies, though, are supportive of racial threat perspectives, with some work finding null or even opposite effects for the influence of racial composition on punishment (cf. Bridges and Crutchfield 1988; Feldmeyer and Ulmer 2011). In part, this may reflect a number of empirical limitations that are common in this research tradition. First, racial threat is typically captured by aggregate population measures rather than more proximate, perceptual threat measures. As Karen Parker, Brian Stults, and Stephen Rice (2005) note, “dependency on percent black as the main indicator of racial threat” has led to “inconsistency in findings” (p. 1111). Some research suggests that perceptual measures of group threat demonstrate stronger and more consistent relationships with social control outcomes (Chiricos et al. 1997). Second, tests of group threat theory focus predominantly on the African American population with considerably less attention devoted to the growing Latino population (Eitle and Taylor 2008). This is despite the fact that Latinos are now the largest and one of the fastest growing racial/ethnic groups in the United States (Pew Research Center 2016). Given the theoretical salience of population growth in group threat perspectives, there is a clear need to incorporate Latinos into broader multiethnic examinations of punitive sentiment. Third, prior work has been conducted at various levels of analysis (Parker et al. 2005), and relatively few studies employ nationally representative data or examine changes in racial composition over time. Finally, prior work makes it clear that social control efforts are often tied specifically to criminal threat. The concurrence of race and crime in public discourse has elevated the salience of criminal threat as a vehicle for enhanced social control efforts targeted at minority defendants (Pickett and Chiricos 2012), yet direct measures of criminal threat are often absent from research in the group threat tradition (Eitle, D’Alessio, and Stolzenberg 2002). Moreover, no prior work has considered the role that victim race plays in perceptions of criminal threat. The current work uses national survey data to investigate both objective and subjective indicators of group threat. It examines threat processes associated with both black and Latino populations, and it incorporates unique information on victim/offender racial dyads into the study of public support for more punitive treatment of minority defendants in the criminal justice system. Criminal Threat, Victim Race, and Punitive Sentiment Although group threat theories were originally formulated in relation to socioeconomic threats, race and crime have gained increasingly prominent roles in these perspectives (Hawkins 1987). Racial threat theory argues that the exercise of social control in society varies with the racial composition of the population because members of the white racial majority feel threatened by large or growing minority populations (Blalock 1967). Community-level racial demographics feed into perceptions of social threat that are rooted in racial prejudice and intergroup conflict. Fueled in part by the emergence of race and crime as a dominant political theme, white Americans increasingly associated criminal threats with growing minority populations (Chiricos et al. 2001). In response, enhanced social control efforts may be mobilized that target minority defendants in the criminal justice system. From this perspective, the racial composition of an area affects the perceived threat of crime, which in turn contributes to both formal and informal mechanisms of social control in society (King and Wheelock 2007). Although racial group threat can take multiple forms that include economic, political, and criminal threat, the latter appears to be the driving force behind punitive responses of the criminal justice system. This is not surprising given the clear association between criminal threat and punishment. In fact, recent work argues that political and economic threats have largely been “replaced by the black male criminal in the iconography of racial threat” (Crawford, Chiricos, and Kleck 1998:483), and it suggests that “The conflation of race and criminal threat is [now] so well established that some regard popular discourse about crime and punishment to be part of the rhetorical code of ‘modern racism’” (Chiricos et al. 2001:323). Research demonstrates that stereotypes tying race and ethnicity to perceptions of crime and violence are widespread in American society (Brown 2010; Devine 1989), and it shows that respondents frequently express greater fear of victimization by black strangers than white strangers (St. John and Heald-Moore 1996). Additional evidence for the salience of criminal threat comes from research that compares different threat mechanisms. David Eitle, Stewart D’Alessio, and Lisa Stolzenberg (2002), for example, tested the relative importance of political, economic, and criminal threat explanations of arrest rates and concluded that “the findings taken together furnish strong support for the threat of black crime hypothesis” (p. 557). Research on racial typification of crime in the media also supports the criminal threat hypothesis. Franklin Gilliam and Shanto Iyengar (2000) report that local news often utilizes a “crime news script” that routinizes the association between racial minorities, crime, and violence, and Daniel Mears and Eric Stewart (2010) note that both “media accounts and policy discourse” have “increasingly equated crime with blacks” (p. 35). Racialized social constructions of criminal threat, then, are often tied to broader patterns of racial demography in society and may be used to promote dominant group interests in core institutions of law (Chambliss and Seidman 1971; Hetey and Eberhardt 2014). Community-level racial dynamics can feed into perceptions of threat and fuel public sentiment for enhanced punishment of minority defendants (Eitle et al. 2002; Johnson et al. 2011). Negative racial stereotypes and fear of black crime increase public support for punitive measures (Barkan and Cohn 2005; Hetey and Eberhardt 2014), and large minority populations heighten punitive attitudes among whites (King and Wheelock 2007). As Liska and Mitchell Chamlin (1984) summarized it, large minority populations produce “an emergent property” tied to the perceived threat of crime, which increases social pressures to control crime (p. 384). Although contemporary formulations of racial threat theory generally focus on black/white comparisons, recent shifts in the demographic landscape of American society suggest group threat may be increasingly associated with Latino defendants. This reflects what Eduardo Bonilla-Silva (2004) has called the “Latin Americanization” of race relations in the United States. Current population estimates suggest more than 55 million Americans self-identify as Latino, accounting for over 17 percent of the population and making them the largest minority group in the United States (Pew Research Center 2016). Latino growth has been disproportionate, accounting for more than half of the total population growth in the country over the past decade (Ennis, Ríos-Vargas, and Albert 2011). Public opinion polls indicate a growing concern over Latino growth among whites (Lane and Meeker 2003). Richard Alba, Ruben Rumbaut, and Karen Marotz (2005), for instance, found that approximately 3 out of 4 survey respondents believed that more immigrants are likely to cause higher crime rates, despite empirical evidence to the contrary (Ousey and Kubrin 2009; Ramey 2013). Shifting population demographics, then, may contribute to perceptions of criminal threat and fear of crime being increasingly associated with Latino groups in the United States. (Eitle and Taylor 2008). Empirical studies that incorporate ethnic threat measures generally suggest that similar threat processes characterize both blacks and Latinos (e.g., Jackson 1989; Jacobs and Carmichael 2001; Johnson et al. 2011; Ulmer and Johnson 2004). Wang (2012), for instance, recently demonstrated that the perceived size of the undocumented immigrant population in the Southwest was positively associated with perceptions of them as a criminally threatening group, and this mattered more than actual immigrant populations or local economic conditions. Moreover, some recent work suggests an even stronger role for ethnic threat relative to racial threat, at least in some social contexts (Chiricos et al. 2001). In one of the few studies to examine both subjective and objective measures of Latino threat, Brian Johnson and colleagues (2011) reported that support for “use of ethnicity” in sentencing increased in areas with greater Hispanic population growth and in areas where individuals perceived there to be greater economic and criminal threat. Victim Race and Group Threat Theory One element of racial group threat theories that has gone largely uninvestigated is the potentially important role that victim characteristics play in perceptions of criminal threat. Liska and Chamlin (1984) have argued that crime committed against minority victims will be diminished in the eyes of the white majority because it generates less perceived threat of personal victimization—a process they refer to as “benign neglect.” There is some evidence that supports this hypothesis. Aggregate research on race-specific arrest rates has found that larger minority populations may actually reduce arrests (Chamlin and Liska 1992; Liska and Chamlin 1984). Parker and colleagues (2005), for example, show that both percent black and percent black immigrant are negatively related to black arrest rates. They suggest this is because crime is more likely to be intra-racial when black populations are large, so it is discounted by social control agents. Few studies, however, directly examine race of victim effects in group threat processes. Eitle and colleagues (2002) examined separate indicators of black-on-white and white-on-white crime and found that only the former was significantly related to levels of arrest. They note that this result is consistent with the criminal threat hypothesis, which deals specifically with perceived threats to white victims, and they suggest that victim race may therefore play an important role in group threat processes. From this perspective, minority crimes involving minority victims are expected to be devalued because they pose no immediate threat to the established social order. Such notions are also consistent with long-standing theoretical arguments about the behavior of law in society (Black 1976) as well as broader discussions of the formulation of contemporary crime policy in America (Garland 2001; Simon 2007). Donald Black (1976) argues that social stratification, or the unequal distribution of resources, directly affects the quantity of law that is exercised. He places particular emphasis on the relationship between offenders and victims, noting that crimes committed by lower status offenders against higher status victims, what he refers to as “upward crimes,” will be judged as relatively more serious and result in greater mobilization of formal social control efforts. Because the state acts on behalf of victims, criminal offenses involving higher status victims result in greater legal recourse. Integrating his ideas into racial group threat theory, this implies that perceptions of criminal threat will be strongest, and support for punitive responses to crime will be greatest, in situations that involve victim-offender dyads consisting of white victims and minority suspects. In a similar vein, Jonathan Simon (2007) has argued that “victim identity is deeply racialized. It is not all victims, but primarily white, suburban, middle-class victims, whose exposure has driven waves of crime legislation” and against which “crime, poverty and … minority demographics are pushing” (p. 76). He suggests that broad shifts in criminal justice policy can be linked specifically to racialized patterns of victimization. In particular, the racial typification of crime in the media places greater emphasis on offenses involving white victims and minority offenders (Chiricos and Eschholz 2002), which may lead to enhanced support for punitive measures (Chiricos, Welch, and Gertz 2004). As Justin Pickett and Ted Chiricos (2012) argue, over time whites have come “to believe … that victims of violent crimes tend to be white” (p. 676). Travis Dixon and Daniel Linz (2000) suggest this trend is emblematic of a broader “ethnic blame discourse” in which stereotypical images of morality portray whites as law enforcers and minorities as law breakers. To the extent that racial dyads shape perceptions of criminal threat, then, they may also contribute to punitive sentiment toward minority defendants. Empirical evidence for the salience of racial dyads in criminal justice decision making is well established (e.g., Eberhardt et al. 2006; LaFree 1980; Myers 1979; Williams, Demuth, and Holcom 2007). In general, crimes that target higher status and more socially valued victims tend to be punished more severely (Baldus, Pulaski, and Woodworth 1983; Franklin and Fearn 2008). In their examination of victim characteristics in homicide, for example, Eric Baumer, Steven Messner, and Richard Felson (2000) concluded that “killings of disreputable or stigmatized victims tend to be treated more leniently by the justice system” (p. 304). A number of studies demonstrate that homicide offenses involving minority victims are punished more leniently and are less likely to eventuate in death sentences (Paternoster 1984). Indeed, Jennifer Eberhardt and colleagues (2006) observed that defendants who were perceived to be more “stereotypically” black were more likely to be sentenced to death only when their victims were white. Similar results have been demonstrated for other types of crimes and for related punishment decisions (e.g., Curry 2010; Johnson et al. 2011; Spohn and Spears 1996). Although some studies report mixed results for victim effects (Auerhahn 2007; Wooldredge et al. 2011), the weight of the evidence clearly supports their centrality for understanding racial disparities in the justice system. As Cassia Spohn (2000) summarized, “criminal punishment is contingent on the race of the victim as well as the race of the offender” (p. 469). The importance of racial dyads in criminal punishment has direct implications for our understanding of group threat perspectives. Punitive sentiments that are mobilized by perceptions of minority criminal threat are also likely to be affected by the racial characteristics of victims. The social significance of victim race may even depend on the broader racial context of the local environment. To the extent that large or growing minority populations heighten perceptions of “ecological vulnerability” on the part of whites (Covington and Taylor 1991), they may also amplify criminal threat perceptions and lead to increased support for punitiveness when minority suspects target white victims. This implies that victim race will have direct effects on punitive sentiments, and that this influence will be conditioned by the broader racial context of local social environments. Taken together, then, prior work emphasizes the importance of examining the link between the size and growth of ethnic minority populations in the United States and local perceptions of criminal threat. It highlights the significance of going beyond traditional conceptualizations of racial threat to examine ethnic threat tied to the growing Latino population in the United States. And it suggests a potentially important role for victim race and ethnicity in group threat theories and in support for punitive responses of the criminal justice system. The current study addresses these interrelated issues by delineating and testing specific theoretical hypotheses that relate group threat processes to victim race and punitive sentiment. Summary and Hypotheses Collectively, prior research and theorizing suggest a number of predictions about the role of group threat, victim race/ethnicity, and public support for enhanced punishment of minority defendants. In line with group threat perspectives, we expect the absolute size of black and Latino populations, and their recent growth, to influence punitive sentiment toward minority defendants. We incorporate both static and dynamic measures that capture the absolute size and recent change in racial and ethnic minority populations, as well as more subjective measures of perceived criminal threat from minority groups. Specifically, we expect the following: H1: The objective size and recent growth of the black and Latino populations will increase punitive sentiment among white respondents. H2: Subjective perceptions of black and Latino criminal threat will increase punitive sentiment among white respondents. Given the theoretical centrality of fear of victimization in perceptions of criminal threat, it is likely that punitive sentiment will be substantially conditioned by victim race. Crimes involving minority victims are likely to be viewed as less egregious, less demanding of public censure, and less deserving of harsh punishment because they are judged to be relatively less serious and because they pose less of a threat to the established social order. Crimes perpetrated by minority offenders against white victims, on the other hand, will incite greater racial fear and lead to increased punitiveness on the part of white respondents. These arguments are consistent with Black’s (1976) racial stratification thesis that suggests greater mobilization of legal resources will attach to victims of higher social status, particularly when social distance is greatest. The implication is that crimes involving white victims and minority offenders will increase punitive sentiment, whereas crimes involving minority victims will be devalued. Specifically, we expect the following: H3: Crimes involving white victims that are committed by black or Latino offenders will increase punitive sentiment among white respondents. H4: Crimes involving black or Latino victims that are committed by black or Latino offenders will not increase punitive sentiment among white respondents. In addition to these main effects, there are also persuasive theoretical reasons to expect that the individual effects of victim race will be conditioned by broader racial and ethnic social contexts. As others have noted, the social meaning of victim race varies by place (Xie and Lauritsen 2012). Contexts characterized by greater perceptions of criminal threat are likely to produce stronger race of victim effects. In neighborhoods where racial minorities are viewed as particularly threatening, the salience of minority-on-white crime should be enhanced, resulting in stronger emotional reactions and greater support for punitiveness toward minority defendants. Increases in both objective and subjective measures of threat are expected to significantly increase the influence of victim race on punitive sentiment. We therefore investigate the following: H5: The effect of victim race on punitive sentiment will be stronger in social contexts characterized by large or growing black and Latino populations. H6: The effect of victim race on punitive sentiment will be stronger in social contexts characterized by greater subjective perceptions of black and Latino criminal threat. DATA AND METHOD The data analyzed in this study were gathered through a national random telephone survey of 2,736 American adults (18 and older), using random-digit dialing and computer assisted telephone interviewing (CATI) to ensure accuracy in recording data. The sampling frame includes households with either landlines or cellular phones. The telephone surveys were conducted during the spring, summer, and fall of 2013. The survey focused primarily on respondents’ attitudes about punishment, crime, residential preferences, ethnicity, and immigration. The data set contains a rich variety of information and offers a unique opportunity to examine punitive sentiment as a result of group threat processes and victim/offender relationships, something that has largely been unaddressed in prior studies. As such, these survey data are uniquely suited to our research questions. A two-stage modified Mitofsky–Waksberg sampling design was utilized to develop the random-digit dialing sample (Tourangeau 2004). Respondents were limited to one adult resident per household who was 18 years or older. From each household sampled, the adult respondent with the most recent birthday was selected, an efficient way to randomly choose adults within households (Kish 1965). Trained interviewers conducted the telephone interviews and were closely monitored by supervisors. Additionally, to minimize interviewer error, supervisors reviewed 10 percent of completed interviews for accuracy by comparing selected responses to digitally recorded excerpts of interviews. There was 93 percent agreement between supervisors and interviewers. In the 7 percent of cases where there was not agreement, the supervisors and interviewers met to reconcile the discrepancy. A five call-back rule was employed before replacement of households. Using the definition recommended by the American Association for Public Opinion Research (AAPOR 2008), we obtained a 60.8 percent response rate among all contacts with eligible respondents.1 Cases of unknown eligibility, such as answering machines, busy signals, no answer, and known ineligibility, such as disconnected numbers, businesses, and fax numbers, were excluded from this calculation as recommended by AAPOR (2008). The response rate is comparable to studies that use rigorous survey methodologies (Pew Research Center 2004), as well as other recent studies utilizing telephone surveys (e.g., McCarty et al. 2006). Additionally, 94 percent of all surveys initiated were completed. This completion rate was substantially higher than the 60 percent average for national telephone interviews (Weisberg, Krosnick, and Bowen 1989). The final sample consisted of 2,408 non-Latino white respondents.2 About 44 percent of the sample was male. The age of the sample ranged from 18 to 81 with an average of 42 years. As for educational attainment, 46 percent of the sample graduated from college. The breakdown for annual household income was as follows: about 34 percent of the sample reported earning less than $50,000; around 31 percent of the respondents earned between $50,000 and $75,000; 20 percent of participants earned between $75,000 and $100,000; and about 15 percent of the sample reported earning more than $100,000. The mean family income of the sample was $63,539. Approximately 53 percent of the participants reported being married. About 69 percent of the respondents reported owning their home. In regard to geographic census region, 55 percent of the sample lived in the South, 12 percent in the Northeast, 18 percent in the Midwest, and 15 percent in the West. There is an overrepresentation of white, female, older, and higher-income respondents when compared to the 2010 Census, which is not uncommon in telephone surveys (Lavrakas 1987).3 To assess the effects of black and Latino population contexts on whites’ punitiveness, we matched respondents to the 168 counties where they resided and appended to the individual-level records, county-level data from the U.S. Census Bureau. The number of respondents in a county averaged about 14 and ranged from 9 to 31. Dependent Variables Our two dependent variables, punitive-black sentiment and punitive-Latino sentiment, are measured using six questions for each measure. Specifically, the questions ask respondents whether or not they agree with the following statements: “The U.S. criminal justice system is too lenient with [black/Latino] offenders; The U.S. criminal justice system needs tougher prison sentences on [black/Latino] offenders; The U.S. criminal justice system needs tougher prison sentences on [black/Latino] repeat offenders; In the U.S., [black/Latino] offenders should be punished severely for violating misdemeanor laws; In the U.S., [black/Latino] offenders convicted of a violent crime (i.e., murder, aggravated assault, rape, or robbery) should receive the death penalty; In the U.S., [black/Latino] offenders convicted of a property crime (i.e., burglary, larceny theft, or auto theft) should receive the death penalty.”4 The response options were coded so that higher scores indicated stronger agreement with each statement (0 = strongly disagree, 1 = disagree, 2 = agree, and 3 = strongly agree). The items were combined into two indexes that yielded alpha values of .93 for punitive-black sentiment and .89 for punitive-Latino sentiment. Thus, the dependent variables range from 0 (low) through 18 (high). Descriptive statistics for the dependent variables, as well as all of the study variables, are provided in Table 1. Table 1. Descriptive Statistics for Study Variables Mean SD Dependent variables  Punitive-black sentiment 9.97 3.29  Punitive-Latino sentiment 9.04 3.12 Independent variables  Victim/offender dyad   White victim/black offender .35 .41   Black victim/black offender .05 .22   Latino victim/black offender .06 .24  Victim/offender dyad   White victim/Latino offender .31 .43   Black victim/Latino offender .04 .20   Latino victim/Latino offender .05 .22  Perceived threat   Black criminal threat 7.93 3.05   Latino criminal threat 5.87 2.63  Population context   Percent Latino .18 .14   Latino growth .05 .03   Percent black .20 .13   Black growth .04 .03  County controls   Homicide rate (per 100,000) 9.26 4.12   Concentrated disadvantage 3.59 1.37   Percent Republican .49 .12   Population structure 5.89 .75  Demographic controls   Age 42.17 15.09   Male .44 .50   Married .53 .31   Education level (college graduate) .46 .31   Family income $63,539 $12,145   Employed .49 .50   Political conservative .51 .30   Own home .69 .33   Southwest .20 .32   Northeast .12 .35   Midwest .18 .43   West .15 .38   South .55 .32   General punitive attitudes 6.99 2.04 N1 = 2,408 individuals N2 = 168 counties Mean SD Dependent variables  Punitive-black sentiment 9.97 3.29  Punitive-Latino sentiment 9.04 3.12 Independent variables  Victim/offender dyad   White victim/black offender .35 .41   Black victim/black offender .05 .22   Latino victim/black offender .06 .24  Victim/offender dyad   White victim/Latino offender .31 .43   Black victim/Latino offender .04 .20   Latino victim/Latino offender .05 .22  Perceived threat   Black criminal threat 7.93 3.05   Latino criminal threat 5.87 2.63  Population context   Percent Latino .18 .14   Latino growth .05 .03   Percent black .20 .13   Black growth .04 .03  County controls   Homicide rate (per 100,000) 9.26 4.12   Concentrated disadvantage 3.59 1.37   Percent Republican .49 .12   Population structure 5.89 .75  Demographic controls   Age 42.17 15.09   Male .44 .50   Married .53 .31   Education level (college graduate) .46 .31   Family income $63,539 $12,145   Employed .49 .50   Political conservative .51 .30   Own home .69 .33   Southwest .20 .32   Northeast .12 .35   Midwest .18 .43   West .15 .38   South .55 .32   General punitive attitudes 6.99 2.04 N1 = 2,408 individuals N2 = 168 counties Table 1. Descriptive Statistics for Study Variables Mean SD Dependent variables  Punitive-black sentiment 9.97 3.29  Punitive-Latino sentiment 9.04 3.12 Independent variables  Victim/offender dyad   White victim/black offender .35 .41   Black victim/black offender .05 .22   Latino victim/black offender .06 .24  Victim/offender dyad   White victim/Latino offender .31 .43   Black victim/Latino offender .04 .20   Latino victim/Latino offender .05 .22  Perceived threat   Black criminal threat 7.93 3.05   Latino criminal threat 5.87 2.63  Population context   Percent Latino .18 .14   Latino growth .05 .03   Percent black .20 .13   Black growth .04 .03  County controls   Homicide rate (per 100,000) 9.26 4.12   Concentrated disadvantage 3.59 1.37   Percent Republican .49 .12   Population structure 5.89 .75  Demographic controls   Age 42.17 15.09   Male .44 .50   Married .53 .31   Education level (college graduate) .46 .31   Family income $63,539 $12,145   Employed .49 .50   Political conservative .51 .30   Own home .69 .33   Southwest .20 .32   Northeast .12 .35   Midwest .18 .43   West .15 .38   South .55 .32   General punitive attitudes 6.99 2.04 N1 = 2,408 individuals N2 = 168 counties Mean SD Dependent variables  Punitive-black sentiment 9.97 3.29  Punitive-Latino sentiment 9.04 3.12 Independent variables  Victim/offender dyad   White victim/black offender .35 .41   Black victim/black offender .05 .22   Latino victim/black offender .06 .24  Victim/offender dyad   White victim/Latino offender .31 .43   Black victim/Latino offender .04 .20   Latino victim/Latino offender .05 .22  Perceived threat   Black criminal threat 7.93 3.05   Latino criminal threat 5.87 2.63  Population context   Percent Latino .18 .14   Latino growth .05 .03   Percent black .20 .13   Black growth .04 .03  County controls   Homicide rate (per 100,000) 9.26 4.12   Concentrated disadvantage 3.59 1.37   Percent Republican .49 .12   Population structure 5.89 .75  Demographic controls   Age 42.17 15.09   Male .44 .50   Married .53 .31   Education level (college graduate) .46 .31   Family income $63,539 $12,145   Employed .49 .50   Political conservative .51 .30   Own home .69 .33   Southwest .20 .32   Northeast .12 .35   Midwest .18 .43   West .15 .38   South .55 .32   General punitive attitudes 6.99 2.04 N1 = 2,408 individuals N2 = 168 counties Independent Variables Victim/Offender Dyads To capture the racial and ethnic composition of the victim/offender dyads, our questions deal directly with scenarios that involve offenders who are either black or Latino. In addition, we systematically vary the race and ethnicity of the victims in the survey. Respondents were asked whether or not they thought violent crimes were more serious if the offender was black or Latino and the victims were either white, black or Latino. For example, in the punitive-black sentiment equations, to measure the white victim/black offender dyad, respondents answered the following questions: “In the U.S., are criminal acts more serious when a black offender commits a violent crime against a white victim?”5 The black victim/black offender dyad was measured by the following question: “In the U.S., are criminal acts more serious when a black offender commits a violent crime against a black victim?” The Latino victim/black offender dyad was measured by the following question: “In the U.S., are criminal acts more serious when a black offender commits a violent crime against a Hispanic victim?” For the models in which punitive-Latino sentiment was the outcome, the white victim/Latino offender dyad was measured by the following question: “In the U.S., are criminal acts more serious when a Hispanic offender commits a violent crime against a white victim?” The black victim/Latino offender dyad was measured by the following question: “In the U.S., are criminal acts more serious when a Hispanic offender commits a violent crime against a black victim?” The Latino victim/Latino offender dyad was measured by the following question: “In the U.S., are criminal acts more serious when a Hispanic offender commits a violent crime against a Hispanic victim?” The response categories were coded as 1 if the respondents marked yes to the question and 0 if they indicated no. Each item was treated as orthogonal in measurement and also in the statistical analysis below. This approach allows us to not only hold the race and ethnicity of the offender constant, but also to vary the racial and ethnic profile of the victims. By doing so, we are able to leverage respondents’ sensitivity to the race and ethnicity of the victim and offender in levels of public support for severity of punishment. Perceived Threat Measures We examine two subjective measures of threat: black criminal threat and Latino criminal threat. Both constructs are measured using four statements respectively: ‘‘[blacks/Latinos] pose a greater threat to public order and safety than other racial or ethnic groups; [blacks/Latinos] hurt the U.S. by committing more violent crimes than other racial or ethnic groups; [blacks/Latinos] commit most of the crime in the United States; and The United States needs to put more police on the streets, to protect law abiding citizens from [blacks/Latinos].’’ There are four response categories for the statements (0 = strongly disagree, 1 = disagree, 2 = agree, and 3 = strongly agree). The alpha coefficient for black criminal threat is .86, while it is .87 for Latino criminal threat. Importantly, these measures of perceived threat are similar to those used in prior studies that assess racial and ethnic threat processes, making them consistent with previous research (e.g., Johnson et al. 2011; Stewart et al. 2015; Stults and Baumer 2007). Objective Population Contexts Consistent with group threat perspectives, we incorporate objective indicators of racial and ethnic population contexts into the study of non-Latino white survey respondents. To capture black population dynamics, we used two county-level indicators: percent black and black growth. Both items are drawn from the U.S. county census data. Percent black was drawn from the 2010 Census and represents percentage of blacks who resided in the respondents’ counties. We also incorporated an indicator of change—black growth. This measure represents the difference between the percentage of residents identified as black in 2000 and 2010 in sample members’ counties (Green, Strolovitch, and Wong 1998). Latino population context is measured by incorporating two county-level indicators: percent Latino and Latino growth. Both items are drawn from the U.S. county census data. The measure of percent Latino was drawn from the 2010 U.S. county census data. We used 2000 decennial census data and the 2010 census data to measure Latino growth, an indicator that represents the difference between the percent of residents identified as Hispanic in 2000 and 2010 in respondents’ counties (Green et al. 1998; Stewart et al. 2015). Together these measures capture both the relative size of black and Latino populations and recent population changes in counties where our respondents are located. Control Variables We controlled for a host of additional county- and individual-level factors that have been linked to punitive attitudes and punishment outcomes in prior work (Holmes 2000; Kent and Jacobs 2005; Smith and Holmes 2003). At the county level, these factors include the following: homicide rate, concentrated disadvantage, percent Republican, and population structure. At the individual level, we controlled for the following factors: age, marital status, education level, family income, employment status, political conservative, homeownership, Northeast, Midwest, West, South, Southwest, and general punitive attitudes. Appendix Table A1 provides the full details about the metrics and definitions for these variables.6 Analytic Strategy We estimated multilevel linear models to examine how victim/offender racial and ethnic dyads, subjective perceptions of criminal threat, and objective population contexts are related to punitive sentiment among white survey respondents. Multilevel models are useful for dealing with the non-independence of observations within higher order structural groupings. Multilevel models are appropriate here because we are interested in individual-level outcomes that may be influenced by both individual- and contextual-level characteristics. In our analysis, respondents are nested within counties, and ignoring this clustering could underestimate standard errors of parameter estimates possibly leading to Type I error in which the wrong conclusions are observed for nonexistent relationships (Raudenbush and Bryk 2002). Multilevel modeling accounts for this form of non-independence and produces correct estimates of the standard errors (Raudenbush and Bryk 2002). This technique also is useful because it allows us to isolate the independent effects of both individual- and county-level variables, as well as test for cross-level interaction effects.7 All variables are grand-mean centered and estimated using the multilevel function in the STATA 14 program (Rabe-Hesketh and Skrondal 2008).8 Our analysis proceeds in three stages. We first estimate two-level random-intercept models of punitive-black sentiment and punitive-Latino sentiment that include subjective perceptions of criminal threat and objective indicators of population context, along with both individual- and county-level control variables. Second, we build on these multivariate models by incorporating racial and ethnic victim/offender dyads into models of both punitive-black and punitive-Latino sentiment. Third, we test whether the influence of the victim/offender dyads on punitive-black sentiment and punitive-Latino sentiment is conditioned by perceived criminal threat and county-level black and Latino population contexts. We estimate random slope models in which the slopes for victim/offender dyads are allowed to vary across counties and are modeled as a function of perceived criminal threat and black and Latino population contexts respectively. RESULTS Table 1 reports descriptive statistics for the variables used in the analysis. Punitive sentiment towards blacks and Latinos is moderately high, averaging between 9 and 10 points on the 18-point scale. With regard to the victim/offender dyads, the perceived severity of criminal acts varied starkly by race of the victim. When the victim was white and the offender was black, 35 percent of the respondents viewed criminal acts as more serious. A similar pattern was observed when the victim was white and the offender was Latino. The percentages were much lower for scenarios involving minority victims. Indeed, no more than 6 percent of the respondents viewed the offense as relatively more serious when both the victim and offender were black or Latino.9 To investigate whether victim/offender racial and ethnic dyads, perceptions of threat, and population contexts predict punitive sentiment, we turn to our multilevel analysis. The results of multilevel linear regression models aimed at identifying the factors that help to explain punitive-black sentiment are presented in Table 2. Before turning to our main substantive results, we note that several control variables emerged as significant predictors of punitive-black sentiment. In particular, punitive sentiment is more likely among males, homeowners, political conservatives, and those who hold more general punitive attitudes. In addition, respondents in counties with high violence rates and a large republican voting base have higher levels of punitive-black sentiment. On the other hand, we found no evidence that concentrated disadvantage or population structure affect punitive-black sentiment. Overall, these results are largely consistent with those reported in previous studies that have examined formal social control (e.g., Johnson et al. 2011). Table 2. Multilevel Models of Punitive-Black Sentiment Regressed on Victim/Offender Dyad and Threat Indicators Model 1 Model 2 Model 3 Model 4 Variables b SE B SE b SE b SE Victim/offender dyad  White victim/black offender − − .224* .061 − − − −  Black Victim/black offender − − − − −.287 .461 − −  Latino Victim/black offender − − − − − − −.109 .416 Perceived black threat  Black criminal threat .197* .031 .189* .025 .196* .025 .196* .025 Black population context  Percent black −.748 .902 −.748 .902 −.748 .902 −.748 .902  Black growth .137* .046 .132* .046 .134* .046 .135* .046 County controls  Homicide rate (per 100,000) .042* .014 .038* .014 .044* .014 .044* .014  Concentrated disadvantage −.175 .248 −.175 .248 −.175 .248 −.175 .248  Percent Republican .349* .142 .338* .142 .349* .142 .349* .142  Population structure −.097 .171 −.097 .171 −.097 .171 −.097 .171 Demographic controls  Age −.006 .008 −.006 .008 −.006 .008 −.006 .008  Male .551* .191 .549* .191 .551* .191 .551* .191  Married .362 .224 .362 .224 .362 .224 .362 .224  Education level −.365 .229 −.365 .229 −.365 .229 −.365 .229  Family income −.152 .109 −.152 .109 −.152 .109 −.152 .109  Employed −.303 .215 −.303 .215 −.303 .215 −.303 .215  Political conservative .437* .175 .431* .175 .437* .175 .437* .175  Own home .113* .027 .108* .027 .113* .027 .113* .027  Southwest .452 .312 .452 .312 .452 .312 .452 .312  Northeast −.988* .293 −.988* .293 −.988* .293 −.988* .293  Midwest −.261 .299 −.261 .299 −.261 .299 −.261 .299  West −.763* .273 −.763* .273 −.763* .273 −.763* .273  General punitive attitudes .319* .038 .314* .038 .319* .038 .319* .038 Intercept 9.475* .618 9.433* .618 9.475* .618 9.471* .618 Total variance explained (%) 18.4 21.1 18.2 18.4 N1 = 2,408 individuals N2 = 168 counties Model 1 Model 2 Model 3 Model 4 Variables b SE B SE b SE b SE Victim/offender dyad  White victim/black offender − − .224* .061 − − − −  Black Victim/black offender − − − − −.287 .461 − −  Latino Victim/black offender − − − − − − −.109 .416 Perceived black threat  Black criminal threat .197* .031 .189* .025 .196* .025 .196* .025 Black population context  Percent black −.748 .902 −.748 .902 −.748 .902 −.748 .902  Black growth .137* .046 .132* .046 .134* .046 .135* .046 County controls  Homicide rate (per 100,000) .042* .014 .038* .014 .044* .014 .044* .014  Concentrated disadvantage −.175 .248 −.175 .248 −.175 .248 −.175 .248  Percent Republican .349* .142 .338* .142 .349* .142 .349* .142  Population structure −.097 .171 −.097 .171 −.097 .171 −.097 .171 Demographic controls  Age −.006 .008 −.006 .008 −.006 .008 −.006 .008  Male .551* .191 .549* .191 .551* .191 .551* .191  Married .362 .224 .362 .224 .362 .224 .362 .224  Education level −.365 .229 −.365 .229 −.365 .229 −.365 .229  Family income −.152 .109 −.152 .109 −.152 .109 −.152 .109  Employed −.303 .215 −.303 .215 −.303 .215 −.303 .215  Political conservative .437* .175 .431* .175 .437* .175 .437* .175  Own home .113* .027 .108* .027 .113* .027 .113* .027  Southwest .452 .312 .452 .312 .452 .312 .452 .312  Northeast −.988* .293 −.988* .293 −.988* .293 −.988* .293  Midwest −.261 .299 −.261 .299 −.261 .299 −.261 .299  West −.763* .273 −.763* .273 −.763* .273 −.763* .273  General punitive attitudes .319* .038 .314* .038 .319* .038 .319* .038 Intercept 9.475* .618 9.433* .618 9.475* .618 9.471* .618 Total variance explained (%) 18.4 21.1 18.2 18.4 N1 = 2,408 individuals N2 = 168 counties * p < .05 (two-tailed test) Table 2. Multilevel Models of Punitive-Black Sentiment Regressed on Victim/Offender Dyad and Threat Indicators Model 1 Model 2 Model 3 Model 4 Variables b SE B SE b SE b SE Victim/offender dyad  White victim/black offender − − .224* .061 − − − −  Black Victim/black offender − − − − −.287 .461 − −  Latino Victim/black offender − − − − − − −.109 .416 Perceived black threat  Black criminal threat .197* .031 .189* .025 .196* .025 .196* .025 Black population context  Percent black −.748 .902 −.748 .902 −.748 .902 −.748 .902  Black growth .137* .046 .132* .046 .134* .046 .135* .046 County controls  Homicide rate (per 100,000) .042* .014 .038* .014 .044* .014 .044* .014  Concentrated disadvantage −.175 .248 −.175 .248 −.175 .248 −.175 .248  Percent Republican .349* .142 .338* .142 .349* .142 .349* .142  Population structure −.097 .171 −.097 .171 −.097 .171 −.097 .171 Demographic controls  Age −.006 .008 −.006 .008 −.006 .008 −.006 .008  Male .551* .191 .549* .191 .551* .191 .551* .191  Married .362 .224 .362 .224 .362 .224 .362 .224  Education level −.365 .229 −.365 .229 −.365 .229 −.365 .229  Family income −.152 .109 −.152 .109 −.152 .109 −.152 .109  Employed −.303 .215 −.303 .215 −.303 .215 −.303 .215  Political conservative .437* .175 .431* .175 .437* .175 .437* .175  Own home .113* .027 .108* .027 .113* .027 .113* .027  Southwest .452 .312 .452 .312 .452 .312 .452 .312  Northeast −.988* .293 −.988* .293 −.988* .293 −.988* .293  Midwest −.261 .299 −.261 .299 −.261 .299 −.261 .299  West −.763* .273 −.763* .273 −.763* .273 −.763* .273  General punitive attitudes .319* .038 .314* .038 .319* .038 .319* .038 Intercept 9.475* .618 9.433* .618 9.475* .618 9.471* .618 Total variance explained (%) 18.4 21.1 18.2 18.4 N1 = 2,408 individuals N2 = 168 counties Model 1 Model 2 Model 3 Model 4 Variables b SE B SE b SE b SE Victim/offender dyad  White victim/black offender − − .224* .061 − − − −  Black Victim/black offender − − − − −.287 .461 − −  Latino Victim/black offender − − − − − − −.109 .416 Perceived black threat  Black criminal threat .197* .031 .189* .025 .196* .025 .196* .025 Black population context  Percent black −.748 .902 −.748 .902 −.748 .902 −.748 .902  Black growth .137* .046 .132* .046 .134* .046 .135* .046 County controls  Homicide rate (per 100,000) .042* .014 .038* .014 .044* .014 .044* .014  Concentrated disadvantage −.175 .248 −.175 .248 −.175 .248 −.175 .248  Percent Republican .349* .142 .338* .142 .349* .142 .349* .142  Population structure −.097 .171 −.097 .171 −.097 .171 −.097 .171 Demographic controls  Age −.006 .008 −.006 .008 −.006 .008 −.006 .008  Male .551* .191 .549* .191 .551* .191 .551* .191  Married .362 .224 .362 .224 .362 .224 .362 .224  Education level −.365 .229 −.365 .229 −.365 .229 −.365 .229  Family income −.152 .109 −.152 .109 −.152 .109 −.152 .109  Employed −.303 .215 −.303 .215 −.303 .215 −.303 .215  Political conservative .437* .175 .431* .175 .437* .175 .437* .175  Own home .113* .027 .108* .027 .113* .027 .113* .027  Southwest .452 .312 .452 .312 .452 .312 .452 .312  Northeast −.988* .293 −.988* .293 −.988* .293 −.988* .293  Midwest −.261 .299 −.261 .299 −.261 .299 −.261 .299  West −.763* .273 −.763* .273 −.763* .273 −.763* .273  General punitive attitudes .319* .038 .314* .038 .319* .038 .319* .038 Intercept 9.475* .618 9.433* .618 9.475* .618 9.471* .618 Total variance explained (%) 18.4 21.1 18.2 18.4 N1 = 2,408 individuals N2 = 168 counties * p < .05 (two-tailed test) We now turn to our substantive focus. Model 1 provides an assessment of both perceptions of black criminal threat and population context on punitive black sentiment, net of demographic and county control variables. The results show that both perceived black criminal threat and black population growth are significantly associated with punitive black sentiment while the relative size of the black population is not significant.10 Specifically, a one-unit increase in perceived black criminal threat increases punitiveness by .197 points. Similarly, a unit increase in black population growth increases punitive-black sentiment among whites by .137 points. Moving to the next set of models, we incorporate victim/offender dyads into the predictive equations to assess their independent effects on punitive-black sentiment. Because our fundamental interest is in the influence of victim and offender race on punitiveness, we report separate models disaggregated by victim race/ethnicity while holding the race of the offender constant.11 Comparing results across Models 2 through 4 in Table 2 makes clear the important role that victim race plays in public support for punitiveness toward black offenders. When subjects were asked about scenarios involving white victims and black offenders, they were substantially more likely to support harsher punishment toward blacks. Involvement of a white victim and a black offender increased punitive-black sentiment by .224 points. In contrast, crimes involving black or Latino victims had no significant effect on levels of support for punitive-black sentiment. Moreover, we observed a similar set of patterns with regard to punitive-Latino sentiment as the same set of predictors are significant in Table 3 across the various model specifications. We focus on the more substantive findings that guide our research questions. In Model 1, both perceptions of Latino criminal threat and Latino population growth are significant predictors of punitiveness toward Latino offenders. For example, a one unit increases in both perceived Latino criminal threat and Latino population growth are associated with increases in punitive sentiment of .192 and .194 points respectively. In the subsequent models (2 through 4), we include the victim/offender dyad measures. In these models, we vary the race and ethnicity of the victim while keeping the offender ethnicity constant. Similar to the results above, scenarios involving white victims and Latino offenders increased punitive-Latino sentiment by .137 points. None of the other victim contrasts emerged as significant predictors of punitive-Latino sentiment. As observed across both Tables 2 and 3, these findings suggest that attitudes favoring increased punishment for black and Latino offenders are closely tied to victim characteristics. This suggests that black and Latino offenders who target white victims are perceived to be more threatening and more deserving of punishment than blacks and Latinos who target members of other racial or ethnic minority groups. Table 3. Multilevel Models of Punitive-Latino Sentiment Regressed on Victim/Offender Dyad and Threat Indicators Model 1 Model 2 Model 3 Model 4 Variables b SE b SE b SE. b SE Victim/offender dyad  White victim/Latino offender − − .137* .023 − − − −  Black victim/Latino offender − − − − −.416 .423 − −  Latino victim/Latino offender − − − − − − −.037 .383 Perceived Latino threat  Latino criminal threat .192* .041 .198* .041 .191* .041 .192* .041 Latino population context  Percent Latino −.005 .006 −.005 .006 −.006 .006 −.006 .006  Latino growth .194* .061 .189* .061 .195* .061 .193* .061 County controls  Homicide rate (per 100,000) .064* .026 .064* .026 .064* .026 .064* .026  Concentrated disadvantage −.092 .189 −.092 .189 −.097 .189 −.097 .189  Percent Republican .355* .142 .349* .142 .354* .142 .356* .142  Population structure −.109 .168 −.109 .168 −.109 .168 −.109 .168 Demographic controls  Age .029* .011 .025* .011 .029* .011 .029* .011  Male .561* .191 .543* .191 .559* .191 .562* .191  Married .357 .255 .343 .255 .357 .255 .357 .255  Education level −.155 .208 −.153 .208 −.155 .208 −.155 .208  Family income −.214* .064 −.208 .061 −.212 .061 −.212 .061  Employed −.651* .285 −.651* .285 −.651* .285 −.651* .285  Political conservative .421* .188 .425* .188 .421* .188 .423* .188  Own home .689* .245 .683* .245 .689* .245 .689* .245  Southwest .413 .285 .413 .285 .403 .285 .403 .285  Northeast −.589* .254 −.581* .254 −.589* .254 −.589* .254  Midwest −.051 .249 −.051 .249 −.051 .249 −.051 .249  West −.348 .269 −.348 .269 −.348 .269 −.348 .269  General punitive attitudes .335* .038 .325* .038 .335* .038 .335* .038 Intercept 8.936* .461 8.901* .461 8.936* .461 8.936* .461 Total variance explained (%) 18.8 22.4 18.6 18.8 N1 = 2,408 individuals N2 = 168 counties Model 1 Model 2 Model 3 Model 4 Variables b SE b SE b SE. b SE Victim/offender dyad  White victim/Latino offender − − .137* .023 − − − −  Black victim/Latino offender − − − − −.416 .423 − −  Latino victim/Latino offender − − − − − − −.037 .383 Perceived Latino threat  Latino criminal threat .192* .041 .198* .041 .191* .041 .192* .041 Latino population context  Percent Latino −.005 .006 −.005 .006 −.006 .006 −.006 .006  Latino growth .194* .061 .189* .061 .195* .061 .193* .061 County controls  Homicide rate (per 100,000) .064* .026 .064* .026 .064* .026 .064* .026  Concentrated disadvantage −.092 .189 −.092 .189 −.097 .189 −.097 .189  Percent Republican .355* .142 .349* .142 .354* .142 .356* .142  Population structure −.109 .168 −.109 .168 −.109 .168 −.109 .168 Demographic controls  Age .029* .011 .025* .011 .029* .011 .029* .011  Male .561* .191 .543* .191 .559* .191 .562* .191  Married .357 .255 .343 .255 .357 .255 .357 .255  Education level −.155 .208 −.153 .208 −.155 .208 −.155 .208  Family income −.214* .064 −.208 .061 −.212 .061 −.212 .061  Employed −.651* .285 −.651* .285 −.651* .285 −.651* .285  Political conservative .421* .188 .425* .188 .421* .188 .423* .188  Own home .689* .245 .683* .245 .689* .245 .689* .245  Southwest .413 .285 .413 .285 .403 .285 .403 .285  Northeast −.589* .254 −.581* .254 −.589* .254 −.589* .254  Midwest −.051 .249 −.051 .249 −.051 .249 −.051 .249  West −.348 .269 −.348 .269 −.348 .269 −.348 .269  General punitive attitudes .335* .038 .325* .038 .335* .038 .335* .038 Intercept 8.936* .461 8.901* .461 8.936* .461 8.936* .461 Total variance explained (%) 18.8 22.4 18.6 18.8 N1 = 2,408 individuals N2 = 168 counties * p < .05 (two-tailed test) Table 3. Multilevel Models of Punitive-Latino Sentiment Regressed on Victim/Offender Dyad and Threat Indicators Model 1 Model 2 Model 3 Model 4 Variables b SE b SE b SE. b SE Victim/offender dyad  White victim/Latino offender − − .137* .023 − − − −  Black victim/Latino offender − − − − −.416 .423 − −  Latino victim/Latino offender − − − − − − −.037 .383 Perceived Latino threat  Latino criminal threat .192* .041 .198* .041 .191* .041 .192* .041 Latino population context  Percent Latino −.005 .006 −.005 .006 −.006 .006 −.006 .006  Latino growth .194* .061 .189* .061 .195* .061 .193* .061 County controls  Homicide rate (per 100,000) .064* .026 .064* .026 .064* .026 .064* .026  Concentrated disadvantage −.092 .189 −.092 .189 −.097 .189 −.097 .189  Percent Republican .355* .142 .349* .142 .354* .142 .356* .142  Population structure −.109 .168 −.109 .168 −.109 .168 −.109 .168 Demographic controls  Age .029* .011 .025* .011 .029* .011 .029* .011  Male .561* .191 .543* .191 .559* .191 .562* .191  Married .357 .255 .343 .255 .357 .255 .357 .255  Education level −.155 .208 −.153 .208 −.155 .208 −.155 .208  Family income −.214* .064 −.208 .061 −.212 .061 −.212 .061  Employed −.651* .285 −.651* .285 −.651* .285 −.651* .285  Political conservative .421* .188 .425* .188 .421* .188 .423* .188  Own home .689* .245 .683* .245 .689* .245 .689* .245  Southwest .413 .285 .413 .285 .403 .285 .403 .285  Northeast −.589* .254 −.581* .254 −.589* .254 −.589* .254  Midwest −.051 .249 −.051 .249 −.051 .249 −.051 .249  West −.348 .269 −.348 .269 −.348 .269 −.348 .269  General punitive attitudes .335* .038 .325* .038 .335* .038 .335* .038 Intercept 8.936* .461 8.901* .461 8.936* .461 8.936* .461 Total variance explained (%) 18.8 22.4 18.6 18.8 N1 = 2,408 individuals N2 = 168 counties Model 1 Model 2 Model 3 Model 4 Variables b SE b SE b SE. b SE Victim/offender dyad  White victim/Latino offender − − .137* .023 − − − −  Black victim/Latino offender − − − − −.416 .423 − −  Latino victim/Latino offender − − − − − − −.037 .383 Perceived Latino threat  Latino criminal threat .192* .041 .198* .041 .191* .041 .192* .041 Latino population context  Percent Latino −.005 .006 −.005 .006 −.006 .006 −.006 .006  Latino growth .194* .061 .189* .061 .195* .061 .193* .061 County controls  Homicide rate (per 100,000) .064* .026 .064* .026 .064* .026 .064* .026  Concentrated disadvantage −.092 .189 −.092 .189 −.097 .189 −.097 .189  Percent Republican .355* .142 .349* .142 .354* .142 .356* .142  Population structure −.109 .168 −.109 .168 −.109 .168 −.109 .168 Demographic controls  Age .029* .011 .025* .011 .029* .011 .029* .011  Male .561* .191 .543* .191 .559* .191 .562* .191  Married .357 .255 .343 .255 .357 .255 .357 .255  Education level −.155 .208 −.153 .208 −.155 .208 −.155 .208  Family income −.214* .064 −.208 .061 −.212 .061 −.212 .061  Employed −.651* .285 −.651* .285 −.651* .285 −.651* .285  Political conservative .421* .188 .425* .188 .421* .188 .423* .188  Own home .689* .245 .683* .245 .689* .245 .689* .245  Southwest .413 .285 .413 .285 .403 .285 .403 .285  Northeast −.589* .254 −.581* .254 −.589* .254 −.589* .254  Midwest −.051 .249 −.051 .249 −.051 .249 −.051 .249  West −.348 .269 −.348 .269 −.348 .269 −.348 .269  General punitive attitudes .335* .038 .325* .038 .335* .038 .335* .038 Intercept 8.936* .461 8.901* .461 8.936* .461 8.936* .461 Total variance explained (%) 18.8 22.4 18.6 18.8 N1 = 2,408 individuals N2 = 168 counties * p < .05 (two-tailed test) Furthermore, regardless of the race of the victim, there is clear evidence that both perceptual and objective measures of racial and ethnic group threat influence punitive sentiment toward black and Latino offenders. In all models, perceptual measures of both black and Latino criminal threats have significant effects on the outcomes. Although the absolute size of the black and Latino populations are not significant predictors of support for harsher punishment, both black and Latino growth demonstrate strong effects. Punitive sentiment is significantly greater in counties that experienced a recent influx of black or Latino residents. Notably, these effects are independent of all the individual and contextual level controls included in the model. In addition to the main effects for victim race in punishment, prior theorizing also suggests these influences may be conditioned by characteristics of the broader surrounding social context. Tables 4 and 5 investigate the extent to which victim effects are contingent upon perceptions of black and Latino criminal threats and objective indicators of black and Latino population growth. These analyses include the same battery of variables in Models 2 from Tables 2 and 3 but report only the focal coefficients in the interest of space.12 Given that concerns over fear of crime and risk of victimization are closely tied to the race of the victim, we expect stronger victim race effects among whites when racial and ethnic threat is high. Table 4. Multilevel Models of Punitive-Black Sentiment Regressed on Interactions Between Victim/Offender Dyad and Threat Indicators Model 1 Model 2 Variables b SE b SE Victim/offender dyad  White victim/black offender .179* .061 .177* .061 Perceived black threat  Black criminal threat .209* .028 .103* .025 Black population context  Black growth .131* .046 .129* .046 Black threat and victim/offender dyad  Black criminal threat * white victim/black offender .254* .071 − − Black population context and victim/offender dyad  Black growth * white victim/black offender − − .197* .068 Intercept 9.433* .618 9.333* .618 Total variance explained (%) 23.8 22.9 N1 = 2,408 individuals N2 = 168 counties Model 1 Model 2 Variables b SE b SE Victim/offender dyad  White victim/black offender .179* .061 .177* .061 Perceived black threat  Black criminal threat .209* .028 .103* .025 Black population context  Black growth .131* .046 .129* .046 Black threat and victim/offender dyad  Black criminal threat * white victim/black offender .254* .071 − − Black population context and victim/offender dyad  Black growth * white victim/black offender − − .197* .068 Intercept 9.433* .618 9.333* .618 Total variance explained (%) 23.8 22.9 N1 = 2,408 individuals N2 = 168 counties Note: Models include all variables reported in Model 2 of Table 2. * p < .05 (two-tailed test) Table 4. Multilevel Models of Punitive-Black Sentiment Regressed on Interactions Between Victim/Offender Dyad and Threat Indicators Model 1 Model 2 Variables b SE b SE Victim/offender dyad  White victim/black offender .179* .061 .177* .061 Perceived black threat  Black criminal threat .209* .028 .103* .025 Black population context  Black growth .131* .046 .129* .046 Black threat and victim/offender dyad  Black criminal threat * white victim/black offender .254* .071 − − Black population context and victim/offender dyad  Black growth * white victim/black offender − − .197* .068 Intercept 9.433* .618 9.333* .618 Total variance explained (%) 23.8 22.9 N1 = 2,408 individuals N2 = 168 counties Model 1 Model 2 Variables b SE b SE Victim/offender dyad  White victim/black offender .179* .061 .177* .061 Perceived black threat  Black criminal threat .209* .028 .103* .025 Black population context  Black growth .131* .046 .129* .046 Black threat and victim/offender dyad  Black criminal threat * white victim/black offender .254* .071 − − Black population context and victim/offender dyad  Black growth * white victim/black offender − − .197* .068 Intercept 9.433* .618 9.333* .618 Total variance explained (%) 23.8 22.9 N1 = 2,408 individuals N2 = 168 counties Note: Models include all variables reported in Model 2 of Table 2. * p < .05 (two-tailed test) Table 5. Multilevel Models of Punitive-Latino Sentiment Regressed on Interactions Between Victim/Offender Dyad and Threat Indicators Model 1 Model 2 Variables B SE B SE Victim/offender dyad  White victim/Latino offender .154* .026 .159* .025 Perceived Latino threat  Latino criminal threat .182* .045 .188* .045 Latino population context  Latino growth .189* .061 .183* .061 Latino threat and victim/offender dyad  Latino criminal threat * white victim/Latino offender .241* .076 − − Latino population context and victim/offender dyad  Latino growth * white victim/Latino offender − − .189* .067 Intercept 8.911* .461 8.919* .461 Total variance explained (%) 24.7 25.4 N1 = 2,408 individuals N2 = 168 counties Model 1 Model 2 Variables B SE B SE Victim/offender dyad  White victim/Latino offender .154* .026 .159* .025 Perceived Latino threat  Latino criminal threat .182* .045 .188* .045 Latino population context  Latino growth .189* .061 .183* .061 Latino threat and victim/offender dyad  Latino criminal threat * white victim/Latino offender .241* .076 − − Latino population context and victim/offender dyad  Latino growth * white victim/Latino offender − − .189* .067 Intercept 8.911* .461 8.919* .461 Total variance explained (%) 24.7 25.4 N1 = 2,408 individuals N2 = 168 counties Note: Models include all control variables reported in in Model 2 of Table 3. * p < .05 (two-tailed test) Table 5. Multilevel Models of Punitive-Latino Sentiment Regressed on Interactions Between Victim/Offender Dyad and Threat Indicators Model 1 Model 2 Variables B SE B SE Victim/offender dyad  White victim/Latino offender .154* .026 .159* .025 Perceived Latino threat  Latino criminal threat .182* .045 .188* .045 Latino population context  Latino growth .189* .061 .183* .061 Latino threat and victim/offender dyad  Latino criminal threat * white victim/Latino offender .241* .076 − − Latino population context and victim/offender dyad  Latino growth * white victim/Latino offender − − .189* .067 Intercept 8.911* .461 8.919* .461 Total variance explained (%) 24.7 25.4 N1 = 2,408 individuals N2 = 168 counties Model 1 Model 2 Variables B SE B SE Victim/offender dyad  White victim/Latino offender .154* .026 .159* .025 Perceived Latino threat  Latino criminal threat .182* .045 .188* .045 Latino population context  Latino growth .189* .061 .183* .061 Latino threat and victim/offender dyad  Latino criminal threat * white victim/Latino offender .241* .076 − − Latino population context and victim/offender dyad  Latino growth * white victim/Latino offender − − .189* .067 Intercept 8.911* .461 8.919* .461 Total variance explained (%) 24.7 25.4 N1 = 2,408 individuals N2 = 168 counties Note: Models include all control variables reported in in Model 2 of Table 3. * p < .05 (two-tailed test) Tables 4 and 5 provide clear evidence for these expectations. The white victim effects observed in previous analyses are particularly pronounced in social contexts characterized by greater perceived criminal threat for both blacks (γ = .254) and Latinos (γ = .241).13 In line with our theoretical expectations, then, support for enhanced social control directed at black and Latino offenders who target white victims is greatest where fear of black and Latino crime is most pronounced. Furthermore, Models 2 in Tables 4 and 5 show that the effect of victim race on support for punitiveness is also significantly conditioned by recent growth in both the black (γ = .197) and Latino (γ = .189) populations. Social contexts characterized by a recent inflow of black and Latino residents tend to place greater emphasis on severity of punishment for black and Latino offenders who target white victims. The magnitude of these effects is substantial and is perhaps best demonstrated by Figures 1 through 4, which plot the predicted values for our significant interaction effects. Figures 1 and 2 present the effects of victim race on punitive sentiment respectively by black and Latino criminal threats. In Figures 3 and 4, we illustrate these same effects for black and Latino population growth respectively. As the figures show, public support for punitiveness toward black and Latino offenders is highest in contexts where residents perceive minorities to be the most criminally threatening and in counties that experienced recent growth in the black and Latino populations. Further, these effects are dramatically more pronounced in cases involving white victims rather than minority victims (none of the minority victim effects were statistically significant). Overall, these results are consistent with expectations derived from an integration of group threat theory and contemporary perspectives on the importance of victim characteristics in punishment. Figure 1. View largeDownload slide The Effect of White Victim/Black Offender Dyad on Punitive-Black Sentiment by Perceived Black Criminal Threat Figure 1. View largeDownload slide The Effect of White Victim/Black Offender Dyad on Punitive-Black Sentiment by Perceived Black Criminal Threat Figure 2. View largeDownload slide The Effect of White Victim/Latino Offender Dyad on Punitive-Latino Sentiment by Perceived Latino Criminal Threat Figure 2. View largeDownload slide The Effect of White Victim/Latino Offender Dyad on Punitive-Latino Sentiment by Perceived Latino Criminal Threat Figure 3. View largeDownload slide The Effect of White Victim/Black Offender Dyad on Punitive-Black Sentiment by Black Population Growth Figure 3. View largeDownload slide The Effect of White Victim/Black Offender Dyad on Punitive-Black Sentiment by Black Population Growth Figure 4. View largeDownload slide The Effect of White Victim/Latino Offender Dyad on Punitive-Latino Sentiment by Latino Population Growth Figure 4. View largeDownload slide The Effect of White Victim/Latino Offender Dyad on Punitive-Latino Sentiment by Latino Population Growth Supplemental Analyses We conducted a series of supplemental analyses, specifying punitive-white sentiment as the dependent variable, to further examine the robustness of our results. To create this outcome, we relied on the six identical questions used to form the other dependent variables except whites were specified as the target group (e.g., “The U.S. criminal justice system is too lenient with [white] offenders; The U.S. criminal justice system needs tougher prison sentences on [white] offenders.”). We also created a white criminal threat measure relying on the four statements that captured perceived criminal threat but using whites as the target group (e.g., “[Whites] pose a greater threat to public order and safety than other racial or ethnic groups.”). Furthermore, we incorporated dynamic and static measures of white population size. Measures of the racial and ethnic victim/offender dyads were created specifying whites as offenders and the other racial and ethnic groups as victims. We estimated equivalent models to those in Tables 2 and 3. Only three variables emerged as statistically significant (p < .05) predictors of punitive white sentiment: general punitive attitudes, black victim/white offender, and Latino victim/white offender. General punitive attitudes increased white punitive sentiment (b = .271 SE = .053, p < .05 and b = .268 SE = .055, p < .05). On the other hand, black victim/white offender (b = −.197 SE = .068, p < .05) and Latino victim/white offender dyads significantly decreased punitive white sentiment (b = −.189 SE = .065, p < .05). These findings, in combination with those reported above, indicate that white respondents are less punitive when faced with scenarios that involve minority victims rather than white victims. Thus, even when racial and ethnic minorities are victims of crimes perpetrated by whites, the perception among white citizens is that they should not be punished as severely.14 DISCUSSION Prior research in the sociology of punishment suggests that racial group threat dynamics are related to public support for the exercise of formal social control in society (Eitle et al. 2002; Liska et al. 1985). One largely uninvestigated issue, however, is the role that victim characteristics play in perceptions of criminal threat and in the formation of punitive sentiment. Some research suggests a process of victim discounting or “benign neglect” may occur in cases that involve minority victims (Liska and Chamlin 1984). Such arguments are consistent with work that demonstrates the salience of racial dyads in criminal punishment (Swigert and Farrell 1977). To date, though, no prior research has explicitly integrated victim characteristics into the study of group threat processes. The current study addressed this issue, using nationally representative survey data to examine the effects of victim-offender racial dyads on punitive sentiment toward minority defendants in the justice system. We incorporate both objective population contexts and subjective perceptions of criminal threat, test for both stable and dynamic indicators of group threat, and examine criminal threat associated with both the black and Latino populations in the United States. Our findings suggest key roles for subjective perceptions of criminal threat, aggregate shifts in racial population dynamics, and individual victim/offender relations. Overall, we find substantial support among white respondents for punitive sentiment aimed at black and Latino offenders. Moreover, more than one third of the sample agreed that criminal acts were more serious if they involved a white victim and a black or Latino offender. This speaks to a general punitive sentiment directed at minority offenders and it provides prima facie evidence for the salience of victim/offender dyads in subjective perceptions of criminal threat. It also raises important questions about the role that the criminal justice system plays in reinforcing racial stereotypes and in shaping contemporary race relations (Kennedy 1997; Wilson 2009). Notably, these findings are also consistent with empirical research in other social domains beyond punishment disparities. For instance, negative stereotypes of racial and ethnic minority students by teachers and administrators have been linked to poor educational outcomes for minority students such as elevated levels of school failure and punishment (Dee 2005). Similarly, scholars have argued that negative stereotypical assessments of racial and ethnic minorities are linked to persistent disparities in the quality of health care and health outcomes for racial and ethnic minorities (Goodwin and Duke 2012). Thus, our findings are consistent with a broader literature on racial and ethnic disparities in other social domains, and they extend this work to suggest that negative perceptions of racial and ethnic minorities are essential for understanding and addressing these disparities. As such, studies should continue to investigate the underlying sources of individual-level perceptions and their consequences for societal inequalities. In terms of racial population contexts, aggregate patterns of punitiveness were not significantly related to the absolute size of black or Latino populations, but they were associated substantially with recent population growth for both groups. White survey respondents in counties with growing black or Latino populations expressed greater punitive sentiment toward black and Latino defendants. These results are largely consistent with prior work that suggests racial demographics are tied to group threat processes, and they support recent arguments about the importance of including dynamic measures of population change in addition to static indicators of population size (Chamlin 1989; Jacobs and O’Brien 1998; Johnson et al. 2011). These findings are congruent with “defended neighborhoods” perspectives that suggest racial group threat becomes particularly heightened in contexts characterized by rapid in-migration of minority groups, especially in racially homogenous white localities (Green et al. 1998). Whites in communities with rapidly growing minority populations may view these groups as more criminally threatening, or alternatively, individuals who view minorities as more threatening may be more sensitive to demographic population shifts. The latter explanation is largely consistent with work that finds racial threat perceptions are also related to assessments of minority group size (Gallagher 2003). With regard to subjective perceptions of group threat, we find strong and consistent evidence that criminal threat is positively related to punitive sentiment toward black and Latino offenders. Importantly, these effects emerge net of individual and aggregate controls, including general punitive attitudes and broader population demographics. Individuals who view blacks and Latinos as more criminally involved and greater threats to public safety are more likely to support punitive measures that specifically target them in the criminal justice system. In addition to highlighting the salience of criminal threat as a key theoretical mechanism in racial group threat theory (Eitle et al. 2002), these results also highlight the importance of combatting racial stereotypes in popular discourse. The racialization of crime and criminality in popular media has been well documented (Chiricos and Eschholz 2002; Dixon and Linz 2000), and may feed into implicit racial associations that contribute to inequalities in justice outcomes (Smith and Levinson 2012). Prior work, for instance, suggests that even among judges who profess strong egalitarian beliefs, implicit racial biases can still affect punishment decisions (Rachlinski et al. 2009). Thus, an essential step in trying to reduce race-based disparities in the justice system might be to train law enforcement, prosecutors, judges, juries, and others about the salience of stereotypes and how they can be consequential for the application of punishment for minorities (Correll et al. 2007; Tonry 2010). Our research findings also highlight the importance of victim/offender racial characteristics in group threat theories and in public support for punitive justice responses. We find that victim race plays an important role in punitive sentiment toward black and Latino offenders. Specifically, white respondents report greater punitiveness when asked about scenarios involving white victims and minority perpetrators. The same is not true, however, for situations involving minority victims. One in three respondents in our sample viewed criminal acts as less serious when they involved minority victims. Moreover, white respondents express less punitiveness toward white offenders who target minority victims. For a nontrivial fraction of the white majority, this suggests an ethos of victim discounting with regard to minority victims (Baldus et al. 1983). These results also bolster prior work on the perceived risk of criminal victimization in group threat processes. White respondents may express greater punitive sentiment for minority offenders when there is a white victim, in part, because it elevates their own estimates of the risk of victimization. This process reflects what Lincoln Quillian and Devah Pager (2010) referred to as “stereotype amplification”—the distorted assessment of individual risk based on racialized fears and stereotypical social cues. Finally, we investigated potential interactions among key social predictors in our model. Specifically, we expected that victim race would have stronger effects on punitive sentiment in contexts characterized by elevated perceptions of minority criminal threat and in areas with recent minority population growth. Both of these expectations were supported in the data. The effect of a white victim was notably stronger in areas where black criminal threat was high and black population growth was more rapid. This effect also increased significantly when Latino criminal threat was high and Latino population growth increased. Taken together, these results provide strong evidence that race of victim effects are conditioned by broader social contexts involving perceived criminal threat and objective population growth. Again, these various group threat dynamics operate independently of individual attitudinal and demographic predictors and other county-level controls. Although this study provides significant new insights into group threat and social control, it is not without its limitations. First, we focus on punitive sentiment. We show that group threat variables interact with victim race to shape punitive attitudes, but future work is needed that demonstrates this relationship for additional social control outcomes, such as arrest and incarceration rates. Second, we examine group threat processes that attach to black and Latino populations, yet some important differences might exist within these broad racial and ethnic categorizations. In particular, there may be unique threat mechanisms that operate for black or Latino immigrant subgroups (Johnson et al. 2011; Parker et al. 2005; Wang 2012). Unique group threat dynamics could also be at play among members of different racial and ethnic minority groups. Hubert Blumer’s (1958) original formulation of group threat theory suggested that as members of a racial group become increasingly oppressed, they are more likely to see other groups as potential threats (Bobo and Hutchings 1996). This implies that racial threat theory should apply to inter-group conflict among underrepresented minority groups as well. Replicating the current findings with larger samples of black and Latino respondents, as well as members of other underrepresented groups, could therefore prove to be enlightening. Third, although we investigate the impact of black and Latino criminal threats on punitiveness, we are not able to disentangle the unique historical legacies of racism that are associated with different racial and ethnic groups. Future research might consider incorporating racial and ethnic group histories in punishment given the tenuous historical conflict around race in the United States (Bobo and Hutchings 1996; Bonilla-Silva 2006). Fourth, we focus primarily on criminal threat, which is closely tied to social control efforts in the criminal justice system (Eitle et al. 2002), but future work might expand on this study by investigating additional threat mechanisms, such as economic competition or the relative political power of minority groups. In addition, cultural indicators of group threat, such as changing social identity, multiculturalism, and exposure to foreign language (Citrin, Reingold, and Green 1990; Newman 2013; Newman, Hartman, and Taber 2012), could also provide productive avenues for future research. Further, while we were able to examine the racial/ethnic nature of the victim/offender dyads in a general sense, we were not able to measure more nuanced racial, ethnic, and gendered relationships within victim/offender dyads. For example, it is possible that social control is strongly impacted if the offender is a minority male and the victim is a white female (Hawkins 1987; LaFree 1989). Future research could expand on our work by using survey vignettes to examine more fine-grained relationships within racial, ethnic, and gendered victim/offender dyads. Finally, we are somewhat limited in our ability to examine the underlying theoretical processes that feed into subjective perceptions of criminal threat. Ideally, these processes should be studied with longitudinal data that allows for causal ordering to be established between minority population growth, subjective perceptions of criminal threat, and changing attitudes towards crime and punishment. In conclusion, this study combined both perceptual and objective measures of group threat, examined punitive sentiment toward both black and Latino offenders, and incorporated unique measures of the race of victims into the study of group threat and social control. We find compelling evidence that both perceived criminal threat and racial and ethnic population growth are related to punitiveness among whites, and we show that these relationships are closely tied to victim race. These results are important for several reasons. They inform ongoing debates over modern racism and colorblind justice (Alexander 2010). They highlight the essential role that criminal threat and fear of victimization play in public support for tough on crime policies. And they emphasize the ongoing need to investigate the complex ways that aggregate population dynamics condition punitive attitudes regarding race, crime, and punishment. As such, research that replicates and expands this work could prove invaluable for not only advancing empirical scholarship on race and crime but also for furthering our theoretical understanding of underlying sources of racial and ethnic inequality in the American criminal justice system. Footnotes 1 The response rate was calculated using the following formula: completes/(completes + terminals + refusals). This formula is supported and developed by the American Association for Public Opinion Research (2008). 2 With regard to racial and ethnic background of the original sample (2,736), 88 percent of the sample was non-Latino white, 7 percent was black, 3 percent was Latino, and 2 percent was Asian. Since our primary interest is in how the dominant group, in our case whites, views blacks and Latinos, we restricted the sample to non-Latino white respondents. 3 The sampling frame includes American households with either landlines or cellular phones, but as with other telephone survey research, households without either form of telecommunication may be underrepresented, which could account for the overrepresentation of white, female, older, and higher-income respondents. To determine if our results are biased, we re-estimated the models by weighting the sample by the 2010 U.S. Census. Our weighting procedures produced coefficients that mirrored the ones presented in our tables. Based on these supplemental analyses, we do not see evidence that our findings are impacted in any way because we observed identical patterns of results. 4 In an effort to reduce confusion in the racial and ethnic categories in our study design, interviewers explained to respondents that we were referring exclusively to blacks/African Americans in the survey questions. Respondents were also told that we were referring exclusively to Hispanic ethnicity in the survey questions. The interviewers took tremendous care to distinguish the two racial and ethnic categories. Given our efforts to clarify the differences between racial and ethnic categories, we are confident that our measures capture support for increased social control against blacks and Latinos respectively. However, we cannot rule out the possibility that some respondents may not have interpreted the survey questions around racial and ethnic categories as mutually exclusive. 5 To ensure standardization across respondents, the interviewers explained that our use of the term violent crime was in reference to the following crimes: robbery, aggravated assaults, and homicides. 6 We estimated confirmatory factor analysis (CFA), to evaluate several of our constructs: punitive-black sentiment, punitive-Latino sentiment, black criminal threat, Latino criminal threat, concentrated disadvantage, population structure, and general punitive attitudes. We estimated the models with Analysis of Moment Structures (AMOS) version 22 (Arbuckle 2013). In each case, the factor loadings for the constructs were relatively high ranging from .61 to .88. The fit indices also indicated that the data fit the model specifications well: AGFI = .98; RMSEA = .012; and CN = 803. 7 Estimation of the intraclass correlation (ICC) for our data illuminates the utility of multilevel analysis. Specifically, the overall variance in punitive-black sentiment was 3.27, with 2.55 lying within counties and .72 between counties. This implies that about 78 percent of the variance in punitive-black sentiment lies within counties, while the remaining 22 percent falls between counties. There was also significant variation in the average level of punitive-black sentiment across counties (χ2=284, p < .05). Similarly, the total variance in punitive-Latino sentiment was 3.10, with 2.45 within counties and .65 between counties. This translates into about 79 percent of the variance in punitive-Latino sentiment being located within counties and 21 percent being located between counties. Further, there is significant variation in the average level of punitive-Latino sentiment across counties (χ2=271, p < .05). These results imply that there is sufficient variance in both punitive-black sentiment and punitive-Latino sentiment between- and within-counties to support a multilevel analysis. The intercept reliability estimates were .76 for punitive-black sentiment and .78 for punitive-Latino sentiment, indicating that the data are sufficient to generate reliable estimates. 8 Specifically, we used the XTMIXED function in STATA 14 to estimate our multilevel linear models. Moreover, to assess multicollinearity among the predictor variables, we examined the variance inflation factor (VIF). Multicollinearity does not appear to be a problem, as most of the predictors have VIFs near 1 and none has a VIF greater than 2.4, suggesting that the variables are theoretically and empirically distinct constructs. 9 Similar patterns were observed for scenarios involving Asian victims. Only six of white respondents reported that crimes were more serious when black or Latino offenders targeted Asian victims. 10 We also assessed the possibility of nonlinear effects of percent black, black growth, percent Latino, and Latino growth on our punitive sentiment outcomes. The results showed no evidence of significant nonlinear relationships between the four predictors and our outcome measures. Additionally, we explored nonlinear effects of perceived criminal threat predictors on our outcomes, which also yielded nonsignificant results. 11 We initially entered all of the victim/offender dyads into a single equation to predict punitiveness, but this resulted in multicollinearity among the victim/offender dyads. Thus, we report estimates from separate models for each victim/offender relationship. 12 In the interest of space, we only focus on the white victim and minority offender dyads because none of the other combinations were significant in either the main effects or interaction analyses. Full results are available by request. 13 We allowed the slopes for the white victim/black offender dyad (slope variance = .199, p < .05) and the white victim/Latino offender dyad (slope variance = .192, p < .05) to vary across counties. Indeed the slopes varied significantly, indicating that the relationship between racial and ethnic victim/offender dyads and punitive sentiment varies significantly across county contexts. 14 We attempted to estimate models in which racial and ethnic minorities (blacks and Latinos) were the respondents. However, the models were not able to converge to provide reliable estimates because the cell sizes were too small to estimate the specified models. APPENDIX Table A1. Metrics and Definitions for Control Variables Variable Definition Homicide rate A variable that captured the reported rate of homicides for each county during the years of 2008, 2009, and 2010. The homicide rate is calculated per 100,000 residents in the county. Concentrated disadvantage A variable measured by three county-level census indicators in 2010: percentage of persons on public assistance, percentage of households below the poverty level, and percentage of persons unemployed. These items were standardized and combined to form a measure of disadvantage. The alpha coefficient was .87. Percent Republican A variable measured as the percent of the population voting for the Republican presidential candidate in 2012. Population structure A variable measured by county population size and population density for each county. The correlation between these two items is .83. Age A variable measured in years for each respondent. Male A dichotomous variable that captured whether respondents reported being male (coded “1”) or female (coded “0”). Married A dichotomous variable that captured whether respondents reported being married (coded “1”) or not (coded “0”). Education level A dichotomous measure that captured whether respondents reported having a college degree (coded “1”) or not (coded “0”). Family income A variable measured as the total amount of household income. This variable is measured as follows: less than $15,000; $15,000 to $34,999; $35,000 to $49,999; $50,000 to $74,999; $75,000 to $99,999; and $100,000 or more. Employed A dichotomous measure that captured whether respondents reported being employed (coded “1”) or not (coded “0”). Political conservative A dichotomous measure that captured whether respondents reported being politically conservative (coded “1”) or non-politically conservative (coded “0”). Own home A dichotomous measure that captured whether respondents reported being a homeowner (coded “1”) or not (coded “0”). Census region These variables were measured by a set of dichotomous variables where 1 = Northeast, 1 = Midwest, and 1 = West with respondents who reside in the South as the reference designation. Southwest A dichotomous measure that indicated whether or not the respondent lived in the Southwest United States (coded as “1”) or not (coded as “0”) General punitive attitudes A variable measured by four questions (0 = strongly disagree to 3 = strongly agree): “Sentences for violent offenders should be more severe than they are today; Sentences for nonviolent offenders should be more severe than they are today; Sentences for repeat offenders should be more severe than they are today; The U.S. courts are too lenient with offenders.” The alpha coefficient is .88. Variable Definition Homicide rate A variable that captured the reported rate of homicides for each county during the years of 2008, 2009, and 2010. The homicide rate is calculated per 100,000 residents in the county. Concentrated disadvantage A variable measured by three county-level census indicators in 2010: percentage of persons on public assistance, percentage of households below the poverty level, and percentage of persons unemployed. These items were standardized and combined to form a measure of disadvantage. The alpha coefficient was .87. Percent Republican A variable measured as the percent of the population voting for the Republican presidential candidate in 2012. Population structure A variable measured by county population size and population density for each county. The correlation between these two items is .83. Age A variable measured in years for each respondent. Male A dichotomous variable that captured whether respondents reported being male (coded “1”) or female (coded “0”). Married A dichotomous variable that captured whether respondents reported being married (coded “1”) or not (coded “0”). Education level A dichotomous measure that captured whether respondents reported having a college degree (coded “1”) or not (coded “0”). Family income A variable measured as the total amount of household income. This variable is measured as follows: less than $15,000; $15,000 to $34,999; $35,000 to $49,999; $50,000 to $74,999; $75,000 to $99,999; and $100,000 or more. Employed A dichotomous measure that captured whether respondents reported being employed (coded “1”) or not (coded “0”). Political conservative A dichotomous measure that captured whether respondents reported being politically conservative (coded “1”) or non-politically conservative (coded “0”). Own home A dichotomous measure that captured whether respondents reported being a homeowner (coded “1”) or not (coded “0”). Census region These variables were measured by a set of dichotomous variables where 1 = Northeast, 1 = Midwest, and 1 = West with respondents who reside in the South as the reference designation. Southwest A dichotomous measure that indicated whether or not the respondent lived in the Southwest United States (coded as “1”) or not (coded as “0”) General punitive attitudes A variable measured by four questions (0 = strongly disagree to 3 = strongly agree): “Sentences for violent offenders should be more severe than they are today; Sentences for nonviolent offenders should be more severe than they are today; Sentences for repeat offenders should be more severe than they are today; The U.S. courts are too lenient with offenders.” The alpha coefficient is .88. Table A1. Metrics and Definitions for Control Variables Variable Definition Homicide rate A variable that captured the reported rate of homicides for each county during the years of 2008, 2009, and 2010. The homicide rate is calculated per 100,000 residents in the county. Concentrated disadvantage A variable measured by three county-level census indicators in 2010: percentage of persons on public assistance, percentage of households below the poverty level, and percentage of persons unemployed. These items were standardized and combined to form a measure of disadvantage. The alpha coefficient was .87. Percent Republican A variable measured as the percent of the population voting for the Republican presidential candidate in 2012. Population structure A variable measured by county population size and population density for each county. The correlation between these two items is .83. Age A variable measured in years for each respondent. Male A dichotomous variable that captured whether respondents reported being male (coded “1”) or female (coded “0”). Married A dichotomous variable that captured whether respondents reported being married (coded “1”) or not (coded “0”). Education level A dichotomous measure that captured whether respondents reported having a college degree (coded “1”) or not (coded “0”). Family income A variable measured as the total amount of household income. This variable is measured as follows: less than $15,000; $15,000 to $34,999; $35,000 to $49,999; $50,000 to $74,999; $75,000 to $99,999; and $100,000 or more. Employed A dichotomous measure that captured whether respondents reported being employed (coded “1”) or not (coded “0”). Political conservative A dichotomous measure that captured whether respondents reported being politically conservative (coded “1”) or non-politically conservative (coded “0”). Own home A dichotomous measure that captured whether respondents reported being a homeowner (coded “1”) or not (coded “0”). Census region These variables were measured by a set of dichotomous variables where 1 = Northeast, 1 = Midwest, and 1 = West with respondents who reside in the South as the reference designation. Southwest A dichotomous measure that indicated whether or not the respondent lived in the Southwest United States (coded as “1”) or not (coded as “0”) General punitive attitudes A variable measured by four questions (0 = strongly disagree to 3 = strongly agree): “Sentences for violent offenders should be more severe than they are today; Sentences for nonviolent offenders should be more severe than they are today; Sentences for repeat offenders should be more severe than they are today; The U.S. courts are too lenient with offenders.” The alpha coefficient is .88. Variable Definition Homicide rate A variable that captured the reported rate of homicides for each county during the years of 2008, 2009, and 2010. The homicide rate is calculated per 100,000 residents in the county. Concentrated disadvantage A variable measured by three county-level census indicators in 2010: percentage of persons on public assistance, percentage of households below the poverty level, and percentage of persons unemployed. These items were standardized and combined to form a measure of disadvantage. The alpha coefficient was .87. Percent Republican A variable measured as the percent of the population voting for the Republican presidential candidate in 2012. Population structure A variable measured by county population size and population density for each county. The correlation between these two items is .83. Age A variable measured in years for each respondent. Male A dichotomous variable that captured whether respondents reported being male (coded “1”) or female (coded “0”). Married A dichotomous variable that captured whether respondents reported being married (coded “1”) or not (coded “0”). Education level A dichotomous measure that captured whether respondents reported having a college degree (coded “1”) or not (coded “0”). Family income A variable measured as the total amount of household income. This variable is measured as follows: less than $15,000; $15,000 to $34,999; $35,000 to $49,999; $50,000 to $74,999; $75,000 to $99,999; and $100,000 or more. Employed A dichotomous measure that captured whether respondents reported being employed (coded “1”) or not (coded “0”). Political conservative A dichotomous measure that captured whether respondents reported being politically conservative (coded “1”) or non-politically conservative (coded “0”). Own home A dichotomous measure that captured whether respondents reported being a homeowner (coded “1”) or not (coded “0”). Census region These variables were measured by a set of dichotomous variables where 1 = Northeast, 1 = Midwest, and 1 = West with respondents who reside in the South as the reference designation. Southwest A dichotomous measure that indicated whether or not the respondent lived in the Southwest United States (coded as “1”) or not (coded as “0”) General punitive attitudes A variable measured by four questions (0 = strongly disagree to 3 = strongly agree): “Sentences for violent offenders should be more severe than they are today; Sentences for nonviolent offenders should be more severe than they are today; Sentences for repeat offenders should be more severe than they are today; The U.S. courts are too lenient with offenders.” The alpha coefficient is .88. REFERENCES Alba Richard , Rumbaut Ruben G. , Marotz Karen . 2005 . “ A Distorted Nation: Perceptions of Racial/Ethnic Group Sizes and Attitudes Toward Immigrants and Other Minorities.” Social Forces 84 : 901 - 19 . Google Scholar CrossRef Search ADS Alexander Michelle. 2010 . The New Jim Crow: Mass Incarceration in the Age of Colorblindness . New York : The New Press . American Association for Public Opinion Research (AAPOR) . 2008 . Standard Definitions: Final Dispositions of Case Codes and Outcome Rates for Surveys . Ann Arbor, MI : American Association for Public Opinion Research . Arbuckle James L. 2013 . AMOS. Version 22. Chicago : IBM SPSS Inc . Auerhahn Kathleen. 2007 . “ Adjudication Outcomes in Intimate and Non-intimate Homicides.” Homicide Studies 11 : 213 - 30 . Google Scholar CrossRef Search ADS Baik Ellen. 2012 . “Gender, Religion, and National Origin: Latinos' Attitude Toward Capital Punishment.” Journal of Social Sciences 8 : 79 - 84 . Google Scholar CrossRef Search ADS Baldus David C. , Pulaski Charles , Woodworth George . 1983 . “ Comparative Review of Death Sentences: An Empirical Study of the Georgia Experience.” Journal of Criminal Law and Criminology 74 : 661 - 753 . Google Scholar CrossRef Search ADS Barkan Steven E. , Cohn Steven F . 2005 . “ Why Whites Favor Spending More Money to Fight Crime: The Role of Racial Prejudice.” Social Problems 52 : 300 - 14 . Google Scholar CrossRef Search ADS Baumer Eric P. , Messner Steven F. , Felson Richard B . 2000 . “ The Role of Victim Characteristics in the Disposition of Murder Cases.” Justice Quarterly 17 : 281 - 307 . Google Scholar CrossRef Search ADS Baumer Eric P. , Messner Steven F. , Rosenfeld Richard . 2003 . “ Explaining Spatial Variation in Support for Capital Punishment: A Multilevel Analysis.” American Journal of Sociology 108 : 844 - 75 . Google Scholar CrossRef Search ADS Black Donald. 1976 . The Behavior of Law . New York : Academic Press . Blalock Hubert M. Jr . 1967 . Toward a Theory of Minority-Group Relations . New York : John Wiley & Sons . Blumer Herbert. 1958 . “ Race Prejudice as a Sense of Group Position.” Pacific Sociological Review 1 : 3 - 7 . Google Scholar CrossRef Search ADS Bobo Lawrence , Hutchings Vincent L . 1996 . “ Perceptions of Racial Group Competition: Extending Blumer’s Theory of Group Position to a Multiracial Social Context.” American Sociological Review 61 : 951 - 72 . Google Scholar CrossRef Search ADS Bobo Lawrence D. , Johnson Devon . 2004 . “ A Taste for Punishment: Black and White Americans’ Views on the Death Penalty and the War on Drugs.” Du Bois Review: Social Science Research on Race 1 : 151 - 80 . Bonilla-Silva Eduardo. 2004 . “ From Bi-Racial to Tri-Racial: Towards a New System of Racial Stratification in the USA.” Ethnic and Racial Studies 27 : 931 - 50 . Google Scholar CrossRef Search ADS Bonilla-Silva Eduardo. 2006 . Racism Without Racists: Color-Blind Racism and the Persistence of Racial Inequality in the United States . New York : Rowman and Littlefield . Brewer Rose M. , Heitzeg Nancy A . 2008 . “ The Racialization of Crime and Punishment: Criminal Justice, Color-Blind Racism, and the Political Economy of the Prison Industrial Complex.” American Behavioral Scientist 51 : 625 - 44 . Google Scholar CrossRef Search ADS Bridges George S. , Crutchfield Robert D . 1988 . “ Law, Social Standing, and Racial Disparities in Imprisonment.” Social Forces 66 : 699 - 724 . Google Scholar CrossRef Search ADS Bridges George S. , Crutchfield Robert D. , Simpson Edith E . 1987 . “ Crime, Social Structure, and Criminal Punishment: White and Nonwhite Rates of Imprisonment.” Social Problems 34 : 345 - 61 . Google Scholar CrossRef Search ADS Britt Chester L. 2000 . “ Social Context and Racial Disparities in Punishment Decisions.” Justice Quarterly 17 : 707 - 32 . Google Scholar CrossRef Search ADS Brooks Richard R. W. , Jeon-Slaughter Haekyung . 2001 . “ Race, Income, and Perceptions of the U.S. Court System.” Behavioral Sciences and the Law 29 : 249 - 64 . Google Scholar CrossRef Search ADS Brown Rupert. 2010 . Prejudice: Its Social Psychology . Oxford, UK : Wiley-Blackwell . Brunson Rod K. 2007 . “‘ Police Don’t Like Black People’: African-American Young Men’s Accumulated Police Experiences.” Criminology & Public Policy 6 : 71 - 102 . Google Scholar CrossRef Search ADS Chambliss William J. , Seidman Robert . 1971 . Law, Order, and Power . Reading, MA : Addison-Wesley . Chamlin Mitchell B. 1989 . “ A Macro Social Analysis of Change in Police Force Size, 1972-1982: Controlling for Static and Dynamic Influences.” Sociological Quarterly 30 : 615 - 24 . Google Scholar CrossRef Search ADS Chamlin Mitchell B. , Liska Allen E . 1992 . “Social Structure and Crime Control Revisited: The Declining Significance of Intergroup Threat.” Pp. 103-12 in Social Threat and Social Control , edited by Liska Allen E . Albany : State University of New York Press . Chiricos Ted , Welch Kelly , Gertz Marc . 2004 . “Racial Typification of Crime and Support for Punitive Measures.” Criminology 42 : 358 - 89 . Google Scholar CrossRef Search ADS Chiricos Ted , Hogan Michael , Gertz Marc . 1997 . “Racial Composition of Neighborhood and Fear of Crime.” Criminology 35 : 107 - 31 . Google Scholar CrossRef Search ADS Chiricos Ted , McEntire Ranee , Gertz Marc . 2001 . “ Perceived Racial and Ethnic Composition of Neighborhood and Perceived Risk of Crime.” Social Problems 48 : 322 - 40 . Google Scholar CrossRef Search ADS Chiricos Ted , Eschholz Sarah . 2002 . “The Racial and Ethnic Typification of Crime and the Criminal Typification of Race and Ethnicity in Local Television News.” Journal of Research in Crime and Delinquency 39 : 400 - 20 . Google Scholar CrossRef Search ADS Citrin Jack , Reingold Beth , Green Donald P . 1990 . “ American Identity and the Politics of Change.” The Journal of Politics 52 : 1124 - 54 . Google Scholar CrossRef Search ADS Correll Joshua , Wittenbrink Bernd , Park Bernadette , Judd Charles M. , Keesee Tracie , Sadler Melody S . 2007 . “ Across the Thin Blue Line: Police Officers and Racial Bias in the Decision to Shoot.” Journal of Personality and Social Psychology 92 : 1006 - 23 . Google Scholar CrossRef Search ADS Covington Jeannette , Taylor Ralph B . 1991 . “ Fear of Crime in Urban Residential Neighborhoods: Implications of Between- and Within-Neighborhood Sources for Current Models.” The Sociological Quarterly 32 : 231 - 49 . Google Scholar CrossRef Search ADS Crawford Charles , Chiricos Ted , Kleck Gary . 1998 . “ Race, Racial Threat, and Sentencing of Habitual Offenders.” Criminology 36 : 481 - 512 . Google Scholar CrossRef Search ADS Curry Theodore R. 2010 . “ The Conditional Effects of Victim and Offender Ethnicity and Victim Gender on Sentences for Non-Capital Cases.” Punishment & Society 12 : 438 - 62 . Google Scholar CrossRef Search ADS Dee Thomas S. 2005 . “A Teacher Like Me: Does Race, Ethnicity, or Gender Matter?” The American Economic Review 95 : 158 - 65 . Google Scholar CrossRef Search ADS Devine Patricia G. 1989 . “ Stereotypes and Prejudice: Their Automatic and Controlled Components.” Journal of Personality and Social Psychology 56 : 5 - 18 . Google Scholar CrossRef Search ADS Dixon Travis L. , Linz Daniel . 2000 . “ Overrepresentation and Underrepresentation of African Americans and Latinos as Lawbreakers on Television News.” Journal of Communication 50 : 131 - 54 . Google Scholar CrossRef Search ADS Eberhardt Jennifer L. , Davies Paul G. , Purdie-Vaughns Valerie J. , Johnson Sheri Lynn . 2006 . “ Looking Deathworthy: Perceived Stereotypicality of Black Defendants Predicts Capital-Sentencing Outcomes.” Psychological Science 17 : 383 - 86 . Google Scholar CrossRef Search ADS Eitle David , Taylor John . 2008 . “Are Hispanics the New ‘Threat?’ Minority Group Threat and Fear of Crime in Miami-Dade County.” Social Science Research 37 : 1102 - 15 . Google Scholar CrossRef Search ADS Eitle David , D’Alessio Stewart J. , Stolzenberg Lisa . 2002 . “ Racial Threat and Social Control: A Test of the Political, Economic, and Threat of Black Crime Hypotheses.” Social Forces 81 : 557 - 76 . Google Scholar CrossRef Search ADS Ennis Sharon R. , Ríos-Vargas Merarys , Albert Nora G . 2011 . The Hispanic Population: 2010 . Washington, DC : U.S. Department of Commerce, Economics and Statistics Administration, U.S. Census Bureau . Feldmeyer Ben , Ulmer Jeffery T . 2011 . “ Racial/Ethnic Threat and Federal Sentencing.” Journal of Research in Crime and Delinquency 48 : 238 - 70 . Google Scholar CrossRef Search ADS Franklin Cortney A. , Fearn Noelle E . 2008 . “Gender, Race, and Formal Court Decision-Making Outcomes: Chivalry/Paternalism, Conflict Theory or Gender Conflict?” Journal of Criminal Justice 36 : 279 - 90 . Google Scholar CrossRef Search ADS Gallagher Charles A. 2003 . “ Miscounting Race: Explaining Whites’ Misperceptions of Racial Group Size.” Sociological Perspectives 46 : 381 - 96 . Google Scholar CrossRef Search ADS Garland David. 2001 . The Culture of Control: Crime and Social Order in Contemporary Society . Chicago : University of Chicago Press . Gilliam Franklin D. Jr. , Iyengar Shanto . 2000 . “ Prime Suspects: The Influence of Local Television News on the Viewing Public.” American Journal of Political Science 44 : 560 - 73 . Google Scholar CrossRef Search ADS Goodwin Michele , Duke Naomi N . 2012 . “Health Law: Cognitive Bias in Medical Decision-Making.” Pp. 95 - 112 in Implicit Racial Bias Across the Law , edited by Levinson J. D. , Smith R. J . New York : Cambridge University Press . Google Scholar CrossRef Search ADS Green Donald P. , Strolovitch Dara Z. , Wong Janelle S . 1998 . “ Defended Neighborhoods, Integration, and Racially Motivated Crime.” American Journal of Sociology 104 : 372 - 403 . Google Scholar CrossRef Search ADS Greenberg David F. , West Valerie . 2001 . “ State Prison Populations and Their Growth, 1971-1991.” Criminology 39 : 615 - 53 . Google Scholar CrossRef Search ADS Hagan John , Shedd Carla , Payne Monique R . 2005 . “Race, Ethnicity, and Youth Perceptions of Criminal Injustice.” American Sociological Review 70 : 381 - 407 . Google Scholar CrossRef Search ADS Hagan John , Albonetti Celesta . 1982 . “ Race, Class, and the Perception of Criminal Injustice in America.” American Journal of Sociology 88 : 329 - 55 . Google Scholar CrossRef Search ADS Harer Miles D. , Steffensmeier Darrell . 1992 . “ The Differing Effects of Economic Inequality on Black and White Rates of Violence.” Social Forces 70 : 1035 - 54 . Google Scholar CrossRef Search ADS Hawkins Darnell F. 1987 . “ Beyond Anomalies: Rethinking the Conflict Perspective on Race and Criminal Punishment.” Social Forces 65 : 719 - 45 . Google Scholar CrossRef Search ADS Hetey Rebecca C. , Eberhardt Jennifer L . 2014 . “ Racial Disparities in Incarceration Increase Acceptance of Punitive Policies.” Psychological Science 25 : 1949 - 54 . Google Scholar CrossRef Search ADS Holmes Malcolm D. 2000 . “ Minority Threat and Police Brutality: Determinants of Civil Rights Criminal Complaints in U.S. Municipalities.” Criminology 38 : 343 - 67 . Google Scholar CrossRef Search ADS Jackson Pamela I. 1989 . Minority Group Threat, Crime, and Policing: Social Context and Social Control . New York : Praeger . Jacobs David , Carmichael Jason T . 2001 . “ The Politics of Punishment Across Time and Space: A Pooled Time-Series Analysis of Imprisonment Rates.” Social Forces 80 : 61 - 89 . Google Scholar CrossRef Search ADS Jacobs David , Carmichael Jason T. 2002 . “The Political Sociology of the Death Penalty: A Pooled Time-Series Analysis.” American Sociological Review 67 : 109 - 31 . Google Scholar CrossRef Search ADS Jacobs David , O’Brien Robert M . 1998 . “ The Determinants of Deadly Force: A Structural Analysis of Police Violence.” American Journal of Sociology 103 : 837 - 62 . Google Scholar CrossRef Search ADS Johnson Brian D. 2006 . “ The Multilevel Context of Criminal Sentencing: Integrating Judge-and County-Level Influences.” Criminology 44 : 259 - 98 . Google Scholar CrossRef Search ADS Johnson Brian D. , Stewart Eric A. , Pickett Justin , Gertz Marc . 2011 . “ Ethnic Threat and Social Control: Examining Public Support for Judicial Use of Ethnicity in Punishment.” Criminology 49 : 401 - 41 . Google Scholar CrossRef Search ADS Kennedy Randall. 1997 . Race, Crime & the Law . New York : Vintage Publishing . Kent Stephanie L. , Jacobs David . 2005 . “ Minority Threat and Police Strength from 1980 to 2000: A Fixed-Effects Analysis of Nonlinear and Interactive Effects in Large U.S . Cities.” Criminology 43 : 731 - 60 . Google Scholar CrossRef Search ADS King Ryan D. , Wheelock Darren . 2007 . “ Group Threat and Social Control: Race, Perceptions of Minorities and the Desire to Punish.” Social Forces 85 : 1255 - 80 . Google Scholar CrossRef Search ADS Kish Leslie. 1965 . Survey Sampling . New York : John Wiley & Sons . LaFree Gary D. 1980 . “ Variables Affecting Guilty Pleas and Convictions in Rape Cases: Toward a Social Theory of Rape Processing.” Social Forces 58 : 833 - 50 . Google Scholar CrossRef Search ADS LaFree Gary D. 1989 . Rape and Criminal Justice: The Social Construction of Sexual Assault . Belmont, CA : Wadsworth . Lane Jodi , Meeker James W . 2003 . “ Fear of Gang Crime: A Look at Three Theoretical Models.” Law & Society Review 37 : 425 - 56 . Google Scholar CrossRef Search ADS Lavrakas Paul J. 1987. Telephone Survey Methods: Sampling, Selection, and Supervision . Newbury Park, CA : Sage Publications . Liska Allen E. 1992 . Social Threat and Social Control . Albany : State University of New York Press . Liska Allen E. , Lawrence Joseph J. , Sanchirico Andrew . 1982 . “Fear of Crime as a Social Fact.” Social Forces 60 : 760 - 70 . Google Scholar CrossRef Search ADS Liska Allen E. , Chamlin Mitchell B . 1984 . “ Social Structure and Crime Control Among Macrosocial Units.” American Journal of Sociology 90 : 383 - 95 . Google Scholar CrossRef Search ADS Liska Allen E. , Chamlin Mitchell B. , Reed Mark D . 1985 . “ Testing the Economic Production and Conflict Models of Crime Control.” Social Forces 64 : 119 - 38 . Google Scholar CrossRef Search ADS McCarty Christopher , House Mark , Harman Jeffrey , Richards Scott . 2006 . “ Effort in Phone Survey Response Rates: The Effects of Vendor and Client-Controlled Factors.” Field Methods 18 : 172 - 88 . Google Scholar CrossRef Search ADS Mears Daniel P. , Stewart Eric A . 2010 . “ Interracial Contact and Fear of Crime.” Journal of Criminal Justice 38 : 34 - 41 . Google Scholar CrossRef Search ADS Myers Martha A. 1979 . “ Offended Parties and Official Reactions: Victims and the Sentencing of Criminal Defendants.” The Sociological Quarterly 20 : 529 - 40 . Google Scholar CrossRef Search ADS Newman Benjamin J. 2013 . “ Acculturating Contexts and Anglo Opposition to Immigration in the United States.” American Journal of Political Science 57 : 374 - 90 . Google Scholar CrossRef Search ADS Newman Benjamin J. , Hartman Todd K. , Taber Charles S . 2012 . “ Foreign Language Exposure, Cultural Threat, and Opposition to Immigration.” Political Psychology 33 : 635 - 57 . Google Scholar CrossRef Search ADS Ousey Graham C. , Kubrin Charis E . 2009 . “Exploring the Connection Between Immigration and Violent Crime Rates in U.S. Cities, 1980–2000.” Social Problems 56 : 447 - 73 . Google Scholar CrossRef Search ADS Parker Karen F. , Stults Brian J. , Rice Stephen K . 2005 . “ Racial Threat, Concentrated Disadvantage, and Social Control: Considering the Macro-Level Sources of Variation in Arrests.” Criminology 43 : 1111 - 34 . Google Scholar CrossRef Search ADS Paternoster Raymond. 1984 . “ Prosecutorial Discretion in Requesting the Death Penalty: A Case of Victim-Based Racial Discrimination.” Law and Society Review 18 : 437 - 78 . Google Scholar CrossRef Search ADS Pew Research Center . 2004 . Polls Face Growing Resistance, but Still Representative: Survey Experiment Shows . Washington, DC : Pew Research Center . Pew Research Center . 2016 . Statistical Portrait of Hispanics in the United States . Washington, DC : Pew Hispanic Center and Pew Research Center . Pickett Justin T. , Chiricos Ted . 2012 . “ Controlling Other People’s Children: Racialized Views of Delinquency and Whites’ Punitive Attitudes Toward Juvenile Offenders.” Criminology 50 : 673 - 710 . Google Scholar CrossRef Search ADS Quillian Lincoln. 1996 . “ Group Threat and Regional Change in Attitudes Toward African-Americans.” American Journal of Sociology 102 : 816 - 60 . Google Scholar CrossRef Search ADS Quillian Lincoln , Pager Devah . 2010 . “ Estimating Risk Stereotype Amplification and the Perceived Risk of Criminal Victimization.” Social Psychology Quarterly 73 : 79 - 104 . Google Scholar CrossRef Search ADS Quinney Richard. 1970 . The Social Reality of Crime . Boston, MA : Little Brown . Rabe-Hesketh Sophia , Skrondal Anders . 2008 . Multilevel and Longitudinal Modeling Using Stata . College Station, TX : Stata Press . Rachlinski Jeffrey J. , Lynn Johnson Sheri , Wistrich Andrew J. , Guthrie Chris . 2009 . “Does Unconscious Racial Bias Affect Trial Judges?” Notre Dame Law Review 84 : 1195 - 1246 . Ramey David M. 2013 . “ Immigrant Revitalization and Neighborhood Violent Crime in Established and New Destination Cities.” Social Forces 92 : 597 - 629 . Google Scholar CrossRef Search ADS Raudenbush Stephen W. , Bryk Anthony S . 2002 . Hierarchical Linear Models . Thousand Oaks, CA : Sage Publications . Russell Kathryn K. 1999 . “‘Driving While Black’: Corollary Phenomena and Collateral Consequences.” Boston College Law Review 40 : 717 - 31 . Simon Jonathan. 2007 . Governing Through Crime: How the War on Crime Transformed American Democracy and Created a Culture of Fear . Oxford, UK : Oxford University Press . Smith Brad W. , Holmes Malcolm D . 2003 . “ Community Accountability, Minority Threat, and Police Brutality: An Examination of Civil Rights Criminal Complaints.” Criminology 41 : 1035 - 63 . Google Scholar CrossRef Search ADS Smith Robert J. , Levinson Justin D . 2012 . “The Impact of Implicit Racial Bias on the Exercise of Prosecutorial Discretion.” Seattle University Law Review 35 : 795 - 826 . Spohn Cassia. 2000 . “Thirty Years of Sentencing Reform: The Quest for a Racially Neutral Sentencing Process.” Criminal Justice 3 : 427 - 501 . Spohn Cassia , Spears Jeffrey . 1996 . “ The Effect of Offender and Victim Characteristics on Sexual Assault Case Processing Decisions.” Justice Quarterly 13 : 649 - 79 . Google Scholar CrossRef Search ADS John Craig St. , Heald‐Moore Tamara . 1996 . “ Racial Prejudice and Fear of Criminal Victimization by Strangers in Public Settings.” Sociological Inquiry 66 : 267 - 84 . Google Scholar CrossRef Search ADS Stewart Eric A. , Martinez Romiro Jr. , Baumer Eric P. , Gertz Marc . 2015 . “ The Social Context of Latino Threat and Punitive Latino Sentiment.” Social Problems 62 : 68 - 92 . Google Scholar CrossRef Search ADS Stults Brian J. , Baumer Eric P . 2007 . “ Racial Context and Police Force Size: Evaluating the Empirical Validity of the Minority Threat Perspective.” American Journal of Sociology 113 : 507 - 46 . Google Scholar CrossRef Search ADS Swigert Victoria Lynn , Farrell Ronald A . 1977 . “ Normal Homicides and the Law.” American Sociological Review 42 : 16 - 32 . Google Scholar CrossRef Search ADS Taylor Marylee C. 1998 . “ How White Attitudes Vary with the Racial Composition of Local Populations: Numbers Count.” American Sociological Review 63 : 512 - 35 . Google Scholar CrossRef Search ADS Taylor Ralph B. , Covington Jeanette . 1993 . “ Community Structural Change and Fear of Crime.” Social Problems 40 : 374 - 97 . Google Scholar CrossRef Search ADS Tonry Michael. 2010 . “The Social, Psychological, and Political Causes of Racial Disparities in the American Criminal Justice System.” Crime & Justice 39 : 273 - 312 . Google Scholar CrossRef Search ADS Tourangeau Roger. 2004 . “ Survey Research and Societal Change.” Annual Review of Psychology 55 : 775 - 801 . Google Scholar CrossRef Search ADS Ulmer Jeffery T. , Johnson Brian D . 2004 . “ Sentencing in Context: A Multilevel Analysis.” Criminology 42 : 137 - 77 . Google Scholar CrossRef Search ADS Unnever James D. 2008 . “ Two Worlds Far Apart: Black-White Differences in Beliefs About Why African American Men are Disproportionately Imprisoned.” Criminology 46 : 511 - 38 . Google Scholar CrossRef Search ADS Unnever James D. , Cullen Francis T . 2007 . “ Reassessing the Racial Divide in Support for Capital Punishment.” Journal of Research in Crime and Delinquency 44 : 124 - 58 . Google Scholar CrossRef Search ADS Wang Xia. 2012 . “ Undocumented Immigrants as Perceived Criminal Threat: A Test of the Minority Threat Perspective.” Criminology 50 : 743 - 76 . Google Scholar CrossRef Search ADS Wang Xia , Mears Daniel P . 2010 . “ A Multilevel Test of Minority Threat Effects on Sentencing.” Journal of Quantitative Criminology 26 : 191 - 215 . Google Scholar CrossRef Search ADS Weisberg Herbert F. , Krosnick Jon A. , Bowen Bruce D . 1989 . An Introduction to Survey Research and Data Analysis . Glenview, IL : Scott Foresman . Weitzer Ronald , Tuch Steven A . 2005 . “ Racially Biased Policing: Determinants of Citizen Perceptions.” Social Forces 83 : 1009 - 30 . Google Scholar CrossRef Search ADS Williams Marian R. , Demuth Stephen , Holcom Jefferson E . 2007 . “Understanding the Influence of Victim Gender in Death Penalty Cases: The Importance of Victim Race, Sex‐Related Victimization, and Jury Decision Making.” Criminology 45 : 865 - 91 . Google Scholar CrossRef Search ADS Wilson William Julius. 2009 . More Than Just Race: Being Black and Poor in the Inner City . New York : W.W. Norton & Company . Wooldredge John , Griffin Timothy , Thistlethwaite Amy , Rauschenberg Fritz . 2011 . “ Victim‐Based Effects on Racially Disparate Sentencing in Ohio.” Journal of Empirical Legal Studies 8 : 85 - 117 . Google Scholar CrossRef Search ADS Xie Min , Lauritsen Janet L . 2012 . “ Racial Context and Crime Reporting: A Test of Black’s Stratification Hypothesis.” Journal of Quantitative Criminology 28 : 265 - 93 . Google Scholar CrossRef Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for the Study of Social Problems. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Social Problems Oxford University Press

The Social Context of Criminal Threat, Victim Race, and Punitive Black and Latino Sentiment

Loading next page...
 
/lp/ou_press/the-social-context-of-criminal-threat-victim-race-and-punitive-black-M8fx0hZ4zD
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the Society for the Study of Social Problems. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
ISSN
0037-7791
eISSN
1533-8533
D.O.I.
10.1093/socpro/spy003
Publisher site
See Article on Publisher Site

Abstract

Abstract A well-established body of research focuses on the relationship between criminal threat and the exercise of formal social control, and a largely separate literature examines the effects of victim race in criminal punishment. Despite their close association, few attempts have been made to integrate these related lines of empirical inquiry in the sociology of punishment. In this article, we address this issue by examining relationships among criminal threat, victim race, and punitive sentiment toward black and Latino defendants. We analyze nationally representative survey data that include both subjective and objective measures of criminal threat, and we incorporate unique information on victim/offender dyads to test research questions about the that role victim race plays in the formation of anti-black and anti-Latino sentiment in the criminal justice system. The results indicate that both subjective perceptions of criminal threat and minority population growth are significantly related to punitiveness among whites, and that punitive sentiment is enhanced in situations that involve minority offenders and white victims. Moreover, we show that aggregate indicators of racial threat strongly condition the effect of victim race on punitive attitudes. Implications of these findings are discussed in relation to racial group threat theories and current perspectives on the exercise of state-sponsored social control. criminal threat, victim race, punitive sentiment, social control; victim-offender dyads The formal exercise of social control is a topic of enduring sociological interest, particularly as it relates to racial and ethnic inequality in society. Theoretical treatments of race and punishment are routinely cast in terms of racial group threat perspectives that suggest large and growing minority populations generate group prejudice and hostility on the part of the white majority, which is then translated into enhanced social control efforts in the criminal justice system (Blalock 1967; Liska 1992; Quinney 1970). At the same time, sociological formulations of the behavior of law argue that the capacity for government social control is directly related to the relative social status of affected parties in the justice system—larger quantities of law are expected when social distance is greater between the structural positions of victims and offenders (Black 1976:2). Because race is a key component of social stratification in American society, stronger legal responses are expected in situations involving white victims and racial minority offenders (Xie and Lauritsen 2012). Taken together, group threat theories suggest a key role for minority group size in the exercise of social control in society, whereas theoretical propositions about the behavior of law highlight the importance of victim and offender race. The current research unites these arguments, examining the complementary and interactive influences of aggregate population contexts, localized perceptions of criminal threat, and race of victim effects in public support for enhanced social control directed at minority offenders. This study contributes to prior work on group threat and punitiveness in several important ways. First, it integrates insights about the importance of victim race into contemporary formulations of racial group threat theory. Second, it expands the ken of prior work by examining group threat associated with both black and Latino populations in the United States. Third, it incorporates both objective and subjective measures of criminal threat and, finally, it focuses on a unique measure of social control that captures punitive sentiment targeting black and Latino defendants in the American criminal court system. The article proceeds by first considering the evidence linking racial threat to social control outcomes. It then introduces relevant scholarship on the role that victim race plays in social control processes, before outlining and testing specific hypotheses about group threat, victim race, and punitive sentiment. BACKGROUND Racial Threat and Social Control Race, crime, and punishment are intertwined in a complex historical legacy that involves past racial oppression and contemporary social stereotypes that link minority groups to enhanced punishment in the criminal justice system (Bonilla-Silva 2006; Brewer and Heitzeg 2008). Public opinions regarding race, crime, and public policy remain sharply divided along racial and ethnic lines in American society (Bobo and Johnson 2004). In particular, racial and ethnic minorities tend to be less supportive of punitive justice policies than majority whites (Hagan and Albonetti 1982; Hagan, Shedd, and Payne 2005). This racial divide is maintained across various domains of the criminal justice system, including policing (Brunson 2007; Russell 1999; Weitzer and Tuch 2005), criminal courts (Brooks and Jeon-Slaughter 2001), and corrections (Baumer, Messner, and Rosenfeld 2003; Unnever and Cullen 2007). James Unnever (2008), for instance, has recently demonstrated that black respondents are more than twice as likely as white respondents to identify police bias and unfairness in the courts as major reasons for the disproportionate incarceration of African American men. Although less research focuses on ethnic differences, Latinos also tend to be less punitive than whites, expressing higher levels of support for rehabilitation and lower levels of support for capital punishment (Baik 2012). Much of the empirical research examining the influence of racial group dynamics in the justice system finds evidence that perceptions of racial threat are spatially patterned. Communities with relatively large or increasing minority populations tend to be associated with higher levels of racial prejudice and increased support for harsh punishment (Bobo and Hutchings 1996; Quillian 1996; Stewart et al. 2015; Taylor 1998). Ryan King and Darren Wheelock (2007), for instance, have shown that spatial variations in racialized perceptions of economic and criminal threat are directly related to individual punitive attitudes. Related work reveals significant associations among minority population size, self-reported fear of crime, and subjective perceptions of the risk of criminal victimization (Quillian and Pager 2010; Taylor and Covington 1993). Over three decades ago, Allen Liska, Joseph Lawrence, and Andrew Sanchirico (1982) demonstrated that the relative size of the non-white population was associated with greater fear of crime across U.S. cities. Similarly, Jeannette Covington and Ralph Taylor (1991) found that when the racial composition of the neighborhood was a predominantly black neighborhood, residents were more fearful of crime. More recent work indicates that individual perceptions of racial demographics are often more salient than objective indicators of population size (Chiricos, Hogan, and Gertz 1997), and it demonstrates that similar findings occur with regard to proximity to Latino populations (Chiricos, McEntire, and Gertz 2001). Indicators of minority group size have also been linked explicitly to formal social control efforts in the criminal justice system. Policing research shows that as the size of the minority population grows so does the size and funding of local police departments (Chamlin 1989), as well as race-specific arrest rates across local communities (Harer and Steffensmeier 1992; Liska, Chamlin, and Reed 1985). Rates of police killings of minority suspects are also higher in cities with large or increasing African American populations (Jacobs and O’Brien 1998). Parallel work on American prison populations links racial threat indicators to spatial and temporal variations in incarceration rates (Greenberg and West 2001). George Bridges, Robert Crutchfield, and Edith Simpson (1987), for instance, found that counties with larger non-white populations also had higher rates of non-white incarceration, and David Greenberg and Valerie West (2001) reported that states with larger black populations not only experienced higher incarceration rates but also greater prison population growth in recent decades. Some research also suggests individual sentencing decisions are affected by broader racial dynamics in the community (Johnson 2006; Ulmer and Johnson 2004). For example, Chester Britt (2000) found that the proportion of black residents in a county increased the individual probability of incarceration, and Xia Wang and Daniel Mears (2010) reported evidence that custodial sentences were more likely when racial group threat was high and increasing in an area. The relative size of the black population has even been tied to the likelihood of receiving the death penalty (Jacobs and Carmichael 2002). Not all studies, though, are supportive of racial threat perspectives, with some work finding null or even opposite effects for the influence of racial composition on punishment (cf. Bridges and Crutchfield 1988; Feldmeyer and Ulmer 2011). In part, this may reflect a number of empirical limitations that are common in this research tradition. First, racial threat is typically captured by aggregate population measures rather than more proximate, perceptual threat measures. As Karen Parker, Brian Stults, and Stephen Rice (2005) note, “dependency on percent black as the main indicator of racial threat” has led to “inconsistency in findings” (p. 1111). Some research suggests that perceptual measures of group threat demonstrate stronger and more consistent relationships with social control outcomes (Chiricos et al. 1997). Second, tests of group threat theory focus predominantly on the African American population with considerably less attention devoted to the growing Latino population (Eitle and Taylor 2008). This is despite the fact that Latinos are now the largest and one of the fastest growing racial/ethnic groups in the United States (Pew Research Center 2016). Given the theoretical salience of population growth in group threat perspectives, there is a clear need to incorporate Latinos into broader multiethnic examinations of punitive sentiment. Third, prior work has been conducted at various levels of analysis (Parker et al. 2005), and relatively few studies employ nationally representative data or examine changes in racial composition over time. Finally, prior work makes it clear that social control efforts are often tied specifically to criminal threat. The concurrence of race and crime in public discourse has elevated the salience of criminal threat as a vehicle for enhanced social control efforts targeted at minority defendants (Pickett and Chiricos 2012), yet direct measures of criminal threat are often absent from research in the group threat tradition (Eitle, D’Alessio, and Stolzenberg 2002). Moreover, no prior work has considered the role that victim race plays in perceptions of criminal threat. The current work uses national survey data to investigate both objective and subjective indicators of group threat. It examines threat processes associated with both black and Latino populations, and it incorporates unique information on victim/offender racial dyads into the study of public support for more punitive treatment of minority defendants in the criminal justice system. Criminal Threat, Victim Race, and Punitive Sentiment Although group threat theories were originally formulated in relation to socioeconomic threats, race and crime have gained increasingly prominent roles in these perspectives (Hawkins 1987). Racial threat theory argues that the exercise of social control in society varies with the racial composition of the population because members of the white racial majority feel threatened by large or growing minority populations (Blalock 1967). Community-level racial demographics feed into perceptions of social threat that are rooted in racial prejudice and intergroup conflict. Fueled in part by the emergence of race and crime as a dominant political theme, white Americans increasingly associated criminal threats with growing minority populations (Chiricos et al. 2001). In response, enhanced social control efforts may be mobilized that target minority defendants in the criminal justice system. From this perspective, the racial composition of an area affects the perceived threat of crime, which in turn contributes to both formal and informal mechanisms of social control in society (King and Wheelock 2007). Although racial group threat can take multiple forms that include economic, political, and criminal threat, the latter appears to be the driving force behind punitive responses of the criminal justice system. This is not surprising given the clear association between criminal threat and punishment. In fact, recent work argues that political and economic threats have largely been “replaced by the black male criminal in the iconography of racial threat” (Crawford, Chiricos, and Kleck 1998:483), and it suggests that “The conflation of race and criminal threat is [now] so well established that some regard popular discourse about crime and punishment to be part of the rhetorical code of ‘modern racism’” (Chiricos et al. 2001:323). Research demonstrates that stereotypes tying race and ethnicity to perceptions of crime and violence are widespread in American society (Brown 2010; Devine 1989), and it shows that respondents frequently express greater fear of victimization by black strangers than white strangers (St. John and Heald-Moore 1996). Additional evidence for the salience of criminal threat comes from research that compares different threat mechanisms. David Eitle, Stewart D’Alessio, and Lisa Stolzenberg (2002), for example, tested the relative importance of political, economic, and criminal threat explanations of arrest rates and concluded that “the findings taken together furnish strong support for the threat of black crime hypothesis” (p. 557). Research on racial typification of crime in the media also supports the criminal threat hypothesis. Franklin Gilliam and Shanto Iyengar (2000) report that local news often utilizes a “crime news script” that routinizes the association between racial minorities, crime, and violence, and Daniel Mears and Eric Stewart (2010) note that both “media accounts and policy discourse” have “increasingly equated crime with blacks” (p. 35). Racialized social constructions of criminal threat, then, are often tied to broader patterns of racial demography in society and may be used to promote dominant group interests in core institutions of law (Chambliss and Seidman 1971; Hetey and Eberhardt 2014). Community-level racial dynamics can feed into perceptions of threat and fuel public sentiment for enhanced punishment of minority defendants (Eitle et al. 2002; Johnson et al. 2011). Negative racial stereotypes and fear of black crime increase public support for punitive measures (Barkan and Cohn 2005; Hetey and Eberhardt 2014), and large minority populations heighten punitive attitudes among whites (King and Wheelock 2007). As Liska and Mitchell Chamlin (1984) summarized it, large minority populations produce “an emergent property” tied to the perceived threat of crime, which increases social pressures to control crime (p. 384). Although contemporary formulations of racial threat theory generally focus on black/white comparisons, recent shifts in the demographic landscape of American society suggest group threat may be increasingly associated with Latino defendants. This reflects what Eduardo Bonilla-Silva (2004) has called the “Latin Americanization” of race relations in the United States. Current population estimates suggest more than 55 million Americans self-identify as Latino, accounting for over 17 percent of the population and making them the largest minority group in the United States (Pew Research Center 2016). Latino growth has been disproportionate, accounting for more than half of the total population growth in the country over the past decade (Ennis, Ríos-Vargas, and Albert 2011). Public opinion polls indicate a growing concern over Latino growth among whites (Lane and Meeker 2003). Richard Alba, Ruben Rumbaut, and Karen Marotz (2005), for instance, found that approximately 3 out of 4 survey respondents believed that more immigrants are likely to cause higher crime rates, despite empirical evidence to the contrary (Ousey and Kubrin 2009; Ramey 2013). Shifting population demographics, then, may contribute to perceptions of criminal threat and fear of crime being increasingly associated with Latino groups in the United States. (Eitle and Taylor 2008). Empirical studies that incorporate ethnic threat measures generally suggest that similar threat processes characterize both blacks and Latinos (e.g., Jackson 1989; Jacobs and Carmichael 2001; Johnson et al. 2011; Ulmer and Johnson 2004). Wang (2012), for instance, recently demonstrated that the perceived size of the undocumented immigrant population in the Southwest was positively associated with perceptions of them as a criminally threatening group, and this mattered more than actual immigrant populations or local economic conditions. Moreover, some recent work suggests an even stronger role for ethnic threat relative to racial threat, at least in some social contexts (Chiricos et al. 2001). In one of the few studies to examine both subjective and objective measures of Latino threat, Brian Johnson and colleagues (2011) reported that support for “use of ethnicity” in sentencing increased in areas with greater Hispanic population growth and in areas where individuals perceived there to be greater economic and criminal threat. Victim Race and Group Threat Theory One element of racial group threat theories that has gone largely uninvestigated is the potentially important role that victim characteristics play in perceptions of criminal threat. Liska and Chamlin (1984) have argued that crime committed against minority victims will be diminished in the eyes of the white majority because it generates less perceived threat of personal victimization—a process they refer to as “benign neglect.” There is some evidence that supports this hypothesis. Aggregate research on race-specific arrest rates has found that larger minority populations may actually reduce arrests (Chamlin and Liska 1992; Liska and Chamlin 1984). Parker and colleagues (2005), for example, show that both percent black and percent black immigrant are negatively related to black arrest rates. They suggest this is because crime is more likely to be intra-racial when black populations are large, so it is discounted by social control agents. Few studies, however, directly examine race of victim effects in group threat processes. Eitle and colleagues (2002) examined separate indicators of black-on-white and white-on-white crime and found that only the former was significantly related to levels of arrest. They note that this result is consistent with the criminal threat hypothesis, which deals specifically with perceived threats to white victims, and they suggest that victim race may therefore play an important role in group threat processes. From this perspective, minority crimes involving minority victims are expected to be devalued because they pose no immediate threat to the established social order. Such notions are also consistent with long-standing theoretical arguments about the behavior of law in society (Black 1976) as well as broader discussions of the formulation of contemporary crime policy in America (Garland 2001; Simon 2007). Donald Black (1976) argues that social stratification, or the unequal distribution of resources, directly affects the quantity of law that is exercised. He places particular emphasis on the relationship between offenders and victims, noting that crimes committed by lower status offenders against higher status victims, what he refers to as “upward crimes,” will be judged as relatively more serious and result in greater mobilization of formal social control efforts. Because the state acts on behalf of victims, criminal offenses involving higher status victims result in greater legal recourse. Integrating his ideas into racial group threat theory, this implies that perceptions of criminal threat will be strongest, and support for punitive responses to crime will be greatest, in situations that involve victim-offender dyads consisting of white victims and minority suspects. In a similar vein, Jonathan Simon (2007) has argued that “victim identity is deeply racialized. It is not all victims, but primarily white, suburban, middle-class victims, whose exposure has driven waves of crime legislation” and against which “crime, poverty and … minority demographics are pushing” (p. 76). He suggests that broad shifts in criminal justice policy can be linked specifically to racialized patterns of victimization. In particular, the racial typification of crime in the media places greater emphasis on offenses involving white victims and minority offenders (Chiricos and Eschholz 2002), which may lead to enhanced support for punitive measures (Chiricos, Welch, and Gertz 2004). As Justin Pickett and Ted Chiricos (2012) argue, over time whites have come “to believe … that victims of violent crimes tend to be white” (p. 676). Travis Dixon and Daniel Linz (2000) suggest this trend is emblematic of a broader “ethnic blame discourse” in which stereotypical images of morality portray whites as law enforcers and minorities as law breakers. To the extent that racial dyads shape perceptions of criminal threat, then, they may also contribute to punitive sentiment toward minority defendants. Empirical evidence for the salience of racial dyads in criminal justice decision making is well established (e.g., Eberhardt et al. 2006; LaFree 1980; Myers 1979; Williams, Demuth, and Holcom 2007). In general, crimes that target higher status and more socially valued victims tend to be punished more severely (Baldus, Pulaski, and Woodworth 1983; Franklin and Fearn 2008). In their examination of victim characteristics in homicide, for example, Eric Baumer, Steven Messner, and Richard Felson (2000) concluded that “killings of disreputable or stigmatized victims tend to be treated more leniently by the justice system” (p. 304). A number of studies demonstrate that homicide offenses involving minority victims are punished more leniently and are less likely to eventuate in death sentences (Paternoster 1984). Indeed, Jennifer Eberhardt and colleagues (2006) observed that defendants who were perceived to be more “stereotypically” black were more likely to be sentenced to death only when their victims were white. Similar results have been demonstrated for other types of crimes and for related punishment decisions (e.g., Curry 2010; Johnson et al. 2011; Spohn and Spears 1996). Although some studies report mixed results for victim effects (Auerhahn 2007; Wooldredge et al. 2011), the weight of the evidence clearly supports their centrality for understanding racial disparities in the justice system. As Cassia Spohn (2000) summarized, “criminal punishment is contingent on the race of the victim as well as the race of the offender” (p. 469). The importance of racial dyads in criminal punishment has direct implications for our understanding of group threat perspectives. Punitive sentiments that are mobilized by perceptions of minority criminal threat are also likely to be affected by the racial characteristics of victims. The social significance of victim race may even depend on the broader racial context of the local environment. To the extent that large or growing minority populations heighten perceptions of “ecological vulnerability” on the part of whites (Covington and Taylor 1991), they may also amplify criminal threat perceptions and lead to increased support for punitiveness when minority suspects target white victims. This implies that victim race will have direct effects on punitive sentiments, and that this influence will be conditioned by the broader racial context of local social environments. Taken together, then, prior work emphasizes the importance of examining the link between the size and growth of ethnic minority populations in the United States and local perceptions of criminal threat. It highlights the significance of going beyond traditional conceptualizations of racial threat to examine ethnic threat tied to the growing Latino population in the United States. And it suggests a potentially important role for victim race and ethnicity in group threat theories and in support for punitive responses of the criminal justice system. The current study addresses these interrelated issues by delineating and testing specific theoretical hypotheses that relate group threat processes to victim race and punitive sentiment. Summary and Hypotheses Collectively, prior research and theorizing suggest a number of predictions about the role of group threat, victim race/ethnicity, and public support for enhanced punishment of minority defendants. In line with group threat perspectives, we expect the absolute size of black and Latino populations, and their recent growth, to influence punitive sentiment toward minority defendants. We incorporate both static and dynamic measures that capture the absolute size and recent change in racial and ethnic minority populations, as well as more subjective measures of perceived criminal threat from minority groups. Specifically, we expect the following: H1: The objective size and recent growth of the black and Latino populations will increase punitive sentiment among white respondents. H2: Subjective perceptions of black and Latino criminal threat will increase punitive sentiment among white respondents. Given the theoretical centrality of fear of victimization in perceptions of criminal threat, it is likely that punitive sentiment will be substantially conditioned by victim race. Crimes involving minority victims are likely to be viewed as less egregious, less demanding of public censure, and less deserving of harsh punishment because they are judged to be relatively less serious and because they pose less of a threat to the established social order. Crimes perpetrated by minority offenders against white victims, on the other hand, will incite greater racial fear and lead to increased punitiveness on the part of white respondents. These arguments are consistent with Black’s (1976) racial stratification thesis that suggests greater mobilization of legal resources will attach to victims of higher social status, particularly when social distance is greatest. The implication is that crimes involving white victims and minority offenders will increase punitive sentiment, whereas crimes involving minority victims will be devalued. Specifically, we expect the following: H3: Crimes involving white victims that are committed by black or Latino offenders will increase punitive sentiment among white respondents. H4: Crimes involving black or Latino victims that are committed by black or Latino offenders will not increase punitive sentiment among white respondents. In addition to these main effects, there are also persuasive theoretical reasons to expect that the individual effects of victim race will be conditioned by broader racial and ethnic social contexts. As others have noted, the social meaning of victim race varies by place (Xie and Lauritsen 2012). Contexts characterized by greater perceptions of criminal threat are likely to produce stronger race of victim effects. In neighborhoods where racial minorities are viewed as particularly threatening, the salience of minority-on-white crime should be enhanced, resulting in stronger emotional reactions and greater support for punitiveness toward minority defendants. Increases in both objective and subjective measures of threat are expected to significantly increase the influence of victim race on punitive sentiment. We therefore investigate the following: H5: The effect of victim race on punitive sentiment will be stronger in social contexts characterized by large or growing black and Latino populations. H6: The effect of victim race on punitive sentiment will be stronger in social contexts characterized by greater subjective perceptions of black and Latino criminal threat. DATA AND METHOD The data analyzed in this study were gathered through a national random telephone survey of 2,736 American adults (18 and older), using random-digit dialing and computer assisted telephone interviewing (CATI) to ensure accuracy in recording data. The sampling frame includes households with either landlines or cellular phones. The telephone surveys were conducted during the spring, summer, and fall of 2013. The survey focused primarily on respondents’ attitudes about punishment, crime, residential preferences, ethnicity, and immigration. The data set contains a rich variety of information and offers a unique opportunity to examine punitive sentiment as a result of group threat processes and victim/offender relationships, something that has largely been unaddressed in prior studies. As such, these survey data are uniquely suited to our research questions. A two-stage modified Mitofsky–Waksberg sampling design was utilized to develop the random-digit dialing sample (Tourangeau 2004). Respondents were limited to one adult resident per household who was 18 years or older. From each household sampled, the adult respondent with the most recent birthday was selected, an efficient way to randomly choose adults within households (Kish 1965). Trained interviewers conducted the telephone interviews and were closely monitored by supervisors. Additionally, to minimize interviewer error, supervisors reviewed 10 percent of completed interviews for accuracy by comparing selected responses to digitally recorded excerpts of interviews. There was 93 percent agreement between supervisors and interviewers. In the 7 percent of cases where there was not agreement, the supervisors and interviewers met to reconcile the discrepancy. A five call-back rule was employed before replacement of households. Using the definition recommended by the American Association for Public Opinion Research (AAPOR 2008), we obtained a 60.8 percent response rate among all contacts with eligible respondents.1 Cases of unknown eligibility, such as answering machines, busy signals, no answer, and known ineligibility, such as disconnected numbers, businesses, and fax numbers, were excluded from this calculation as recommended by AAPOR (2008). The response rate is comparable to studies that use rigorous survey methodologies (Pew Research Center 2004), as well as other recent studies utilizing telephone surveys (e.g., McCarty et al. 2006). Additionally, 94 percent of all surveys initiated were completed. This completion rate was substantially higher than the 60 percent average for national telephone interviews (Weisberg, Krosnick, and Bowen 1989). The final sample consisted of 2,408 non-Latino white respondents.2 About 44 percent of the sample was male. The age of the sample ranged from 18 to 81 with an average of 42 years. As for educational attainment, 46 percent of the sample graduated from college. The breakdown for annual household income was as follows: about 34 percent of the sample reported earning less than $50,000; around 31 percent of the respondents earned between $50,000 and $75,000; 20 percent of participants earned between $75,000 and $100,000; and about 15 percent of the sample reported earning more than $100,000. The mean family income of the sample was $63,539. Approximately 53 percent of the participants reported being married. About 69 percent of the respondents reported owning their home. In regard to geographic census region, 55 percent of the sample lived in the South, 12 percent in the Northeast, 18 percent in the Midwest, and 15 percent in the West. There is an overrepresentation of white, female, older, and higher-income respondents when compared to the 2010 Census, which is not uncommon in telephone surveys (Lavrakas 1987).3 To assess the effects of black and Latino population contexts on whites’ punitiveness, we matched respondents to the 168 counties where they resided and appended to the individual-level records, county-level data from the U.S. Census Bureau. The number of respondents in a county averaged about 14 and ranged from 9 to 31. Dependent Variables Our two dependent variables, punitive-black sentiment and punitive-Latino sentiment, are measured using six questions for each measure. Specifically, the questions ask respondents whether or not they agree with the following statements: “The U.S. criminal justice system is too lenient with [black/Latino] offenders; The U.S. criminal justice system needs tougher prison sentences on [black/Latino] offenders; The U.S. criminal justice system needs tougher prison sentences on [black/Latino] repeat offenders; In the U.S., [black/Latino] offenders should be punished severely for violating misdemeanor laws; In the U.S., [black/Latino] offenders convicted of a violent crime (i.e., murder, aggravated assault, rape, or robbery) should receive the death penalty; In the U.S., [black/Latino] offenders convicted of a property crime (i.e., burglary, larceny theft, or auto theft) should receive the death penalty.”4 The response options were coded so that higher scores indicated stronger agreement with each statement (0 = strongly disagree, 1 = disagree, 2 = agree, and 3 = strongly agree). The items were combined into two indexes that yielded alpha values of .93 for punitive-black sentiment and .89 for punitive-Latino sentiment. Thus, the dependent variables range from 0 (low) through 18 (high). Descriptive statistics for the dependent variables, as well as all of the study variables, are provided in Table 1. Table 1. Descriptive Statistics for Study Variables Mean SD Dependent variables  Punitive-black sentiment 9.97 3.29  Punitive-Latino sentiment 9.04 3.12 Independent variables  Victim/offender dyad   White victim/black offender .35 .41   Black victim/black offender .05 .22   Latino victim/black offender .06 .24  Victim/offender dyad   White victim/Latino offender .31 .43   Black victim/Latino offender .04 .20   Latino victim/Latino offender .05 .22  Perceived threat   Black criminal threat 7.93 3.05   Latino criminal threat 5.87 2.63  Population context   Percent Latino .18 .14   Latino growth .05 .03   Percent black .20 .13   Black growth .04 .03  County controls   Homicide rate (per 100,000) 9.26 4.12   Concentrated disadvantage 3.59 1.37   Percent Republican .49 .12   Population structure 5.89 .75  Demographic controls   Age 42.17 15.09   Male .44 .50   Married .53 .31   Education level (college graduate) .46 .31   Family income $63,539 $12,145   Employed .49 .50   Political conservative .51 .30   Own home .69 .33   Southwest .20 .32   Northeast .12 .35   Midwest .18 .43   West .15 .38   South .55 .32   General punitive attitudes 6.99 2.04 N1 = 2,408 individuals N2 = 168 counties Mean SD Dependent variables  Punitive-black sentiment 9.97 3.29  Punitive-Latino sentiment 9.04 3.12 Independent variables  Victim/offender dyad   White victim/black offender .35 .41   Black victim/black offender .05 .22   Latino victim/black offender .06 .24  Victim/offender dyad   White victim/Latino offender .31 .43   Black victim/Latino offender .04 .20   Latino victim/Latino offender .05 .22  Perceived threat   Black criminal threat 7.93 3.05   Latino criminal threat 5.87 2.63  Population context   Percent Latino .18 .14   Latino growth .05 .03   Percent black .20 .13   Black growth .04 .03  County controls   Homicide rate (per 100,000) 9.26 4.12   Concentrated disadvantage 3.59 1.37   Percent Republican .49 .12   Population structure 5.89 .75  Demographic controls   Age 42.17 15.09   Male .44 .50   Married .53 .31   Education level (college graduate) .46 .31   Family income $63,539 $12,145   Employed .49 .50   Political conservative .51 .30   Own home .69 .33   Southwest .20 .32   Northeast .12 .35   Midwest .18 .43   West .15 .38   South .55 .32   General punitive attitudes 6.99 2.04 N1 = 2,408 individuals N2 = 168 counties Table 1. Descriptive Statistics for Study Variables Mean SD Dependent variables  Punitive-black sentiment 9.97 3.29  Punitive-Latino sentiment 9.04 3.12 Independent variables  Victim/offender dyad   White victim/black offender .35 .41   Black victim/black offender .05 .22   Latino victim/black offender .06 .24  Victim/offender dyad   White victim/Latino offender .31 .43   Black victim/Latino offender .04 .20   Latino victim/Latino offender .05 .22  Perceived threat   Black criminal threat 7.93 3.05   Latino criminal threat 5.87 2.63  Population context   Percent Latino .18 .14   Latino growth .05 .03   Percent black .20 .13   Black growth .04 .03  County controls   Homicide rate (per 100,000) 9.26 4.12   Concentrated disadvantage 3.59 1.37   Percent Republican .49 .12   Population structure 5.89 .75  Demographic controls   Age 42.17 15.09   Male .44 .50   Married .53 .31   Education level (college graduate) .46 .31   Family income $63,539 $12,145   Employed .49 .50   Political conservative .51 .30   Own home .69 .33   Southwest .20 .32   Northeast .12 .35   Midwest .18 .43   West .15 .38   South .55 .32   General punitive attitudes 6.99 2.04 N1 = 2,408 individuals N2 = 168 counties Mean SD Dependent variables  Punitive-black sentiment 9.97 3.29  Punitive-Latino sentiment 9.04 3.12 Independent variables  Victim/offender dyad   White victim/black offender .35 .41   Black victim/black offender .05 .22   Latino victim/black offender .06 .24  Victim/offender dyad   White victim/Latino offender .31 .43   Black victim/Latino offender .04 .20   Latino victim/Latino offender .05 .22  Perceived threat   Black criminal threat 7.93 3.05   Latino criminal threat 5.87 2.63  Population context   Percent Latino .18 .14   Latino growth .05 .03   Percent black .20 .13   Black growth .04 .03  County controls   Homicide rate (per 100,000) 9.26 4.12   Concentrated disadvantage 3.59 1.37   Percent Republican .49 .12   Population structure 5.89 .75  Demographic controls   Age 42.17 15.09   Male .44 .50   Married .53 .31   Education level (college graduate) .46 .31   Family income $63,539 $12,145   Employed .49 .50   Political conservative .51 .30   Own home .69 .33   Southwest .20 .32   Northeast .12 .35   Midwest .18 .43   West .15 .38   South .55 .32   General punitive attitudes 6.99 2.04 N1 = 2,408 individuals N2 = 168 counties Independent Variables Victim/Offender Dyads To capture the racial and ethnic composition of the victim/offender dyads, our questions deal directly with scenarios that involve offenders who are either black or Latino. In addition, we systematically vary the race and ethnicity of the victims in the survey. Respondents were asked whether or not they thought violent crimes were more serious if the offender was black or Latino and the victims were either white, black or Latino. For example, in the punitive-black sentiment equations, to measure the white victim/black offender dyad, respondents answered the following questions: “In the U.S., are criminal acts more serious when a black offender commits a violent crime against a white victim?”5 The black victim/black offender dyad was measured by the following question: “In the U.S., are criminal acts more serious when a black offender commits a violent crime against a black victim?” The Latino victim/black offender dyad was measured by the following question: “In the U.S., are criminal acts more serious when a black offender commits a violent crime against a Hispanic victim?” For the models in which punitive-Latino sentiment was the outcome, the white victim/Latino offender dyad was measured by the following question: “In the U.S., are criminal acts more serious when a Hispanic offender commits a violent crime against a white victim?” The black victim/Latino offender dyad was measured by the following question: “In the U.S., are criminal acts more serious when a Hispanic offender commits a violent crime against a black victim?” The Latino victim/Latino offender dyad was measured by the following question: “In the U.S., are criminal acts more serious when a Hispanic offender commits a violent crime against a Hispanic victim?” The response categories were coded as 1 if the respondents marked yes to the question and 0 if they indicated no. Each item was treated as orthogonal in measurement and also in the statistical analysis below. This approach allows us to not only hold the race and ethnicity of the offender constant, but also to vary the racial and ethnic profile of the victims. By doing so, we are able to leverage respondents’ sensitivity to the race and ethnicity of the victim and offender in levels of public support for severity of punishment. Perceived Threat Measures We examine two subjective measures of threat: black criminal threat and Latino criminal threat. Both constructs are measured using four statements respectively: ‘‘[blacks/Latinos] pose a greater threat to public order and safety than other racial or ethnic groups; [blacks/Latinos] hurt the U.S. by committing more violent crimes than other racial or ethnic groups; [blacks/Latinos] commit most of the crime in the United States; and The United States needs to put more police on the streets, to protect law abiding citizens from [blacks/Latinos].’’ There are four response categories for the statements (0 = strongly disagree, 1 = disagree, 2 = agree, and 3 = strongly agree). The alpha coefficient for black criminal threat is .86, while it is .87 for Latino criminal threat. Importantly, these measures of perceived threat are similar to those used in prior studies that assess racial and ethnic threat processes, making them consistent with previous research (e.g., Johnson et al. 2011; Stewart et al. 2015; Stults and Baumer 2007). Objective Population Contexts Consistent with group threat perspectives, we incorporate objective indicators of racial and ethnic population contexts into the study of non-Latino white survey respondents. To capture black population dynamics, we used two county-level indicators: percent black and black growth. Both items are drawn from the U.S. county census data. Percent black was drawn from the 2010 Census and represents percentage of blacks who resided in the respondents’ counties. We also incorporated an indicator of change—black growth. This measure represents the difference between the percentage of residents identified as black in 2000 and 2010 in sample members’ counties (Green, Strolovitch, and Wong 1998). Latino population context is measured by incorporating two county-level indicators: percent Latino and Latino growth. Both items are drawn from the U.S. county census data. The measure of percent Latino was drawn from the 2010 U.S. county census data. We used 2000 decennial census data and the 2010 census data to measure Latino growth, an indicator that represents the difference between the percent of residents identified as Hispanic in 2000 and 2010 in respondents’ counties (Green et al. 1998; Stewart et al. 2015). Together these measures capture both the relative size of black and Latino populations and recent population changes in counties where our respondents are located. Control Variables We controlled for a host of additional county- and individual-level factors that have been linked to punitive attitudes and punishment outcomes in prior work (Holmes 2000; Kent and Jacobs 2005; Smith and Holmes 2003). At the county level, these factors include the following: homicide rate, concentrated disadvantage, percent Republican, and population structure. At the individual level, we controlled for the following factors: age, marital status, education level, family income, employment status, political conservative, homeownership, Northeast, Midwest, West, South, Southwest, and general punitive attitudes. Appendix Table A1 provides the full details about the metrics and definitions for these variables.6 Analytic Strategy We estimated multilevel linear models to examine how victim/offender racial and ethnic dyads, subjective perceptions of criminal threat, and objective population contexts are related to punitive sentiment among white survey respondents. Multilevel models are useful for dealing with the non-independence of observations within higher order structural groupings. Multilevel models are appropriate here because we are interested in individual-level outcomes that may be influenced by both individual- and contextual-level characteristics. In our analysis, respondents are nested within counties, and ignoring this clustering could underestimate standard errors of parameter estimates possibly leading to Type I error in which the wrong conclusions are observed for nonexistent relationships (Raudenbush and Bryk 2002). Multilevel modeling accounts for this form of non-independence and produces correct estimates of the standard errors (Raudenbush and Bryk 2002). This technique also is useful because it allows us to isolate the independent effects of both individual- and county-level variables, as well as test for cross-level interaction effects.7 All variables are grand-mean centered and estimated using the multilevel function in the STATA 14 program (Rabe-Hesketh and Skrondal 2008).8 Our analysis proceeds in three stages. We first estimate two-level random-intercept models of punitive-black sentiment and punitive-Latino sentiment that include subjective perceptions of criminal threat and objective indicators of population context, along with both individual- and county-level control variables. Second, we build on these multivariate models by incorporating racial and ethnic victim/offender dyads into models of both punitive-black and punitive-Latino sentiment. Third, we test whether the influence of the victim/offender dyads on punitive-black sentiment and punitive-Latino sentiment is conditioned by perceived criminal threat and county-level black and Latino population contexts. We estimate random slope models in which the slopes for victim/offender dyads are allowed to vary across counties and are modeled as a function of perceived criminal threat and black and Latino population contexts respectively. RESULTS Table 1 reports descriptive statistics for the variables used in the analysis. Punitive sentiment towards blacks and Latinos is moderately high, averaging between 9 and 10 points on the 18-point scale. With regard to the victim/offender dyads, the perceived severity of criminal acts varied starkly by race of the victim. When the victim was white and the offender was black, 35 percent of the respondents viewed criminal acts as more serious. A similar pattern was observed when the victim was white and the offender was Latino. The percentages were much lower for scenarios involving minority victims. Indeed, no more than 6 percent of the respondents viewed the offense as relatively more serious when both the victim and offender were black or Latino.9 To investigate whether victim/offender racial and ethnic dyads, perceptions of threat, and population contexts predict punitive sentiment, we turn to our multilevel analysis. The results of multilevel linear regression models aimed at identifying the factors that help to explain punitive-black sentiment are presented in Table 2. Before turning to our main substantive results, we note that several control variables emerged as significant predictors of punitive-black sentiment. In particular, punitive sentiment is more likely among males, homeowners, political conservatives, and those who hold more general punitive attitudes. In addition, respondents in counties with high violence rates and a large republican voting base have higher levels of punitive-black sentiment. On the other hand, we found no evidence that concentrated disadvantage or population structure affect punitive-black sentiment. Overall, these results are largely consistent with those reported in previous studies that have examined formal social control (e.g., Johnson et al. 2011). Table 2. Multilevel Models of Punitive-Black Sentiment Regressed on Victim/Offender Dyad and Threat Indicators Model 1 Model 2 Model 3 Model 4 Variables b SE B SE b SE b SE Victim/offender dyad  White victim/black offender − − .224* .061 − − − −  Black Victim/black offender − − − − −.287 .461 − −  Latino Victim/black offender − − − − − − −.109 .416 Perceived black threat  Black criminal threat .197* .031 .189* .025 .196* .025 .196* .025 Black population context  Percent black −.748 .902 −.748 .902 −.748 .902 −.748 .902  Black growth .137* .046 .132* .046 .134* .046 .135* .046 County controls  Homicide rate (per 100,000) .042* .014 .038* .014 .044* .014 .044* .014  Concentrated disadvantage −.175 .248 −.175 .248 −.175 .248 −.175 .248  Percent Republican .349* .142 .338* .142 .349* .142 .349* .142  Population structure −.097 .171 −.097 .171 −.097 .171 −.097 .171 Demographic controls  Age −.006 .008 −.006 .008 −.006 .008 −.006 .008  Male .551* .191 .549* .191 .551* .191 .551* .191  Married .362 .224 .362 .224 .362 .224 .362 .224  Education level −.365 .229 −.365 .229 −.365 .229 −.365 .229  Family income −.152 .109 −.152 .109 −.152 .109 −.152 .109  Employed −.303 .215 −.303 .215 −.303 .215 −.303 .215  Political conservative .437* .175 .431* .175 .437* .175 .437* .175  Own home .113* .027 .108* .027 .113* .027 .113* .027  Southwest .452 .312 .452 .312 .452 .312 .452 .312  Northeast −.988* .293 −.988* .293 −.988* .293 −.988* .293  Midwest −.261 .299 −.261 .299 −.261 .299 −.261 .299  West −.763* .273 −.763* .273 −.763* .273 −.763* .273  General punitive attitudes .319* .038 .314* .038 .319* .038 .319* .038 Intercept 9.475* .618 9.433* .618 9.475* .618 9.471* .618 Total variance explained (%) 18.4 21.1 18.2 18.4 N1 = 2,408 individuals N2 = 168 counties Model 1 Model 2 Model 3 Model 4 Variables b SE B SE b SE b SE Victim/offender dyad  White victim/black offender − − .224* .061 − − − −  Black Victim/black offender − − − − −.287 .461 − −  Latino Victim/black offender − − − − − − −.109 .416 Perceived black threat  Black criminal threat .197* .031 .189* .025 .196* .025 .196* .025 Black population context  Percent black −.748 .902 −.748 .902 −.748 .902 −.748 .902  Black growth .137* .046 .132* .046 .134* .046 .135* .046 County controls  Homicide rate (per 100,000) .042* .014 .038* .014 .044* .014 .044* .014  Concentrated disadvantage −.175 .248 −.175 .248 −.175 .248 −.175 .248  Percent Republican .349* .142 .338* .142 .349* .142 .349* .142  Population structure −.097 .171 −.097 .171 −.097 .171 −.097 .171 Demographic controls  Age −.006 .008 −.006 .008 −.006 .008 −.006 .008  Male .551* .191 .549* .191 .551* .191 .551* .191  Married .362 .224 .362 .224 .362 .224 .362 .224  Education level −.365 .229 −.365 .229 −.365 .229 −.365 .229  Family income −.152 .109 −.152 .109 −.152 .109 −.152 .109  Employed −.303 .215 −.303 .215 −.303 .215 −.303 .215  Political conservative .437* .175 .431* .175 .437* .175 .437* .175  Own home .113* .027 .108* .027 .113* .027 .113* .027  Southwest .452 .312 .452 .312 .452 .312 .452 .312  Northeast −.988* .293 −.988* .293 −.988* .293 −.988* .293  Midwest −.261 .299 −.261 .299 −.261 .299 −.261 .299  West −.763* .273 −.763* .273 −.763* .273 −.763* .273  General punitive attitudes .319* .038 .314* .038 .319* .038 .319* .038 Intercept 9.475* .618 9.433* .618 9.475* .618 9.471* .618 Total variance explained (%) 18.4 21.1 18.2 18.4 N1 = 2,408 individuals N2 = 168 counties * p < .05 (two-tailed test) Table 2. Multilevel Models of Punitive-Black Sentiment Regressed on Victim/Offender Dyad and Threat Indicators Model 1 Model 2 Model 3 Model 4 Variables b SE B SE b SE b SE Victim/offender dyad  White victim/black offender − − .224* .061 − − − −  Black Victim/black offender − − − − −.287 .461 − −  Latino Victim/black offender − − − − − − −.109 .416 Perceived black threat  Black criminal threat .197* .031 .189* .025 .196* .025 .196* .025 Black population context  Percent black −.748 .902 −.748 .902 −.748 .902 −.748 .902  Black growth .137* .046 .132* .046 .134* .046 .135* .046 County controls  Homicide rate (per 100,000) .042* .014 .038* .014 .044* .014 .044* .014  Concentrated disadvantage −.175 .248 −.175 .248 −.175 .248 −.175 .248  Percent Republican .349* .142 .338* .142 .349* .142 .349* .142  Population structure −.097 .171 −.097 .171 −.097 .171 −.097 .171 Demographic controls  Age −.006 .008 −.006 .008 −.006 .008 −.006 .008  Male .551* .191 .549* .191 .551* .191 .551* .191  Married .362 .224 .362 .224 .362 .224 .362 .224  Education level −.365 .229 −.365 .229 −.365 .229 −.365 .229  Family income −.152 .109 −.152 .109 −.152 .109 −.152 .109  Employed −.303 .215 −.303 .215 −.303 .215 −.303 .215  Political conservative .437* .175 .431* .175 .437* .175 .437* .175  Own home .113* .027 .108* .027 .113* .027 .113* .027  Southwest .452 .312 .452 .312 .452 .312 .452 .312  Northeast −.988* .293 −.988* .293 −.988* .293 −.988* .293  Midwest −.261 .299 −.261 .299 −.261 .299 −.261 .299  West −.763* .273 −.763* .273 −.763* .273 −.763* .273  General punitive attitudes .319* .038 .314* .038 .319* .038 .319* .038 Intercept 9.475* .618 9.433* .618 9.475* .618 9.471* .618 Total variance explained (%) 18.4 21.1 18.2 18.4 N1 = 2,408 individuals N2 = 168 counties Model 1 Model 2 Model 3 Model 4 Variables b SE B SE b SE b SE Victim/offender dyad  White victim/black offender − − .224* .061 − − − −  Black Victim/black offender − − − − −.287 .461 − −  Latino Victim/black offender − − − − − − −.109 .416 Perceived black threat  Black criminal threat .197* .031 .189* .025 .196* .025 .196* .025 Black population context  Percent black −.748 .902 −.748 .902 −.748 .902 −.748 .902  Black growth .137* .046 .132* .046 .134* .046 .135* .046 County controls  Homicide rate (per 100,000) .042* .014 .038* .014 .044* .014 .044* .014  Concentrated disadvantage −.175 .248 −.175 .248 −.175 .248 −.175 .248  Percent Republican .349* .142 .338* .142 .349* .142 .349* .142  Population structure −.097 .171 −.097 .171 −.097 .171 −.097 .171 Demographic controls  Age −.006 .008 −.006 .008 −.006 .008 −.006 .008  Male .551* .191 .549* .191 .551* .191 .551* .191  Married .362 .224 .362 .224 .362 .224 .362 .224  Education level −.365 .229 −.365 .229 −.365 .229 −.365 .229  Family income −.152 .109 −.152 .109 −.152 .109 −.152 .109  Employed −.303 .215 −.303 .215 −.303 .215 −.303 .215  Political conservative .437* .175 .431* .175 .437* .175 .437* .175  Own home .113* .027 .108* .027 .113* .027 .113* .027  Southwest .452 .312 .452 .312 .452 .312 .452 .312  Northeast −.988* .293 −.988* .293 −.988* .293 −.988* .293  Midwest −.261 .299 −.261 .299 −.261 .299 −.261 .299  West −.763* .273 −.763* .273 −.763* .273 −.763* .273  General punitive attitudes .319* .038 .314* .038 .319* .038 .319* .038 Intercept 9.475* .618 9.433* .618 9.475* .618 9.471* .618 Total variance explained (%) 18.4 21.1 18.2 18.4 N1 = 2,408 individuals N2 = 168 counties * p < .05 (two-tailed test) We now turn to our substantive focus. Model 1 provides an assessment of both perceptions of black criminal threat and population context on punitive black sentiment, net of demographic and county control variables. The results show that both perceived black criminal threat and black population growth are significantly associated with punitive black sentiment while the relative size of the black population is not significant.10 Specifically, a one-unit increase in perceived black criminal threat increases punitiveness by .197 points. Similarly, a unit increase in black population growth increases punitive-black sentiment among whites by .137 points. Moving to the next set of models, we incorporate victim/offender dyads into the predictive equations to assess their independent effects on punitive-black sentiment. Because our fundamental interest is in the influence of victim and offender race on punitiveness, we report separate models disaggregated by victim race/ethnicity while holding the race of the offender constant.11 Comparing results across Models 2 through 4 in Table 2 makes clear the important role that victim race plays in public support for punitiveness toward black offenders. When subjects were asked about scenarios involving white victims and black offenders, they were substantially more likely to support harsher punishment toward blacks. Involvement of a white victim and a black offender increased punitive-black sentiment by .224 points. In contrast, crimes involving black or Latino victims had no significant effect on levels of support for punitive-black sentiment. Moreover, we observed a similar set of patterns with regard to punitive-Latino sentiment as the same set of predictors are significant in Table 3 across the various model specifications. We focus on the more substantive findings that guide our research questions. In Model 1, both perceptions of Latino criminal threat and Latino population growth are significant predictors of punitiveness toward Latino offenders. For example, a one unit increases in both perceived Latino criminal threat and Latino population growth are associated with increases in punitive sentiment of .192 and .194 points respectively. In the subsequent models (2 through 4), we include the victim/offender dyad measures. In these models, we vary the race and ethnicity of the victim while keeping the offender ethnicity constant. Similar to the results above, scenarios involving white victims and Latino offenders increased punitive-Latino sentiment by .137 points. None of the other victim contrasts emerged as significant predictors of punitive-Latino sentiment. As observed across both Tables 2 and 3, these findings suggest that attitudes favoring increased punishment for black and Latino offenders are closely tied to victim characteristics. This suggests that black and Latino offenders who target white victims are perceived to be more threatening and more deserving of punishment than blacks and Latinos who target members of other racial or ethnic minority groups. Table 3. Multilevel Models of Punitive-Latino Sentiment Regressed on Victim/Offender Dyad and Threat Indicators Model 1 Model 2 Model 3 Model 4 Variables b SE b SE b SE. b SE Victim/offender dyad  White victim/Latino offender − − .137* .023 − − − −  Black victim/Latino offender − − − − −.416 .423 − −  Latino victim/Latino offender − − − − − − −.037 .383 Perceived Latino threat  Latino criminal threat .192* .041 .198* .041 .191* .041 .192* .041 Latino population context  Percent Latino −.005 .006 −.005 .006 −.006 .006 −.006 .006  Latino growth .194* .061 .189* .061 .195* .061 .193* .061 County controls  Homicide rate (per 100,000) .064* .026 .064* .026 .064* .026 .064* .026  Concentrated disadvantage −.092 .189 −.092 .189 −.097 .189 −.097 .189  Percent Republican .355* .142 .349* .142 .354* .142 .356* .142  Population structure −.109 .168 −.109 .168 −.109 .168 −.109 .168 Demographic controls  Age .029* .011 .025* .011 .029* .011 .029* .011  Male .561* .191 .543* .191 .559* .191 .562* .191  Married .357 .255 .343 .255 .357 .255 .357 .255  Education level −.155 .208 −.153 .208 −.155 .208 −.155 .208  Family income −.214* .064 −.208 .061 −.212 .061 −.212 .061  Employed −.651* .285 −.651* .285 −.651* .285 −.651* .285  Political conservative .421* .188 .425* .188 .421* .188 .423* .188  Own home .689* .245 .683* .245 .689* .245 .689* .245  Southwest .413 .285 .413 .285 .403 .285 .403 .285  Northeast −.589* .254 −.581* .254 −.589* .254 −.589* .254  Midwest −.051 .249 −.051 .249 −.051 .249 −.051 .249  West −.348 .269 −.348 .269 −.348 .269 −.348 .269  General punitive attitudes .335* .038 .325* .038 .335* .038 .335* .038 Intercept 8.936* .461 8.901* .461 8.936* .461 8.936* .461 Total variance explained (%) 18.8 22.4 18.6 18.8 N1 = 2,408 individuals N2 = 168 counties Model 1 Model 2 Model 3 Model 4 Variables b SE b SE b SE. b SE Victim/offender dyad  White victim/Latino offender − − .137* .023 − − − −  Black victim/Latino offender − − − − −.416 .423 − −  Latino victim/Latino offender − − − − − − −.037 .383 Perceived Latino threat  Latino criminal threat .192* .041 .198* .041 .191* .041 .192* .041 Latino population context  Percent Latino −.005 .006 −.005 .006 −.006 .006 −.006 .006  Latino growth .194* .061 .189* .061 .195* .061 .193* .061 County controls  Homicide rate (per 100,000) .064* .026 .064* .026 .064* .026 .064* .026  Concentrated disadvantage −.092 .189 −.092 .189 −.097 .189 −.097 .189  Percent Republican .355* .142 .349* .142 .354* .142 .356* .142  Population structure −.109 .168 −.109 .168 −.109 .168 −.109 .168 Demographic controls  Age .029* .011 .025* .011 .029* .011 .029* .011  Male .561* .191 .543* .191 .559* .191 .562* .191  Married .357 .255 .343 .255 .357 .255 .357 .255  Education level −.155 .208 −.153 .208 −.155 .208 −.155 .208  Family income −.214* .064 −.208 .061 −.212 .061 −.212 .061  Employed −.651* .285 −.651* .285 −.651* .285 −.651* .285  Political conservative .421* .188 .425* .188 .421* .188 .423* .188  Own home .689* .245 .683* .245 .689* .245 .689* .245  Southwest .413 .285 .413 .285 .403 .285 .403 .285  Northeast −.589* .254 −.581* .254 −.589* .254 −.589* .254  Midwest −.051 .249 −.051 .249 −.051 .249 −.051 .249  West −.348 .269 −.348 .269 −.348 .269 −.348 .269  General punitive attitudes .335* .038 .325* .038 .335* .038 .335* .038 Intercept 8.936* .461 8.901* .461 8.936* .461 8.936* .461 Total variance explained (%) 18.8 22.4 18.6 18.8 N1 = 2,408 individuals N2 = 168 counties * p < .05 (two-tailed test) Table 3. Multilevel Models of Punitive-Latino Sentiment Regressed on Victim/Offender Dyad and Threat Indicators Model 1 Model 2 Model 3 Model 4 Variables b SE b SE b SE. b SE Victim/offender dyad  White victim/Latino offender − − .137* .023 − − − −  Black victim/Latino offender − − − − −.416 .423 − −  Latino victim/Latino offender − − − − − − −.037 .383 Perceived Latino threat  Latino criminal threat .192* .041 .198* .041 .191* .041 .192* .041 Latino population context  Percent Latino −.005 .006 −.005 .006 −.006 .006 −.006 .006  Latino growth .194* .061 .189* .061 .195* .061 .193* .061 County controls  Homicide rate (per 100,000) .064* .026 .064* .026 .064* .026 .064* .026  Concentrated disadvantage −.092 .189 −.092 .189 −.097 .189 −.097 .189  Percent Republican .355* .142 .349* .142 .354* .142 .356* .142  Population structure −.109 .168 −.109 .168 −.109 .168 −.109 .168 Demographic controls  Age .029* .011 .025* .011 .029* .011 .029* .011  Male .561* .191 .543* .191 .559* .191 .562* .191  Married .357 .255 .343 .255 .357 .255 .357 .255  Education level −.155 .208 −.153 .208 −.155 .208 −.155 .208  Family income −.214* .064 −.208 .061 −.212 .061 −.212 .061  Employed −.651* .285 −.651* .285 −.651* .285 −.651* .285  Political conservative .421* .188 .425* .188 .421* .188 .423* .188  Own home .689* .245 .683* .245 .689* .245 .689* .245  Southwest .413 .285 .413 .285 .403 .285 .403 .285  Northeast −.589* .254 −.581* .254 −.589* .254 −.589* .254  Midwest −.051 .249 −.051 .249 −.051 .249 −.051 .249  West −.348 .269 −.348 .269 −.348 .269 −.348 .269  General punitive attitudes .335* .038 .325* .038 .335* .038 .335* .038 Intercept 8.936* .461 8.901* .461 8.936* .461 8.936* .461 Total variance explained (%) 18.8 22.4 18.6 18.8 N1 = 2,408 individuals N2 = 168 counties Model 1 Model 2 Model 3 Model 4 Variables b SE b SE b SE. b SE Victim/offender dyad  White victim/Latino offender − − .137* .023 − − − −  Black victim/Latino offender − − − − −.416 .423 − −  Latino victim/Latino offender − − − − − − −.037 .383 Perceived Latino threat  Latino criminal threat .192* .041 .198* .041 .191* .041 .192* .041 Latino population context  Percent Latino −.005 .006 −.005 .006 −.006 .006 −.006 .006  Latino growth .194* .061 .189* .061 .195* .061 .193* .061 County controls  Homicide rate (per 100,000) .064* .026 .064* .026 .064* .026 .064* .026  Concentrated disadvantage −.092 .189 −.092 .189 −.097 .189 −.097 .189  Percent Republican .355* .142 .349* .142 .354* .142 .356* .142  Population structure −.109 .168 −.109 .168 −.109 .168 −.109 .168 Demographic controls  Age .029* .011 .025* .011 .029* .011 .029* .011  Male .561* .191 .543* .191 .559* .191 .562* .191  Married .357 .255 .343 .255 .357 .255 .357 .255  Education level −.155 .208 −.153 .208 −.155 .208 −.155 .208  Family income −.214* .064 −.208 .061 −.212 .061 −.212 .061  Employed −.651* .285 −.651* .285 −.651* .285 −.651* .285  Political conservative .421* .188 .425* .188 .421* .188 .423* .188  Own home .689* .245 .683* .245 .689* .245 .689* .245  Southwest .413 .285 .413 .285 .403 .285 .403 .285  Northeast −.589* .254 −.581* .254 −.589* .254 −.589* .254  Midwest −.051 .249 −.051 .249 −.051 .249 −.051 .249  West −.348 .269 −.348 .269 −.348 .269 −.348 .269  General punitive attitudes .335* .038 .325* .038 .335* .038 .335* .038 Intercept 8.936* .461 8.901* .461 8.936* .461 8.936* .461 Total variance explained (%) 18.8 22.4 18.6 18.8 N1 = 2,408 individuals N2 = 168 counties * p < .05 (two-tailed test) Furthermore, regardless of the race of the victim, there is clear evidence that both perceptual and objective measures of racial and ethnic group threat influence punitive sentiment toward black and Latino offenders. In all models, perceptual measures of both black and Latino criminal threats have significant effects on the outcomes. Although the absolute size of the black and Latino populations are not significant predictors of support for harsher punishment, both black and Latino growth demonstrate strong effects. Punitive sentiment is significantly greater in counties that experienced a recent influx of black or Latino residents. Notably, these effects are independent of all the individual and contextual level controls included in the model. In addition to the main effects for victim race in punishment, prior theorizing also suggests these influences may be conditioned by characteristics of the broader surrounding social context. Tables 4 and 5 investigate the extent to which victim effects are contingent upon perceptions of black and Latino criminal threats and objective indicators of black and Latino population growth. These analyses include the same battery of variables in Models 2 from Tables 2 and 3 but report only the focal coefficients in the interest of space.12 Given that concerns over fear of crime and risk of victimization are closely tied to the race of the victim, we expect stronger victim race effects among whites when racial and ethnic threat is high. Table 4. Multilevel Models of Punitive-Black Sentiment Regressed on Interactions Between Victim/Offender Dyad and Threat Indicators Model 1 Model 2 Variables b SE b SE Victim/offender dyad  White victim/black offender .179* .061 .177* .061 Perceived black threat  Black criminal threat .209* .028 .103* .025 Black population context  Black growth .131* .046 .129* .046 Black threat and victim/offender dyad  Black criminal threat * white victim/black offender .254* .071 − − Black population context and victim/offender dyad  Black growth * white victim/black offender − − .197* .068 Intercept 9.433* .618 9.333* .618 Total variance explained (%) 23.8 22.9 N1 = 2,408 individuals N2 = 168 counties Model 1 Model 2 Variables b SE b SE Victim/offender dyad  White victim/black offender .179* .061 .177* .061 Perceived black threat  Black criminal threat .209* .028 .103* .025 Black population context  Black growth .131* .046 .129* .046 Black threat and victim/offender dyad  Black criminal threat * white victim/black offender .254* .071 − − Black population context and victim/offender dyad  Black growth * white victim/black offender − − .197* .068 Intercept 9.433* .618 9.333* .618 Total variance explained (%) 23.8 22.9 N1 = 2,408 individuals N2 = 168 counties Note: Models include all variables reported in Model 2 of Table 2. * p < .05 (two-tailed test) Table 4. Multilevel Models of Punitive-Black Sentiment Regressed on Interactions Between Victim/Offender Dyad and Threat Indicators Model 1 Model 2 Variables b SE b SE Victim/offender dyad  White victim/black offender .179* .061 .177* .061 Perceived black threat  Black criminal threat .209* .028 .103* .025 Black population context  Black growth .131* .046 .129* .046 Black threat and victim/offender dyad  Black criminal threat * white victim/black offender .254* .071 − − Black population context and victim/offender dyad  Black growth * white victim/black offender − − .197* .068 Intercept 9.433* .618 9.333* .618 Total variance explained (%) 23.8 22.9 N1 = 2,408 individuals N2 = 168 counties Model 1 Model 2 Variables b SE b SE Victim/offender dyad  White victim/black offender .179* .061 .177* .061 Perceived black threat  Black criminal threat .209* .028 .103* .025 Black population context  Black growth .131* .046 .129* .046 Black threat and victim/offender dyad  Black criminal threat * white victim/black offender .254* .071 − − Black population context and victim/offender dyad  Black growth * white victim/black offender − − .197* .068 Intercept 9.433* .618 9.333* .618 Total variance explained (%) 23.8 22.9 N1 = 2,408 individuals N2 = 168 counties Note: Models include all variables reported in Model 2 of Table 2. * p < .05 (two-tailed test) Table 5. Multilevel Models of Punitive-Latino Sentiment Regressed on Interactions Between Victim/Offender Dyad and Threat Indicators Model 1 Model 2 Variables B SE B SE Victim/offender dyad  White victim/Latino offender .154* .026 .159* .025 Perceived Latino threat  Latino criminal threat .182* .045 .188* .045 Latino population context  Latino growth .189* .061 .183* .061 Latino threat and victim/offender dyad  Latino criminal threat * white victim/Latino offender .241* .076 − − Latino population context and victim/offender dyad  Latino growth * white victim/Latino offender − − .189* .067 Intercept 8.911* .461 8.919* .461 Total variance explained (%) 24.7 25.4 N1 = 2,408 individuals N2 = 168 counties Model 1 Model 2 Variables B SE B SE Victim/offender dyad  White victim/Latino offender .154* .026 .159* .025 Perceived Latino threat  Latino criminal threat .182* .045 .188* .045 Latino population context  Latino growth .189* .061 .183* .061 Latino threat and victim/offender dyad  Latino criminal threat * white victim/Latino offender .241* .076 − − Latino population context and victim/offender dyad  Latino growth * white victim/Latino offender − − .189* .067 Intercept 8.911* .461 8.919* .461 Total variance explained (%) 24.7 25.4 N1 = 2,408 individuals N2 = 168 counties Note: Models include all control variables reported in in Model 2 of Table 3. * p < .05 (two-tailed test) Table 5. Multilevel Models of Punitive-Latino Sentiment Regressed on Interactions Between Victim/Offender Dyad and Threat Indicators Model 1 Model 2 Variables B SE B SE Victim/offender dyad  White victim/Latino offender .154* .026 .159* .025 Perceived Latino threat  Latino criminal threat .182* .045 .188* .045 Latino population context  Latino growth .189* .061 .183* .061 Latino threat and victim/offender dyad  Latino criminal threat * white victim/Latino offender .241* .076 − − Latino population context and victim/offender dyad  Latino growth * white victim/Latino offender − − .189* .067 Intercept 8.911* .461 8.919* .461 Total variance explained (%) 24.7 25.4 N1 = 2,408 individuals N2 = 168 counties Model 1 Model 2 Variables B SE B SE Victim/offender dyad  White victim/Latino offender .154* .026 .159* .025 Perceived Latino threat  Latino criminal threat .182* .045 .188* .045 Latino population context  Latino growth .189* .061 .183* .061 Latino threat and victim/offender dyad  Latino criminal threat * white victim/Latino offender .241* .076 − − Latino population context and victim/offender dyad  Latino growth * white victim/Latino offender − − .189* .067 Intercept 8.911* .461 8.919* .461 Total variance explained (%) 24.7 25.4 N1 = 2,408 individuals N2 = 168 counties Note: Models include all control variables reported in in Model 2 of Table 3. * p < .05 (two-tailed test) Tables 4 and 5 provide clear evidence for these expectations. The white victim effects observed in previous analyses are particularly pronounced in social contexts characterized by greater perceived criminal threat for both blacks (γ = .254) and Latinos (γ = .241).13 In line with our theoretical expectations, then, support for enhanced social control directed at black and Latino offenders who target white victims is greatest where fear of black and Latino crime is most pronounced. Furthermore, Models 2 in Tables 4 and 5 show that the effect of victim race on support for punitiveness is also significantly conditioned by recent growth in both the black (γ = .197) and Latino (γ = .189) populations. Social contexts characterized by a recent inflow of black and Latino residents tend to place greater emphasis on severity of punishment for black and Latino offenders who target white victims. The magnitude of these effects is substantial and is perhaps best demonstrated by Figures 1 through 4, which plot the predicted values for our significant interaction effects. Figures 1 and 2 present the effects of victim race on punitive sentiment respectively by black and Latino criminal threats. In Figures 3 and 4, we illustrate these same effects for black and Latino population growth respectively. As the figures show, public support for punitiveness toward black and Latino offenders is highest in contexts where residents perceive minorities to be the most criminally threatening and in counties that experienced recent growth in the black and Latino populations. Further, these effects are dramatically more pronounced in cases involving white victims rather than minority victims (none of the minority victim effects were statistically significant). Overall, these results are consistent with expectations derived from an integration of group threat theory and contemporary perspectives on the importance of victim characteristics in punishment. Figure 1. View largeDownload slide The Effect of White Victim/Black Offender Dyad on Punitive-Black Sentiment by Perceived Black Criminal Threat Figure 1. View largeDownload slide The Effect of White Victim/Black Offender Dyad on Punitive-Black Sentiment by Perceived Black Criminal Threat Figure 2. View largeDownload slide The Effect of White Victim/Latino Offender Dyad on Punitive-Latino Sentiment by Perceived Latino Criminal Threat Figure 2. View largeDownload slide The Effect of White Victim/Latino Offender Dyad on Punitive-Latino Sentiment by Perceived Latino Criminal Threat Figure 3. View largeDownload slide The Effect of White Victim/Black Offender Dyad on Punitive-Black Sentiment by Black Population Growth Figure 3. View largeDownload slide The Effect of White Victim/Black Offender Dyad on Punitive-Black Sentiment by Black Population Growth Figure 4. View largeDownload slide The Effect of White Victim/Latino Offender Dyad on Punitive-Latino Sentiment by Latino Population Growth Figure 4. View largeDownload slide The Effect of White Victim/Latino Offender Dyad on Punitive-Latino Sentiment by Latino Population Growth Supplemental Analyses We conducted a series of supplemental analyses, specifying punitive-white sentiment as the dependent variable, to further examine the robustness of our results. To create this outcome, we relied on the six identical questions used to form the other dependent variables except whites were specified as the target group (e.g., “The U.S. criminal justice system is too lenient with [white] offenders; The U.S. criminal justice system needs tougher prison sentences on [white] offenders.”). We also created a white criminal threat measure relying on the four statements that captured perceived criminal threat but using whites as the target group (e.g., “[Whites] pose a greater threat to public order and safety than other racial or ethnic groups.”). Furthermore, we incorporated dynamic and static measures of white population size. Measures of the racial and ethnic victim/offender dyads were created specifying whites as offenders and the other racial and ethnic groups as victims. We estimated equivalent models to those in Tables 2 and 3. Only three variables emerged as statistically significant (p < .05) predictors of punitive white sentiment: general punitive attitudes, black victim/white offender, and Latino victim/white offender. General punitive attitudes increased white punitive sentiment (b = .271 SE = .053, p < .05 and b = .268 SE = .055, p < .05). On the other hand, black victim/white offender (b = −.197 SE = .068, p < .05) and Latino victim/white offender dyads significantly decreased punitive white sentiment (b = −.189 SE = .065, p < .05). These findings, in combination with those reported above, indicate that white respondents are less punitive when faced with scenarios that involve minority victims rather than white victims. Thus, even when racial and ethnic minorities are victims of crimes perpetrated by whites, the perception among white citizens is that they should not be punished as severely.14 DISCUSSION Prior research in the sociology of punishment suggests that racial group threat dynamics are related to public support for the exercise of formal social control in society (Eitle et al. 2002; Liska et al. 1985). One largely uninvestigated issue, however, is the role that victim characteristics play in perceptions of criminal threat and in the formation of punitive sentiment. Some research suggests a process of victim discounting or “benign neglect” may occur in cases that involve minority victims (Liska and Chamlin 1984). Such arguments are consistent with work that demonstrates the salience of racial dyads in criminal punishment (Swigert and Farrell 1977). To date, though, no prior research has explicitly integrated victim characteristics into the study of group threat processes. The current study addressed this issue, using nationally representative survey data to examine the effects of victim-offender racial dyads on punitive sentiment toward minority defendants in the justice system. We incorporate both objective population contexts and subjective perceptions of criminal threat, test for both stable and dynamic indicators of group threat, and examine criminal threat associated with both the black and Latino populations in the United States. Our findings suggest key roles for subjective perceptions of criminal threat, aggregate shifts in racial population dynamics, and individual victim/offender relations. Overall, we find substantial support among white respondents for punitive sentiment aimed at black and Latino offenders. Moreover, more than one third of the sample agreed that criminal acts were more serious if they involved a white victim and a black or Latino offender. This speaks to a general punitive sentiment directed at minority offenders and it provides prima facie evidence for the salience of victim/offender dyads in subjective perceptions of criminal threat. It also raises important questions about the role that the criminal justice system plays in reinforcing racial stereotypes and in shaping contemporary race relations (Kennedy 1997; Wilson 2009). Notably, these findings are also consistent with empirical research in other social domains beyond punishment disparities. For instance, negative stereotypes of racial and ethnic minority students by teachers and administrators have been linked to poor educational outcomes for minority students such as elevated levels of school failure and punishment (Dee 2005). Similarly, scholars have argued that negative stereotypical assessments of racial and ethnic minorities are linked to persistent disparities in the quality of health care and health outcomes for racial and ethnic minorities (Goodwin and Duke 2012). Thus, our findings are consistent with a broader literature on racial and ethnic disparities in other social domains, and they extend this work to suggest that negative perceptions of racial and ethnic minorities are essential for understanding and addressing these disparities. As such, studies should continue to investigate the underlying sources of individual-level perceptions and their consequences for societal inequalities. In terms of racial population contexts, aggregate patterns of punitiveness were not significantly related to the absolute size of black or Latino populations, but they were associated substantially with recent population growth for both groups. White survey respondents in counties with growing black or Latino populations expressed greater punitive sentiment toward black and Latino defendants. These results are largely consistent with prior work that suggests racial demographics are tied to group threat processes, and they support recent arguments about the importance of including dynamic measures of population change in addition to static indicators of population size (Chamlin 1989; Jacobs and O’Brien 1998; Johnson et al. 2011). These findings are congruent with “defended neighborhoods” perspectives that suggest racial group threat becomes particularly heightened in contexts characterized by rapid in-migration of minority groups, especially in racially homogenous white localities (Green et al. 1998). Whites in communities with rapidly growing minority populations may view these groups as more criminally threatening, or alternatively, individuals who view minorities as more threatening may be more sensitive to demographic population shifts. The latter explanation is largely consistent with work that finds racial threat perceptions are also related to assessments of minority group size (Gallagher 2003). With regard to subjective perceptions of group threat, we find strong and consistent evidence that criminal threat is positively related to punitive sentiment toward black and Latino offenders. Importantly, these effects emerge net of individual and aggregate controls, including general punitive attitudes and broader population demographics. Individuals who view blacks and Latinos as more criminally involved and greater threats to public safety are more likely to support punitive measures that specifically target them in the criminal justice system. In addition to highlighting the salience of criminal threat as a key theoretical mechanism in racial group threat theory (Eitle et al. 2002), these results also highlight the importance of combatting racial stereotypes in popular discourse. The racialization of crime and criminality in popular media has been well documented (Chiricos and Eschholz 2002; Dixon and Linz 2000), and may feed into implicit racial associations that contribute to inequalities in justice outcomes (Smith and Levinson 2012). Prior work, for instance, suggests that even among judges who profess strong egalitarian beliefs, implicit racial biases can still affect punishment decisions (Rachlinski et al. 2009). Thus, an essential step in trying to reduce race-based disparities in the justice system might be to train law enforcement, prosecutors, judges, juries, and others about the salience of stereotypes and how they can be consequential for the application of punishment for minorities (Correll et al. 2007; Tonry 2010). Our research findings also highlight the importance of victim/offender racial characteristics in group threat theories and in public support for punitive justice responses. We find that victim race plays an important role in punitive sentiment toward black and Latino offenders. Specifically, white respondents report greater punitiveness when asked about scenarios involving white victims and minority perpetrators. The same is not true, however, for situations involving minority victims. One in three respondents in our sample viewed criminal acts as less serious when they involved minority victims. Moreover, white respondents express less punitiveness toward white offenders who target minority victims. For a nontrivial fraction of the white majority, this suggests an ethos of victim discounting with regard to minority victims (Baldus et al. 1983). These results also bolster prior work on the perceived risk of criminal victimization in group threat processes. White respondents may express greater punitive sentiment for minority offenders when there is a white victim, in part, because it elevates their own estimates of the risk of victimization. This process reflects what Lincoln Quillian and Devah Pager (2010) referred to as “stereotype amplification”—the distorted assessment of individual risk based on racialized fears and stereotypical social cues. Finally, we investigated potential interactions among key social predictors in our model. Specifically, we expected that victim race would have stronger effects on punitive sentiment in contexts characterized by elevated perceptions of minority criminal threat and in areas with recent minority population growth. Both of these expectations were supported in the data. The effect of a white victim was notably stronger in areas where black criminal threat was high and black population growth was more rapid. This effect also increased significantly when Latino criminal threat was high and Latino population growth increased. Taken together, these results provide strong evidence that race of victim effects are conditioned by broader social contexts involving perceived criminal threat and objective population growth. Again, these various group threat dynamics operate independently of individual attitudinal and demographic predictors and other county-level controls. Although this study provides significant new insights into group threat and social control, it is not without its limitations. First, we focus on punitive sentiment. We show that group threat variables interact with victim race to shape punitive attitudes, but future work is needed that demonstrates this relationship for additional social control outcomes, such as arrest and incarceration rates. Second, we examine group threat processes that attach to black and Latino populations, yet some important differences might exist within these broad racial and ethnic categorizations. In particular, there may be unique threat mechanisms that operate for black or Latino immigrant subgroups (Johnson et al. 2011; Parker et al. 2005; Wang 2012). Unique group threat dynamics could also be at play among members of different racial and ethnic minority groups. Hubert Blumer’s (1958) original formulation of group threat theory suggested that as members of a racial group become increasingly oppressed, they are more likely to see other groups as potential threats (Bobo and Hutchings 1996). This implies that racial threat theory should apply to inter-group conflict among underrepresented minority groups as well. Replicating the current findings with larger samples of black and Latino respondents, as well as members of other underrepresented groups, could therefore prove to be enlightening. Third, although we investigate the impact of black and Latino criminal threats on punitiveness, we are not able to disentangle the unique historical legacies of racism that are associated with different racial and ethnic groups. Future research might consider incorporating racial and ethnic group histories in punishment given the tenuous historical conflict around race in the United States (Bobo and Hutchings 1996; Bonilla-Silva 2006). Fourth, we focus primarily on criminal threat, which is closely tied to social control efforts in the criminal justice system (Eitle et al. 2002), but future work might expand on this study by investigating additional threat mechanisms, such as economic competition or the relative political power of minority groups. In addition, cultural indicators of group threat, such as changing social identity, multiculturalism, and exposure to foreign language (Citrin, Reingold, and Green 1990; Newman 2013; Newman, Hartman, and Taber 2012), could also provide productive avenues for future research. Further, while we were able to examine the racial/ethnic nature of the victim/offender dyads in a general sense, we were not able to measure more nuanced racial, ethnic, and gendered relationships within victim/offender dyads. For example, it is possible that social control is strongly impacted if the offender is a minority male and the victim is a white female (Hawkins 1987; LaFree 1989). Future research could expand on our work by using survey vignettes to examine more fine-grained relationships within racial, ethnic, and gendered victim/offender dyads. Finally, we are somewhat limited in our ability to examine the underlying theoretical processes that feed into subjective perceptions of criminal threat. Ideally, these processes should be studied with longitudinal data that allows for causal ordering to be established between minority population growth, subjective perceptions of criminal threat, and changing attitudes towards crime and punishment. In conclusion, this study combined both perceptual and objective measures of group threat, examined punitive sentiment toward both black and Latino offenders, and incorporated unique measures of the race of victims into the study of group threat and social control. We find compelling evidence that both perceived criminal threat and racial and ethnic population growth are related to punitiveness among whites, and we show that these relationships are closely tied to victim race. These results are important for several reasons. They inform ongoing debates over modern racism and colorblind justice (Alexander 2010). They highlight the essential role that criminal threat and fear of victimization play in public support for tough on crime policies. And they emphasize the ongoing need to investigate the complex ways that aggregate population dynamics condition punitive attitudes regarding race, crime, and punishment. As such, research that replicates and expands this work could prove invaluable for not only advancing empirical scholarship on race and crime but also for furthering our theoretical understanding of underlying sources of racial and ethnic inequality in the American criminal justice system. Footnotes 1 The response rate was calculated using the following formula: completes/(completes + terminals + refusals). This formula is supported and developed by the American Association for Public Opinion Research (2008). 2 With regard to racial and ethnic background of the original sample (2,736), 88 percent of the sample was non-Latino white, 7 percent was black, 3 percent was Latino, and 2 percent was Asian. Since our primary interest is in how the dominant group, in our case whites, views blacks and Latinos, we restricted the sample to non-Latino white respondents. 3 The sampling frame includes American households with either landlines or cellular phones, but as with other telephone survey research, households without either form of telecommunication may be underrepresented, which could account for the overrepresentation of white, female, older, and higher-income respondents. To determine if our results are biased, we re-estimated the models by weighting the sample by the 2010 U.S. Census. Our weighting procedures produced coefficients that mirrored the ones presented in our tables. Based on these supplemental analyses, we do not see evidence that our findings are impacted in any way because we observed identical patterns of results. 4 In an effort to reduce confusion in the racial and ethnic categories in our study design, interviewers explained to respondents that we were referring exclusively to blacks/African Americans in the survey questions. Respondents were also told that we were referring exclusively to Hispanic ethnicity in the survey questions. The interviewers took tremendous care to distinguish the two racial and ethnic categories. Given our efforts to clarify the differences between racial and ethnic categories, we are confident that our measures capture support for increased social control against blacks and Latinos respectively. However, we cannot rule out the possibility that some respondents may not have interpreted the survey questions around racial and ethnic categories as mutually exclusive. 5 To ensure standardization across respondents, the interviewers explained that our use of the term violent crime was in reference to the following crimes: robbery, aggravated assaults, and homicides. 6 We estimated confirmatory factor analysis (CFA), to evaluate several of our constructs: punitive-black sentiment, punitive-Latino sentiment, black criminal threat, Latino criminal threat, concentrated disadvantage, population structure, and general punitive attitudes. We estimated the models with Analysis of Moment Structures (AMOS) version 22 (Arbuckle 2013). In each case, the factor loadings for the constructs were relatively high ranging from .61 to .88. The fit indices also indicated that the data fit the model specifications well: AGFI = .98; RMSEA = .012; and CN = 803. 7 Estimation of the intraclass correlation (ICC) for our data illuminates the utility of multilevel analysis. Specifically, the overall variance in punitive-black sentiment was 3.27, with 2.55 lying within counties and .72 between counties. This implies that about 78 percent of the variance in punitive-black sentiment lies within counties, while the remaining 22 percent falls between counties. There was also significant variation in the average level of punitive-black sentiment across counties (χ2=284, p < .05). Similarly, the total variance in punitive-Latino sentiment was 3.10, with 2.45 within counties and .65 between counties. This translates into about 79 percent of the variance in punitive-Latino sentiment being located within counties and 21 percent being located between counties. Further, there is significant variation in the average level of punitive-Latino sentiment across counties (χ2=271, p < .05). These results imply that there is sufficient variance in both punitive-black sentiment and punitive-Latino sentiment between- and within-counties to support a multilevel analysis. The intercept reliability estimates were .76 for punitive-black sentiment and .78 for punitive-Latino sentiment, indicating that the data are sufficient to generate reliable estimates. 8 Specifically, we used the XTMIXED function in STATA 14 to estimate our multilevel linear models. Moreover, to assess multicollinearity among the predictor variables, we examined the variance inflation factor (VIF). Multicollinearity does not appear to be a problem, as most of the predictors have VIFs near 1 and none has a VIF greater than 2.4, suggesting that the variables are theoretically and empirically distinct constructs. 9 Similar patterns were observed for scenarios involving Asian victims. Only six of white respondents reported that crimes were more serious when black or Latino offenders targeted Asian victims. 10 We also assessed the possibility of nonlinear effects of percent black, black growth, percent Latino, and Latino growth on our punitive sentiment outcomes. The results showed no evidence of significant nonlinear relationships between the four predictors and our outcome measures. Additionally, we explored nonlinear effects of perceived criminal threat predictors on our outcomes, which also yielded nonsignificant results. 11 We initially entered all of the victim/offender dyads into a single equation to predict punitiveness, but this resulted in multicollinearity among the victim/offender dyads. Thus, we report estimates from separate models for each victim/offender relationship. 12 In the interest of space, we only focus on the white victim and minority offender dyads because none of the other combinations were significant in either the main effects or interaction analyses. Full results are available by request. 13 We allowed the slopes for the white victim/black offender dyad (slope variance = .199, p < .05) and the white victim/Latino offender dyad (slope variance = .192, p < .05) to vary across counties. Indeed the slopes varied significantly, indicating that the relationship between racial and ethnic victim/offender dyads and punitive sentiment varies significantly across county contexts. 14 We attempted to estimate models in which racial and ethnic minorities (blacks and Latinos) were the respondents. However, the models were not able to converge to provide reliable estimates because the cell sizes were too small to estimate the specified models. APPENDIX Table A1. Metrics and Definitions for Control Variables Variable Definition Homicide rate A variable that captured the reported rate of homicides for each county during the years of 2008, 2009, and 2010. The homicide rate is calculated per 100,000 residents in the county. Concentrated disadvantage A variable measured by three county-level census indicators in 2010: percentage of persons on public assistance, percentage of households below the poverty level, and percentage of persons unemployed. These items were standardized and combined to form a measure of disadvantage. The alpha coefficient was .87. Percent Republican A variable measured as the percent of the population voting for the Republican presidential candidate in 2012. Population structure A variable measured by county population size and population density for each county. The correlation between these two items is .83. Age A variable measured in years for each respondent. Male A dichotomous variable that captured whether respondents reported being male (coded “1”) or female (coded “0”). Married A dichotomous variable that captured whether respondents reported being married (coded “1”) or not (coded “0”). Education level A dichotomous measure that captured whether respondents reported having a college degree (coded “1”) or not (coded “0”). Family income A variable measured as the total amount of household income. This variable is measured as follows: less than $15,000; $15,000 to $34,999; $35,000 to $49,999; $50,000 to $74,999; $75,000 to $99,999; and $100,000 or more. Employed A dichotomous measure that captured whether respondents reported being employed (coded “1”) or not (coded “0”). Political conservative A dichotomous measure that captured whether respondents reported being politically conservative (coded “1”) or non-politically conservative (coded “0”). Own home A dichotomous measure that captured whether respondents reported being a homeowner (coded “1”) or not (coded “0”). Census region These variables were measured by a set of dichotomous variables where 1 = Northeast, 1 = Midwest, and 1 = West with respondents who reside in the South as the reference designation. Southwest A dichotomous measure that indicated whether or not the respondent lived in the Southwest United States (coded as “1”) or not (coded as “0”) General punitive attitudes A variable measured by four questions (0 = strongly disagree to 3 = strongly agree): “Sentences for violent offenders should be more severe than they are today; Sentences for nonviolent offenders should be more severe than they are today; Sentences for repeat offenders should be more severe than they are today; The U.S. courts are too lenient with offenders.” The alpha coefficient is .88. Variable Definition Homicide rate A variable that captured the reported rate of homicides for each county during the years of 2008, 2009, and 2010. The homicide rate is calculated per 100,000 residents in the county. Concentrated disadvantage A variable measured by three county-level census indicators in 2010: percentage of persons on public assistance, percentage of households below the poverty level, and percentage of persons unemployed. These items were standardized and combined to form a measure of disadvantage. The alpha coefficient was .87. Percent Republican A variable measured as the percent of the population voting for the Republican presidential candidate in 2012. Population structure A variable measured by county population size and population density for each county. The correlation between these two items is .83. Age A variable measured in years for each respondent. Male A dichotomous variable that captured whether respondents reported being male (coded “1”) or female (coded “0”). Married A dichotomous variable that captured whether respondents reported being married (coded “1”) or not (coded “0”). Education level A dichotomous measure that captured whether respondents reported having a college degree (coded “1”) or not (coded “0”). Family income A variable measured as the total amount of household income. This variable is measured as follows: less than $15,000; $15,000 to $34,999; $35,000 to $49,999; $50,000 to $74,999; $75,000 to $99,999; and $100,000 or more. Employed A dichotomous measure that captured whether respondents reported being employed (coded “1”) or not (coded “0”). Political conservative A dichotomous measure that captured whether respondents reported being politically conservative (coded “1”) or non-politically conservative (coded “0”). Own home A dichotomous measure that captured whether respondents reported being a homeowner (coded “1”) or not (coded “0”). Census region These variables were measured by a set of dichotomous variables where 1 = Northeast, 1 = Midwest, and 1 = West with respondents who reside in the South as the reference designation. Southwest A dichotomous measure that indicated whether or not the respondent lived in the Southwest United States (coded as “1”) or not (coded as “0”) General punitive attitudes A variable measured by four questions (0 = strongly disagree to 3 = strongly agree): “Sentences for violent offenders should be more severe than they are today; Sentences for nonviolent offenders should be more severe than they are today; Sentences for repeat offenders should be more severe than they are today; The U.S. courts are too lenient with offenders.” The alpha coefficient is .88. Table A1. Metrics and Definitions for Control Variables Variable Definition Homicide rate A variable that captured the reported rate of homicides for each county during the years of 2008, 2009, and 2010. The homicide rate is calculated per 100,000 residents in the county. Concentrated disadvantage A variable measured by three county-level census indicators in 2010: percentage of persons on public assistance, percentage of households below the poverty level, and percentage of persons unemployed. These items were standardized and combined to form a measure of disadvantage. The alpha coefficient was .87. Percent Republican A variable measured as the percent of the population voting for the Republican presidential candidate in 2012. Population structure A variable measured by county population size and population density for each county. The correlation between these two items is .83. Age A variable measured in years for each respondent. Male A dichotomous variable that captured whether respondents reported being male (coded “1”) or female (coded “0”). Married A dichotomous variable that captured whether respondents reported being married (coded “1”) or not (coded “0”). Education level A dichotomous measure that captured whether respondents reported having a college degree (coded “1”) or not (coded “0”). Family income A variable measured as the total amount of household income. This variable is measured as follows: less than $15,000; $15,000 to $34,999; $35,000 to $49,999; $50,000 to $74,999; $75,000 to $99,999; and $100,000 or more. Employed A dichotomous measure that captured whether respondents reported being employed (coded “1”) or not (coded “0”). Political conservative A dichotomous measure that captured whether respondents reported being politically conservative (coded “1”) or non-politically conservative (coded “0”). Own home A dichotomous measure that captured whether respondents reported being a homeowner (coded “1”) or not (coded “0”). Census region These variables were measured by a set of dichotomous variables where 1 = Northeast, 1 = Midwest, and 1 = West with respondents who reside in the South as the reference designation. Southwest A dichotomous measure that indicated whether or not the respondent lived in the Southwest United States (coded as “1”) or not (coded as “0”) General punitive attitudes A variable measured by four questions (0 = strongly disagree to 3 = strongly agree): “Sentences for violent offenders should be more severe than they are today; Sentences for nonviolent offenders should be more severe than they are today; Sentences for repeat offenders should be more severe than they are today; The U.S. courts are too lenient with offenders.” The alpha coefficient is .88. Variable Definition Homicide rate A variable that captured the reported rate of homicides for each county during the years of 2008, 2009, and 2010. The homicide rate is calculated per 100,000 residents in the county. Concentrated disadvantage A variable measured by three county-level census indicators in 2010: percentage of persons on public assistance, percentage of households below the poverty level, and percentage of persons unemployed. These items were standardized and combined to form a measure of disadvantage. The alpha coefficient was .87. Percent Republican A variable measured as the percent of the population voting for the Republican presidential candidate in 2012. Population structure A variable measured by county population size and population density for each county. The correlation between these two items is .83. Age A variable measured in years for each respondent. Male A dichotomous variable that captured whether respondents reported being male (coded “1”) or female (coded “0”). Married A dichotomous variable that captured whether respondents reported being married (coded “1”) or not (coded “0”). Education level A dichotomous measure that captured whether respondents reported having a college degree (coded “1”) or not (coded “0”). Family income A variable measured as the total amount of household income. This variable is measured as follows: less than $15,000; $15,000 to $34,999; $35,000 to $49,999; $50,000 to $74,999; $75,000 to $99,999; and $100,000 or more. Employed A dichotomous measure that captured whether respondents reported being employed (coded “1”) or not (coded “0”). Political conservative A dichotomous measure that captured whether respondents reported being politically conservative (coded “1”) or non-politically conservative (coded “0”). Own home A dichotomous measure that captured whether respondents reported being a homeowner (coded “1”) or not (coded “0”). Census region These variables were measured by a set of dichotomous variables where 1 = Northeast, 1 = Midwest, and 1 = West with respondents who reside in the South as the reference designation. Southwest A dichotomous measure that indicated whether or not the respondent lived in the Southwest United States (coded as “1”) or not (coded as “0”) General punitive attitudes A variable measured by four questions (0 = strongly disagree to 3 = strongly agree): “Sentences for violent offenders should be more severe than they are today; Sentences for nonviolent offenders should be more severe than they are today; Sentences for repeat offenders should be more severe than they are today; The U.S. courts are too lenient with offenders.” The alpha coefficient is .88. REFERENCES Alba Richard , Rumbaut Ruben G. , Marotz Karen . 2005 . “ A Distorted Nation: Perceptions of Racial/Ethnic Group Sizes and Attitudes Toward Immigrants and Other Minorities.” Social Forces 84 : 901 - 19 . Google Scholar CrossRef Search ADS Alexander Michelle. 2010 . The New Jim Crow: Mass Incarceration in the Age of Colorblindness . New York : The New Press . American Association for Public Opinion Research (AAPOR) . 2008 . Standard Definitions: Final Dispositions of Case Codes and Outcome Rates for Surveys . Ann Arbor, MI : American Association for Public Opinion Research . Arbuckle James L. 2013 . AMOS. Version 22. Chicago : IBM SPSS Inc . Auerhahn Kathleen. 2007 . “ Adjudication Outcomes in Intimate and Non-intimate Homicides.” Homicide Studies 11 : 213 - 30 . Google Scholar CrossRef Search ADS Baik Ellen. 2012 . “Gender, Religion, and National Origin: Latinos' Attitude Toward Capital Punishment.” Journal of Social Sciences 8 : 79 - 84 . Google Scholar CrossRef Search ADS Baldus David C. , Pulaski Charles , Woodworth George . 1983 . “ Comparative Review of Death Sentences: An Empirical Study of the Georgia Experience.” Journal of Criminal Law and Criminology 74 : 661 - 753 . Google Scholar CrossRef Search ADS Barkan Steven E. , Cohn Steven F . 2005 . “ Why Whites Favor Spending More Money to Fight Crime: The Role of Racial Prejudice.” Social Problems 52 : 300 - 14 . Google Scholar CrossRef Search ADS Baumer Eric P. , Messner Steven F. , Felson Richard B . 2000 . “ The Role of Victim Characteristics in the Disposition of Murder Cases.” Justice Quarterly 17 : 281 - 307 . Google Scholar CrossRef Search ADS Baumer Eric P. , Messner Steven F. , Rosenfeld Richard . 2003 . “ Explaining Spatial Variation in Support for Capital Punishment: A Multilevel Analysis.” American Journal of Sociology 108 : 844 - 75 . Google Scholar CrossRef Search ADS Black Donald. 1976 . The Behavior of Law . New York : Academic Press . Blalock Hubert M. Jr . 1967 . Toward a Theory of Minority-Group Relations . New York : John Wiley & Sons . Blumer Herbert. 1958 . “ Race Prejudice as a Sense of Group Position.” Pacific Sociological Review 1 : 3 - 7 . Google Scholar CrossRef Search ADS Bobo Lawrence , Hutchings Vincent L . 1996 . “ Perceptions of Racial Group Competition: Extending Blumer’s Theory of Group Position to a Multiracial Social Context.” American Sociological Review 61 : 951 - 72 . Google Scholar CrossRef Search ADS Bobo Lawrence D. , Johnson Devon . 2004 . “ A Taste for Punishment: Black and White Americans’ Views on the Death Penalty and the War on Drugs.” Du Bois Review: Social Science Research on Race 1 : 151 - 80 . Bonilla-Silva Eduardo. 2004 . “ From Bi-Racial to Tri-Racial: Towards a New System of Racial Stratification in the USA.” Ethnic and Racial Studies 27 : 931 - 50 . Google Scholar CrossRef Search ADS Bonilla-Silva Eduardo. 2006 . Racism Without Racists: Color-Blind Racism and the Persistence of Racial Inequality in the United States . New York : Rowman and Littlefield . Brewer Rose M. , Heitzeg Nancy A . 2008 . “ The Racialization of Crime and Punishment: Criminal Justice, Color-Blind Racism, and the Political Economy of the Prison Industrial Complex.” American Behavioral Scientist 51 : 625 - 44 . Google Scholar CrossRef Search ADS Bridges George S. , Crutchfield Robert D . 1988 . “ Law, Social Standing, and Racial Disparities in Imprisonment.” Social Forces 66 : 699 - 724 . Google Scholar CrossRef Search ADS Bridges George S. , Crutchfield Robert D. , Simpson Edith E . 1987 . “ Crime, Social Structure, and Criminal Punishment: White and Nonwhite Rates of Imprisonment.” Social Problems 34 : 345 - 61 . Google Scholar CrossRef Search ADS Britt Chester L. 2000 . “ Social Context and Racial Disparities in Punishment Decisions.” Justice Quarterly 17 : 707 - 32 . Google Scholar CrossRef Search ADS Brooks Richard R. W. , Jeon-Slaughter Haekyung . 2001 . “ Race, Income, and Perceptions of the U.S. Court System.” Behavioral Sciences and the Law 29 : 249 - 64 . Google Scholar CrossRef Search ADS Brown Rupert. 2010 . Prejudice: Its Social Psychology . Oxford, UK : Wiley-Blackwell . Brunson Rod K. 2007 . “‘ Police Don’t Like Black People’: African-American Young Men’s Accumulated Police Experiences.” Criminology & Public Policy 6 : 71 - 102 . Google Scholar CrossRef Search ADS Chambliss William J. , Seidman Robert . 1971 . Law, Order, and Power . Reading, MA : Addison-Wesley . Chamlin Mitchell B. 1989 . “ A Macro Social Analysis of Change in Police Force Size, 1972-1982: Controlling for Static and Dynamic Influences.” Sociological Quarterly 30 : 615 - 24 . Google Scholar CrossRef Search ADS Chamlin Mitchell B. , Liska Allen E . 1992 . “Social Structure and Crime Control Revisited: The Declining Significance of Intergroup Threat.” Pp. 103-12 in Social Threat and Social Control , edited by Liska Allen E . Albany : State University of New York Press . Chiricos Ted , Welch Kelly , Gertz Marc . 2004 . “Racial Typification of Crime and Support for Punitive Measures.” Criminology 42 : 358 - 89 . Google Scholar CrossRef Search ADS Chiricos Ted , Hogan Michael , Gertz Marc . 1997 . “Racial Composition of Neighborhood and Fear of Crime.” Criminology 35 : 107 - 31 . Google Scholar CrossRef Search ADS Chiricos Ted , McEntire Ranee , Gertz Marc . 2001 . “ Perceived Racial and Ethnic Composition of Neighborhood and Perceived Risk of Crime.” Social Problems 48 : 322 - 40 . Google Scholar CrossRef Search ADS Chiricos Ted , Eschholz Sarah . 2002 . “The Racial and Ethnic Typification of Crime and the Criminal Typification of Race and Ethnicity in Local Television News.” Journal of Research in Crime and Delinquency 39 : 400 - 20 . Google Scholar CrossRef Search ADS Citrin Jack , Reingold Beth , Green Donald P . 1990 . “ American Identity and the Politics of Change.” The Journal of Politics 52 : 1124 - 54 . Google Scholar CrossRef Search ADS Correll Joshua , Wittenbrink Bernd , Park Bernadette , Judd Charles M. , Keesee Tracie , Sadler Melody S . 2007 . “ Across the Thin Blue Line: Police Officers and Racial Bias in the Decision to Shoot.” Journal of Personality and Social Psychology 92 : 1006 - 23 . Google Scholar CrossRef Search ADS Covington Jeannette , Taylor Ralph B . 1991 . “ Fear of Crime in Urban Residential Neighborhoods: Implications of Between- and Within-Neighborhood Sources for Current Models.” The Sociological Quarterly 32 : 231 - 49 . Google Scholar CrossRef Search ADS Crawford Charles , Chiricos Ted , Kleck Gary . 1998 . “ Race, Racial Threat, and Sentencing of Habitual Offenders.” Criminology 36 : 481 - 512 . Google Scholar CrossRef Search ADS Curry Theodore R. 2010 . “ The Conditional Effects of Victim and Offender Ethnicity and Victim Gender on Sentences for Non-Capital Cases.” Punishment & Society 12 : 438 - 62 . Google Scholar CrossRef Search ADS Dee Thomas S. 2005 . “A Teacher Like Me: Does Race, Ethnicity, or Gender Matter?” The American Economic Review 95 : 158 - 65 . Google Scholar CrossRef Search ADS Devine Patricia G. 1989 . “ Stereotypes and Prejudice: Their Automatic and Controlled Components.” Journal of Personality and Social Psychology 56 : 5 - 18 . Google Scholar CrossRef Search ADS Dixon Travis L. , Linz Daniel . 2000 . “ Overrepresentation and Underrepresentation of African Americans and Latinos as Lawbreakers on Television News.” Journal of Communication 50 : 131 - 54 . Google Scholar CrossRef Search ADS Eberhardt Jennifer L. , Davies Paul G. , Purdie-Vaughns Valerie J. , Johnson Sheri Lynn . 2006 . “ Looking Deathworthy: Perceived Stereotypicality of Black Defendants Predicts Capital-Sentencing Outcomes.” Psychological Science 17 : 383 - 86 . Google Scholar CrossRef Search ADS Eitle David , Taylor John . 2008 . “Are Hispanics the New ‘Threat?’ Minority Group Threat and Fear of Crime in Miami-Dade County.” Social Science Research 37 : 1102 - 15 . Google Scholar CrossRef Search ADS Eitle David , D’Alessio Stewart J. , Stolzenberg Lisa . 2002 . “ Racial Threat and Social Control: A Test of the Political, Economic, and Threat of Black Crime Hypotheses.” Social Forces 81 : 557 - 76 . Google Scholar CrossRef Search ADS Ennis Sharon R. , Ríos-Vargas Merarys , Albert Nora G . 2011 . The Hispanic Population: 2010 . Washington, DC : U.S. Department of Commerce, Economics and Statistics Administration, U.S. Census Bureau . Feldmeyer Ben , Ulmer Jeffery T . 2011 . “ Racial/Ethnic Threat and Federal Sentencing.” Journal of Research in Crime and Delinquency 48 : 238 - 70 . Google Scholar CrossRef Search ADS Franklin Cortney A. , Fearn Noelle E . 2008 . “Gender, Race, and Formal Court Decision-Making Outcomes: Chivalry/Paternalism, Conflict Theory or Gender Conflict?” Journal of Criminal Justice 36 : 279 - 90 . Google Scholar CrossRef Search ADS Gallagher Charles A. 2003 . “ Miscounting Race: Explaining Whites’ Misperceptions of Racial Group Size.” Sociological Perspectives 46 : 381 - 96 . Google Scholar CrossRef Search ADS Garland David. 2001 . The Culture of Control: Crime and Social Order in Contemporary Society . Chicago : University of Chicago Press . Gilliam Franklin D. Jr. , Iyengar Shanto . 2000 . “ Prime Suspects: The Influence of Local Television News on the Viewing Public.” American Journal of Political Science 44 : 560 - 73 . Google Scholar CrossRef Search ADS Goodwin Michele , Duke Naomi N . 2012 . “Health Law: Cognitive Bias in Medical Decision-Making.” Pp. 95 - 112 in Implicit Racial Bias Across the Law , edited by Levinson J. D. , Smith R. J . New York : Cambridge University Press . Google Scholar CrossRef Search ADS Green Donald P. , Strolovitch Dara Z. , Wong Janelle S . 1998 . “ Defended Neighborhoods, Integration, and Racially Motivated Crime.” American Journal of Sociology 104 : 372 - 403 . Google Scholar CrossRef Search ADS Greenberg David F. , West Valerie . 2001 . “ State Prison Populations and Their Growth, 1971-1991.” Criminology 39 : 615 - 53 . Google Scholar CrossRef Search ADS Hagan John , Shedd Carla , Payne Monique R . 2005 . “Race, Ethnicity, and Youth Perceptions of Criminal Injustice.” American Sociological Review 70 : 381 - 407 . Google Scholar CrossRef Search ADS Hagan John , Albonetti Celesta . 1982 . “ Race, Class, and the Perception of Criminal Injustice in America.” American Journal of Sociology 88 : 329 - 55 . Google Scholar CrossRef Search ADS Harer Miles D. , Steffensmeier Darrell . 1992 . “ The Differing Effects of Economic Inequality on Black and White Rates of Violence.” Social Forces 70 : 1035 - 54 . Google Scholar CrossRef Search ADS Hawkins Darnell F. 1987 . “ Beyond Anomalies: Rethinking the Conflict Perspective on Race and Criminal Punishment.” Social Forces 65 : 719 - 45 . Google Scholar CrossRef Search ADS Hetey Rebecca C. , Eberhardt Jennifer L . 2014 . “ Racial Disparities in Incarceration Increase Acceptance of Punitive Policies.” Psychological Science 25 : 1949 - 54 . Google Scholar CrossRef Search ADS Holmes Malcolm D. 2000 . “ Minority Threat and Police Brutality: Determinants of Civil Rights Criminal Complaints in U.S. Municipalities.” Criminology 38 : 343 - 67 . Google Scholar CrossRef Search ADS Jackson Pamela I. 1989 . Minority Group Threat, Crime, and Policing: Social Context and Social Control . New York : Praeger . Jacobs David , Carmichael Jason T . 2001 . “ The Politics of Punishment Across Time and Space: A Pooled Time-Series Analysis of Imprisonment Rates.” Social Forces 80 : 61 - 89 . Google Scholar CrossRef Search ADS Jacobs David , Carmichael Jason T. 2002 . “The Political Sociology of the Death Penalty: A Pooled Time-Series Analysis.” American Sociological Review 67 : 109 - 31 . Google Scholar CrossRef Search ADS Jacobs David , O’Brien Robert M . 1998 . “ The Determinants of Deadly Force: A Structural Analysis of Police Violence.” American Journal of Sociology 103 : 837 - 62 . Google Scholar CrossRef Search ADS Johnson Brian D. 2006 . “ The Multilevel Context of Criminal Sentencing: Integrating Judge-and County-Level Influences.” Criminology 44 : 259 - 98 . Google Scholar CrossRef Search ADS Johnson Brian D. , Stewart Eric A. , Pickett Justin , Gertz Marc . 2011 . “ Ethnic Threat and Social Control: Examining Public Support for Judicial Use of Ethnicity in Punishment.” Criminology 49 : 401 - 41 . Google Scholar CrossRef Search ADS Kennedy Randall. 1997 . Race, Crime & the Law . New York : Vintage Publishing . Kent Stephanie L. , Jacobs David . 2005 . “ Minority Threat and Police Strength from 1980 to 2000: A Fixed-Effects Analysis of Nonlinear and Interactive Effects in Large U.S . Cities.” Criminology 43 : 731 - 60 . Google Scholar CrossRef Search ADS King Ryan D. , Wheelock Darren . 2007 . “ Group Threat and Social Control: Race, Perceptions of Minorities and the Desire to Punish.” Social Forces 85 : 1255 - 80 . Google Scholar CrossRef Search ADS Kish Leslie. 1965 . Survey Sampling . New York : John Wiley & Sons . LaFree Gary D. 1980 . “ Variables Affecting Guilty Pleas and Convictions in Rape Cases: Toward a Social Theory of Rape Processing.” Social Forces 58 : 833 - 50 . Google Scholar CrossRef Search ADS LaFree Gary D. 1989 . Rape and Criminal Justice: The Social Construction of Sexual Assault . Belmont, CA : Wadsworth . Lane Jodi , Meeker James W . 2003 . “ Fear of Gang Crime: A Look at Three Theoretical Models.” Law & Society Review 37 : 425 - 56 . Google Scholar CrossRef Search ADS Lavrakas Paul J. 1987. Telephone Survey Methods: Sampling, Selection, and Supervision . Newbury Park, CA : Sage Publications . Liska Allen E. 1992 . Social Threat and Social Control . Albany : State University of New York Press . Liska Allen E. , Lawrence Joseph J. , Sanchirico Andrew . 1982 . “Fear of Crime as a Social Fact.” Social Forces 60 : 760 - 70 . Google Scholar CrossRef Search ADS Liska Allen E. , Chamlin Mitchell B . 1984 . “ Social Structure and Crime Control Among Macrosocial Units.” American Journal of Sociology 90 : 383 - 95 . Google Scholar CrossRef Search ADS Liska Allen E. , Chamlin Mitchell B. , Reed Mark D . 1985 . “ Testing the Economic Production and Conflict Models of Crime Control.” Social Forces 64 : 119 - 38 . Google Scholar CrossRef Search ADS McCarty Christopher , House Mark , Harman Jeffrey , Richards Scott . 2006 . “ Effort in Phone Survey Response Rates: The Effects of Vendor and Client-Controlled Factors.” Field Methods 18 : 172 - 88 . Google Scholar CrossRef Search ADS Mears Daniel P. , Stewart Eric A . 2010 . “ Interracial Contact and Fear of Crime.” Journal of Criminal Justice 38 : 34 - 41 . Google Scholar CrossRef Search ADS Myers Martha A. 1979 . “ Offended Parties and Official Reactions: Victims and the Sentencing of Criminal Defendants.” The Sociological Quarterly 20 : 529 - 40 . Google Scholar CrossRef Search ADS Newman Benjamin J. 2013 . “ Acculturating Contexts and Anglo Opposition to Immigration in the United States.” American Journal of Political Science 57 : 374 - 90 . Google Scholar CrossRef Search ADS Newman Benjamin J. , Hartman Todd K. , Taber Charles S . 2012 . “ Foreign Language Exposure, Cultural Threat, and Opposition to Immigration.” Political Psychology 33 : 635 - 57 . Google Scholar CrossRef Search ADS Ousey Graham C. , Kubrin Charis E . 2009 . “Exploring the Connection Between Immigration and Violent Crime Rates in U.S. Cities, 1980–2000.” Social Problems 56 : 447 - 73 . Google Scholar CrossRef Search ADS Parker Karen F. , Stults Brian J. , Rice Stephen K . 2005 . “ Racial Threat, Concentrated Disadvantage, and Social Control: Considering the Macro-Level Sources of Variation in Arrests.” Criminology 43 : 1111 - 34 . Google Scholar CrossRef Search ADS Paternoster Raymond. 1984 . “ Prosecutorial Discretion in Requesting the Death Penalty: A Case of Victim-Based Racial Discrimination.” Law and Society Review 18 : 437 - 78 . Google Scholar CrossRef Search ADS Pew Research Center . 2004 . Polls Face Growing Resistance, but Still Representative: Survey Experiment Shows . Washington, DC : Pew Research Center . Pew Research Center . 2016 . Statistical Portrait of Hispanics in the United States . Washington, DC : Pew Hispanic Center and Pew Research Center . Pickett Justin T. , Chiricos Ted . 2012 . “ Controlling Other People’s Children: Racialized Views of Delinquency and Whites’ Punitive Attitudes Toward Juvenile Offenders.” Criminology 50 : 673 - 710 . Google Scholar CrossRef Search ADS Quillian Lincoln. 1996 . “ Group Threat and Regional Change in Attitudes Toward African-Americans.” American Journal of Sociology 102 : 816 - 60 . Google Scholar CrossRef Search ADS Quillian Lincoln , Pager Devah . 2010 . “ Estimating Risk Stereotype Amplification and the Perceived Risk of Criminal Victimization.” Social Psychology Quarterly 73 : 79 - 104 . Google Scholar CrossRef Search ADS Quinney Richard. 1970 . The Social Reality of Crime . Boston, MA : Little Brown . Rabe-Hesketh Sophia , Skrondal Anders . 2008 . Multilevel and Longitudinal Modeling Using Stata . College Station, TX : Stata Press . Rachlinski Jeffrey J. , Lynn Johnson Sheri , Wistrich Andrew J. , Guthrie Chris . 2009 . “Does Unconscious Racial Bias Affect Trial Judges?” Notre Dame Law Review 84 : 1195 - 1246 . Ramey David M. 2013 . “ Immigrant Revitalization and Neighborhood Violent Crime in Established and New Destination Cities.” Social Forces 92 : 597 - 629 . Google Scholar CrossRef Search ADS Raudenbush Stephen W. , Bryk Anthony S . 2002 . Hierarchical Linear Models . Thousand Oaks, CA : Sage Publications . Russell Kathryn K. 1999 . “‘Driving While Black’: Corollary Phenomena and Collateral Consequences.” Boston College Law Review 40 : 717 - 31 . Simon Jonathan. 2007 . Governing Through Crime: How the War on Crime Transformed American Democracy and Created a Culture of Fear . Oxford, UK : Oxford University Press . Smith Brad W. , Holmes Malcolm D . 2003 . “ Community Accountability, Minority Threat, and Police Brutality: An Examination of Civil Rights Criminal Complaints.” Criminology 41 : 1035 - 63 . Google Scholar CrossRef Search ADS Smith Robert J. , Levinson Justin D . 2012 . “The Impact of Implicit Racial Bias on the Exercise of Prosecutorial Discretion.” Seattle University Law Review 35 : 795 - 826 . Spohn Cassia. 2000 . “Thirty Years of Sentencing Reform: The Quest for a Racially Neutral Sentencing Process.” Criminal Justice 3 : 427 - 501 . Spohn Cassia , Spears Jeffrey . 1996 . “ The Effect of Offender and Victim Characteristics on Sexual Assault Case Processing Decisions.” Justice Quarterly 13 : 649 - 79 . Google Scholar CrossRef Search ADS John Craig St. , Heald‐Moore Tamara . 1996 . “ Racial Prejudice and Fear of Criminal Victimization by Strangers in Public Settings.” Sociological Inquiry 66 : 267 - 84 . Google Scholar CrossRef Search ADS Stewart Eric A. , Martinez Romiro Jr. , Baumer Eric P. , Gertz Marc . 2015 . “ The Social Context of Latino Threat and Punitive Latino Sentiment.” Social Problems 62 : 68 - 92 . Google Scholar CrossRef Search ADS Stults Brian J. , Baumer Eric P . 2007 . “ Racial Context and Police Force Size: Evaluating the Empirical Validity of the Minority Threat Perspective.” American Journal of Sociology 113 : 507 - 46 . Google Scholar CrossRef Search ADS Swigert Victoria Lynn , Farrell Ronald A . 1977 . “ Normal Homicides and the Law.” American Sociological Review 42 : 16 - 32 . Google Scholar CrossRef Search ADS Taylor Marylee C. 1998 . “ How White Attitudes Vary with the Racial Composition of Local Populations: Numbers Count.” American Sociological Review 63 : 512 - 35 . Google Scholar CrossRef Search ADS Taylor Ralph B. , Covington Jeanette . 1993 . “ Community Structural Change and Fear of Crime.” Social Problems 40 : 374 - 97 . Google Scholar CrossRef Search ADS Tonry Michael. 2010 . “The Social, Psychological, and Political Causes of Racial Disparities in the American Criminal Justice System.” Crime & Justice 39 : 273 - 312 . Google Scholar CrossRef Search ADS Tourangeau Roger. 2004 . “ Survey Research and Societal Change.” Annual Review of Psychology 55 : 775 - 801 . Google Scholar CrossRef Search ADS Ulmer Jeffery T. , Johnson Brian D . 2004 . “ Sentencing in Context: A Multilevel Analysis.” Criminology 42 : 137 - 77 . Google Scholar CrossRef Search ADS Unnever James D. 2008 . “ Two Worlds Far Apart: Black-White Differences in Beliefs About Why African American Men are Disproportionately Imprisoned.” Criminology 46 : 511 - 38 . Google Scholar CrossRef Search ADS Unnever James D. , Cullen Francis T . 2007 . “ Reassessing the Racial Divide in Support for Capital Punishment.” Journal of Research in Crime and Delinquency 44 : 124 - 58 . Google Scholar CrossRef Search ADS Wang Xia. 2012 . “ Undocumented Immigrants as Perceived Criminal Threat: A Test of the Minority Threat Perspective.” Criminology 50 : 743 - 76 . Google Scholar CrossRef Search ADS Wang Xia , Mears Daniel P . 2010 . “ A Multilevel Test of Minority Threat Effects on Sentencing.” Journal of Quantitative Criminology 26 : 191 - 215 . Google Scholar CrossRef Search ADS Weisberg Herbert F. , Krosnick Jon A. , Bowen Bruce D . 1989 . An Introduction to Survey Research and Data Analysis . Glenview, IL : Scott Foresman . Weitzer Ronald , Tuch Steven A . 2005 . “ Racially Biased Policing: Determinants of Citizen Perceptions.” Social Forces 83 : 1009 - 30 . Google Scholar CrossRef Search ADS Williams Marian R. , Demuth Stephen , Holcom Jefferson E . 2007 . “Understanding the Influence of Victim Gender in Death Penalty Cases: The Importance of Victim Race, Sex‐Related Victimization, and Jury Decision Making.” Criminology 45 : 865 - 91 . Google Scholar CrossRef Search ADS Wilson William Julius. 2009 . More Than Just Race: Being Black and Poor in the Inner City . New York : W.W. Norton & Company . Wooldredge John , Griffin Timothy , Thistlethwaite Amy , Rauschenberg Fritz . 2011 . “ Victim‐Based Effects on Racially Disparate Sentencing in Ohio.” Journal of Empirical Legal Studies 8 : 85 - 117 . Google Scholar CrossRef Search ADS Xie Min , Lauritsen Janet L . 2012 . “ Racial Context and Crime Reporting: A Test of Black’s Stratification Hypothesis.” Journal of Quantitative Criminology 28 : 265 - 93 . Google Scholar CrossRef Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for the Study of Social Problems. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Social ProblemsOxford University Press

Published: Mar 12, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off