Hate Crimes Hurt More,but so Do Co-Offenders: Separating the Influence of Co-Offending and Bias on Hate-Motivated Physical Injury

Abstract

One of the primary motivations for hate crime laws is that hate crimes “hurt more.” But hate crimes are often committed by groups, and research indicates that crimes committed by groups are also more violent than other crimes. This research focuses on one type of harm, physical injury, asking, are hate crimes more violent because they involve co-offenders or because of the bias motivation behind the incident? Results using data from the National Incident-Based Reporting System (NIBRS) indicate that hate crimes are positively associated with serious injury, but that this association is partially driven by co-offenders. More importantly, co-offending moderates this relationship: Incidents involving bias and co-offending are especially violent. Anti-sexual orientation incidents were an exception to this pattern, however, and are likely to be violent regardless of co-offending. These results suggest that hate crimes do hurt more, but that this relationship is partially attributable to the influence of co-offenders.

Keywords

hate crime bias crime violence co-offending NIBRS

A number of scholars have argued that hate crimes, which are crimes wherein a victim is targeted because of real or perceived group membership (i.e., race, religion, sexual orientation, etc.), are qualitatively different from other crimes, causing extra psychological and physical harm to victims and to victim’s communities (e.g., Iganski & Lagou, 2014; Lawrence, 1994, 2002, 2009; Levin, 1999; Levin & McDevitt, 1993). Indeed, the primary argument for hate crime legislation is that the increased harm associated with hate crime is worthy of increased punishment. A number of studies have supported these arguments regarding physical harm (Fetzer & Pezzella, 2016; Herek, Gillis, & Cogan, 1999; Herek, Gillis, Cogan, & Glunt, 1997; Iganski, 2001; McDevitt, Balboni, Garcia, & Gu, 2001; Perry, 2001). Importantly, however, a number of scholars have also noted that hate crimes are significantly more likely than other crimes to involve the participation of multiple offenders or co-offending (e.g., Dunbar, 1999; Levin, 1999). In addition, recent research has indicated that groups of offenders commit more violent offenses than solo offenders and are significantly more likely than solo offenders to seriously injure victims (Lantz, 2018; McGloin & Piquero, 2009).

Taken together, this research suggests two important research questions

Research Question 1: Are hate crimes, indeed, more violent than nonbias crimes?

Research Question 2: Given that group offenses are more violent than solo offenses, and hate crimes are more likely than other crimes to involve groups, is hate crime violence a result of increased brutality associated with bias, or a result of the presence of multiple offenders? That is, are hate crimes more violent because they involve co-offending groups or are they more violent because of the bias motivation behind the incident?

Literature Review

In 2001, Iganski notably argued that hate crimes “hurt more,” and enhanced punishment for hate-motivated violence has often been justified based on the seriousness of its consequences (Iganski, 2001; McDevitt et al., 2001; Meyer, 2010; Pezzella & Fetzer, 2017). More specifically, hate crimes may cause greater harm in three different ways. First, hate crimes may have a greater psychological and emotional impact on victims than other crimes (Herek et al., 1997). Second, hate crimes may negatively affect the greater community, spreading beyond the individual to those who know the victim or learn of the incident. That is, hate crimes may serve as message crimes, indicating a general threat of more violence to members of the victim’s group or the community (Iganski, 2001; Weinstein, 1992)

Finally, hate-motivated violence may hurt more by exacerbating the physical consequences of victimization (Levin, 1999). Supporting this argument, a number of studies have demonstrated that, along with other serious consequences (e.g., psychological harms, harmful effects to the community), hate crime injuries are qualitatively more severe than other hate crimes (Fetzer & Pezzella, 2016; Iganski, 2001; Levin, 1999; Levin & McDevitt, 1993; Messner, McHugh, & Felson, 2004; Perry, 2012; Strom, 2001; Weisburd & Levin, 1993). Levin (1999), for example, noted that hate crime incidents were twice as likely as other crimes to result in injury to the victim and 4 times more likely than other incidents to require victim hospitalization. Levin and McDevitt (2002) also examined a sample of police hate crime records in Boston and found that 50% of hate crime cases involved a serious physical injury that required hospitalization.

Using National Crime Victimization Survey (NCVS) data, Fetzer and Pezzella (2016) also tested whether hate-motivated violence was associated with an increased likelihood of serious physical injuries. Their results indicated that the odds of victim serious physical injury were 23% greater for bias-motivated assault than nonbias assault, net of controls. Several studies have also found evidence that hate crimes are more likely than other crimes to result in serious victim injury using data from the National Incident-Based Reporting System (NIBRS). Strom (2001) analyzed NIBRS data on aggravated assaults and results showed that 60% of bias crimes resulted in serious injuries. Messner and colleagues (2004) also analyzed assault cases and found that compared to nonbias assaults, bias assaults were nearly three times more likely to result in serious victim injury.

Hate crimes may be associated with increased physical violence for several reasons. One reason that hate crimes may be more violent than other crimes may be that the inherent animus involved in hate crimes increases the brutality of hate-motivated violence, leading to more serious physical injuries than nonbias violence. According to Levin and McDevitt (1993, p. 11), hate-motivated offenders are more violent in their actions than other offenders because of the hateful views that they hold. Similarly, Perry (2001) argued that the purpose of hate crimes was to subordinate and intimidate a victim, suggesting increased brutality. Messner and colleagues (2004) also found support for the assertion that hate crimes can be “excessively brutal,” and point out that hate crime offenders might have stronger violent and antisocial tendencies than conventional offenders because they commit violence without victim provocation. Taken together, research on bias crime commonly indicates that hate crime offenders are more likely to use excessive violence, and that hate crimes may lead to more severe victim injuries (Dunbar, 2003; Levin & McDevitt, 1993; Messner et al., 2004).

It is also important to note, however, that a number of scholars have demonstrated that bias-motivated violence is often committed by multiple offenders (Craig, 2002; Dunbar, 1999; Garofalo, 1991; Levin, 1999; Levin & McDevitt, 1993; Ruback, Gladfelter, & Lantz, 2015). Dunbar (1999), for example, found that nearly two-thirds of hate crimes reported in the NCVS involved multiple perpetrators. Levin (1999) examined data from Maryland and also determined that hate crimes frequently involve multiple offenders: 49% of hate crimes reported involved multiple offenders, while only 25% of all crimes involved more than one offender. Most recently, Pezzella and Fetzer (2017) found that incidents involving two offenders were roughly 25% more likely to be motivated by bias than incidents involving only one offender and incidents involving three or more offenders were roughly 73% more likely to be motivated by bias than incidents involving only one offender.

Levin and McDevitt’s (1993) hate crime typology provides one explanation for high rates of hate crime co-offending. According to the typology, the majority of hate crimes are thrill-seeking crimes that involve peers; that is, they are thrill-seeking activities among young offender groups “who regard hatred as cool” (Levin & McDevitt, 1993, p. 65). During thrill-seeking hate crimes, only one or two members of a group may be motivated by hatred, but they may directly or indirectly influence the behavior of their co-offenders. This conceptualization of most hate crime as thrill-seeking is further supported by Messner et al. (2004), who find that hate crime offenders are often versatile offenders, rather than specialists, and are more likely to use alcohol or drugs in groups during the crime. In these cases, hate crimes can occur as an attempt to prove masculinity to peers, and peers can exacerbate violent behavior (Lantz, 2018; Warr, 2002).

Following this, another potential explanation for the relationship between bias and severity of physical violence may be the high prevalence of hate crime co-offending. That is, research has indicated that group offenses are associated with an increased likelihood of violence compared with solo offenses (Conway & McCord, 2002; Lantz, 2018; McGloin & Piquero, 2009). Shaw and McKay (1931) first noted this pattern in their case study of a young offender when they noted that he escalated to violent offending behavior after affiliating with serious violent criminals. More recently, Carrington (2002) found that groups were more likely than solo offenders to commit more serious crimes: 34% of incidents involving co-offenders involved an injury of some sort, while only 17% of incidents involving co-offenders resulted in no injury.

McGloin and Piquero (2009) similarly found that violent offenses involved a higher average number of offenders per offense than nonviolent offenses and that the number of violent offenses increased as the number of accomplices increased. This research argues that the presence of co-offenders increases violence through a number of mechanisms which allow individuals to engage in behavior that they would never have engaged in had they been alone (see also Warr, 2002). More specifically, co-offenders may facilitate changes in offending behavior because of processes inherent to the group, not because of the individual characteristics (i.e., individual levels of bias) of the members of a co-offending group (McGloin & Piquero 2009; Warr, 2002). Being in the presence of others can, for example, diffuse responsibility and reduce personal blame for behavior.

McGloin and Rowan (2015) recently argued that individuals have different thresholds for engaging in a behavior, like violence, and that the presence of accomplices impacts these thresholds. In the case of violence, their research would suggest that individuals with a high propensity for violence will participate in violence alone, but that those with lower propensities for such behavior require the presence of more co-offenders to induce their participation. Recent research by McGloin and Thomas (2016) further supported the proposition that the presence of other offenders impacts the individual offender’s decision-making process. In a series of vignettes, they found that as offender group size increased respondents reported a decreased anticipated risk of punishment and informal sanctions.

McGloin and Piquero (2009) point out that these mechanisms are also not necessarily specific to violent behavior and that the potential risks associated with different crime types vary (Cornish & Clarke, 1986). That said, these potential risks are more severe when the criminal behavior is violent, as is the case with violent hate crime. As such, given that there are relatively high rates of co-offending for bias-motivated crimes compared with nonbias crimes, co-offending is one potential explanation for a relationship between bias and offense severity. In the context of a group hate crime, many of which may be thrill-seeking in nature, violence may be escalated not because of high levels of the bias involved but because of group influence. In other words, individuals may engage in violence for a number of alternative explanations related to the presence of co-offenders, such as a diffusion of responsibility (McGloin & Piquero, 2009) or fear of peer ridicule should they refuse to participate (Warr, 2002).

Taken together, prior research suggests a potentially complex relationship between bias crime, co-offending, and physical violence. Some research has indicated the bias crimes are more likely than other crimes to be physically violent or brutal in nature (e.g., Levin & McDevitt, 1993; Messner et al., 2004; Perry, 2001). But research on co-offending has demonstrated a similar relationship between groups and violence (e.g., Lantz, 2018; McGloin & Piquero, 2009), and a number of scholars have noted that hate crimes have higher rates of co-offending than other crimes (Craig, 2002; Dunbar, 1999; Levin, 1999; Ruback et al., 2015). Following this, this research suggests the need to disentangle this relationship.

This co-offending explanation could also potentially explain the small number of studies that have indicated that hate-motivated violence is less likely than other types of violence to result in serious physical injuries or that have yielded mixed results (Corcoran, Lader, & Smith, 2015; Garcia, McDevitt, Gu, & Balboni, 1999; Iganski & Lagou, 2014; Martin, 1996; Pezzella & Fetzer, 2017). The majority of these studies were descriptive comparisons of the percentage of physical injuries between hate-motivated assaults and nonbiased assaults and did not account for the presence of multiple offenders in hate crime offenses (Corcoran et al., 2015; Garcia et al., 1999; Iganski & Lagou, 2014; Martin, 1996). One study conducted by Pezzella and Fetzer (2017), however, did examine the relationship between bias and injury in a multivariate framework, while also accounting for co-offending. They found no relationship between most bias types and injury. Importantly, however, they control for the number of offenders in their analyses; such an approach would result in a null estimate if the relationship between bias and severity of violence was attributable to the influence of co-offending.

Current Study

The current study examines the relationship between hate crimes and severity of physical violence. Based on prior research, we suggest three possible explanations for the relationship. First, hate crimes are more violent than other crimes because the bias behind the offense increases the brutality of the violence (e.g., Levin & McDevitt, 1993). In this case, hate crimes should be more likely to involve serious victim injury regardless of the number of co-offenders involved in the offense. Second, hate crimes are more violent because they involve higher rates of co-offending and, in such offenses, individuals are more likely to act violently regardless of bias toward the victim (e.g., Lantz, 2018; McGloin & Piquero, 2009). In such a case, co-offending would act as a mediating mechanism in the relationship between bias motivation and victim injury. In peer groups, for example, the presence of a single bias-motivated offender can escalate to hate-motivated co-offending violence as a result of the influence of group processes, as bias offenders receive social support and encouragement from peers, who then act as co-offenders (Messner et al., 2004). Group dynamics promote conformity (Warr, 2002), and Hamm (1993) characterized hate crime offenders as “hyperconformists.” McDevitt et al. (2002) similarly noted that some individuals, referred to as “fellow travelers,” may also follow the lead and conform to their hate crime co-offenders. As such, group influence may play a particularly important role in hate crime offending (Franklin, 2000), increasing the level of physical violence toward the victim(s). Finally, a third explanation for the relationship is that both bias motivation and co-offending are important factors in the production of hate-motivated violence. In this case, both bias motivation and co-offending would be significantly and independently related to victim injury, but co-offending would also have a moderating effect on the relationship such that injury would be most likely in cases that involve both bias motivation and co-offending. Taken together, these three possibilities make clear the need for mediation and moderation analyses of the relationship.

Following this, the current study examines the joint relationship between bias motivation, co-offending, and severity of physical violence in three steps. First, we examine differences between bias-motivated incidents and nonbias-motivated incidents at the bivariate level. Second, the relationship between bias, co-offending, and serious victim injury is examined in a multivariate framework. This potential relationship is presented in Figure 1.

Figure 1:

Theoretical Relationship Between Bias Motivation, Co-Offending, and Victim Injury

In the current study, we conduct formal mediation analyses (Baron & Kenny, 1986) wherein we examine (a) the relationship between bias and serious injury (path c); (b) the relationship between bias motivation and co-offending (path a); and (c) the relationship between bias, co-offending, and serious injury. In this way, we test the hypothesis that the relationship between bias and injury is mediated by the influence of co-offending.¹ We also argue that bias motivation and co-offending may both matter in the production of hate crime violence. That is, it is also possible that co-offending moderates the relationship between bias motivation and victim injury (path d). If this is the case, bias motivation and co-offending would both be related to the likelihood of serious injury, but incidents involving both bias and co-offending would be more likely than other incidents to be especially violent in their outcome. In the third step of this research, we examine this possibility by introducing an interaction term to disentangle the joint (and separate) influence of bias motivation and co-offending on hate crime violence.

Method

The current research uses data from the NIBRS. For the current study, NIBRS provides incident-level data on violent offending, bias motivation, the number of offenders involved in an offense, and victim injury. Data on violent incidents from the offender, offense, and victim segments are used. Because bias crimes are relatively uncommon, we use 10 years of data (2003-2012), resulting in 14,760 violent bias-motivated assaults, robbery offenses, and sexual assault offenses. Homicide offenses are excluded, given that all homicide offenses involve serious injury. Because the use of all nonbias violent incidents for these years would result in an unnecessarily large sample that would unduly inflate statistical power (over 11,000,000 incidents), we follow Lyons and Roberts (2014) and select a random sample of nonbias incidents equal to 5 times the number of bias incidents (N = 73,785) using random sampling procedures in Stata. As such, the final sample includes a total of 88,545 total violent bias and nonbias incidents.

Measures

The primary dependent measure in the current research is serious victim injury. An incident is coded as serious injury if any victim suffered from broken bones, internal injury, loss of teeth, severe lacerations, unconsciousness, or other major injuries. If the victim(s) were not injured, or only suffered a minor injury, the incident was coded as not involving a serious injury (Felson & Lantz, 2016). Because this outcome is dichotomous, logistic regression is used.

The current study includes two primary independent measures: whether the incident was motivated by bias, and whether the incident involved a group of offenders (i.e., co-offending). Following this, a dichotomous measure was created indicating whether the incident was motivated by bias (1 = yes). For supplementary analyses, the type of bias motivation was also disaggregated into five different dummy measures: anti-racial bias motivation, anti-religious bias motivation, anti-ethnic or national origin bias motivation, anti-sexual orientation bias motivation, and anti-disability bias motivation.² Co-offending is measured using a dichotomous measure of whether there was more than one offender involved in the offense (1 = yes).³ In the final stage of the analysis, an additional interaction measure is created between the dichotomous bias motivation measure and the group offending measure to measure whether the relationship between bias and serious victim injury varies according to whether the incident involved multiple offenders or only a single offender.

Control Measures

The current research includes several control measures at the offender, victim, and offense level. First, the analysis includes control measures for the gender, age, and race of both the offenders and victims. The same coding scheme is used for both offenders and victims. Both hate crimes (McDevitt, Levin, & Bennett, 2002) and co-offenses (Lantz & Ruback, 2017) are more likely than other offenses to involve young people. Age is coded as the mean of all offenders and of all victims involved. Gender is coded into three dummy measures indicating whether the incident involves male, female, or mixed gender co-offending partners. If the incident involved a single offender or victim, the gender of that individual is coded. If multiple individuals were involved, the incident is coded as male when all offenders were male and coded as female when all offenders were female. Otherwise, the offenders, or victims, are coded as mixed gender. A similar coding scheme was used for the coding of race, which was coded into four measures: White, Black, other race, and mixed race. If the incident involved only a single offender, or a single victim, the person’s race is coded. If multiple offenders (or victims) were involved, the incident is coded as White when all offenders were White, Black when all offenders were Black, and so on. If the offenders or victims were of different races, then the incident is coded as mixed race. The prior relationship between the victim and offender is coded into a dichotomous measure indicating whether any of the offenders knew the victim in any way. If not, the incident is coded as involving strangers (1 = yes).

At the offense level, because alcohol and drug use are associated with increased levels of violent behavior (Laufer, Harel, & Molcho, 2006) and hate crime offending (Messner et al., 2004), the current analysis uses a dummy measure for alcohol use (1 = yes) and drug use (1 = yes) during any offense. In addition, to ensure that any group estimate is not the result of gang membership, a dummy measure indicating whether any of the offenses were known to involve gang activity (1 = yes) is also included. Finally, because weapon use is likely significantly associated with the likelihood of victim injury, the current research also controls for incident weapon use. Weapon use is coded as a dichotomous measure indicating whether any weapon, including an automatic weapon, handgun, rifle, shotgun, other firearm, knife or cutting instrument, blunt object, motor vehicle, poison, explosive, and incendiary device, was used in the course of an incident. Other incidents were coded as no weapon. Finally, offenses were coded into three different dichotomous measures indicating the nature of the violent offense type: assault, robbery, or sexual assault.

Results

The current study proceeds in three steps. First, we examine differences between bias-motivated incidents and nonbias-motivated incidents at the bivariate level. Second, the relationship between bias, co-offending, and serious victim injury is examined in a multivariate framework and mediation analyses are conducted. Finally, we introduce an interaction term between bias and group offending to disentangle the joint influence of these measures on hate crime violence. A bivariate comparison of means between incidents motivated by bias and other incidents is presented in Table 1.

Table 1:

Bivariate Comparison of Means Between Bias and Nonbias-Motivated Offenses

	Nonbias motivated	Bias motivated	χ²	t test	p
Variable	%	%	χ²	t test	p
Serious injury	4.52	6.05	63.58		***
Minor injury	37.76	27.39	573.90		***
Co-offending	14.07	20.99	456.24		***
Offense characteristics
Assault	88.51	95.81	710.12		***
Robbery	6.21	3.32	190.54		***
Sex offense	5.44	1.14	506.04		***
Weapon use	17.15	17.53	NS
Alcohol	10.88	13.31	72.22		***
Drug use	1.69	1.67	NS
Gang	0.22	1.13	274.63		***
Offender characteristics	M	M
Proportion male	0.745	0.780		−8.58	***
Age (SD)	30.444 (12.349)	29.517 (13.732)		7.23	***
Proportion White	0.580	0.644		−13.61	***
Proportion Black	0.362	0.248		25.46	***
Proportion other race	0.013	0.020		−5.95	***
Victim characteristics
Victim age (SD)	30.513 (13.868)	30.211 (13.839)		2.38	*
Number of victims (SD)	1.203 (.545)	1.304 (0.733)		−19.37	***
Proportion male	0.403	0.641		−56.52	***
Proportion White	0.658	0.555		24.38	***
Proportion Black	0.282	0.351		−17.01	***
Proportion other race	0.013	0.034		−18.21	***
Proportion Hispanic	0.059	0.080		−9.56	***

p < .05. ***p < .001.

There are two primary differences worth noting in Table 1. First, there are important differences in the likelihood of injury between bias and nonbias incidents. Bias-motivated incidents are actually significantly less likely than nonbias incidents to result in minor injury (27.3% vs. 37.8%, p < .001). Importantly, however, bias incidents are significantly more likely than nonbias incidents to result in serious victim injury (6.1% vs. 4.5%, p < .001). That is, while hate crime incidents are less likely than other incidents to involve a minor injury, they are significantly more likely to involve victim broken bones, internal injury, loss of teeth, severe lacerations, unconsciousness, or other major injuries. Second, and consistent with the argument that co-offending may partially explain increased hate crime violence, bias crimes are significantly more likely than nonbias incidents to involve multiple offenders (p < .001), such that roughly 21% of bias crimes involve multiple offenders, while only about 14% of nonbias crimes involve multiple offenders. The relationship between bias, co-offending, and injury is examined in a multivariate framework in Table 2.

Table 2:

The Relationship Between Bias Injury, Co-Offending, and Serious Victim Injury, Logistic Regression

	Mediation analyses
	Model 1: Bias and serious injury			Model 2: Bias and co-offending			Model 3: Full model			Model 4: Bias and co-offending interaction
Variable	Coefficient	SE	p	Coefficient	SE	p	Coefficient	SE	p	Coefficient	SE	p
Bias	0.163	0.049	**	0.430	0.029	***	0.127	0.049	*	0.018	0.060
Co-offending							0.532	0.046	***	0.450	0.053	***
Offender characteristics
Age mean	−0.008	0.002	***	−0.031	0.001	***	−0.005	0.002	**	−0.005	0.002	**
Proportion male	0.408	0.052	***	−0.718	0.026	***	0.435	0.053	***	0.434	0.053	***
Proportion Black	0.458	0.047	***	0.124	0.029	***	0.451	0.048	***	0.448	0.048	***
Proportion other race	0.316	0.152	*	0.071	0.098		0.320	0.153	*	0.318	0.153	*
Offense characteristics
Weapon	1.947	0.039	***	−0.056	0.029		1.951	0.039	***	1.953	0.039	***
Alcohol	0.697	0.049	***	0.059	0.035		0.695	0.049	***	0.695	0.049	***
Drug use	0.205	0.115		−0.337	0.088	***	0.223	0.115		0.229	0.115	*
Gang	0.695	0.203	**	0.843	0.146	***	0.615	0.204	**	0.604	0.205	**
Robbery	−0.594	0.080	***	1.509	0.045	***	−0.759	0.082	***	−0.740	0.082	***
Sexual assault	0.680	0.092	***	−0.138	0.066	*	0.682	0.092	***	0.677	0.092	***
Victim characteristics
Number of victims	0.002	0.026		1.161	0.018	***	−0.081	0.028	**	−0.078	0.028	**
Age mean	0.004	0.002	*	−0.004	0.001	***	0.004	0.002	**	0.004	0.002	**
Proportion male	0.855	0.043	***	0.809	0.025	***	0.804	0.044	***	0.809	0.044	***
Proportion Black	−0.200	0.050	***	0.152	0.030	***	−0.210	0.050	***	−0.205	0.050	***
Proportion other race	−0.029	0.151		0.041	0.095		−0.027	0.152		−0.019	0.152
Proportion Hispanic	0.136	0.073		0.086	0.047		0.131	0.073		0.136	0.073
Proportion resident	−0.098	0.044	*	0.063	0.026	*	−0.107	0.044	*	−0.110	0.044	*
Stranger	0.070	0.051		0.150	0.032	***	0.057	0.052		0.057	0.052
Bias × Co-Offending										0.320	0.098	***

Note. N = 75,216 incidents.

p < .05. **p < .01. ***p < .001.

Models 1 through 3 in Table 2 present formal mediation analyses. First, while such an effect is not necessarily required for mediation analysis (Rucker, Preacher, Tormala, & Petty, 2011), we examine the direct relationship between bias and serious injury. Model 1 presents a logistic regression of serious victim injury on bias, net of controls (path c, Figure 1). Bias is significantly and positively associated with the likelihood of serious victim injury (b = .163, p = .001), such that, net of controls, incidents involving bias motivation are associated with roughly an 18% increase in the likelihood of injury compared with nonbias incidents. Second, Model 2 presents a logistic regression of co-offending on bias (path a, Figure 1). Again, the results support the hypothesis that bias is significantly associated with the likelihood of co-offending. More specifically, incidents involving bias motivation are significantly more likely than nonbias incidents to involve multiple offenders (b = .430, p < .001). Put another way, bias-motivated incidents are roughly 54% more likely than nonbias incidents to involve instances of co-offending.

Taken together, these results indicate that bias is significantly associated with both serious injury and co-offending. Model 3 presents the relationship between bias and serious victim injury after accounting for the presence of co-offenders. As hypothesized, the relationship between bias and injury is, indeed, partially mediated, such that accounting for the impact of co-offending results in a 22% reduction in the estimated relationship between bias and injury. This mediational relationship is presented in Figure 2. Specifically, these results indicate that bias is associated with a roughly 54% increase in the likelihood of co-offending (a = .430, SE = .029), and the likelihood of serious injury is increased by roughly 26% as a function of such an increase in co-offending.⁴ That said, bias motivation still has an indirect effect on serious victim injury, increasing the risk of serious injury by roughly 13.5% (c = .127, SE = .046), net of controls. These results suggest that, in the current data, the relationship between bias motivation and victim injury is at least partially attributable to the involvement of multiple offenders in these offenses.

Figure 2:

Observed Mediation Relationship Between Bias Motivation, Co-Offending, and Victim Injury

That said, Model 3 still indicates that bias motivation and co-offending are both associated with the likelihood of serious injury. Following this, Model 4 examines the joint relationship between bias motivation, co-offending, and injury using an additional interaction term. Essentially, this model tests whether co-offending moderates the relationship between bias motivation and injury (path d, Figure 1) or whether the relationship between bias and injury varies significantly according to the number of offenders involved in an offense. The results indicate that co-offending does, indeed, significantly moderate the relationship between bias and serious injury. More specifically, there are three important findings to note. First, there is a significant interaction between bias motivation and co-offending (b = .320, p < .001), indicating that incidents involving bias and more than one offender are especially likely to involve serious victim injury, compared with other incidents. Substantively, this result means incidents involving bias motivation and multiple offenders are especially likely to involve serious victim injury compared with other incidents, net of controls. Second, the conditional effect of co-offending remains positive and significant (b = .450, p < .001), meaning that co-offending is also significantly associated with the likelihood of victim injury in nonbias incidents as well. Finally, the conditional effect of bias is not significant, indicating that bias is not significantly associated with the likelihood of victim injury in incidents involving only a single offender. Based on these regression results, the predicted probability of serious victim injury is presented in Figure 3, according to whether the incident involved bias motivation and/or multiple offenders.

Figure 3:

Predicted Probability of Serious Injury, by Bias and Group

As demonstrated in Figure 3, the presence of multiple offenders is an important aspect of the relationship between bias motivation and serious victim injury. Incidents involving only a single offender do not vary substantively in the likelihood of serious victim injury, regardless of whether the incident was motivated by bias or not. There is, however, an independent co-offending effect, such that incidents involving co-offenders are more likely than other incidents to result in serious victim injury, both when the incidents are motivated by bias and when they are not. Most importantly, however, incidents involving bias motivation and co-offending are particularly likely to involve serious injury, suggesting that co-offending may have an exacerbating influence on the relationship between bias and injury.

Supplementary Analysis

In supplementary analyses, we disaggregated the analyses presented in Table 2 by bias type. These results are presented in Table 3.

Table 3:

Disaggregation of Relationship Between Bias, Co-Offending, and Injury by Bias Type

Variable	Coefficient	SE	p
Anti-racial	0.075	0.062
Anti-religious	−0.204	0.203
Anti-ethnicity	0.003	0.115
Anti-sexual orientation	0.424	0.089	***
Anti-disability	0.207	0.283
Co-offending	0.529	0.046	***
Offender characteristics
Age mean	−0.005	0.002	**
Proportion male	0.436	0.053	***
Proportion Black	0.444	0.048	***
Proportion other race	0.322	0.153	*
Offense characteristics
Weapon	1.954	0.039	***
Alcohol	0.691	0.049	***
Drug use	0.213	0.116
Gang	0.642	0.205	**
Robbery	−0.760	0.082	***
Sexual assault	0.678	0.092	***
Victim characteristics
Number of victims	−0.077	0.028	**
Age mean	0.004	0.002	**
Proportion male	0.800	0.044	***
Proportion Black	−0.204	0.051	***
Proportion other race	−0.008	0.152
Proportion Hispanic	0.156	0.076	*
Proportion resident	−0.113	0.044	*
Stranger	0.061	0.052

Note. Model includes all controls presented in Table 2 (not presented). Reference group is nonbias crime. N = 75,216.

p < .05. **p < .01. ***p < .001.

There are two important findings of note. First, the results indicate that, for most bias types, the inclusion of co-offending in the model reduces the relationship between bias and serious victim injury to nonsignificance. That is, after accounting for the presence of multiple offenders in an incident, anti-racial bias, anti-religious bias, anti-ethnicity bias, and anti-disability bias are not significantly associated with the likelihood of serious victim injury. Second, the results indicate that anti-sexual orientation incidents are unique, such that incidents involving this bias type are significantly associated with the likelihood of serious victim injury even after accounting for the presence of multiple offenders. More specifically, anti-sexual orientation bias incidents are roughly 53% more likely than nonbias incidents to involve serious victim injury. Taken together, these results suggest that anti-sexual orientation hate crimes may be especially physically violent, regardless of the presence of co-offenders.

Discussion

One of the primary arguments for hate crime legislation is that bias-motivated crimes are qualitatively more severe than nonbias-motivated crime and, thus, warrant more severe punishment. Following this, a number of scholars have demonstrated that hate crimes are indeed more likely than other crimes to involve greater psychological and physical harm (Fetzer & Pezzella, 2016; Herek et al., 1999; Iganski, 2001; Levin, 1999; Levin & McDevitt, 1993, 2002; McDevitt et al., 2001; Perry, 2001). The current research adds to this literature by focusing on one of these types of greater harm, physical violence, and suggesting one potential explanation for this relationship: co-offending.

Several studies have noted that hate crimes involve an increased prevalence of co-offending (Craig, 2002; Dunbar, 1999; Levin, 1999; Pezzella & Fetzer, 2017). Even more importantly, however, there is a growing body of research suggesting that co-offending is itself related to physical violence (Conway & McCord, 2002; Lantz, 2018; McGloin & Piquero, 2009). To this point, these literatures have developed largely independent of one another. The current research, however, suggested that, because (a) co-offending has been demonstrated to be significantly associated with serious injury and (b) bias-motivated crimes have high rates of co-offending compared with other crimes, then (c) co-offending may play an important role in the relationship between bias motivation and quality of violence.

More specifically, the current study suggested that co-offending may act as a mediator and/or moderator in the relationship between bias motivation and violence, and demonstrated evidence for both processes. Regarding the former process, we conducted a mediation analysis wherein we demonstrated (a) a significant association between bias motivation and serious victim injury; (b) a significant association between bias motivation and co-offending; and (c) a reduced relationship between bias motivation and serious victim injury after accounting for co-offending. That is, incidents involving bias motivation were significantly more likely than other incidents to involve both co-offending and serious injury. Accounting for co-offending, however, resulted in a 22% reduction in the relationship between bias and serious injury, suggesting that co-offending partially mediated the relationship.

Importantly, however, the results also indicated that co-offending acts as a moderator in the relationship between bias and serious injury. That is, the results indicate that the relationship between bias motivation and injury varies significantly according to whether or not multiple offenders were involved in the offense. More specifically, there were three important findings from this analysis. First, bias is not significantly associated with increased injury risk in incidents involving only single offenders. Second, co-offending is significantly and robustly related to increased injury risk regardless of the motivation behind the offense. Finally, and most importantly, the likelihood of serious victim injury is highest in incidents that involve both bias motivation and co-offending.

So, why might bias-motivated offenses involving co-offenders be particularly violent in their outcome? Prior research has suggested that the inherent bias involved in hate crimes increases the brutality of an offense, leading to more violence (e.g., Levin & McDevitt, 1993; Perry, 2001). In other words, the hate motivation itself leads to increased violence. Research on group influence and offending, however, has also suggested that accomplices change the situation of an offense, facilitating changes in behavior that would not have occurred had an individual been alone (Lantz, 2018; McGloin & Piquero, 2009). Co-offenders change thresholds for engaging in behavior by diffusing responsibility and reducing the personal blame associated with each individual behavior (McGloin & Rowan, 2015; McGloin & Thomas, 2016). To this point, these two lines of research (i.e., bias crime and co-offending) have developed largely independent of one another. Following this, one of the primary contributions of this research is the theoretical integration of these two areas of research: Group processes may play an especially important role in promoting bias crime violence. If accomplices facilitate more extreme behavior, and bias has the potential to itself be violent, bias-motivated co-offenses may have an especially high potential for violence.

Some research has even suggested that group behavior may change when only a single individual feels less inhibited, and this may be especially important to consider in the case of bias motivation. Heider (1958), for example, demonstrated that people tend to equate liking someone with agreeing with them. Similarly, Kiesler and Kiesler (1969) argued that people believe they should agree with those that they like and that they should like those with whom they agree. This tendency means that even in groups where individuals hold differing positions, as might be the case in co-offending groups, individuals tend to move toward consensus. Even when consensus is not reached, it is common for individuals to imitate agreement with others to avoid rejection (e.g., Jones, 1964). Consequently, actions can suggest consensus within a group, even when individuals hold disparate opinions.⁵ In the case of bias motivation, this means that it is possible that accomplices may engage in a violent hate crime even when only a single offender holds strong bias opinions.

In general, the current results suggest that accounting for the impact of co-offending partially explains the relationship between bias and physical violence. It is important to note, however, that these results also indicate that anti-sexual orientation violence is different and that these incidents may be unique in their violence. More specifically, incidents involving anti-sexual orientation bias were roughly 53% more likely than nonbias incidents to involve serious victim injury, regardless of the number of offenders involved in the offense. This finding is not without precedent. Gruenewald and Allison (2018), for example, found that anti-lesbian, gay, bisexual, and transgender (LGBT) homicides were more likely than other bias homicides to involve single offenders, which is consistent with the current results indicating that anti-sexual orientation bias crimes are associated with increased violence, even after controlling for co-offending. Furthermore, Cheng, Ickes, and Kenworthy (2013) argued that anti-sexual orientation hate crimes were more likely to be directed against persons and that they were more likely to involve simple assault than intimidation. Moreover, they noted that rates of anti-sexual orientation hate crimes involving aggravated assaults were increasing, and becoming more severe in recent years.

One potential explanation for this increased brutality is the role that masculinity may play in such offenses. Masculinity is tied to aggression (e.g., DeKeseredy & Schwartz, 2005; Messerschmidt, 1993), and past research has suggested that antigay violence may occur as an assertion of masculinity (Bufkin, 1999). Similarly, victims of these crimes may be perceived by offenders as transgressing traditional gender roles, and research has suggested that hate crime violence may function as a mechanism for offenders to attack those who operate outside of traditional social roles or challenge the status quo (Gruenewald & Allison, 2018; Perry, 2001). That is, offenders seek to send a message to victims of these crimes that they are different or less than human (Gruenewald, 2012; Gruenewald & Kelley, 2014). This research also indicates that anti-LGBT hate crimes frequently occur in intimate settings (Gruenewald & Allison, 2018), which may facilitate further violence. Future research should further investigate the unique nature of anti-sexual orientation hate crime violence and the role that co-offending plays in sexual orientation hate crimes in greater depth. In addition, given this variation by bias type, future research should also consider more in-depth examinations of this relationship by hate crime characteristics, including bias type and victim/offender characteristics.

Taken together, the current results suggest at least three important policy implications. First, the present research indicated that incidents involving anti-sexual orientation bias were more likely than any other incidents to involve serious victim injury. This finding is especially important to consider given the current state of hate crime legislation in the United States. Forty-five states currently have a law that addresses hate or bias crimes, but only 30 of those states include sexual orientation as a protected status. Moreover, only 18 states include gender identity as a protected group. If one of the primary rationales for having hate crime laws is that hate crimes hurt more, and anti-sexual orientation hate crimes are especially physically severe as the current research suggests, policymakers must consider addressing the lack of protection for this victim group in the future.

Second, it is important to note that, while the current research indicates that high co-offending rates may partially explain the relationship between bias motivation and serious victim injury, this finding is not meant to suggest that hate crime laws are less than important. Indeed, this research does not indicate that hate crimes do not hurt more, only that co-offending may be one explanation for why hate crimes hurt more physically. Moreover, the presence of co-offending, while important, only partially mediates the relationship between bias and physical violence; bias is still significantly associated with increased likelihood of serious injury after controlling for co-offending. Even further, qualitative differences in hate crime violence are one important justification for hate crime laws, but they are not the only justification. A number of studies have argued that hate crimes hurt more psychologically than other crimes: hate crime victims suffer increased anger and fear (Barnes & Ephross, 1994), stress (Ehrlich, Larcom, & Purvis, 1994), depression (Herek et al., 1999), and general psychological trauma (Levin & McDevitt, 1993) than other victims. Hate crime laws also have symbolic effects, the importance of which cannot be overstated. Hate crime laws express social condemnation of bias-motivated crimes, affirm the social value of victim groups, and reinforce community commitment to equality (Beale, 2000). The current research suggests that not only should these laws be extended to protect more victim groups but, given the violence involved in these offenses, that they should be enforced with greater regularity.

Finally, given the role that co-offenders may play in hate crime violence, it may be especially important to consider devoting special resources toward hate crime group offenders. In the NIBRS data, these offenses committed by these offenders are more violent, and some research on co-offending suggests that co-offenders may have longer offending careers (Carrington, 2009; Lantz & Hutchison, 2015). Conway and McCord (2002) also suggested that offenders may learn violence through co-offending. Following this, it is possible that hate crime group offenders could not only learn violence through co-offending, but also hate. Prejudice is often learned, and those who participate in hate crime group offenses may learn hate and bias in this way. Future research should consider whether such co-offending can “pull” offenders into hate crime careers or hate group membership.

Relatedly, future research might consider these findings in the context of McDevitt et al.’s (2002) typology of group offenders. McDevitt and colleagues (2002) noted that a significant proportion of hate crimes involved “thrill-seeking” crimes, in which peer dynamics played an important role (see also Franklin, 2000). Importantly, however, they also argued that there are varying levels of individual culpability within these groups. In other words, offenders frequently play different roles in a hate crime, and these roles suggest different levels of culpability. More specifically, they posit that there are three levels of culpability within hate crime co-offending groups: the leader, the fellow traveler, and the unwilling participant. The leader instigates the violence and is most culpable. The fellow traveler does not instigate the offense, but is happy to willingly participate. These offenders, McDevitt et al. argue, are nearly as culpable as the leaders. Finally, in some incidents an offender may actually disagree with the behavior of their co-offenders, but not know how to remove themselves from the situation without losing the respect of their peers (see also, Warr, 2002); these offenders are referred to as unwilling participants and, while still somewhat culpable, should be considered less culpable than the other offender types. It is likely that the level of violence associated with an incident might vary according to the distribution of these roles within a group; groups with more fellow travelers, for example, compared with unwilling participants, commit significantly more violent hate crimes; future research should consider the impact that the distribution of these roles might play in hate crime violence.

There are, of course, limitations to the use of NIBRS to analyze hate crime data. First, NIBRS is not a nationally representative sample (Addington, 2004). Thus, it is possible that the patterns observed in these data are less generalizable to highly urbanized areas than they are to less urban areas. Given that bias crime varies significantly across space (Gladfelter, Lantz, & Ruback, 2017), future research should consider variation in hate crime violence by context as well. Second, while NIBRS includes detailed information on physical trauma, it does not include any details on psychological trauma. Following this, our research only focused on one way in which hate crimes may hurt more; it is possible that hate crime co-offending may also be related to levels of victim psychological injury as well, and future research should consider this possibility.

Third, these data are measures of official crime data, meaning they are limited to incidents known to law enforcement. Hate crimes are especially likely to be underreported (Lantz, Gladfelter, & Ruback, 2017; Zaykowski, 2010). As a result, these data exclude unreported hate crime, meaning the prevalence rates of hate crime are likely underestimated. In particular, recent research has demonstrated that those hate crimes most likely to come to the attention of the police are those that fit the stereotypical conception of a hate crime: those involving White offenders and Black victims (Lantz et al., 2017; Lyons, 2008). As such, these offenses may be overrepresented in the NIBRS data. That said, Lantz et al. (2017) also found that violent hate crimes, compared with other hate crimes, are particularly likely to come to the attention of police. Therefore, because this research focuses on only violent hate crimes, the data used in the current study may be less likely than other crimes to be affected by issues with underreporting. Given these potential limitations, however, future research should consider these relationships in other data sources not based solely on official records. Recent research has demonstrated the validity and reliability of data based on victim and media reports (e.g., Levin & Reichelmann, 2015; Ruback, Gladfelter, & Lantz, 2018), as well as open-source data like the Extremist Crime Database (ECDB; Chermak, Freilich, Parkin, & Lynch, 2012; Klein & Allison, 2018), and future research should consider the use of these data sources to examine the role of groups in hate crime offending.

Fourth, the official nature of the NIBRS data makes it difficult to precisely test many of the mechanisms outlined in this research. As a result, we cannot test potential mechanisms that may explain the remaining relationship between bias and severity of violence, after controlling for co-offending. Prior research has suggested a number of mechanisms, including the presence of animus (Levin & McDevitt, 1993), and the desire to send a message by subordinating and intimidating victim(s) (Perry, 2001); future research should investigate other potential explanations for the relationship between bias and physical violence. Furthermore, while we suggest that mechanisms of group influence, like diffusion of responsibility, may explain the relationship between hate crime co-offending and violence, we cannot directly test these mechanisms. The purpose of the current research was to demonstrate the importance of co-offenders in explaining the relationship between bias motivation and violence; future research should examine, in greater detail, why co-offenders are important.

Conclusion

While recent research has argued that there is “insufficient support for enhanced penalties for hate crimes based on the severity of physical injury” (Pezzella & Fetzer, 2017, p. 724), the current research finds that hate crimes do hurt more, albeit for somewhat different reasons than previously suggested. More specifically, this research finds that hate crimes are associated with increased likelihood of serious victim injury and that co-offending may be a mechanism through which hate crime increases physical violence. This is, of course, with an important caveat: anti-sexual orientation hate crimes do hurt more, no matter how many offenders are involved, suggesting that future hate crime legislation should devote special attention to this population. Other hate crimes, however, are qualitatively more severe when they involve co-offenders. In this way, this research represents an important next step in the theoretical integration of research on bias crimes and co-offending violence. As a result, we suggest that instead of hate crimes hurt more, it may actually be that group crimes hurt more, and bias group crimes hurt most.

Footnotes

Authors’ Note:

The authors would like to thank Marin Wenger for helpful comments on earlier versions of this article, as well as three anonymous reviewers and the editorial office at Criminal Justice & Behavior for constructive reviews. Brendan Lantz is the primary contact (e-mail: blantz@fsu.edu).

Notes

ORCID iD

Brendan Lantz

Brendan Lantz is an assistant professor of criminology and criminal justice at Florida State University. He received his MA and PhD in criminology from The Pennsylvania State University and was a graduate research fellow for the Bureau of Justice Statistics (BJS). His research focuses on co-offending, violence, and hate crime, and has appeared in a number of research outlets, including Journal of Research in Crime and Delinquency, Justice Quarterly, Criminology & Public Policy, Aggressive Behavior, Journal of Interpersonal Violence, and Crime & Delinquency.

Joonggon Kim is a senior inspector of Korea National Police Agency. He received his doctoral degree in criminology and criminal justice at Florida State University in 2018. His research interests include life-course perspectives, criminological theories, hate crime, and policing.

References

Addington

L. A.

(2004). The effect of NIBRS reporting on item missing data in murder cases. Homicide Studies, 8, 193-213.

Barnes

Ephross

P. H.

(1994). The impact of hate violence on victims: Emotional and behavioral responses to attacks. Social Work, 39, 247-251.

Baron

R. M.

Kenny

D. A.

(1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173-1182.

Beale

S. S.

(2000). Federalizing hate crimes: Symbolic politics, expressive law, or tool for criminal enforcement. Boston University Law Review, 80, 1227-1281.

Bufkin

(1999). Bias crime as gendered behavior. Social Justice, 26, 155-170.

Carrington

P. J.

(2002). Group crime in Canada. Canadian Journal of Criminology, 44, 277-316.

Carrington

P. J.

(2009). Co-offending and the development of the delinquent career. Criminology, 47, 1295-1329.

Cheng

Ickes

Kenworthy

J. B.

(2013). The phenomenon of hate crimes in the United States. Journal of Applied Social Psychology, 43, 761-794.

Chermak

S. M.

Freilich

J. D.

Parkin

Lynch

J. P.

(2012). American terrorism and extremist crime data sources and selectivity bias: An investigation focusing on homicide events committed by far-right extremists. Journal of Quantitative Criminology, 28, 191-218.

10.

Conway

K. P.

McCord

(2002). A longitudinal examination of the relation between co-offending with violent accomplices and violent crime. Aggressive Behavior, 28, 97-108.

11.

Corcoran

Lader

Smith

(2015). Hate crime, England and Wales, 2014/15. London, England: Home Office Statistical Bulletin.

12.

Cornish

Clarke

(1986). Situational prevention, displacement of crime and rational choice theory. In Heal

Laycock

G. K.

(Eds.), Situational crime prevention: From theory into practice (pp. 1-16). Richmond, UK: Her Majesty’s Stationery Office.

13.

Craig

K. M.

(2002). Examining hate-motivated aggression: A review of the social psychological literature on hate crimes as a distinct form of aggression. Aggression and Violent Behavior, 7, 85-101.

14.

DeKeseredy

Schwartz

M. D.

(2005). Masculinities and interpersonal violence. In Kimmel

M. S.

Hearn

Connell

R. W.

(Eds.), Handbook of studies on men and masculinities (pp. 353-366). Thousand Oaks, CA: SAGE.

15.

Dunbar

(1999). Defending the indefensible: A critique and analysis of psycholegal defense arguments of hate crime perpetrators. Journal of Contemporary Criminal Justice, 15, 64-77.

16.

Dunbar

(2003). Symbolic, relational, and ideological signifiers of bias-motivated offenders: Toward a strategy of assessment. American Journal of Orthopsychiatry, 73, 203-211.

17.

Ehrlich

Larcom

Purvis

(1994). The traumatic effects of ethnoviolence. In Perry

(Ed.), Hate and bias crime: A reader (pp. 153-167). Abingdon, UK: Routledge.

18.

Felson

Lantz

(2016). When are victims unlikely to cooperate with the police? Aggressive Behavior, 42, 97-108.

19.

Fetzer

M. D.

Pezzella

F. S.

(2016). The nature of bias crime injuries: A comparative analysis of physical and psychological victimization effects. Journal of Interpersonal Violence. Advance online publication. doi:10.1177/0886260516672940

20.

Franklin

(2000). Antigay behaviors among young adults: Prevalence, patterns and motivators in a noncriminal population. Journal of Interpersonal Violence, 15, 339-362.

21.

Garcia

McDevitt

Balboni

(1999). The psychological and behavioral effects of bias and non-bias motivated assault. Washington, DC: U.S. Department of Justice, National Institute of Justice.

22.

Garofalo

(1991). Racially motivated crimes in New York City. In Lynch

M. J.

Paterson

E. B.

(Eds.), Race and criminal justice (pp. 161-173). Albany, NY: Harrow and Heston.

23.

Gladfelter

A. S.

Lantz

Ruback

R. B.

(2017). The complexity of hate crime and bias activity: Variation across contexts and types of bias. Justice Quarterly, 34, 55-83.

24.

Gruenewald

(2012). Are anti-LGBT homicides in the United States unique? Journal of Interpersonal Violence, 27, 3601-3623.

25.

Gruenewald

Allison

(2018). Examining differences in bias homicide across victim groups. Crime & Delinquency, 64, 316-341.

26.

Gruenewald

Kelley

(2014). Typifying anti-LGBT homicide events by mode of victim selection. Criminal Justice and Behavior, 41, 1130-1142.

27.

Hamm

M. S.

(1993). American skinheads. Westport, CT: Praeger.

28.

Heider

(1958). The psychology of interpersonal relations. London, England: Psychology Press.

29.

Herek

G. M.

Gillis

J. R.

Cogan

J. C.

(1999). Psychological sequelae of hate-crime victimization among lesbian, gay, and bisexual adults. Journal of Consulting and Clinical Psychology, 67, 945-951.

30.

Herek

G. M.

Gillis

J. R.

Cogan

J. C.

Glunt

E. K.

(1997). Hate crime victimization among lesbian, gay, and bisexual adults. Journal of Interpersonal Violence, 12, 195-215.

31.

Iganski

(2001). Hate crimes hurt more. American Behavioral Scientist, 45, 626-638.

32.

Iganski

Lagou

(2014). The personal injuries of “hate crime.” In Hall

Corb

Giannasi

Grieve

(Eds.), The routledge international handbook on hate crime (pp. 34-46). Abingdon, UK: Routledge.

33.

Jones

E. E.

(1964). Ingratiation, a social psychological analysis. New York, NY: Appleton-Century-Crofts.

34.

Kiesler

C. A.

Kiesler

S. B.

(1969). Conformity. Boston, MA: Addison-Wesley.

35.

Klein

B. R.

Allison

(2018). Accomplishing difference: How do anti-race/ethnicity bias homicides compare to average homicides in the United States? Justice Quarterly, 35, 977-1003. doi:10.1080/07418825.2017.1351576

36.

Lantz

(2018). The consequences of crime in company: Co-offending, victim-offender relationship, and quality of violence. Journal of Interpersonal Violence. Advance online publication. doi:10.1177/0886260518786497

37.

Lantz

Gladfelter

A. S.

Ruback

R. B.

(2017). Stereotypical hate crimes and criminal justice processing: A multi-dataset comparison of bias crime arrest patterns by offender and victim race. Justice Quarterly. Advance online publication. doi:10.1080/07418825.2017.1399211

38.

Lantz

Hutchison

(2015). Co-offender ties and the criminal career: The relationship between co-offender group structure and the individual offender. Journal of Research in Crime and Delinquency, 52, 658-690.

39.

Lantz

Ruback

R. B.

(2017). The relationship between co-offending, age, and experience using a sample of adult burglary offenders. Journal of Developmental and Life-course Criminology, 3, 76-97.

40.

Laufer

Harel

Molcho

(2006). Daring, substance use and involvement in violence among school children: Exploring a path model. Journal of School Violence, 5, 71-88.

41.

Lawrence

F. M.

(1994). The punishment of hate: Toward a normative theory of bias-motivated crimes. Michigan Law Review, 93, 320-381.

42.

Lawrence

F. M.

(2002). Racial violence on a “small island”: Bias crime in a multicultural society. In Burney

(Ed.), The hate debate: Should hate be punished as a crime. London, England: Profile Books.

43.

Lawrence

F. M.

(2009). Punishing hate. Cambridge, MA: Harvard University Press.

44.

Levin

(1999). Hate crimes: Worse by definition. Journal of Contemporary Criminal Justice, 15, 6-21.

45.

Levin

McDevitt

(1993). Hate crimes: The rising tide of bigotry and Bloodshed. New York, NY: HarperCollins.

46.

Levin

McDevitt

(2002). Hate crimes revisited: America’s war on those who are different. New York, NY: Basic Books.

47.

Levin

Reichelmann

(2015). From thrill to defensive motivation: The role of group threat in the changing nature of hate-motivated assaults. American Behavioral Scientist, 59, 1546-1561.

48.

Lyons

C. J.

(2008). Individual perceptions and the social construction of hate crimes: A factorial survey. The Social Science Journal, 45, 107-131.

49.

Lyons

C. J.

Roberts

(2014). The difference “hate” makes in clearing crime: An event history analysis of incident factors. Journal of Contemporary Criminal Justice, 30, 268-289.

50.

Martin

S. E.

(1996). Investigating hate crimes: Case characteristics and law enforcement responses. Justice Quarterly, 13, 455-480.

51.

Matza

(1964). Delinquency and drift. Piscataway, NJ: Transaction Publishers.

52.

McDevitt

Balboni

Garcia

(2001). Consequences for victims: A comparison of bias-and non-bias-motivated assaults. American Behavioral Scientist, 45, 697-713.

53.

McDevitt

Levin

Bennett

(2002). Hate crime offenders: An expanded typology. Journal of Social Issues, 58, 303-317.

54.

McGloin

J. M.

Piquero

A. R.

(2009). “I wasn’t alone”: Collective behaviour and violent delinquency. Australian & New Zealand Journal of Criminology, 42, 336-353.

55.

McGloin

J. M.

Rowan

Z. R.

(2015). A threshold model of collective crime. Criminology, 53, 484-512.

56.

McGloin

J. M.

Thomas

K. J.

(2016). Incentives for collective deviance: Group size and changes in perceived risk, cost, and reward. Criminology, 54, 459-486.

57.

Messerschmidt

J. W.

(1993). Masculinities and crime: Critique and reconceptualization of theory. Lanham, MD: Rowman & Littlefield.

58.

Messner

S. F.

McHugh

Felson

R. B.

(2004). Distinctive characteristics of assaults motivated by bias. Criminology, 42, 585-618.

59.

Meyer

(2010). Evaluating the severity of hate-motivated violence: Intersectional differences among LGBT hate crime victims. Sociology, 44, 980-995.

60.

Perry

(2001). In the name of hate: Understanding hate crimes. London, England: Psychology Press.

61.

Perry

(2012). Hate and bias crime: A reader. Abingdon, UK: Routledge.

62.

Pezzella

F. S.

Fetzer

M. D.

(2017). The likelihood of injury among bias crimes: An analysis of general and specific bias types. Journal of Interpersonal Violence, 32, 703-729.

63.

Ruback

R. B.

Gladfelter

A. S.

Lantz

(2015). An examination of hate and bias incidents in Pennsylvania, 1999-2012. Harrisburg: Center for Rural Pennsylvania.

64.

Ruback

R. B.

Gladfelter

A. S.

Lantz

(2018). Hate crime victimization data in Pennsylvania: A useful complement to the UCR. Violence & Victims, 33, 330-350.

65.

Rucker

D. D.

Preacher

K. J.

Tormala

Z. L.

Petty

R. E.

(2011). Mediation analysis in social psychology: Current practices and new recommendations. Social and Personality Psychology Compass, 5, 359-371.

66.

Shaw

C. R.

McKay

H. D.

(1931). Social factors in juvenile delinquency. Washington, DC: Government Press.

67.

Strom

(2001). Hate crimes reported in NIBRS, 1997-99. Washington, DC: U.S. Department of Justice.

68.

Warr

(2002). Companions in crime: The social aspects of criminal conduct. Cambridge, UK: Cambridge University Press.

69.

Weinstein

(1992). First Amendment challenges to hate crime legislation: Where’s the speech? Criminal Justice Ethics, 11, 6-20.

70.

Weisburd

S. B.

Levin

(1993). On the basis of sex: Recognizing gender-based bias crimes. Stanford Law & Policy Review, 5, 21-48.

71.

Zaykowski

(2010). Racial disparities in hate crime reporting. Violence & Victims, 25, 378-394.