In 2020, a total of 11,129 hate crimes were reported in the United States (U.S. Department of Justice, Federal Bureau of Investigation, n.d.); however, hate crimes are frequently not reported or are categorized differently by law enforcement and as such, the actual number of hate crimes is estimated to be approximately thirty times greater (Pezzella et al., 2019). Hate or bias crimes stem from prejudice against social groups or members of social groups and result in victimization of persons or groups and/or their property (Federal Bureau of Investigation, n.d.). In addition to physical harm, hate crimes can involve intimidation and harassment (Lockwood & Cuevas, 2020). The social groups most often targeted are distinguished by race and ethnicity, gender, country of origin, religion, sexual orientation, and disability (Lockwood & Cuevas, 2020; U.S. Department of Justice, 2017). In terms of race/ethnicity, Black individuals accounted for 48.5% of total hate crime victims in 2019 (Federal Bureau of Investigation, 2020). Further, nearly 17% of hate crimes were motivated by bias regarding sexual orientation in 2019 (Federal Bureau of Investigation, 2020). Hate crimes related to religion, such as anti-Jewish hate crimes, increased by 54% between 2014 and 2018 and have the third highest frequency, following race and ethnicity and sexual orientation (Hodge & Boddie, 2021; U.S. Department of Justice, n.d.). Finally, transgender persons are over four times more likely to be personally victimized and over twice as likely to have their property victimized than cisgender persons (Flores et al., 2021).

Hate crime legislation was first passed by the U.S. government in 1990 (101st Congress, 1990). In subsequent years, governing bodies strengthened this legislation by increasing penalties for hate crimes (Hate Crimes Sentencing Enhancement Act, 1994) and adding protections for disability, race and ethnicity, religion, and sexual orientation, which climaxed in 2009 with the landmark Matthew Shepard and James Byrd, Jr. Hate Crimes Prevention Act. Some states have passed hate crimes laws to increase punishment and fill gaps in federal legislation including increased penalties for non-violent crimes that are motivated by hate (Cabeldue et al., 2018) and elevating degrees of offense when crimes are motivated by hate (e.g., misdemeanor elevated to felony, second degree murder elevated to first degree murder) (Tex. Crim. Pro. Art. § 42.014, 2005; Tex. Penal Code Ann. §12.47, 2017). Of concern, despite enhanced hate crime legislation, reported hate crimes increased in the U.S. by 13.36% between 2019 and 2020 (U.S. Department of Justice, n.d.). Further, during the COVID-19 pandemic, hate crimes against Asian Americans in the US increased by 150% (Edmondson & Tankersley, 2021). In the face of data showing increasing rates of hate crimes, debate has continued regarding the effectiveness and constitutionality of hate crime laws (Cabeldue et al., 2018).

Not only have hate crime rates risen, but US residents’ fears of experiencing hate crimes have grown by 6% in recent years (Brenan, 2021). Beyond fears about hate crimes, beliefs about hate crime laws, offenders, and victims have political bearing. For example, protection of sexual minorities as part of hate crime laws detracts from public support for their implementation (Johnson & Byers, 2003). Moreover, a series of studies (e.g., Cramer et al., 2013a, 2013b, 2013c, 2017) show the importance of attitudes about hate crimes for legal perceptions and decision making. Overall, support for hate crime legislation is associated with more punitive decisions toward offenders (Cramer et al., 2013b). For example, perceived blame toward a hate crime perpetrator (Cramer et al., 2013a) and perceived malice of a victim in a hate crime event (Cramer et al., 2013c) impact punishment decisions. Elaborating on these patterns, Cramer and colleagues (Cramer et al. 2017) assessed narrative explanations of support for hate crime penalty enhancement laws, finding negative views toward offenders yielded favorable views about the laws, whereas persons articulating basic legal explanations (e.g., “all crimes should be treated equally”) were less likely to support these laws.

Additionally, there are several related literatures that help inform our understanding of how hate crime beliefs are formed, ranging from models of prejudice, such as the dual-process model of prejudice (Duckitt, 2001), the justification-suppression model (Crandall & Eshleman, 2003, 2005) and political attitudes (Cramer et al., 2017; Malcom et al., 2022). Most recently, Malcom and colleagues (2022) examined the mediating role of prejudice against Hispanic, Asian, and Black individuals on political affiliation and support for hate crime legislation. The authors findings suggest that individuals who hold more negative views of minorities and are affiliated with more conservative political ideologies, are less likely to support hate crime legislation. As such, it appears that the main determinates relevant to hate crime beliefs are negative attitudes toward minorities and political ideologies, which will be discussed in more detail below as correlates of the HCBS.

Given the above-mentioned findings and legal relevance of hate crime-related attitudes, it stands to reason such beliefs may influence perspectives of constitutionality of hate crime laws, and ultimately impact perceptions of local and national elections where issues of hate are part of the debate. Thus, accurate measurement of assessing hate crime beliefs has implications for research, intervention development, fear reduction, and policy-making.

Measurement of hate crime beliefs

Cabeldue et al. (2018) developed the Hate Crimes Beliefs Scale (HCBS), designed to provide a comprehensive, validated measure of hate crime-related social norms. HCBS items were derived from hate crimes literature that addressed supportive and unsupportive legal, philosophical, and social arguments (e.g., Gerstenfeld, 2011; Glaser, 2005); social and criminological theories (e.g., Chockler & Halpern, 2004; Mikula, 2003); and theories involving blame attribution and prejudice (e.g., Nelson, 2006; Shaver & Drown, 1986). Cabeldue and colleagues (2018) gathered a community sample of adults who completed 50 items used to measure beliefs about hate crimes (e.g., “hate crime perpetrators cause psychological trauma to their victims”). Exploratory factor analysis resulted in ten items being dropped due to lack of factor loadings. Four factors were identified from the remaining 40 items: 1) negative beliefs (negative views related to hate-crime legislation and protection of minority groups; 27 items; α = 0.95), 2) offender punishment (belief in strict hate crime penalties; 5 items; α = 0.84), 3) deterrence (belief that hate crime legislation is effective; 3 items; α = 0.79), and 4) victim harm (belief that hate crimes cause harm to the victim; 5 items; α = 0.74). All four factors were significantly correlated. Negative beliefs were significantly negatively correlated with offender punishment, deterrence, and harm. Deterrence was significantly positively associated with offender punishment. Harm was significantly positively associated with offender punishment and deterrence.

Regarding validity, HCBS subscales were examined for associations with political orientation and several types of prejudice (i.e., racism, transphobia, sexual prejudice, and religious intolerance). The HCBS negative beliefs subscale was most robustly and positively associated with a conservative political identity and all measures of prejudice. The other three subscales displayed positive associations with a liberal political identity, and to a lesser extent, modest significant negative associations with several prejudice metrics. In terms of predictive validity, controlling for demographic factors (e.g., participant race and gender), only the negative views subscale predicted legal decision-making in a hate crime jury decision-making scenario.

Arguing that hate crime beliefs vary across nations, Bacon et al. (2021) modified the HCBS by conducting two studies in the United Kingdom (UK). HCBS-UK modifications included additional population-specific questions (e.g., questions about Muslim and Jewish, persons, as well as persons with disabilities) as well as governmental-related modifications (e.g., freedom of speech vs. First Amendment of the US Constitution; police vs. prosecutors). These additions and modifications resulted in a pool of 46 items, which were administered in two separate studies, one study included a sample of community members and college students and the other included only community members. The four-factor model originally proposed by Cabeldue et al. (2018) did not demonstrate adequate fit among the UK sample data. Thus, Bacon et al. (2021) identified a shorter 20-item version of the HCBS-UK with three factors: 1) compassion (i.e., having understanding and concern for victims and groups affected by hate crimes; 5 items; α = 0.76), 2) denial (i.e., refusing to recognize the seriousness and extent of harm resulting from hate crimes and the need for laws to address them; 10 items; α = 0.90), and 3) sentencing (i.e., supporting punishment of those who commit hate crimes: 5 items; α = 0.89).

Sociopolitical attitudinal correlates

Sociopolitical attitudes provide an appropriate starting point for validity assessment for the HCBS. Bacon et al. (2021) examined the convergent validity of the shortened HCBS-UK. Specifically, using a community sample, they assessed the association of hate crime attitudes with measures of individual differences implicated in the formation of prejudice. Social Dominance Orientation (SDO; i.e., positive attitudes towards social inequality; Pratto et al., 1994) and Right-Wing Authoritarianism (RWA; i.e., authoritarian and obedience beliefs; Altemeyer, 1981) were selected by Bacon and colleagues (2021) given robust evidence of their causal role in prejudice formation and manifestation (Asbrock et al., 2010; Duckitt & Sibley, 2007, 2010; Duckitt et al., 2002; Sibley et al., 2006). The dual-process motivational model of ideology and prejudice (Duckitt, 2001) represents SDO and RWA as ideological attitudes that develop from two separate sets of personality traits, social beliefs, and social environments (Duckitt & Sibley, 2010). SDO is proposed to be formed by competitive worldviews that develop from a tough-minded personality and social contexts characterized by ingroup dominance and superiority (Duckitt & Sibley, 2009, 2010; Sibley & Duckitt, 2013). On the other hand, RWA is proposed to be formed by maladaptive worldviews that develop from high conscientiousness and threatening and unpredictable social contexts (Duckitt & Sibley, 2009, 2010). RWA is characterized by values of social conformity and security (e.g., social conservativism, traditionalism) rather than autonomy (e.g., individualism, openness; Duckitt et al., 2002). Sibley et al. (2006) discuss that levels of RWA are predictive of several types of oppressive attitudes (e.g., homophobia, religious ethnocentrism, benevolent sexism). Thus, Bacon et al. (2021) theorized that the dual-process model of prejudice may translate to hate crime beliefs and found support for this hypothesis. Bacon and colleagues found that RWA and SDO were positively correlated with the denial factor and negatively correlated with the compassion factor. Lastly, SDO was shown to be negatively correlated with the sentencing factor.

Political orientation is another factor warranting attention with the HCBS for several reasons. First, SDO and RWA are associated with political orientation (e.g., Pratto et al., 1994; Van Hiel & Mervielde, 2002). Specifically, SDO and RWA are positively correlated with right-wing party affiliation and conservative attitudes (Balliet et al., 2018; Duckitt & Sibley, 2009; Grina et al., 2016; McCann, 2009; Sidanius et al., 1994). Interestingly, data suggests that higher RWA and SDO have been strongly related to preferences for Republicans and far-right, and less favorable views of Democratic, presidential election candidates (Godbole et al., 2022; Jost et al., 2009; Van Assche et al., 2019). Second, a study examining the link between political orientation and support for hate crime law penalty enhancements highlight the importance of beliefs about hate crimes. Cramer and colleagues (2017) found that political conservatism was associated with less support for punishment aspects of hate crime laws. Further, the link between political conservatism and lack of hate crime law support was explained by narrative legal arguments such as hate crimes being treated as all other crimes. On the contrary, the link between liberal political orientation and support for hate crime laws was explained by narrative offender-focused arguments such as dehumanizing offenders. This body of evidence suggests hate crime related attitudes are politically and legally relevant, and that political identity may be linked to HCBS subscales.

The current study

Further assessment and refinement of the HCBS holds potential to impact research through public opinion and social norms/attitudes assessment. Identification of a reliable and valid measure of hate crime beliefs across countries holds potential to impact policy-making and intervention development. As such, the current study sought to examine the multi-factorial structure of the original HCBS (Cabeldue et al., 2018) in comparison to the multi-factorial structure of the shortened HCBS-UK (Bacon et al., 2021) among a community and college sample of US adults. We hypothesized:

  • H1: HCBS subscales would be significantly associated with RWA and SDO in directions consistent with prior literature. That is, RWA and SDO would be positively associated with more pejorative/less supportive subscales, and negatively associated with supportive views of issues like support for hate crime laws or victims.

  • H2: HCBS subscales would be significantly associated with political party affiliation. Republican party affiliation (compared to Democratic or other party affiliation) would be positively associated with more pejorative/less supportive subscales, and negatively associated with supportive views of issues like support for hate crime laws or victims.

We also explored the following research question:

  • RQ: Does the original factor structure of the HCBS remain more psychometrically sound compared to the shortened version of the HCBS?

Materials and method

Participants

Participants (N = 463) were primarily female (n = 340, 73.4%) and non-Hispanic White race/ethnicity (n = 401, 86.6%), with an average age of 25.55 years (SD = 10.99; see Table 1 for summary of sample demographics). Most participants had a high school education or less (n = 294, 63.6%) and were heterosexual (n = 409, 88.5%). Most participants reported a political affiliation of Republican (n = 139, 30.2%) or Democrat (n = 148, 32.1%). Most participants were of Christian faith, either Protestant (n = 222, 48.3%) or Catholic (n = 96, 20.8%). Finally, participants were primarily from the upper Midwest region of the United States (n = 342, 73.86%).

Table 1 Demographic characteristics of study sample
Full size table

Measures

Demographics

Participants completed a demographic questionnaire assessing gender, race/ethnicity, education, political affiliation, religious affiliation, sexual orientation, and age.

Hate crime beliefs scale

The Hate Crime beliefs Scale (HCBS; Cabeldue et al., 2018) is a forty-item self-report questionnaire used to assess attitudes towards both victims and perpetrators of hate crimes, as well as hate crime legislation. Sample items include “I believe hate crimes receive too much attention” and “Hate-crime perpetrators cause psychological trauma to their victims.” Items are rated on a 5-point Likert scale, ranging from 1 (strongly disagree) to 5 (strongly agree). Initial scale development yielded a four-factor structure of 1) Negative Beliefs (27 items; α = 0.95), 2) Offender Punishment (5 items; α = 0.84), 3) Deterrence (3 items; α = 0.79), and 4) Victim Harm (5 items; α = 0.74). Importantly, the present study used the original 40-item HCBS (Cabeldue et al., 2018), which does not include UK-specific items added by Bacon et al. (2021) regarding Muslim hate crimes. Further, language in the original HCBS referenced African Americans, whereas Bacon et al. (2021) use the term “Black.”

Social dominance orientation scale

The Social Dominance Orientation Scale (SDO; Pratto et al., 1994) is a 16-item self-report questionnaire used to assess one’s attitudes towards social inequality (e.g., “Some groups of people are simply inferior to other groups”, α = 0.91). Responses are indicated on a 7-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). Mean scores are generated via an average of the summed score, with a total possible score ranging from 1 to 7. Higher mean scores indicate greater positive attitudes towards, or preference for, social inequality and group-based dominance beliefs. Internal consistency in the current study was excellent (α = 0.90).

Right wing authoritarianism scale

Participants completed the 14-item version of the Right-Wing Authoritarianism Scale (RWA; Manganelli Rattazzi et al., 2007). The RWA assesses two sub-components: 1) submission and authoritarian aggression (e.g., “Obedience and respect for authority are the most important values children should learn”, α = 0.72) and 2) conservatism (e.g., “It is good that nowadays young people have greater freedom ‘to make their own rules’ and to protest against things they don’t like”, α = 0.75). Responses are indicated on a 7-point scale ranging from -3 (totally disagree) to +3 (totally agree). Items on the conservatism subscale are reverse scored before generating mean subscale scores. Internal consistency in the current study was acceptable for both subscales (submission and authoritarian aggression: α = 0.91; conservatism: α = 0.82).

Procedure

Participants were recruited using two sampling strategies: 1) online snowball sampling via social media platforms (i.e., Facebook and Twitter) and 2) from a student participant pool from a mid-sized Midwestern University (Brickman Bhutta, 2012). Most participants were recruited via the student research pool (n = 332, 71.7%). Interested participants were provided a link to complete the online survey battery via Qualtrics. All participants were provided an online consent form and indicated consent by clicking through to the survey. Along with the study materials, participants were presented with four attention check items to ensure mindful participation. The complete survey battery took approximately 40 to 50 min to complete. Non-student participants were not compensated for their participation, while students were incentivized with one research credit towards their course requirements. All procedures were approved by the Institutional Review Board at [University of North Dakota].

Data analysis

A total of 637 persons started the survey (i.e., clicked the link to initiate data collection). 174 were dropped due to failed attention check items or not providing any data, resulting in a final sample of 463. Data missingness on survey items for the analyzable sample ranged from 0.0% to 0.1%. Multiple imputation (Enders, 2017) was used to supplant missing values on variables of interest. Confirmatory factor analysis was conducted using AMOS v. 24 to compare the original four-factor (Cabeldue et al., 2018) and approximated UK three-factor (Bacon et al., 2021) models of the HCBS. Not all item wording is identical in the UK three-factor model, requiring minor adjustments to the model. HCBS subscales were permitted to correlate consistent with previous HCBS studies (Bacon et al., 2021; Cabeldue et al., 2018). Maximum likelihood estimation was used as the data were normally distributed. Our sample size of 463 satisfied statistical power requirements for both models and items per factor (Wolf et al., 2013). Acceptable model fit was determined through inspection of the Comparative Fit Index (CFI), Tucker-Lewis Index (TLI), Root Mean Square Error of Approximation (RMSEA), and Root Mean Square Residual (RMR). The following guidelines for acceptable model fit were used: CFI > 0.90; TLI > 0.90, RMSEA < 0.08, and SRMR < 0.08 (Hu & Bentler, 1999; Schumaker & Lomax, 2010). We followed recommendations in the statistical literature regarding use of modification indices (e.g., Bentler, 2007, 2010). First, we decided to use post-hoc modifications because they can aid in identifying gaps or limitations of models, especially those needing development and validation. Second, we adopted a conservative cut-off (MI ≥ 25) in order to avoid model misspecification. Third, we made modifications only supported by theory or empirical literature suggesting item content was conceptually related for any error terms allowed to correlate. For example, error terms for HCBS items 36 and 37 were allowed to correlate (see details in results section) because they both pertain to the documented harm done based on anti-religious prejudice (e.g., Nadal et al., 2010). Finally, we reported model fit statistics with and without modifications, as well as the specific details regarding modifications made.

To examine hypotheses, bivariate correlations were used to assess HCBS subscales with RWA and SDO. Analysis of Variance with Bonferroni post-hoc tests and partial eta-squared effect size were used to test political affiliation differences in HCBS subscales. Political affiliation was recoded to ensure large enough sample sizes as follows: Democrat (n = 139; 32%), Republican (n = 148; 30%), and another political affiliation (n = 176; 38%).

Results

RQ1: HCBS factor structure

Model 1 was the HCBS with original four-factor structure (Cabeldue et al., 2018), allowing the following latent factors to correlate: 1) Negative Beliefs and Offender Punishment, 2) Offender Punishment & Deterrence, 3) Deterrence & Victim Harm, and 4) Offender Punishment and Victim Harm. Model 1 indicated poor model fit, (see Table 2 for model fit statistics). Model 2 retained the model 1 structure with post-hoc modifications whereby the following error terms for item pairs were allowed to correlate: negative beliefs subscale item pairs 1 & 2, 2 & 3, 1 & 3, 4 & 1, 4 & 2, 4 & 3, 7 & 8, 10 & 12, 4 & 13, 10 & 13, 12 & 13, 1 & 19, 18 & 19, 20 & 21, 20 & 23, 21 & 23, 24 & 25, 24 & 27, 25 & 27, and 26 & 27; victim harm items 36 & 37, and subscale latent variables of negative beliefs and victim harm. Model 2 indicated poor fit (see Table 2).

Table 2 Model fit indices for confirmatory factor analyses
Full size table

Model 3 was the HCBS with the UK three-factor structure, allowing the latent factors of compassion, denial, and sentencing to correlate. Model 3 indicated borderline model fit (see Table 2). Thus, model 4 allowed for implementation of post-hoc modification indices whereby error terms for item pairs were allowed to correlate as follows: items 33 & 35 and items 21 & 23 on the compassion subscale; items 1 & 2, items 1 & 19, items 2 & 19, items 6 & 19 on the denial subscale. Model 4 indicated good fit to the data (see Table 2). Factor loading values ranged from 0.357 to 0.869 in the three-factor HCBS-UK model (all ps < 0.001; see Table 3). Internal consistency values for the subscales were acceptable: denial (α = 0.94), sentencing (α = 0.80), and compassion (α = 0.73). A visual depiction of the HCBS-UK model can be seen in Fig. 1. We used this model for further hypothesis testing.

Table 3 The final 17-item HCBS-SF and descriptive statistics
Full size table
Fig. 1
figure 1

17-item HCBS three-factor CFA model. Note: HCBS = Hate Crime Beliefs Scale; e = error

Full size image

H1 and H2: Correlations with SDO and RWA

Consistent with expectations, the denial subscale score was significantly negatively associated with both the sentencing (r = -0.59, p < 0.001) and compassion (r = -0.68, p < 0.001) subscale scores, while the sentencing subscale score was significantly positively associated with the compassion subscale score (r = 0.65, p < 0.001). In support of hypotheses, the denial subscale was significantly positively associated with SDO (r = 0.65, p < 0.001) and RWA subscales (authoritarian aggression and submission: r = 0.65, p < 0.001; conservatism: r = 0.58, p < 0.001). Further supporting H1, both the sentencing and compassion subscales were significantly negatively correlated with SDO (sentencing: r = -0.40, p < 0.001; compassion: r = -0.54, p < 0.001) and the RWA subscale scores authoritarian aggression and submission (sentencing: r = -0.40, p < 0.001); compassion: r = -0.50, p < 0.001) and conservatism (sentencing: r = -0.43, p < 0.001; compassion: r = -0.51, p < 0.001).

Table 4 contains HCBS subscale scores by political party. Supporting H2, Republicans indicated significantly higher denial and significantly lower sentencing and compassion beliefs compared to both Democrats and individuals of another political affiliation. Democrats indicated significantly lower denial beliefs and significantly higher sentencing and compassion beliefs compared to both Republicans and individuals of another political affiliation.

Table 4 Hate crime belief scale sub-scales by political affiliation
Full size table

Discussion

The current study examined the multi-factorial structure of the original HCBS (Cabeldue et al., 2018) in comparison to the multi-factorial structure of the shortened HCBS-UK (Bacon et al., 2021) among a community and college sample of US adults. The original four-factor structure of the HCBS demonstrated poor fit to the data. Lack of confirmatory analytic fit for the EFA-derived Cabeldue et al. HCBS factor structure may have resulted from several reasons. First, the nature of beliefs about hate crimes may have shifted in recent years in light of documented rises in hate crimes (e.g., Hodwitz & Massingale, 2021) and socio-political attention to racial injustice (e.g., Reny & Newman, 2021). Alternatively, the complexity of the original model reported by Cabeldue and colleagues (2018) may not hold; that is, the nature of beliefs about hate crimes may be simpler than originally conceived.

On the other hand, the three-factor model demonstrated good fit to the data. Shortened and refined HCBS among a U.S. sample offers further evidence for the possibility that hate crime-related attitudes cut across country borders. Importantly, our pool of items differed from the pool of items administered via the HCBS-UK (Bacon et al., 2021) which was adapted to include items pertinent to the UK population (e.g., focus on hate crimes towards Muslims). Moving forward, we propose a shortened 17-item version of the original HCBS with three factors consistent with the HCBS-UK: 1) denial, 2) sentencing, and 3) compassion. Appendix contains the Hate Crimes Beliefs Scale-Short Form (HCBS-SF). The first factor, denial, contains eight items reflecting an attitude which downplays the impact and nefariousness of hate crimes; this factor exclusively contains items from the negative beliefs subscale of the original HCBS. The second factor, sentencing, reflects one’s support for enhanced legal punishment of those who commit hate crimes. The third factor, compassion, contains six items which reflect greater caring for victims and desire to prevent hate crime victimization. We recommend adopting Bacon et al. (2021)’s language of Black in place of African American to allow for use beyond the United States and more generally we suggest thoughtful adaptations for stigmatized groups when the measure is adopted in other countries.

The initial adaptation of the HCBS-SF provides a reliable and efficient method to capturing attitudes about hate crimes. Such attitudes are particularly salient in light of documented increases of hateful acts (e.g., Hanes & Machin, 2014), especially toward certain minority groups in the COVID-19 era (e.g., Gray & Hansen, 2021). Possible next uses of the HCBS-SF include the following. Federal and non-profit programs exist that provide hate crime training for collegiate, law enforcement, and other persons interfacing with hate crime offenders and victims (e.g., Matthew Shepard Foundation, n.d.; U.S. Department of Justice, n.d.). To the extent training content is aligned with HCBS-SF subscales, our measure can be employed during pre-training attitude assessment and as a metric of program effectiveness. Also, there is a need to determine the extent to which attitudes about hate crimes may impact relevant behaviors. As such, the HCBS-SF may be implemented in a variety of social-cognitive and behavioral contexts such as voting choices, legal decision-making, and commission of hate-motivated behavior. To the degree hate crime-related attitudes affect these outcomes, hate crime attitudes may become a target of political campaigns, jury consulting, or offender rehabilitation efforts.

Our first hypothesis was supported in that the three HCBS factors were significantly related to SDO and RWA in the predicted direction. This replicates the findings of Bacon et al. (2021), except they did not find a significant relationship between RWA and the sentencing factor, while our results revealed a significant negative relationship. The current results also extend the construct validity of the HCBS first demonstrated by Cabeldue et al. (2018) through its relationship with various measures of prejudice. Patterns associated with RWA and SDO have been demonstrated in both student samples (e.g., Duckitt & Sibley, 2007; Duckitt et al., 2002; Thomsen et al., 2008) and community samples (e.g., Sibley & Duckitt, 2013; Sibley et al., 2006). Overall, the HCBS-SF operates as expected in terms of the relationship to measures for authoritarianism and social dominance, pointing to both the political/legal relevance of the dual process model of prejudice and need to continue validity studies of the HCBS regarding factors such as political and prejudiced behavior.

The second hypothesis was supported in that HCBS-SF subscale scores were in line with politically conservative versus liberal party affiliation. Our findings provide a quantitative extension of Cramer and colleagues (2017), who reported qualitative findings about hate crime related attitudes. Political conservatism was particularly linked to legal arguments against hate crime laws, whereas political liberalism was connected to negative views of hate crime perpetrators. We replicated these patterns with regard to the denial and sentencing subscales, respectively. The compassion subscale adds a further domain warranting examination in political and legal decision-making contexts in future research.

Overall, these findings are further support for the data presented by Malcom and colleagues (2022) that highlight the importance of prejudice, political affiliation, and attitudes on hate crime beliefs. The dual process model of prejudice formation (Gerstenfeld 2011) appears to continue informing our understanding of hate crime beliefs. Based on our data, some practical implications would be to consider the attitudes and political beliefs of individuals who draft hate crime related legislation. Inevitably, it will be important to have a diverse group of individuals working together to determine appropriate hate crime laws and legislation.

Our findings should be considered in the context of study limitations. To begin, the convenience sampling strategy limited generalizability. Although both students and community-dwelling adults were sampled to remain consistent with Bacon and colleagues (2021), our sample was quite limited with regard to geographic U.S. region, gender, and race/ethnicity. In light of the diverse array of victimization categories (e.g., by race, sexual orientation, disability status), future research should use the HCBS-SF with demographically diverse groups. Second, while we demonstrated convergent validity with cognitive measures related to socio-political attitudes, we did not examine how hate crime beliefs were related to either enacted hate crime violence or victimization history. Future research should examine how hate crime beliefs may differ among (1) those with and without a history of hate crime victimization and (2) those with and without a history of hate crime perpetration. Statistically, we employed modification indices in order to identify possible points of refinement in identifying the ideal HCBS factor structure. Use of MIs in measure development is debated (Bentler, 2010). In line with recommendations in the literature (e.g., Bentler, 2007) we recommend validation of the final HCBS model in additional samples in order to ensure the generalization of the reduced three-factor structure. Finally, future psychometric work is needed examining the factor structure of the shortened version of the HCBS with more diverse samples, including samples with more men and of older age. The demographics of the current sample differed from the demographics of typical hate crime perpetrators, representing a needed area for future research. Hate crime perpetrators are majority White (55.1%) and primarily male (National Consortium for the Study of Terrorism and Responses to Terrorism [START], 2020); U.S. Department of Justice, n.d.). Future studies with more diverse, larger samples would also allow for examination of measurement invariance across demographic groups, such as gender, race, religion, and sexual orientation.