Abstract
Objectives. To estimate US nonlethal violent victimization rates for lesbian, gay, and bisexual (LGB) males and females aged 16 years and older and to compare disparities among LGB and straight males and females, controlling for other correlates of victimization.
Methods. We used data from the 2017 and 2018 National Crime Victimization Survey (NCVS) to provide nationally representative rates of various forms of violent victimization for self-identified LGB and straight persons. Multivariable models assessed the risk for violence associated with LGB status.
Results. Total violence rates were 2 to 9 times higher among LGB persons compared with heterosexuals. For some forms of violence (e.g., rape and sexual assault, violence with serious injuries, and multiple offender violence) there were notably high disparities between bisexuals and heterosexuals. With adjustment for covariates, LGB orientation was associated with odds ratios nearly 2 to 4 times those of heterosexuals.
Conclusions. This is one of the first known uses of NCVS data to estimate LGB victimization, revealing substantially higher rates of violence directed at LGB individuals.
Public Health Implications. Sexual orientation and gender identity questions in federal surveys such as the NCVS enable monitoring of violent victimization rates and should continue. Collecting these data can help researchers understand victimization risk and guide appropriate resources toward victim services, especially important given the high violent crime levels experienced by LGB individuals.
In recent years, scholars have recognized violence and crime against sexual- and gender-minority groups as a major public health priority in the United States.1 These diverse groups, comprising varied sexual orientations (e.g., lesbian, gay, bi- or pansexual, and others) and gender identities (e.g., transgender, nonbinary, gender variant, queer, and others), represent as much as 5.4% and 0.3% of the US population, respectively.2,3 Unfortunately, both groups appear to be at high risk of violent victimization relative to the general population. Community-based samples suggest that prevalence of sexual assault,4 intimate partner violence,5 property crime,6 and hate crimes6,7 among lesbian, gay, bisexual, and transgender (LGBT) individuals are nearly double those of heterosexuals. Within lesbian, gay, and bisexual (LGB) populations specifically, lesbian and bisexual women are more likely than gay and bisexual men to report lifetime sexual assault, including childhood and intimate partner sexual assault.8
For purposes of monitoring and improving our understanding of the victimization risk of LGBT persons, existing research faces several challenges. Many studies are based on community-based samples, or samples of youths or college students, leaving the generalizability of the findings to the US population uncertain. In addition, the focus of much past research is limited to certain forms of violence, such as sexual assault, intimate partner violence,9 and hate crimes,6,7 while less is known about how often other forms of violence are experienced by LGBT individuals. A notable exception involves a meta-analysis of 386 studies published between 1992 and 2009 that revealed high rates of many types of victimization among LGB persons versus heterosexuals, especially discrimination (effect size [ES]p = 0.41) and verbal harassment (ESp = 0.55).10 Yet many of the studies included were drawn from relatively small community-based samples, and, importantly, nationally representative data covering the years since the 2015 Supreme Court ruling extending marriage rights to LGBT couples in all 50 states are rare.
Recently, the National Crime Victimization Survey (NCVS), the nation’s primary source of information on nonfatal violent victimization, began asking questions about respondents’ sexual orientation and gender identity (SOGI),11 which were added to the survey in July 2016. Although other federal statistical surveys, such as the Behavioral Risk Factor Surveillance System (BRFSS) and National Intimate Partner and Sexual Violence Survey (NISVS) also include SOGI questions, only the NCVS contains detailed measures of multiple types of violent victimization, enabling estimations of victimization rates among LGB persons aged 16 years and older and comparisons of their risks to those of heterosexuals. While the NCVS data collection also includes gender identity questions, the sample size for the transgender population is insufficient at this time for producing reliable victimization rates.
This article describes nonlethal violent victimization rates for LGB populations and compares them with the rates for “straight” populations, hereafter referred to as heterosexual. The NCVS data used here were collected in 2017 and 2018 and provide one of the first known uses of these data to report estimates of nonfatal crime victimization against the LGB population, stratified by sex (as reported by the household respondent before all interviews began) and sexual orientation.
METHODS
The NCVS is an ongoing survey conducted by the Census Bureau for the Bureau of Justice Statistics, with a sample designed to be representative of persons aged 12 years and older living in households in the United States. The data enable estimation of the rate of different types of nonfatal violence over time. SOGI questions are administered to persons aged 16 years and older. NCVS data for 2017 and 2018 were made publically available in March 2020 through the National Archive of Criminal Justice Data.12
The NCVS collects self-report data about individuals’ experiences with violence using a 6-month recall period designed to minimize errors associated with determining when an event occurred. Unlike many surveys of LGBT experiences, the data do not include lifetime estimates of victimization, instead focusing on recent occurrences. Therefore, the rates shown here reflect violence reported in 2017 and 2018. Our analysis was based on interviews with persons aged 16 years and older, resulting in a total unweighted sample size of 463 674 interviews. This large sample size is necessary for producing statistically reliable rates, particularly among smaller population subgroups and subcategories of violence. Response rates for the NCVS are high: in 2017, 76% of sampled NCVS households completed an interview, with a response rate among persons in these households of 84%.9 In 2018, the comparable household and person response rates were 73% and 82%, respectively.13
Victimization is measured through a series of cues and common-language questions, and numerous characteristics of incidents are used to code the event into crime types following the interview. Interviews are conducted in English and Spanish, as well as other languages under some circumstances. The NCVS data also include methodological information about each interview that previous research has shown affects survey reporting,14 such as the bounding of interviews with information from a previous interview, the type of interview (e.g., in-person, telephone), and the number of previous interviews. We took these conditions into account in our multivariable analysis.
Our definitions of violent crime paralleled those used by the Bureau of Justice Statistics in their annual reports on criminal victimization.13,14 Violent crime includes attempted or completed rape and sexual assault, robbery, aggravated assault, and simple assault. Aggravated assault includes attacks or attempted attacks with a weapon, and attacks without a weapon that result in serious bodily injury (e.g., broken bones, internal injuries, loss of consciousness). Simple assault includes attacks or attempted attacks without a weapon that result in minor injury (e.g., bruises) or no injury (e.g., attempts without physical contact). Serious violent crime rates include all violence other than simple assaults, and physically injurious violence is events resulting in physical injuries to the victim.
Victim–offender relationship information is also gathered for victimization events. Incidents in which the victim reported no previous relationship with the offender are coded as stranger violence, while those involving a spouse, ex-spouse, or current or former boyfriend or girlfriend are coded as intimate partner violence. Victimization involving other persons known to the victim (such as friends, acquaintances, and family members) are coded as violence involving other known persons. Rates of multiple offender victimization are also estimated. In incidents involving multiple offenders, the victim–offender relationship is coded according to the most familiar relationship to the victim.
Sexual orientation (and gender identity) items appear at the end of the base screening questionnaire (see Truman et al.,11 for further details). The item asks “Which of the following best represents how you think of yourself?” “[Lesbian or] gay,” “Straight, that is, not [lesbian or] gay,” “Bisexual,” “Something else,” or “I don’t know the answer.” The phrase “lesbian or” is read to the respondent only if they are designated as female on the household roster. We coded persons as lesbian, gay, or bisexual if they responded accordingly, and as heterosexual if they respond “straight, that is, not [lesbian or] gay.” Persons selecting other options or refusing to answer the question were excluded from the analysis. Such cases constituted 9.5% of the total weighted NCVS sample, with refusals constituting the majority of these responses (78.3%).
Victimization rates presented in this article were calculated by using the methodology followed by the Bureau of Justice Statistics.13,14 Rates represent the number of victimizations that occurred during the recall period among persons in each group, divided by the number of persons in each group and multiplied by 1000. Sample weights and design features were used to take into account the multistage complex sample design, which involves stratification (e.g., by location) and clustering (e.g., household and repeated interviews), as well as nonresponse. We generated the weighted sample-based rates by using SPSS version 26 (IBM, Somers, NY), and we estimated standard errors and confidence intervals by using generalized variance function parameters.15 Preliminary analysis determined that it was necessary to pool the NCVS 2017 and 2018 data to provide reliable estimates (with coefficients of variation less than 50%) of the different types of violent victimization among the subgroups.
To assess the extent to which differences in violent victimization between LGB and heterosexual males and females were associated with subgroup differences in sociodemographic characteristics known to be correlated with risk, we estimated survey-weighted logistic regression models in which the outcome consists of whether the respondent was violently victimized (1 = yes; 0 = no) during the recall period. These models included NCVS measures of the respondents’ age, race, ethnicity, sex, and household income, as well as general information about their residential areas (i.e., metropolitan city, surrounding areas, and nonmetropolitan areas). Even with the large pooled sample size, it was necessary to limit the multivariable logistic regression analysis to assessments of total violent victimization, and the subcategory of serious violent victimization to minimize model fitting restrictions associated with sample and cell sizes. We estimated the survey-weighted logistic regression models with Stata version 15 (StataCorp LP, College Station, TX) and also took into account the complex survey design, clustered interviews, and sample weights.
RESULTS
Descriptive statistics for respondents by sexual orientation are provided in Table 1. When the data were weighted and cases with unknown LGB status were excluded, estimates of the gay or lesbian, bisexual, and heterosexual populations were 1.4%, 0.7%, and 97.9%, respectively. The NCVS estimates of the gay or lesbian and bisexual populations were slightly lower than those found in the National Health Interview Survey (1.6% gay or lesbian, and 0.8% bisexual), and the proportion of adults who identified as bisexual were slightly higher in the National Survey on Drug Use and Health and in the National Survey of Family Growth.11 However, the differences in the estimates across these data sources were not large in magnitude and provide reasonable confidence in the external validity of the NCVS for purposes of estimating victimization rates.
TABLE 1—
Descriptive Statistics for Sociodemographic Characteristics by Sexual Orientation: Weighted National Crime Victimization Survey Data: United States, 2017–2018
| Gay or Lesbian (n = 5380),a % (95% CI) | Bisexual (n = 2585),a % (95% CI) | Heterosexual (n = 418 003),a % (95% CI) | |
| Age, y | |||
| 16–24 | 17.1 (15.3, 19.1) | 49.1 (46.3, 51.9) | 15.7 (15.4, 16.1) |
| 25–39 | 32.1 (29.8, 34.5) | 30.6 (28.0, 33.4) | 23.8 (23.5, 24.2) |
| 40–54 | 26.7 (24.5, 29.1) | 12.6 (10.9, 14.5) | 24.1 (23.8, 24.4) |
| ≥ 55 | 24.1 (22.1, 26.1) | 7.7 (6.4, 9.1) | 36.3 (35.8, 36.8) |
| Sex | |||
| Male | 42.9 (40.7, 45.2) | 75.0 (72.0, 77.8) | 51.8 (51.6, 52.1) |
| Female | 57.1 (54.8, 59.3) | 25.0 (22.2, 28.0) | 48.2 (47.9, 48.4) |
| Race/ethnicity | |||
| Non-Hispanic White | 68.8 (66.4, 71.1) | 64.9 (61.3, 68.4) | 63.4 (62.5, 64.3) |
| Non-Hispanic Black | 9.9 (8.6, 11.5) | 10.4 (8.6, 12.50) | 12.0 (11.3, 12.6) |
| Hispanic | 15.0 (13.3, 17.0) | 15.5 (13.1, 18.2) | 16.4 (15.8, 17.0) |
| Non-Hispanic other race | 6.1 (5.0, 7.5) | 9.1 (7.3, 11.3) | 8.1 (7.8, 8.5) |
| Household income, $ | |||
| < 25 000 | 15.1 (13.4, 17.0) | 26.5 (23.9, 29.2) | 13.7 (13.2, 14.2) |
| 25 000–49 999 | 17.3 (15.5, 19.2) | 22.3 (19.9, 25.0) | 19.2 (18.7, 19.7) |
| 50 000–74 999 | 15.3 (13.5, 17.2) | 13.3 (11.3, 15.6) | 13.9 (13.6, 14.2) |
| ≥ 75 000 | 36.3 (34.0, 38.7) | 25.3 (22.7, 28.1) | 28.3 (27.8, 28.9) |
| Unknown | 16.1 (14.3, 18.0) | 12.6 (10.7, 14.8) | 25.0 (24.1, 25.9) |
| Residential area type | |||
| Metropolitan central city | 48.7 (45.7, 51.8) | 45.1 (41.4, 48.9) | 33.1 (31.7, 34.5) |
| Surrounding area | 43.1 (40.1, 46.2) | 45.0 (41.5, 48.6) | 52.5 (50.8, 54.1) |
| Nonmetropolitan area | 8.2 (6.3, 10.6) | 9.9 (6.9, 13.8) | 14.4 (12.1, 17.1) |
| Population percentagea | 1.37 (1.30, 1.46) | 0.69 (0.64, 0.75) | 97.93 (97.83, 98.03) |
Note. CI = confidence interval.
Population percentage and sample size based on persons aged 16 years and older and coded nonmissing on sexual orientation.
The descriptive statistics revealed notable sociodemographic differences across the populations. Those who identified as bisexual were more likely to be younger, female, and in the lowest income category than were others. Persons who identified as gay or lesbian were more likely to report income in the $75 000 and above category than were bisexual and heterosexual persons. Heterosexuals were more likely to be aged 55 years and older, and less likely to live in the central cities of metropolitan areas than were others. We observed no significant differences in racial and ethnic composition across the 3 groups.
The 2017–2018 violent victimization rates for male and female LGB and heterosexual populations are shown in Table 2. Among males, total violent victimization rates were significantly higher among gay (38.6 per 1000; 95% confidence interval [CI] = 26.3, 50.9) and bisexual (76.7 per 1000; 95% CI = 43.6, 109.8) men compared with heterosexual men (19.4 per 1000; 95% CI = 17.4, 21.4). We also observed significant differences in the category of serious violent crime in which the rates among gay (19.2 per 1000; 95% CI = 10.9, 27.5) and bisexual (41.6 per 1000; 95% CI = 17.7, 65.5) males were higher than those of heterosexual males (6.8 per 1000; 95% CI = 5.8, 7.8). Although gay and bisexual males also exhibited higher rates of subcategories of violent victimization, only some of the differences in these rates were statistically significant because of the comparatively small sample sizes of the male and female LGB population and the lower frequency of the subcategories of violence. For example, rates of robbery among gay and bisexual males were between 2.8 and 5.6 times higher, respectively, than those of heterosexual males, yet these differences were not statistically significant at a P value of less than .05. The NCVS data revealed statistically significant differences in violence committed by other known persons (i.e., not strangers or intimate partners) in which the rates were higher among gay (17.9 per 1000; 95% CI = 9.9, 25.8) and bisexual (40.3 per 1000; 95% CI = 16.8, 63.7) males compared with heterosexual males (5.7 per 1000; 95% CI = 4.8, 6.6).
TABLE 2—
Violent Victimization Rates, by Type of Crime, Sex, and Sexual Orientation: Weighted National Crime Victimization Survey Data, United States, 2017–2018
| Gay or Lesbian |
Bisexual |
||||
| Rate (95% CI) | Ratio | Rate (95% CI) | Ratio | Heterosexual, Rate (95% CI) | |
| Male | |||||
| Type of violent victimization | |||||
| Total violence | 38.6* (26.3, 50.9) | 2.0 | 76.7* (43.6, 109.8) | 4.0 | 19.4 (17.4, 21.4) |
| Rape or sexual assault | 6.3* (1.9, 10.7) | 19.2 | 7.5a (0, 17.1) | 22.9 | 0.3 (0.2, 0.5) |
| Robbery | 6.8 (2.2, 11.3) | 2.8 | 13.4 (0.4, 26.4) | 5.6 | 2.4 (1.9, 2.9) |
| Aggravated assault | 6.1 (1.8, 10.5) | 1.5 | 20.7* (4.3, 37.0) | 5.1 | 4.1 (3.4, 4.8) |
| Simple assault | 19.4 (11.1, 27.7) | 1.5 | 35.1* (13.3, 56.9) | 2.8 | 12.6 (11.1, 14.1) |
| Serious violence | 19.2* (10.9, 27.5) | 2.8 | 41.6* (17.7, 65.5) | 6.1 | 6.8 (5.8, 7.8) |
| Characteristics of victimization | |||||
| Intimate partner violence | 2.2a (0, 4.7) | 4.1 | 9.0a (0, 19.5) | 16.4 | 0.5 (0.4, 0.7) |
| Other known persons | 17.9* (9.9, 25.8) | 3.1 | 40.3* (16.8, 63.7) | 7.0 | 5.7 (4.8, 6.6) |
| Stranger violence | 15.4 (8.1, 22.7) | 1.6 | 22.8 (5.5, 40.1) | 2.4 | 9.7 (8.4, 10.9) |
| Violent crime with injury | 16.5* (8.9, 24.0) | 4.9 | 12.9 (0.2, 25.6) | 3.9 | 3.3 (2.7, 4.0) |
| Multiple offender incidents | 4.6 (0.9, 8.3) | 1.6 | 14.3 (0.8, 27.7) | 4.9 | 2.9 (2.3, 3.5) |
| Female | |||||
| Type of violent victimization | |||||
| Total violence | 78.4* (57.9, 98.9) | 3.7 | 189.1* (155.0, 223.3) | 9.0 | 21.0 (18.9, 23.1) |
| Rape or sexual assault | 3.8a (0, 7.6) | 1.2 | 40.9* (25.9, 55.9) | 12.6 | 3.2 (2.6, 3.8) |
| Robbery | 0.8a (0, 2.4) | 0.4 | 24.9* (13.6, 36.3) | 13.7 | 1.8 (1.4, 2.2) |
| Aggravated assault | 5.0 (0.6, 9.3) | 1.6 | 21.7* (11.2, 32.3) | 6.9 | 3.1 (2.6, 3.7) |
| Simple assault | 68.9* (49.8, 88.0) | 5.4 | 101.6* (76.8, 126.4) | 7.9 | 12.8 (11.3, 14.3) |
| Serious violence | 9.6 (3.3, 15.8) | 1.2 | 87.5* (64.6, 110.4) | 10.7 | 8.2 (7.1, 9.3) |
| Characteristics of victimization | |||||
| Intimate partner violence | 6.1 (1.2, 11.0) | 1.2 | 38.7* (24.2, 53.3) | 7.6 | 5.1 (4.3, 5.9) |
| Other known persons | 22.0* (12.0, 32.0) | 3.0 | 50.8* (33.9, 67.8) | 7.0 | 7.3 (6.3, 8.3) |
| Stranger violence | 48.9* (33.2, 64.7) | 8.8 | 87.5* (64.6, 110.4) | 15.8 | 5.5 (4.7, 6.4) |
| Violent crime with injury | 7.8 (2.2, 13.4) | 1.3 | 41.4* (26.3, 56.5) | 6.9 | 6.0 (5.1, 6.9) |
| Multiple offender incidents | 17.0* (8.3, 25.6) | 9.0 | 45.5* (29.6, 61.5) | 24.2 | 1.9 (1.5, 2.3) |
Note. CI = confidence interval.
Male rate is significantly different from heterosexual male rate at P < .05. Adjacent ratio represents rate divided by heterosexual male rate. Female rate is significantly different from heterosexual female rate at P < .05. Adjacent ratio represents rate divided by heterosexual female rate.
Rate and ratio should be interpreted with caution: CI includes zero and the coefficient of variation is greater than 50%.
Comparisons of rates for lesbian, bisexual, and heterosexual females indicated that total violence rates were significantly higher among lesbians (78.4 per 1000; 95% CI = 57.9, 98.9) compared with heterosexual women (21.0 per 1000; 95% CI = 18.9, 23.1), and were notably the highest among bisexual females (189.1 per 1000; 95% CI = 155.0, 223.3). Bisexual females also experienced the highest rate of serious violent victimization (87.5 per 1000; 95% CI = 64.6, 110.4); however, differences in serious violence were not statistically significant between lesbian (9.6 per 1000; 95% CI = 3.3, 15.8) and heterosexual females (8.2 per 1000; 95% CI = 7.1, 9.3). Stranger violence rates were significantly higher among lesbian (48.9 per 1000; 95% CI = 33.2, 64.7) and bisexual females (87.5 per 1000; 95% CI = 64.6, 110.4) compared with heterosexual females (5.5 per 1000; 95% CI = 4.7, 6.4). So, too, were violence rates involving multiple offenders, which was 1.9 per 1000 (95% CI = 1.5, 2.3) among heterosexual females, but 17.0 (95% CI = 8.3. 25.6) and 45.5 (95% CI = 29.6, 61.5) per 1000 among lesbian and bisexual females, respectively.
For each type of violence shown here, bisexual women were found to have significantly higher rates of victimization than heterosexual females, and also higher rates than lesbian females for all crime types except simple assault. Because the bisexual population is significantly younger and of lower income than the other populations, it is necessary to determine the extent to which these differences persist when such correlates are taken into account.16
The multivariable model predicting total violent victimization showed that the differences between heterosexual and LGB persons were statistically significant after we controlled for age, race/ethnicity, household income, residential area type, interview conditions, and missing data on the SOGI questions (Table 3). Compared with heterosexual females, odds of experiencing a violent victimization in 2017 or 2018 were 2.0 times higher among lesbian females (95% CI = 1.34, 2.99) and 3.61 times higher among bisexual females (95% CI = 2.78, 4.68). The patterns in the odds ratios (ORs) among males were generally similar to those among females. Odds of violent victimization were 1.9 times higher among gay males (95% CI = 1.34, 2.69) and 2.66 times higher among bisexual males (95% CI = 1.65, 4.30), with heterosexual males and females exhibiting similar odds of a violent victimization. Although the ORs for violent victimization were higher among the LGB population, the 95% CIs show that differences among bisexual, gay, and lesbian persons were not statistically significant when these additional factors were taken into account. The likelihood ratio (LR) test showed that inclusion of the male and female sexual orientation in the analysis significantly improved model fit (LR test statistic = 75.5; P < .001).
TABLE 3—
Survey-Weighted Logistic Regression Analysis of Victim Characteristics on Violent Victimization: National Crime Victimization Survey Data, United States, 2017–2018
| Total Violent Victimization, OR (95% CI) | Serious Violent Victimization, OR (95% CI) | |
| Sexual orientation by sex | ||
| Heterosexual female (Ref) | 1 | 1 |
| Heterosexual male | 0.98 (0.90, 1.08) | 0.97 (0.84, 1.11) |
| Lesbian female | 2.00 (1.34, 2.99) | 1.37 (0.76, 2.45) |
| Gay male | 1.90 (1.34, 2.69) | 2.41 (1.58, 3.65) |
| Bisexual female | 3.61 (2.78, 4.68) | 4.64 (3.32, 6.51) |
| Bisexual male | 2.66 (1.65, 4.30) | 3.84 (2.10, 7.03) |
| Age, y | ||
| 16–24 | 2.05 (1.79, 2.33) | 2.71 (2.25, 3.26) |
| 25–39 | 2.00 (1.80, 2.22) | 2.34 (1.98, 2.77) |
| 40–54 | 1.79 (1.62, 1.98) | 2.02 (1.68, 2.44) |
| ≥ 55 (Ref) | 1 | 1 |
| Race/ethnicity | ||
| Non-Hispanic White (Ref) | 1 | 1 |
| Non-Hispanic Black | 0.77 (0.66, 0.89) | 0.89 (0.73, 1.08) |
| Hispanic | 0.72 (0.62, 0.83) | 0.86 (0.71, 1.05) |
| Non-Hispanic other race | 0.83 (0.72, 0.97) | 1.00 (0.81, 1.24) |
| Household income, $ | ||
| < 25 000 | 2.67 (2.34, 3.06) | 3.40 (2.80, 4.13) |
| 25 000–49 999 | 1.68 (1.47, 1.92) | 1.94 (1.59, 2.37) |
| 50 000–74 999 | 1.20 (1.02, 1.41) | 1.17 (0.93, 1.48) |
| ≥ 75 000 (Ref) | 1 | 1 |
| Unknown | 0.83 (0.71, 0.97) | 1.06 (0.83, 1.33) |
| Residential area type | ||
| Metropolitan central city | 1 | 1 |
| Surrounding area | 0.74 (0.67, 0.82) | 0.71 (0.64, 0.80) |
| Nonmetropolitan | 0.75 (0.60, 0.93) | 0.77 (0.60, 1.00) |
Note. CI = confidence interval; OR = odds ratio. Outcome is whether any violent victimization occurred during the 6-month recall period based on pooled 2017–2018 National Crime Victimization Survey data. Models controlled for interview bounding, type of interview, number of previous interviews, and missing responses on sexual orientation (results not shown).
Odds of serious violent victimization were 4.64 times higher for bisexual females (95% CI = 3.32, 6.51) compared with heterosexual females, and the difference between lesbian and heterosexual females was not statistically significant once the correlates were included. Odds of serious violence were 2.41 times higher among gay males (95% CI = 1.58, 3.65) and 3.84 times higher among bisexual males (95% CI = 2.10, 7.03), with heterosexual males and females exhibiting similar likelihoods of serious violence. Similar to total violence, ORs for bisexual, gay, and lesbian persons did not differ significantly from one another in the analysis of serious violence, and model fit was significantly improved when sexual orientation was included (LR test statistic = 63.7; P < .001).
We conducted supplementary analyses to consider interactions between race/ethnicity and sexual orientation, but the results were found to be statistically unreliable given the further sample restrictions.
DISCUSSION
Our results indicate that males and females identifying as LGB experienced violent victimization in 2017 to 2018 at significantly higher rates than heterosexual males and females. Rates of serious violence were significantly higher against gay and bisexual males versus heterosexual males, and higher for bisexual females compared with heterosexual females. The multivariable analysis showed that differences in total violence between LGB and heterosexual males and females remained significant when sociodemographic correlates of victimization were taken into account. Similar patterns emerged when we restricted the outcome to serious violence; however, in this analysis, the OR for lesbian females was not significantly different from that of heterosexual females.
Many of these findings beg further inquiry. For example, some of our findings concerning specific forms of violence, such as stranger violence, showed large differences between heterosexual females compared with lesbian and bisexual females whose rates were more than 8 and 15 times higher, respectively. Generally speaking, stranger violence risk is hypothesized to be associated with residential proximity to violent crime areas and variations in routine activities (e.g., work, leisure) that increase exposure to potential offenders.17 The NCVS data do not provide detailed items that would permit assessment of the role of each of these factors. However, it is reasonable to hypothesize that some of the differences in stranger violence that we found would persist even if such factors were taken into account because of the high levels of heterosexist violence (e.g., verbal harassment, being followed) found in previous studies of LGB individuals.10 As future years of NCVS data become available, it will be possible to assess additional features of stranger violence to learn more about the contexts of such incidents.
Furthermore, community and convenience samples of LGB victimization have consistently revealed higher reported rates of violence across the life course, including bullying18 and dating violence19 in adolescence, sexual assault5 and intimate partner violence in adulthood,4,9 sexual orientation bias or hate crimes,6,7,20 and the potential for increased risk of elder abuse.21 Additional years of data will enable a determination of whether these patterns hold true in a diverse, nationally representative population. More data can illuminate how LGB persons may be differentially impacted by violent victimization in different phases of the life span, as early life adversities, reported at consistently higher rates among LGB individuals compared with heterosexuals,22 are associated with increased risk for various types of victimization in adulthood.23
Limitations
Our study represents the first report, to our knowledge, of LGB victimization for years 2017 to 2018 using NCVS data; however, it is not without limitations. One restraint was the limited statistical power to estimate victimization rates for transgender individuals even with a large pooled sample. Although the NCVS includes gender identity questions, we made the decision to focus on sexual orientation only rather than conflating sexual orientation with gender identity, in keeping with expert recommendations by scholars of sexual and gender minority populations.24 The sample size of LGB groups would not permit examination of potential intersectional differences according to race, ethnicity, and sexual orientation. Because the NCVS relies on a household-based sample, interviews were not conducted with persons who were homeless and those in institutional settings such as prisons or jails or nursing homes, and it is unknown how the exclusion of these persons might affect victimization estimates for either LGB or straight persons.
Although the NCVS data permit nationally representative estimates of subgroup rates for many types of violent victimization and do so using a consistent methodology over time, these data are subject to concerns that affect all survey data, such as recall error and underreporting. Potential sources of error in the NCVS have been studied extensively25; however, we are unaware of any research that has examined whether survey reporting error in victimization data varies across LGB and heterosexual male and female samples. Others have noted that the NCVS produces estimates of rape, sexual assault, and intimate partner violence that are lower than those found in other surveys that use different methodological approaches.26 How this might influence the disparities that we reported for these types of victimization is unknown. To our knowledge, the possibility that underreporting of these forms of violence may vary across male and female LGB and heterosexual groups has not been assessed in the NCVS.
Conclusions
In contrast with other probability-based surveys such as the BRFSS and the NISVS, the NCVS permits the examination of multiple forms of violent victimization with detailed measures that more fully estimate the burden of violence among the LGB population. Our findings reveal that LGB persons in the United States experienced significantly higher rates of total violence than heterosexuals in 2017 to 2018, and that, within the LGB population, bisexual women experienced violence at the highest rates. In multivariable models, greater odds of violent victimization were associated with LGB identity, younger age, lower household income, and metropolitan area city residence. While additional research is indicated, what our findings most underscore is the urgent need for multisector violence prevention programs; victim services that are affirming, inclusive, and culturally tailored to various LGB groups; and widespread social norms changes to eliminate discrimination, prejudice, and violence on the basis of sexual orientation.
Public Health Implications
Nationally representative surveys with sufficient sample sizes to estimate risk among SOGI populations are necessary to further investigate and continue monitoring the burden of violent victimization and need for victim support among these groups. As of this writing, the future of SOGI questions among those aged 16 and 17 years is in doubt, and multiple researchers have submitted a petition to the Bureau of Justice Statistics advocating continued inclusion of questions to ascertain sexual orientation and gender identity for youths27—an age group, which, among LGB individuals, reports high rates of violent victimization, including assaults and school bullying,28 compared with heterosexual youths. Our findings emphasize the ongoing need for SOGI questions in the NCVS to estimate victimization risk in all LGB populations, including those younger than 18 years.
CONFLICTS OF INTEREST
The authors have no conflict of interest to disclose.
HUMAN PARTICIPANT PROTECTION
Institutional review board approval was exempt as this was secondary analysis of an existing, public-use data set (available at https://www.icpsr.umich.edu/web/NACJD/studies/37689), which has been de-identified. Neither author had or has access to identifying information or could readily access identifying information for any of the respondents in the survey.
Footnotes
See also p. XXX.
REFERENCES
- 1.Institute of Medicine Committee on Lesbian. The Health of Lesbian, Gay, Bisexual and Transgender People. Washington, DC: National Academies Press; 2011. Gay, Bisexual and Transgender Health Issues and Research Gaps and Opportunities. [PubMed] [Google Scholar]
- 2.Gates GJ. LGBT data collection amid social and demographic shifts of the US LGBT community. Am J Public Health. 2017;107(8):1220–1222. doi: 10.2105/AJPH.2017.303927. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Gates GJ. How many people are lesbian, gay, bisexual and transgender? University of California–Los Angeles School of Law, Williams Institute. 2011. Available at: https://escholarship.org/uc/item/09h684X2. Accessed April 8, 2020.
- 4.Langenderfer-Magruder L, Walls NE, Kattari SK, Whitfield DL, Ramos D. Sexual victimization and subsequent police reporting by gender identity among lesbian, gay, bisexual, transgender, and queer adults. Violence Vict. 2016;31(2):320–331. doi: 10.1891/0886-6708.VV-D-14-00082. [DOI] [PubMed] [Google Scholar]
- 5.Krebs C, Lindquist C, Berzofsky M . Campus climate survey validation study: final technical report. Washington, DC: Bureau of Justice Statistics, Office of Justice Programs; 2016. [Google Scholar]
- 6.Herek GM. Hate crimes and stigma-related experiences among sexual minority adults in the United States: prevalence estimates from a national probability sample. J Interpers Violence. 2009;24(1):54–74. doi: 10.1177/0886260508316477. [DOI] [PubMed] [Google Scholar]
- 7.Coston L. Understanding characteristics of victims and perpetrators of anti-LGBT hate crimes in the United States. Violence Vict. 2018;33(3):453–471. doi: 10.1891/0886-6708.v33.i3.453. [DOI] [PubMed] [Google Scholar]
- 8.Rothman EF, Exner D, Baughman AL. The prevalence of sexual assault against people who identify as gay, lesbian, or bisexual in the United States: a systematic review. Trauma Violence Abuse. 2011;12(2):55–66. doi: 10.1177/1524838010390707. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. National Intimate Partner & Sexual Violence Survey. An overview of 2010 findings on victimization by sexual orientation. Atlanta, GA: National Center for Injury Prevention and Control, Centers for Disease Control and Prevention, 2011.
- 10.Katz-Wise SL, Hyde JS. Victimization experiences of lesbian, gay, and bisexual individuals: a meta-analysis. J Sex Res. 2012;49(2-3):142–167. doi: 10.1080/00224499.2011.637247. [DOI] [PubMed] [Google Scholar]
- 11.Truman JL, Morgan RE, Gilbert T, Vaghela P. Measuring sexual orientation and gender identity in the National Crime Victimization Survey. J Off Stat. 2019;35(4):835–858. doi: 10.2478/jos-2019-0035. [DOI] [Google Scholar]
- 12.Bureau of Justice Statistics. National Crime Victimization Survey, concatenated file, 1992–2018. ICPSR 37322. 2020. Available at: https://www.icpsr.umich.edu/web/NACJD/studies/37689. Accessed March 9, 2020.
- 13.Morgan RE, Truman JL. Criminal victimization. Washington, DC: Bureau of Justice Statistics; 2017. [Google Scholar]
- 14.Morgan RE, Oudekerk BA. Criminal victimization. Washington, DC: Bureau of Justice Statistics; 2018. [Google Scholar]
- 15.Williams RL, Heller D, Couzens G . Evaluation of direct variance estimation, estimate reliability, and confidence intervals for the National Crime Victimization Survey. Washington, DC: Bureau of Justice Statistics, Office of Justice Programs; 2015. [Google Scholar]
- 16.Lauritsen JL, Rezey ML. Victimization trends and correlates: macro- and micro-influences and new directions for research. Annu Rev Criminol. 2018;1(1):103–121. doi: 10.1146/annurev-criminol-032317-092202. [DOI] [Google Scholar]
- 17.Xie M, Heimer K, Lauritsen J. Violence against women in US metropolitan areas: changes in women’s status and risk, 1980–2004. Criminology. 2012;50:961–992. doi: 10.1111/j.1745-9125.2012.00284.x. [DOI] [Google Scholar]
- 18.O’Malley Olsen E, Vivolo-Kantor AM, Kann L, Milligan CN. Trends in school-related victimization of lesbian, gay, and bisexual youths—Massachusetts, 1995–2015. Am J Public Health. 2017;107(7):1116–1118. doi: 10.1111/j.1745-9125.2011.00245.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Dank M, Lachman P, Zweig JM, Yahner J. Dating violence experiences of lesbian, gay, bisexual, and transgender youth. J Youth Adolesc. 2014;43(5):846–857. doi: 10.1007/s10964-013-9975-8. [DOI] [PubMed] [Google Scholar]
- 20.Briones-Robinson R, Powers RA, Socia KM. Sexual orientation bias crimes: examination of reporting, perception of police bias, and differential police response. Crim Justice Behav. 2016;43(12):1688–1709. doi: 10.1177/0093854816660583. [DOI] [Google Scholar]
- 21.Westwood S. Abuse and older lesbian, gay bisexual, and trans (LGBT) people: a commentary and research agenda. J Elder Abuse Negl. 2019;31(2):97–114. doi: 10.1080/08946566.2018.1543624. [DOI] [PubMed] [Google Scholar]
- 22.Andersen JP, Blosnich J. Disparities in adverse childhood experiences among sexual minority and heterosexual adults: results from a multi-state probability-based sample. PLoS One. 2013;8(1):e54691. doi: 10.1371/journal.pone.0054691. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Ports KA, Ford DC, Merrick MT. Adverse childhood experiences and sexual victimization in adulthood. Child Abuse Negl. 2016;51:313–322. doi: 10.1016/j.chiabu.2015.08.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Williams Institute Scholars. Sexual orientation and gender identity (SOGI) adult measures recommendations FAQs. University of California–Los Angeles School of Law, Williams Institute. 2020. Available at: http://thewilliamsins.wpengine.com/wp-content/uploads/SOGI-Measures-FAQ-Mar-2020.pdf. Accessed April 14, 2020.
- 25.Groves RM, Cork DL. Surveying victims: options for conducting the National Crime Victimization Survey. Washington, DC: National Academies Press; 2008. [Google Scholar]
- 26.Kruttschnitt C, Kalsbeek WD, House CC, editors. Estimating the Incidence of Rape and Sexual Assault. Panel on Measuring Rape and Sexual Assault in Bureau of Justice Statistics Household Surveys. Washington, DC: The National Academies Press; 2014. [PubMed] [Google Scholar]
- 27. RE: Number OMB. 1121-0111//83 FR156, Document Number 2018-07448. Letter to Bureau of Justice Statistics. May 11, 2018. Available at: https://nadler.house.gov/sites/nadler.house.gov/files/documents/5.11.18%20letter%20to%20bjs%20re%20ncvs%20sogi.pdf. Accessed November 6, 2020.
- 28.Kahle L. Are sexual minorities more at risk? Bullying victimization among lesbian, gay, bisexual, and questioning youth. J Interpers Violence. 2020;35(21-22):4960–4978. doi: 10.1177/0886260517718830. [DOI] [PubMed] [Google Scholar]