![Sample our Social Sciences journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/110822/TOC-Subject-Sample-2021-Social-Sciences.jpg"})
Abstract
Prior research has begun the task of assessing the applicability of group threat theory to explain sexual orientation hate crime. Drawing upon this perspective, researchers have hypothesized that relatively large and growing gay populations might be perceived as a threat to heterosexual norms, leading to sexual orientation hate crimes. Previous work has also identified an additional source of threat that may be particularly salient: threat to traditional marriage. Using data for a sample of metropolitan areas, we assess the impact of a previously unexamined potential source of threat: same-sex marriage. We also compare the relationships between the relative size and growth of the gay population and hate crimes expressed as incidence and victimization rates. The results reveal that the effects of the gay population’s relative size on incidence vs. victimization rates differ in important respects, and that the level of same-sex marriage is positively associated with both hate crime rates.
Disclosure statement
No potential conflict of interest was by reported the authors.
Notes
1 The prior research on sexual orientation bias crime has referred to the minority population under investigation in various ways, sometimes using the term “gay” and other times using acronyms for a more comprehensive characterization of the population, e.g. LGB or LGBT (Mills, Citation2019; Stacey, Citation2018). In the present study, we use the terms "gay population" and "LGBT population" interchangeably simply as a linguistic convenience, while mindful of the distinctions.
2 When reviewing the literature on sexual orientation hate crime, we focus on the ecological research. As Stacey (Citation2018) and Mills (Citation2019) observed, the majority of research on sexual orientation has been conducted at the individual level, although there are a few exceptions. Green, Strolovitch, Wong, and Bailey (Citation2001) used population data from commercial mailing lists and tabulation of the 1990 Census data to estimate gay and lesbian population in the New York City. They found a positive relationship between the density of the gay and lesbian population and sexual orientation hate crime incidence rates, although they cautioned that their analyses for the lesbian population fared less well. Green, et al. (Citation1998a) examined the relationship between unemployment rates and anti-gay and anti-lesbian hate crimes for the New York City boroughs. They did not detect robust effects.
3 Piatkowska, Messner, and Hövermann, (Citation2019a) contrasted cultural/symbolic threat theory with an alternative perspective that implies an opposite prediction. To the extent that Black outgroup marriages with Whites signify intercultural accommodation in the population at large, there should be a negative relationship between Black outgroup marriages and anti-Black hate crime, as predicted by the intergroup contact hypothesis (Allport, Citation1954). As noted in the text, this hypothesis was disconfirmed.
4 We matched these codes using the 2018 Geographic Correspondence Engine (Missouri Census Data Center, Citation2018). Here, the special value “-99999” is given to places outside of the metro area. These places have been subsequently excluded from analysis. The link between MSA and county is also elucidated by the U.S. Census in Delineation Files. According to the Census Bureau, MSAs are comprised of one or more whole counties (or county equivalents) delineated under classification. Specifically, a county (or counties) is identified as a “central county” (counties) under the condition that at least 50 percent of the population reside within urban areas of 10,000 or more population, or contain at least 5000 people, who reside within a single urban area of 10,000 or more population. The condition for “outlying counties” to be included in the metro area is that they meet specified requirements of commuting to or from the central counties (Census Bureau, Citation2020; Mackun, Citation2009).
5 To address the non-recording issue, in supplementary analysis, we followed an approach adopted in previous studies (Piatkowska, Messner, & Hövermann, Citation2019a; Piatkowska, Messner, & Yang, Citation2019b) and made use of the subsample that consists of counties that reported at least one hate crime of any type, not necessarily crime motivated by sexual orientation, during the period under investigation. The reasoning underlying this approach was that “the absence of a record of a racial hate crime can more plausibly be interpreted as a ‘true zero’ if the law enforcement agency has compiled with reporting requirements as reflected in some kind of reported hate crime” (Piatkowska, Messner, & Yang, Citation2019b, p. 1074). Thus, in this subsample, we assigned a missing value if no hate crime of any type was recorded during the period under investigation, and zero if the recorded sexual orientation hate crime was zero and any other type of hate crime was recorded. The results for the analyses using this subsample, which serves as a robustness check, were substantively the same as those presented in this study. The results of these analyses are presented in the Online Appendix.
6 Hipp, Tita, and Boggess (Citation2009) demonstrated that using the targeted population in the computation of rates of intergroup crimes does not necessarily remove the “opportunity effects” because there is still a built-in relationship between the relative size of the targeted population and such rates. They proposed that one way to circumvent this dependency is to create a rate using as a denominator the number of interactions that could be expected by chance given the relative sizes of the groups of interest. The extent to which there are differences between using their denominator and the denominator in a conventional victimization rate depends upon the relative size of the groups. Given the very small estimated relative size of the gay population in our data, the two rates are empirically interchangeable, as reflected in a nearly perfect correlation, thereby yielding virtually identical results in the regression analyses.
7 Another source of information on the LGBT population is the Gallup Daily tracking survey provided by the Williams Institute (LGBT Demographic Data Interactive, Citation2019), which identifies self-reported LGBT respondents. The Gallup survey offers estimates of LGBT populations at the state level (as the percentage of the total population). To explore the validity of the measure of unmarried, same-sex households as an indicator of the relative size of the gay population, we computed this measure at the state level using ACS data and examined its correlation with the survey-based measure provided by Gallup. The two measures were highly correlated (r = 0.93), suggesting that the two indicators of the LGBT population measure the same concept.
8 As an indicator of a static gay population, Mills (Citation2019) used the percentage of same-sex couples from the 2000 census, whereas to assess the change in the gay population, she calculated the midpoint of 2005 using the 2000 decennial census and the 2008–2012 ACS and then subtracted the 2000 indicator from the 2005 midpoint. Stacey (Citation2018) used the percentage change of same-sex unmarried-partner households using data from the 2000 and 2010 decennial census. Notably, same-sex spouses in the 2000 and 2010 census and 2005–2013 ACS have been changed to same-sex unmarried partners (Census Bureau, Citation2019).
9 The ACS offers five major occupational categories: management, business, science, and arts occupations; service occupations; sales and office occupations; natural resources, construction, and maintenance occupations; and production, transportation, and material moving occupations. The measure of occupational sex segregation represents the percentage of either men or women who would have to change these occupational categories so that the distribution of men and women across these categories would be even.
10 To assess gender inequality, Alden and Parker (Citation2005) constructed an index from two measures: the ratio of male to female median income for persons age 16 and older and the ratio of male to female unemployment rate. In the present analysis, these two measures were only weakly correlated across all datasets (r= −0.07). Accordingly, we decided not to combine the measures in a single index but instead included the ratio of male to female median income only, although results for the variables of primary interest were highly similar when the ratio of male to female unemployment rate was incorporated into the analysis.
11 Preliminary analysis showed that approximately 17% of the sample yielded zero sexual orientation hate crimes. This percentage is smaller than that reported in Stacey’s (Citation2018) research based on county-level data (about 46%), which reflects the fact that sexual orientation hate crimes for our study have been aggregated to the corresponding MSAs.
12 The descriptive statistics for the victimization rates provide a somewhat misleading picture of the actual risks of victimization. Recall that, consistent with past research, our measure of the relative size of the gay population is the percentage of households with same-sex, unmarried couples, which is undoubtedly an underestimate of the actual size of this population. Consequently, the denominator for victimization rates (the estimated count of the gay population) is also an underestimate, thereby inflating the rates. It is not particularly important for the purposes of our analyses to devise accurate point estimates of the gay population, but rather to incorporate measures that capture the variation in this population across the units of analysis. As reported in Endnote 6 above, a comparison of our measure with the Gallup survey data indicates that this assumption is reasonable. To display the spatial distribution of incidence and victimization rates across MSAs, we provide the respective quintile maps for both rates in the Online Appendix (Figures 1.S and 2.S). The overall spatial patterns are highly similar: hate crime incidence and victimization rates are relatively high in the Northeast and Midwest, and particularly in the Southern and Western parts of the West. The overlap in spatial patterning of the two rates is not surprising given a high correlation between these two measures as reported in the Online Appendix (Table 1S r = 0.69).
13 We computed variance inflation factors (VIFs) for the regression equations. The highest VIF is 3.31, revealing that multicollinearity is not a problem (results available upon request).
14 We also considered divorce rates as a more commonly used indicator of distribution of marriage. The results showed that the divorce rates had no effect on our outcome variables. The effects of the remaining variables paralleled those presented in the manuscript (results available upon request).
15 The results of the sensitivity analyses discussed below are available upon request.
16 The six specific types of state legislation are: hate crime law, gay marriage, adoptions laws, anti-bullying statues, protections against discrimination in public accommodation, and employment. These legislations were coded as 0–1, where 1 indicates that the law is present.
Additional information
Notes on contributors
Sylwia J. Piatkowska
Sylwia J. Piatkowska is an assistant professor in the College of Criminology and Criminal Justice at Florida State University. Her areas of interest include hate crime, suicide and suicidal behavior, immigration and crime, and both international and comparative criminology.
Steven F. Messner
Steven F. Messner is Distinguished Teaching Professor of Sociology at the University at Albany, State University of New York. His research focuses on social institutions and crime, understanding spatial and temporal patterns of crime, and crime and social control in China.