Introduction
Racial tensions in America have persisted since the country's founding (Reference AlexanderAlexander 2012; Reference BellBell 2002; Reference McPherson, Smith-Lovin and CookMcPherson et al. 2001; Reference Peffley and HurwitzPeffley and Hurwitz 2010; Reference PetersonPeterson 2017; Reference Unnever, Bucerius and TonryUnnever 2014). Although many such tensions can be identified, the “lynching era” (Reference Tolnay and BeckTolnay and Beck 1995: 17)—which spanned a roughly 50-year period from around 1880 (the end of Reconstruction) to the 1930s—stands out and has been the subject of an emerging body of scholarship that seeks to document and understand the legacy of lynchings in contemporary America.
Research on lynchings has emphasized the salience of lynchings for exemplifying and supporting a culture of racial animus and hostility toward blacks that exerts a persisting influence on race relations in contemporary America (Reference DeFina and HannonDeFina and Hannon 2011; Reference Durso and JacobsDurso and Jacobs 2013; Reference Jacobs, Carmichael and KentJacobs et al. 2005, Reference Jacobs, Malone and Iles2012; Reference King, Messner and BallerKing et al. 2009; Reference Messner, Baller and ZevenbergenMessner et al. 2005; Reference PorterPorter 2011; Reference Porter, Howell and HempelPorter et al. 2014; Reference SmângsSmângs 2016; Reference Stewart, Mears, Warren, Baumer and ArnioStewart et al. 2018). Many studies in this tradition highlight the salience of racial threat for explaining how the racial animus exemplified by lynchings contributes to modern-day whites' views about and reactions to blacks. In so doing, they parallel the research that employs racial threat theory to understand racial disparities in crime policy, law enforcement, punishment, mass incarceration, and Americans' views about blacks and crime (see, e.g., Reference Chiricos, Welch and GertzChiricos et al. 2004; Reference Goidel, Parent and MannGoidel et al. 2011; Reference Smith and HolmesSmith and Holmes 2003; Reference Stults and BaumerStults and Baumer 2007; Reference Tonry, Petersilia and ReitzTonry 2012; Reference Ulmer, Laskorunsky, Blomberg, Brancale, Beaver and BalesUlmer and Laskorunsky 2016; Reference Wang and MearsWang and Mears 2010).
This paper seeks to contribute to scholarship aimed at understanding the historical legacy of lynchings and contemporary racialized views of crime. To this end, it examines whether lynchings influence modern-day views that whites hold of blacks both as criminals and as criminal threats to whites. The touchstone for this theoretical argument stems from literature on the role of lynchings in expressing and supporting a deep-rooted cultural view of blacks that persists in contemporary society and continues to shape how whites perceive blacks. Drawing on this work and on racial threat theory, we hypothesize that whites who reside in areas where lynchings occurred will be more likely to perceive blacks as criminals and, more specifically, as criminal threats to whites. We draw, too, on scholarship that highlights the salience of social class and political ideology to hypothesize that residency in areas of socioeconomic disadvantage and political conservatism will amplify these effects. In what follows, we discuss prior theory and research that provide the context for these hypotheses. We then discuss the data and methods used to test them, the findings, and their implications for scholarship.
Background
The Legacy of Lynchings in Contemporary America
The Civil War marked a turning point in the history of the United States, one that drew attention to changing status of blacks. Passage of the Thirteenth Amendment, which abolished slavery, raised questions about whites' position in society (Reference Acharya, Blackwell and SenAcharya et al. 2016; Reference AlexanderAlexander 2012; Reference DavisDavis 2006; Reference FeaginFeagin 2013). In response, during the post-War upheaval, whites' animosity toward blacks escalated (Reference BrundageBrundage 1993; Reference Tolnay and BeckTolnay and Beck 1995). Lynchings—which peaked during the 1880s and 1890s and then persisted, albeit at increasingly lower levels, for another 50 years—constituted one of the central and most violent results of this animosity.
Research points to a complex array of forces that led to lynchings. As Reference King, Messner and BallerKing et al. (2009: 294) have emphasized, the literature on lynchings “include psychological and psychoanalytic accounts, arguments about the reliance on popular or vigilante justice to compensate for the perceived ineffectiveness of legal institutions, cultural interpretations highlighting southern notions of chivalry and honor, and ‘social threat’ accounts emphasizing the political and economic competition between blacks and whites” (see, e.g., Reference ClarkeClarke 1998; Reference King, Messner and BallerKing et al. 2009; Reference Wells and DusterWells 1970; Reference Messner, Baumer and RosenfeldMessner et al. 2006). Regardless, one consistent theme that surfaces across these accounts is that lynchings emanated from and reflected extreme racial hatred.
In efforts to understand the aftermath of lynchings, scholars have emphasized the entrenched and open pervasiveness of whites' hostility toward blacks. Murders of blacks frequently occurred in the dark of night (Reference Bailey and TolnayBailey and Tolnay 2015; Reference BeckBeck 2015). But they also not infrequently constituted public events, ones in which no effort was made to obscure participants or their actions. Indeed, lynchings in such instances were communal undertakings that involved children and families and were reported in news accounts (Reference AllenAllen 2000; Reference Bailey and TolnayBailey and Tolnay 2015; Reference ClarkeClarke 1998; Reference GinzburgGinzburg 1988; Reference Litwack and AllenLitwack 2000; Reference PfeiferPfeifer 2004; Reference PoolePoole 2009). An illustration of the overt displays of violence can be seen in the 1908 lynching of an African American, Nelse Patton, who had been accused of killing a white woman in Oxford, Mississippi. The lynching was overseen by an ex-U.S. Senator, William V. Sullivan, who openly boasted about the daytime lynching, which, notably, had occurred in the town square (Reference WilliamsonWilliamson 1993). Similar events unfolded elsewhere. In typical fashion, lynchings in such cases not only were public spectacles but also entailed gruesome violence that conveyed the profound animus and resentment that whites felt toward blacks (Reference GinzburgGinzburg 1988; Reference WhiteWhite 1918).
This sensibility was not isolated. Lynchings were part and parcel of the social and cultural fabric of white society. When lynchings occurred, it reflected the view among whites that it was acceptable to do so. It also reflected a view of blacks as “inherently and permanently inferior” and “less than human” (Reference Litwack and AllenLitwack 2000: 12). It reflected more than that, though. The intensity of violence against lynching victims went well beyond what would be required to put someone to death (Reference Tolnay and BeckTolnay and Beck 1995; Reference WhiteWhite 1918). Among whites, a sense of rage existed that a seemingly less-than-human group, one previously enslaved, could occupy the same social status as them (Reference SmângsSmângs 2016; Reference Tolnay and BeckTolnay and Beck 1995; Reference WilliamsonWilliamson 1993).
Lynchings provided, then, an expression of animus and resentment that pervaded white society and, at the same time, reinforced hostility toward blacks. As Reference MarkovitzMarkovitz (2004: xvii) has argued, “lynchings were never entirely confined to the physical realm and were instead always intended to be seen as a metaphor for race relations,” and this metaphor “has resurfaced, changed, and been deployed over time.” In this view, lynchings provided a reflection of a broader racial tension that governed whites' views of blacks. The penetration of this tension in communities in turn created the foundation for a legacy and culture of racial hatred that could resurface in subsequent decades (Reference CarriganCarrigan 2004; Reference ClarkeClarke 1998; Reference Durso and JacobsDurso and Jacobs 2013; Reference FrazierFrazier 2015; Reference Gabriel and TolnayGabriel and Tolnay 2017). Reference King, Messner and BallerKing et al. (2009: 292) have, for example, argued that the “racial antagonism” embodied by lynchings “tends to be deeply ingrained in culture (i.e., it dies hard),” and thus continue to influence modern racial dynamics.
Lynchings and Contemporary Whites' Views of Black Criminal Threat
One such resurfacing can be seen in get-tough criminal justice policies that have disproportionately affected blacks. Scholars argue that these policies have resulted in part from a contemporary fear of blacks as criminal threats (Reference AlexanderAlexander 2012; Reference Pettit and WesternPettit and Western 2004; Reference Unnever and GabbidonUnnever and Gabbidon 2011). Some studies have directly linked the policies to lynchings. Reference Jacobs, Malone and IlesJacobs et al. (2012: 170), for example, have written that “the intense anti-Black fervor exemplified by [lynchings] may persist in sufficient strength to influence contemporary legal punishments.” Similarly, Reference King, Messner and BallerKing et al. (2009: 293) have argued that the “symbolic and cultural framework surrounding lynching” can help to explain current sanctioning policies and practices. Support for this argument can be found in an emerging and diverse body of research that has found a positive association between past lynchings and contemporary capital punishment, prison admissions, noncompliance with federal hate crime law, violence toward blacks, and white supremacist group activity (Reference DeFina and HannonDeFina and Hannon 2011; Reference Durso and JacobsDurso and Jacobs 2013; Reference Jacobs, Carmichael and KentJacobs et al. 2005, Reference Jacobs, Malone and Iles2012; Reference King, Messner and BallerKing et al. 2009; Reference Messner, Baumer and RosenfeldMessner et al. 2006; Reference Petersen and WardPetersen and Ward 2015; Reference PorterPorter 2011; Reference Porter, Howell and HempelPorter et al. 2014).
In seeking to explain how lynchings have influenced contemporary America, researchers have drawn on racial threat theory, which argues that majority groups use informal or formal social control against putative threat groups to control them (Reference BlalockBlalock 1967; Reference BlaunerBlauner 1972; Reference LiskaLiska 1992). Racial threat theory has been widely used to explain variation in arrests, convictions, the incarceration of blacks (Reference GarlandGarland 2001, Reference Garland2013; Reference JacksonJackson 1989; Reference MyersMyers 1990; Reference Pettit and WesternPettit and Western 2004; Reference SampsonSampson 2009, Reference Sampson2012; Reference Smith and HolmesSmith and Holmes 2003; Reference Stults and BaumerStults and Baumer 2007; Reference Ulmer, Laskorunsky, Blomberg, Brancale, Beaver and BalesUlmer and Laskorunsky 2016; Reference Wang and MearsWang and Mears 2010). It has been used as well to explain majority members' prejudice toward minority groups who are perceived to be a threat (Reference Behrens, Uggen and ManzaBehrens et al. 2003; Reference QuillianQuillian 1996). A central departure for racial threat scholarship has been the observation that whites equate criminal threat with blacks. That is, the assumption is that blacks are assumed by whites to be criminal and, not least, to present a direct criminal threat to whites (Reference Chiricos, Welch and GertzChiricos et al. 2004; Reference Gilliam, Valentino and BeckmannEitle et al. 2002; Gilliam Jr. et al. 2002; Reference Mancini, Mears, Stewart, Beaver and PickettKeen and Jacobs 2009; Mancini et al. 2015; Reference Mears, Pickett, Golden, Chiricos and GertzMears et al. 2013; Reference Peffley and HurwitzPeffley and Hurwitz 2010; Reference Pickett, Chiricos, Golden and GertzPickett et al. 2012; Reference Unnever, Bucerius and TonryUnnever 2014; Reference Wang and MearsWang and Mears 2010).
Building on this theoretical tradition, a number of studies have explored how lynchings—and the culturally embedded racial animus that they represent—contribute to hostile and punitive actions toward blacks in contemporary America, such as limited hate crime enforcement (Reference King, Messner and BallerKing et al. 2009), increased participation in white supremacist groups (Reference Durso and JacobsDurso and Jacobs 2013), and greater support for capital punishment (Reference Messner, Baumer and RosenfeldMessner et al. 2006) and for punitive sanctions of black offenders in particular (Reference Stewart, Mears, Warren, Baumer and ArnioStewart et al. 2018). These studies suggest that the legacy of lynching can be seen in violence toward blacks and in efforts to impose punishments. Such responses have been interpreted through the lens of racial threat theory. The legacy entails persistent racial animosity that in modern times finds expression not only in anti-black sentiment but also in concern about blacks as presenting a criminal threat (Reference Tolnay and BeckTolnay and Beck 1995: 17). This idea can readily be seen in analyses of support for the death penalty. Whites who reside in areas where lynchings occurred are more likely to support the death penalty (Reference Messner, Baumer and RosenfeldMessner et al. 2006), which logically implies a concern among whites about blacks as sources of crime. The idea can be seen as well in white supremacist group participation; itself an effort to exercise informal social control. For example, Reference Durso and JacobsDurso and Jacobs (2013: 129) have emphasized that “supremacist rhetoric” emphasizes a concern about violent crime committed by blacks.
These lines of work lend support to racial threat arguments about the effects of lynchings.Footnote 1 The consistent theme that emerges is that deeply rooted racial animus has persisted over time. In addition, this animus has shaped modern-day whites' views about blacks and punitive sanctions, especially those that may serve to control or punish blacks. Research has emphasized more than a global perception of racial threat. It suggests that lynchings have contributed to views among whites that blacks present a potent criminal threat to society and, beyond that, a criminal threat specifically to whites (see, e.g., Reference Durso and JacobsDurso and Jacobs 2013; Reference King, Messner and BallerKing et al. 2009; Reference Stewart, Mears, Warren, Baumer and ArnioStewart et al. 2018). Not least, scholarship on lynchings and racial threat underscores the potential for threat mechanisms to be amplified by social context (see, generally, Reference DeFina and HannonDeFina and Hannon 2011; Reference Gabriel and TolnayGabriel and Tolnay 2017; Reference JacksonJackson 1989; Reference Jacobs, Malone and IlesJacobs et al. 2012; Reference Mears, Cochran and LindseyMears et al. 2016; Reference Welch, Payne, Chiricos and GertzWelch et al. 2011).
Study Questions and Hypotheses
In this study, we seek to contribute to efforts to understand the salience of lynchings and the racialization of crime both historically and in contemporary America. To this end, we draw on prior theory and empirical research to develop and test two interrelated hypotheses.
First, perceived black criminal threat—that is, the perception that blacks are criminal and that they are more likely to commit crimes against whites—will be greater among whites who reside in areas that experienced higher numbers of lynchings (H1). This hypothesis derives from scholarship that has identified “lingering effects” of lynchings (Reference MorrisonMorrison 2006; Reference PorterPorter 2011) and highlights the potential for lynchings to influence how whites in contemporary America view blacks as criminal threats (e.g., Reference Jacobs, Malone and IlesJacobs et al. 2012; Reference King, Messner and BallerKing et al. 2009; Reference Messner, Baumer and RosenfeldMessner et al. 2006). This line of research suggests that whites who reside in areas where lynchings, in the past, were more frequent will be more likely to register greater levels of perceived black criminal threat and, in particular, that they will be more likely to view blacks as a criminal threat to whites.
Second, in contemporary America, this effect will be more pronounced among whites who reside in areas marked by greater social and economic disadvantage (H2a) or areas that are politically conservative (H2b). Hypothesis 2a incorporates insights that scholars have made about the potential for the effect of racial threat to be greater in areas where the majority, whites, may be more likely to live on the social or economic margins. They face a greater risk of exclusion from mainstream society or of failing to conform with conventional middle-class views of success and belonging (Reference Byrd, Hall, Roberts and SotoByrd et al. 2015; Reference Durso and JacobsDurso and Jacobs 2013; Reference Fullerton and DixonFullerton and Dixon 2009; Reference ThompsonThompson 2016). As a result, whites in such areas may experience higher levels of frustration and defeatism and have a heightened sense of being in competition with blacks (see, e.g., Reference HochschildHochschild 2016; see, generally, Reference Wells and DusterWells 1970). This social context in turn may contribute to whites' animosity toward blacks and related concerns about blacks as posing a threat to the safety and order of mainstream society. At the same time, the effect of lynchings—and the anti-black animosity that such lynchings represented—may be amplified among whites who reside in disadvantaged areas. Such individuals may be more likely to view blacks as a criminal threat and as a threat that may directly affect them.
The roots of this potentially more acutely felt threat have their foundation in the aftermath of the Civil War. Reference Tolnay and BeckTolnay and Beck (1995: 4) have observed, for example, that southern whites faced greater competition for employment and in this sense “were casualties of war.” For example, from 1900 to 1930, the number of black tenant farmers in the South increased by 27 percent. This increase contributed to higher unemployment and economic disadvantage among rural whites. As a result, mob violence and lynchings were used as an organized method to displace black workers who represented a threat to whites' economic well-being (Reference RaperRaper 2003). Scholars have documented the economic competition that arose as blacks migrated north during the early part of the twentieth century and were perceived as economic competitors to whites, in turn resulting in exclusionary practices against blacks (Reference HorowitzHorowitz 1985; Reference LiebersonLieberson 1981; Reference OlzakOlzak 1992; Reference WilsonWilson 1996). In contemporary America, however, disadvantage may exert effects that go beyond activating perceptions among whites about competition with blacks that in turn fuels resentment or antipathy toward them (Reference Tolnay and BeckTolnay and Beck 1995). It also may fuel defeatism, frustration, and a broader sense, as Reference HochschildHochschild (2016) has described, “of being shoved back in line” while others get ahead (215), as amounting to “strangers in their own land” (219), and as being “culturally marginalized” (221). Social and economic disadvantage thus may provide a context that provides multiple grounds for amplifying whites' racial animus and hostility toward blacks. In so doing, it provides a setting in which the legacy of lynchings may resonate more strongly and thus amplify perceptions of blacks as criminal threats to whites.
Hypothesis 2b extends this line of reasoning and draws on literature that finds that political conservatives may be more likely to endorse punitive views grounded in opposition to or fear of blacks (Reference Chiricos, Welch and GertzChiricos et al. 2004; Reference Fullerton and DixonFullerton and Dixon 2009; Reference Jacobs, Malone and IlesJacobs et al. 2012; Reference KingKing 2008; Reference McCorkleMcCorkle 1993). Specifically, conservatives are more likely to campaign on law-and-order platforms and support punitive policies that disproportionately affect blacks and other minorities (Reference BeckettBeckett 1997; Reference DaveyDavey 1998; Reference Jacobs and HelmsJacobs and Helms 1996; Reference McVeighMcVeigh et al. 2014; Reference StuckyStucky et al. 2005). Indeed, conservative rhetoric about law and order have been used historically and in contemporary times to increase public awareness of crime and punishment and to link street crime to blacks (Reference AlexanderAlexander 2012; Reference BeckettBeckett 1997; Reference Beckett and SassonBeckett and Sasson 2000; Reference Campbell and SchoenfeldCampbell and Schoenfeld 2013; Reference Chiricos, Welch and GertzChiricos et al. 2004; Reference GarlandGarland 2001; Reference HawkinsHawkins 1987; Reference WardWard 2015).
Some scholars have observed that associating crime with race provides an avenue through which Republican presidential candidates and administrations can appeal to working and lower middle-class whites (Reference BeckettBeckett 1997; Reference Jacobs and HelmsJacobs and Helms 1996; Reference McVeighMcVeigh et al. 2014). For example, Reference BeckettBeckett (1997) and Reference Beckett and SassonBeckett and Sasson (2000) have observed that the Reagan and Bush administrations emphasized a political discourse centered on exclusion of marginalized populations. This discourse led to a view of minorities and the poor as potent threats to social order, a phenomenon that has persisted in subsequent elections (Reference HochschildHochschild 2016; Reference RamirezRamirez 2013). Collectively, such work suggests that in politically conservative communities, whites may be more likely to perceive blacks as criminal threats. More relevant to our focus, however, it suggests that political conservatism may provide an ideological context that amplifies the effects of a historical legacy of anti-black sentiment. We thus hypothesize that the effects of lynchings will be greater among whites who reside in politically conservative communities.
Data and Methods
To test the hypotheses, we constructed a multilevel database using individual level survey data and county-level contextual predictors. The individual level data for the study were gathered through a national random telephone survey of 2736 American adults age 18 or older, using random-digit dialing and Computer Assisted Telephone Interviewing to ensure accuracy in recording data. The sampling frame includes households with either landlines or cellular phones; households without either form of telecommunication may be underrepresented. The telephone surveys were conducted throughout 2013. The survey focused on respondents' attitudes about crime and punishment and collected information on whites' views about black criminals; it thus offers a unique opportunity to examine our research questions.
A two-stage modified Mitofsky-Waksberg sampling design was used to develop the random-digit dialing sample (Reference TourangeauTourangeau 2004). Respondents were limited to one adult resident per household. From each household sampled, the adult respondent with the most recent birthday was selected (Reference KishKish 1965). Trained interviewers conducted the telephone interviews and were closely monitored by supervisors. In addition, 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)—that is, completed calls divided by the sum of completed calls, terminated calls, and refusals—we obtained a 61 percent response rate among all contacts with eligible respondents. 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. The response rate is comparable to studies that use rigorous survey methodologies (Reference McCarty, House, Harman and RichardsMcCarty et al. 2006; Pew Research Center 2004), as well as those in similar studies (e.g., Reference King and WheelockKing and Wheelock 2007). Overall, 94 percent of all surveys initiated were completed; this rate is substantially higher than the 60 percent average for national telephone interviews (see, generally, Reference Weisberg, Krosnick and BowenWeisberg et al. 1989).
We restricted the analyses to the 2408 non-Latino white respondents in the sample; these respondents were distributed across 168 counties. The racial and ethnic background of the original sample was as follows: 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 this case whites, views blacks, we restricted the sample to non-Latino white respondents. To account for the potential influence of contextual characteristics, we geocoded respondents to the 168 counties where they resided and appended to the individual level records county-level data from the U.S. Census Bureau. We employed modern county boundaries to link lynching counts to the identifying codes used by the Census Bureau.
Two sets of data were analyzed. First, the southern sample comes from Reference Beck and TolnayBeck and Tolnay's (2015) southern inventory. It includes lynching data for 11 southern states from 1877 through 1950: Alabama, Arkansas, Florida, Georgia, Kentucky, Louisiana, Mississippi, North Carolina, South Carolina, Tennessee, and Virginia (see also Reference Tolnay, Deane and BeckTolnay et al. 1996). The analyses rely on these data both because lynchings occurred primarily in the South (Reference BrundageBrundage 1993) and because of the well-established validity of the Reference Beck and TolnayBeck and Tolnay (2015) inventory. In total, for the southern sample, there are 1301 respondents across 79 counties in these 11 states.
Second, in addition to analysis of the southern sample, we replicated the analyses using a national sample to assess the robustness of the results. The national sample consists of 2408 respondents who resided in 168 counties across 38 states. In creating the national data file, we augmented the Reference Beck and TolnayBeck and Tolnay (2015) inventory using data from several scholars and organizations. One source comes from Reference PfeiferPfeifer's (2004, Reference Pfeifer2013) lynching inventories. These spanned the years from 1870 to 1947 and provided lynching data for counties in the following states: Illinois, Indiana, Iowa, Michigan, Minnesota, Nebraska, Nevada, New Jersey, New York, Ohio, Oregon, Pennsylvania, Utah, Washington, and Wisconsin. Another source is Reference FrazierFrazier's (2009, Reference Frazier2015) lynching inventories, which covered the years 1870–1932 for Kansas counties and 1870–1950 for Missouri counties. For counties in Maryland, we drew on the lynching inventory, for 1870–1933, from the Maryland State Archives (2017). Reference LeonardLeonard (2002) provided lynching inventory data for Colorado counties for 1870–1919. Reference CarriganCarrigan (2004) provided lynching inventory data for Texas counties for 1870–1916. Reference KiktodeKiktode (2008) provided lynching inventory data for Oklahoma counties for 1870–1930. The National Association for the Advancement of Colored People's (NAACP 1919) lynching inventory, 1889–1918, was used to obtain lynching data for counties in Connecticut, Massachusetts, New Mexico, and Rhode Island.
Time spans covered by the different data sources varied. Accordingly, we computed annualized lynching counts by dividing total lynching counts for by the number of years covered by each data source. To illustrate, the Reference Beck and TolnayBeck and Tolnay (2015) lynching totals were divided by 73 (i.e., the number of years encompassed between 1877 and 1950) and the Reference PfeiferPfeifer (2004, Reference Pfeifer2013) lynching totals were divided by 77 (1870–1947). This procedure was followed for each lynching inventory (see also Reference King, Messner and BallerKing et al. 2009: 297). This approach allows for investigation of the peak era of lynchings (Reference Tolnay and BeckTolnay and Beck 1995: 30) and whether any effect of lynching observed in the southern sample can be replicated in the national sample of counties. More generally, by merging the lynching and survey data, we are able to examine whether racial animus over 100 years ago, as reflected in lynchings, affects contemporary whites' attitudes of black criminal threat. We can investigate, too, whether this relationship is moderated by contemporary county-level concentrated disadvantage and political context, respectively.
Below we describe each of the variables used in the study. Descriptive statistics for all variables are provided in Table 1.
Table 1. Descriptive Statistics for Study Variables
N 1 = 1301 individuals; N 2 = 79 counties.
a Divided by 1000.
Dependent Variables
A unique contribution of this study lies in the focus on testing whether lynchings are associated with perceived black criminal threat. Prior lynching scholarship typically has inferred that contemporary support for punitive actions toward or sanctions of blacks through racial threat processes. Such work typically assumes that these processes involve concerns that whites have about blacks as posing a criminal threat in general or as a criminal threat to whites in particular. These studies, however, have not directly examined whites views of black criminality or, by extension, whether whites view blacks as posing a direct criminal threat to them. To address this research gap, we examine two dependent variables that accord with those used in prior studies that assess racial and ethnic threat processes (see, e.g., Reference Barkan and CohnBarkan and Cohn 1994, Reference Barkan and Cohn2005; Reference King and WheelockKing and Wheelock 2007; Reference Stults and BaumerStults and Baumer 2007; Reference Johnson, Stewart, Pickett and GertzJohnson et al. 2011): Perceived black criminal threat and perceived black-on-white crime.
Black criminal threat is measured using responses to six statements: blacks “pose a greater threat to public order and safety than other racial or ethnic groups; hurt the U.S. by committing more violent crimes than other racial or ethnic groups; hurt the U.S. by committing more property crimes than other racial or ethnic groups; hurt the U.S. by committing more drug crimes than other racial or ethnic groups; commit most of the crime in the United States; commit crimes that are more harmful to society when compared to other racial or ethnic groups.” Response categories were: 0 = strongly disagree, 1 = disagree, 2 = agree, and 3 = strongly agree. The resulting measure ranges from 0 (low) through 18 (high) and the alpha coefficient for it was .86.
Black-on-white crime is measured using five statements: blacks “commit most of the violent crimes against law-abiding Whites than any other racial or ethnic groups; commit most of the property crimes against law-abiding Whites than any other racial or ethnic groups; target law-abiding Whites for victimization more than any other racial or ethnic groups; pose a greater criminal threat to law-abiding Whites than any other racial or ethnic groups; commit most of the crimes in the U.S. against law-abiding Whites than any other racial or ethnic groups.” Response categories ranged from 0 = strongly disagree to 3 = strongly agree.
This measure had an alpha coefficient of .83 and values that ranged from 0 (low) to 15 (high). To assess whether the black criminal threat and black-on-white crime measures were empirically distinct from each other, we estimated both an exploratory factor analysis and confirmatory factor analysis. The analyses showed that perceived black criminal threat and perceived black-on-white crime were moderately correlated (r = .58) and thus cannot be collapsed into a single factor without significant loss of information. In short, the two reflect distinct constructs that are empirically distinguishable.
Independent Variable
Our main theoretical variable is black lynchings, as a measure of racial animus and hostility among whites in the late 1800s and early 1900s. It is the number of lynching events with at least one black victim for the years covered in the data sources, and is a measure used in similar studies of lynching (e.g., Reference King, Messner and BallerKing et al. 2009; Reference Messner, Baller and ZevenbergenMessner et al. 2005; Reference Tolnay and BeckTolnay and Beck 1995). We estimated the number rather than rate of lynchings because the symbolic meaning of lynching is unlikely to be a function of proportionate representation. Lynchings occurred rarely; in many cases, though certainly not all, when they did occur, they were publicized and involved entire communities (Reference King, Messner and BallerKing et al. 2009; Reference Litwack and AllenLitwack 2000; Reference Tolnay, Deane and BeckTolnay et al. 1996; Reference WilliamsonWilliamson 1993). The period covered by the sources represents the height of lynching activity in America (Reference Tolnay and BeckTolnay and Beck 1995: 30). Black lynchings for the main analyses, which rely on the Reference Tolnay and BeckTolnay and Beck (1995) southern inventory, range from 0 (low) to 36 (high).
We also examine two moderating variables that we theorize will condition the effect of lynchings. The first, concentrated disadvantage, is measured by three county-level 2010 decennial Census indicators: the percentage of persons on public assistance, the percentage of households below the poverty level, and the percentage of persons unemployed. Other studies have used such variables to examine social and economic disadvantage (see, e.g., Reference Sampson, Raudenbush and EarlsSampson et al. 1997; Reference Parker and ReckdenwaldParker and Reckdenwald 2008). All three items were standardized and combined to form a measure of disadvantage. We then added a constant (12) to eliminate negative values; this measure ranges from 0 (low) to 11 (high) and had an alpha coefficient of .87. The second moderator is political conservatism, measured here as the percent voting republican. Specifically, it is the proportion of county residents voting for the Republican Presidential candidate, Mitt Romney, in 2012. This measure ranges from 0 (low) to 100 (high).
Control Variables
We controlled for a wide range of individual and county-level factors that scholars have linked to punitive attitudes (see, e.g., Reference Barkan and CohnBarkan and Cohn 1994, Reference Barkan and Cohn2005; Reference Baumer, Messner and RosenfeldBaumer et al. 2003; Reference King and WheelockKing and Wheelock 2007; Reference Smith and HolmesSmith and Holmes 2003; Reference Stults and BaumerStults and Baumer 2007). At the individual level, we controlled for the following factors: age, gender, marital status, education level, family income, employment status, political conservatism, homeownership, and fear of blacks. At the county level, controls included the following: homicide rate, percent black, black growth, population structure, and percent urban. We also included a spatial lag. It is derived from a weighting technique that assumes that black lynchings in spatially proximate counties may influence the black lynchings in a given county. This lag gives less weight to geographically distant counties and more to those nearby to control for the influence of lynchings in neighboring counties. The Appendix provides full details about the definitions and metrics of these variables.
Analytic Strategy
We estimated multilevel linear models, with individuals nested within counties, to examine how lynchings, concentrated disadvantage, and political context are associated with whites' perceptions of blacks as criminal threats. Multilevel models are useful for addressing nonindependence of observations within higher order groupings. In our analysis, the 1301 respondents are nested within the 79 southern counties, with an average of 16 respondents per county. Ignoring this clustering may yield standard errors that are deflated, thus exaggerating tests of statistical significance. Specifically, residual errors are likely to be correlated within counties in nested data. That violates the assumption of independence of observations in regression analysis, which tends to bias standard errors downward and increase the probability of type I (false positive) errors (Reference Raudenbush and BrykRaudenbush and Bryk 2002).
Multilevel modeling accounts for this nonindependence and produces correct estimates of standard errors (Reference Raudenbush and BrykRaudenbush and Bryk 2002). This technique also is useful because it allows us to estimate a model that isolates the independent effects of both individual and county-level variables. All models are estimated using the multilevel function in the Stata 14 program (Reference Rabe-Hesketh and SkrondalRabe-Hesketh and Skrondal 2008).Footnote 2
The analyses proceed through the following steps. First, we regressed perceived black criminal threat among whites on black lynchings, as well as on individual and county-level control variables. Second, we build on this multilevel model by including concentrated disadvantage and percent voting Republican to examine their influence on whites' perceptions of blacks as criminal threats. Third, we test whether the influence of black lynchings on perceived black criminal threat is amplified by concentrated disadvantage and percent voting Republican. Finally, we repeat these analyses with our second dependent variable, black-on-white crime. We do so to assess whether whites differentiate in their views about black threat and whether, in particular, the legacy of lynchings increases whites' views of blacks not only as criminals but also as posing, in particular, a criminal threat to whites.
Results
We begin first with a test of the first hypothesis. In Table 2, model 1 provides an assessment of whether, net of the control variables, black lynchings influence contemporary whites' views of blacks as criminal threats. Before discussing the estimated effect of lynchings, the associations between the control variables and perceived criminal threat warrant discussion. As can be seen in the model, black criminal threat is higher among the non-married, political conservatives, and those who are more fearful of blacks in general. In addition, respondents in counties with greater growth in the black population and in counties with a lower population structure or density reported higher levels of perceived black criminal threat. The black percent and growth measures were not significant, which contrasts with some studies that have examined views of minorities as criminal threats (see, e.g., Reference King and WheelockKing and Wheelock 2007; Reference Unnever and CullenUnnever and Cullen 2007; Reference WangWang 2012).
We now turn to our substantive focus. Inspection of model 1 shows that the black lynchings variable is statistically significant and is positively associated with perceived black criminal threat. This finding suggests that, among whites in contemporary America, levels of perceived black criminal threat are higher in counties where historical patterns of black lynching activities were more frequent. To illustrate the magnitude of this relationship, we predict levels of black criminal threat that corresponded to varying levels of lynching activity, with all covariates set at their means. The predicted value of criminal threat for whites who resided in a county that experienced no lynching is 7.4. By contrast, this level rises to 10.4 for those residing in a county with 10 black lynching victims. Thus, the legacy of a particularly high prevalence of lynching activity increases perceived black criminal threat by 3 points, a 40 percent increase.Footnote 3
In the subsequent model, we include concentrated disadvantage and percent voting Republican. The results displayed in model 2 showed that both variables also are statistically significant and positively associated with criminal threat. Thus, perceptions of black threat are greater in less prosperous counties.Footnote 4 Similarly, the statistically significant coefficient for percent voting Republican indicates that whites who resided in counties with more votes for Mitt Romney, the Republican presidential candidate in the 2012 election, than for Barack Obama are more likely to view blacks as criminally threatening.
Model 2 establishes that concentrated disadvantage and percent Republican are associated with whites' views of black criminal threat and provides the foundation for our second hypothesis—examining whether these variables amplify the effects of black lynchings. Model 3 examines the interaction of lynchings and concentrated disadvantage, and model 4 examines the interaction of lynchings and percent Republican. Both models lend support to the hypothesis, as can be seen in the statistically significant interaction terms (b = .070 in model 3 and b = .008 in model 4). Specifically, the effect of lynchings on perceived black criminal threat is stronger in disadvantaged counties and in counties where the percent voting Republican is greater.
We show the predicted values for the two statistically significant interactions in Figures 1 and 2, respectively. The figures show the influence of lynching on perceived black criminal threat across levels of disadvantage and, separately, levels of percent voting Republican. In the figures, “low” concentrated disadvantage was defined as 2 s.d. below the mean and “high” concentrated disadvantage was defined as 2 s.d. above the mean. Similarly, “low” percent voting Republican was defined as 2 s.d. below the mean and “high” percent voting Republican was defined as 2 s.d. above the mean.
Figure 1. The Effect of Black Lynchings on Perceived Black Criminal Threat by Level of Concentrated Disadvantage. [Color figure can be viewed at wileyonlinelibrary.com]
Figure 2. The Effect of Black Lynchings on Perceived Black Criminal Threat by Percentage of County Residents Voting Republican. [Color figure can be viewed at wileyonlinelibrary.com]
As inspection of Figure 1 highlights, among whites who reside in highly disadvantaged areas, the effects of living in areas that experienced more lynchings is greater. For these individuals, residency in communities that had a greater number of lynchings substantially increases their perceptions of blacks as a criminal threat, and does so considerably more than it does among whites in areas of greater socioeconomic advantage. A similar pattern surfaces among whites residing in Republican strongholds versus those who do not, as can be seen in Figure 2.
To this point, our focus has centered on whites' perceptions of blacks as a criminal threat. This focus is, however, diffused and does not address the possibility of a more targeted effect, such that whites perceive blacks as a criminal threat to whites in particular. A targeted effect is anticipated by scholarship on lynchings, which highlight the animosity that whites felt toward blacks and that was expressed through violence against blacks (Reference Tolnay and BeckTolnay and Beck 1995). It is anticipated, too, by scholarship on racial threat, which anticipates that whites may be more likely to view blacks as criminals and as likely to harm them (Reference Mears, Pickett, Golden, Chiricos and GertzMears et al. 2013). In short, the question arises: Does residing in areas where racial animus and hostility were greater increase the probability that contemporary whites who reside in these areas will view blacks as likely to commit crimes specifically against whites? Table 3 answers this question by revisiting the models above and focusing on our second outcome—perceptions among whites that blacks are more likely to commit crimes against whites.
Table 2. Multilevel Models of Perceived Black Criminal Threat Regressed on Black Lynchings, Southern Sample
*p ≤ .05; +p ≤ .10.
N 1 = 1301 individuals; N 2 = 79 counties.
As can be seen in Table 3, a similar pattern of results surfaces to that in Table 2. In model 1, as expected, the results indicate that lynchings have a statistically significant and positive effect on whites perceptions of black-on-white crime. This finding indicates that whites who resided in counties with a historical legacy of racial violence express a heightened belief that blacks pose a criminal threat specifically against them. To illustrate, the predicted value of black-on-white crime for whites who resided in a county that experienced no lynchings is 6. This level rises 2.8 points, a 47 percent increase, for those residing in a county with 10 black lynching victims. The legacy of lynching thus not only increases whites' perceptions of blacks as criminals but it also increases their perceptions that blacks pose a direct criminal threat to them.
Table 3. Multilevel Models of Perceived Black-On-White Crime Regressed on Black Lynchings, Southern Sample
*p ≤ .05; +p ≤ .10.
N 1 = 1301 individuals; N 2 = 79 counties.
Here, again, consistent with the above analyses, concentrated disadvantage and percent voting Republican were statistically significant predictors of black criminal threat and, in particular, perceived threat of black-on-white crime (model 2). Of particular relevance for our second hypothesis, models 3 and 4 indicate that disadvantage and percent voting Republican amplify the threat-increasing effect of residing in areas that had more lynchings. To facilitate interpretation of the results, we again plot the predicted values for the two interactions in Figures 3 and 4. As shown in Figure 3, residing in areas that had higher levels of lynchings increases the perceptions among contemporary whites that blacks pose a criminal threat to whites; this effect is more pronounced among whites who live in highly disadvantaged communities. Similarly, the effects of residing in areas where the legacy of lynchings was greatest are more pronounced when whites reside in areas that are more likely to be Republican strongholds. These results are not, it should be noted, sensitive to state-level differences.
Figure 3. The Effect of Black Lynchings on Perceived Black-On-White Crime by Level of Concentrated Disadvantage. [Color figure can be viewed at wileyonlinelibrary.com]
Figure 4. The Effect of Black Lynchings on Perceived Black-On-White Crime by Percentage of County Residents Voting Republican. [Color figure can be viewed at wileyonlinelibrary.com]
Our primary focus in the analyses above has centered on the South because of the greater concentration of lynching activity in the South and the higher quality of the Reference Beck and TolnayBeck and Tolnay (2015) data. However, to test if the findings hold beyond the South, we conducted a parallel set of analyses that examined lynching effects nationally. For these analyses, we relied on a data set that merged the Reference Beck and TolnayBeck and Tolnay (2015) inventory with lynching data from the other sources discussed in the previous section. The results are presented in Table 4.
Table 4. Multilevel Models of Perceived Black Criminal Threat and Perceived Black-On-White Crime Regressed on Black Lynchings, National Sample
Models 1–3 focus on whites' perceptions of blacks as a criminal threat, while models 4–6 focus on whites' perception that blacks pose a direct criminal threat to whites. Models 1 and 4, respectively, indicate that residing in high lynching areas is associated with a greater perception among whites in contemporary America that blacks are both a criminal threat in general and a criminal threat specifically to whites. Models 2 and 5, respectively, indicate that this effect is amplified by residing in areas of greater socioeconomic disadvantage. Models 4 and 6, respectively, indicate that the effect is amplified, too, by residing in areas that have a stronger Republican base. In short, the results suggest that the effects of historical lynchings, including the amplifying effects of contemporary concentrated disadvantage and political context, are important factors in explaining perceptions of black threat nationally.Footnote 5
We turn, finally, to an analysis that examined whether the results differed for black respondents. The theoretical logic of the study suggests that lynching effects on perceptions about black criminal threat should arise only for whites, not for blacks. To investigate this prediction, we estimated models consistent with those in Tables 2 and 3 but focused instead on black respondents (n = 200). Black lynchings did not yield a statistically significant effect on black criminal threat or black-on-white crime. The analyses thus suggest that, as anticipated, lynchings may be influential in shaping whites', not blacks', views of black criminal threat.
Conclusion
The lineage of racial tensions in America can be traced to many factors, not least of which is the enslavement of Africans during the country's formation. Throughout the 1800s, blacks were viewed by many whites as less than human and, by extension, as occupying a social position well below that of whites. During the latter half of the nineteenth century, calls to provide blacks greater legal rights challenged this view. In so doing, they created the foundation for racial animus and hostility toward blacks that—through a “cultural and symbolic framework” (Reference King, Messner and BallerKing et al. (2009: 293) built on anti-black sentiment (Reference Jacobs, Malone and IlesJacobs et al. (2012)—scholars have argued has persisted into modern times (Reference Messner, Baumer and RosenfeldMessner et al. 2006; Reference Gabriel and TolnayGabriel and Tolnay 2017; see, generally, Reference AlexanderAlexander 2012; Reference Unnever, Bucerius and TonryUnnever 2014). Arguably, the most vivid expression of hostility came in the form of lynchings in the late 1800s and early 1900s (Reference GinzburgGinzburg 1988; Reference Tolnay and BeckTolnay and Beck 1995).
Building on scholarship on lynchings and racial threat theory, we sought to extend efforts to understand the potentially enduring effects of lynchings and the culture from which they emanated (e.g., Reference DeFina and HannonDeFina and Hannon 2011; Reference Jacobs, Carmichael and KentJacobs et al. 2005, Reference Jacobs, Malone and Iles2012; Reference King, Messner and BallerKing et al. 2009; Reference McVeigh, Cunningham and FarrellMcVeigh and Cunningham 2012; Reference Messner, Baller and ZevenbergenMessner et al. 2005; Reference Tolnay, Deane and BeckTolnay et al. 1996). At the same time, we sought to extend scholarship aimed at understanding why whites in contemporary America may equate blacks with criminality (Reference AlexanderAlexander 2012; Reference Barkan and CohnBarkan and Cohn 2005; Reference Chiricos, Welch and GertzChiricos et al. 2004; Reference Pettit and WesternPettit and Western 2004). To this end, we drew on prior theory and research to hypothesize that perceived racial threat would be greater among whites who resided in areas that had a more entrenched legacy of lynchings. We hypothesized, too, that this effect would be greater among whites who resided in areas of greater social and economic disadvantage or political conservatism. Such areas, we argued, provide conditions within which the cultural legacy and effects of lynchings may be most likely to flourish and thus animate whites' perceptions of blacks as a criminal threat in general and toward whites in particular.
Analyses of contemporary public opinion data and historical lynching data lent support to both hypotheses. Whites who reside in communities with a greater historical legacy of racial animus and hostility, as reflected in and perpetuated by higher levels of lynchings, were more likely to view blacks as a threat. Specifically, whites from such areas were more likely to view blacks as criminal threats and to view blacks as more likely to commit crimes against whites. The perceived threat thus was not just that blacks are more criminal but also that whites are likely to be the victims of black criminal activity. In addition, the analyses indicated that, as hypothesized, these effects were more pronounced among whites residing in areas of greater socioeconomic disadvantage and in more politically conservative communities.
Several implications of these findings warrant emphasis. First, the results lend support and add to a growing body of research that has identified enduring effects of lynchings during the turn of the nineteenth century that have persisted to the present (Reference Jacobs, Malone and IlesJacobs et al. 2012; Reference ZimringZimring 2003). The legacy of lynchings casts, it appears, a long shadow, one relevant for understanding contemporary race relations in America (Reference King, Messner and BallerKing et al. 2009; Reference Messner, Baller and ZevenbergenMessner et al. 2005; Reference SmângsSmângs 2016).
Second, prior work has identified associations between lynchings and capital punishment (Reference Jacobs, Carmichael and KentJacobs et al. 2005), homicides (Reference Messner, Baller and ZevenbergenMessner et al. 2005; Reference Petersen and WardPetersen and Ward 2015), law enforcement responses to hate crimes (Reference King, Messner and BallerKing et al. 2009), prison admissions (Reference Jacobs, Malone and IlesJacobs et al. 2012), white supremacist group activity (Reference Durso and JacobsDurso and Jacobs 2013), and support for more punitive sanctioning of black offenders relative to white offenders (Reference Stewart, Mears, Warren, Baumer and ArnioStewart et al. 2018). Although the present study did not examine whether perceptions of criminal threat contributes to these outcomes, it illuminates that a central theoretical mechanism anticipated by past studies—perceived black criminal threat—is itself associated with lynchings.
Third, the present study and prior work collectively point to the possibility that one legacy of lynchings lies in the “conflation of blackness and crime” (Reference WacquantWacquant 2001: 118) in modern America. This conflation may serve to sustain and reinforce views of blacks as criminal and as likely to victimize whites, and in turn to contribute to racial disparities in the arrest, prosecution, and punishment of blacks (Reference AlexanderAlexander 2012; Reference Chiricos, Welch and GertzChiricos et al. 2004; Reference Mears, Cochran and LindseyMears et al. 2016; Reference Ulmer, Laskorunsky, Blomberg, Brancale, Beaver and BalesUlmer and Laskorunsky 2016; Reference Unnever, Bucerius and TonryUnnever 2014).
Fourth, the results underscore the potential for community context to moderate the effects of lynchings. It underscores, too, the attendant need, as Reference HochschildHochschild (2016) has highlighted, for analyses that identify how a range of social forces may intersect to influence whites' views and, more generally, to influence interracial dynamics (Reference Durso and JacobsDurso and Jacobs 2013; Reference Messner, Baumer and RosenfeldMessner et al. 2006; Reference Porter, Howell and HempelPorter et al. 2014). For example, disadvantaged areas may provide settings, and politically conservative views may provide belief systems that enable stereotypical views of blacks as criminal to flourish. Further investigation of these possibilities is an important, albeit difficult one, to pursue. Reference HochschildHochschild's 2016 work is illustrative—it involved intensive qualitative research aimed at demonstrating the intersection of multiple forces and how they affect American's views. Our analyses focused on relatively simple two-way interactions. In reality, more complicated interactions of three or more forces may contribute to how whites, and citizens more generally, view blacks. Investigating such interactions may go beyond what conventional data sets allow. They are, however, important for unpacking the ways in which lynchings, and the belief systems that may have helped to perpetuate, directly, indirectly, or in interaction with other forces, influence contemporary America.
In a related vein, there is a need for research that isolates the mechanisms that allow for the cultural transmission of views that whites hold about blacks. Reference SmângsSmângs (2016: 1368), for example, has argued that “racial inequalities, whether in the past or present, cannot be understood apart from processes of racial category, boundary, and identify formation.” And Reference Gabriel and TolnayGabriel and Tolnay (2017) have underscored the importance of investigating factors, such as legal rulings and residential heterogeneity, that may attenuate the effects of lynching. Connecting past lynching practices and these processes—including the dynamics and salience of identity formation and the transmission or inhibition of racial animus—to contemporary whites' views of blacks and black criminality constitutes a critical avenue of inquiry for future research (Reference DeFina and HannonDeFina and Hannon 2011; Reference King and WheelockKing and Wheelock 2007; Reference Petersen and WardPetersen and Ward 2015).
Finally, greater attention is needed to ways in which other historical conditions may be relevant in understanding the effects of lynchings, racial threat processes, and contemporary views whites hold toward blacks. Reference Acharya, Blackwell and SenAcharya et al. (2016), for example, have argued for the potential role that slavery, and competition among whites and blacks, may play in influencing whites' racial resentment, political affiliation, and views about affirmative action (see also Reference Wells and DusterWells 1970). Although they found no evidence that slavery effects arose through racial threat processes, the study aligns with studies of lynchings in pointing to the persistent influence of the past. Future research ideally will investigate the ways in which social, economic, and cultural events of the past combine to influence racial dynamics in contemporary America.
There is no time like the present to do so. In addition to the considerable attention scholars have given to the study of lynchings, there is also national interest in understanding and acknowledging the fact of lynchings and their consequences. Perhaps the most prominent example can be seen in the erection in 2018 of the National Memorial for Peace and Justice in Alabama (Reference RobertsonRobertson 2018). Its impacts remain to be seen, but its presence signals the potential for greater efforts to document, understand, and address violence of all kinds.
Appendix: Definitions and Metrics of the Control Variables
