The Policing of Queerness in the United States

The objective of this research is hence to study whether significant disparities and bias exist when agents come into contact with people whom they perceive as being LGBTQ+ and if the interaction of belonging to LGBTQ+ identities with racial minority status exposes people to more severe police targeting in post-stop decisions. By registering the officers’ perception of whether a person is LGBTQ+ or a racial minority, the data used in this study help to identify the mechanism that can lead to discrimination dependent on the individual’s identity: regardless of whether the person stopped really belongs to one of these categories, it is the agent’s perception that will mediate the transformation of its possible prejudice into discriminatory behavior.

It is crucial to understand policing, to study how identities in the spectrum of queerness shape police-civilian interactions and in which directions. The police are oftentimes understood as an institution that is also crucial in maintaining the status quo, reproducing socioeconomic inequalities through the criminalization and the marginalization of vulnerable groups (Rios et al. Reference Rios, Prieto and Ibarra2020; Vitale Reference Vitale2017). Whether it is Alexander (Reference Alexander2010) or Wacquant (Reference Wacquant2001) describing how police and prisons have been crucial in maintaining the racial status quo in the United States; Soss and Weaver (Reference Soss and Weaver2017) speaking about its role in shaping what they call “Race-Class Subjugated Communities”; or Knox et al. (Reference Knox, Lowe and Mummolo2020) framing it as a “a political act in and of itself [that] determine[s] who gets protected, punished, or left to fend for themselves” (621), law enforcement undoubtedly plays a key part in shaping the (mis)fortunes of minoritized communities all around the country, across the spectra of race, gender identity, or sexual orientation.

Considering LGBTQ+ identities, the role of law enforcement (as well as that of prisons or mental institutions) in reproducing their marginalization has been clear throughout history (Stanley and Smith Reference Stanley and Smith2011). Lvovsky (Reference Lvovsky2021) cites, among others, U.S. police departments creating “moral squads” in the 50s, whose objective was to suppress the “public manifestations of queer life” (5). Through institutional practices like this, LGBTQ+ people have been disproportionately targeted by police agents in their spaces and communities for at least a century, with the daily harassment, surveillance, and patrolling of officers culminating in high-profile clashes between the police and the queer community that shaped the future of LGBTQ+ liberation in the country (Lvovsky Reference Lvovsky2021; Stanley and Smith Reference Stanley and Smith2011). Law enforcement played an active part in violently repressing LGBTQ+ spaces in the United States in episodes such as during the raid at the Compton Cafe in 1966 or at the Stonewall Inn in 1969 (Hanhardt Reference Hanhardt2008) or when deploying violence during AIDS protests in the 1990s (Stockdill Reference Stockdill2002). The criminalization of HIV has also been, in the 1980s and 1990s, a means to police and criminalize non-conforming gender and sexualities (Stanley and Smith Reference Stanley and Smith2011). None of this is ancient history either: just in 2003, the police raided a gay club in Detroit, addressed its customers with slurs, and used violence against them (Mogul et al. Reference Mogul, Ritchie and Whitlock2011). It is also relevant to remember that until the same year, homosexuality was still criminalized in many U.S. states, when the Supreme Court’s decision in Lawrence v. Texas struck the remaining so-called “Sodomy Laws” still standing, and that we are recently witnessing a wave of criminalization against transgender identities and gender affirming care in many U.S. states (Movement Advancement Project Reference Project2025).

Criminalization and harassment are hence not a thing of the far past, and police misconduct against LGBTQ+ people is to this day “systemic and commonplace” (Mogul et al. Reference Mogul, Ritchie and Whitlock2011, 5). Recent literature on the topic has consistently found that LGBTQ+ interactions with the police are still characterized by profiling, harassment, and violence (Hodge and Sexton Reference Hodge and Sexton2020; Mallory et al. Reference Mallory, Hasenbush and Sears2015; Mogul et al. Reference Mogul, Ritchie and Whitlock2011). A 2005 report by Amnesty International describes widespread violence, brutality, and abuses in policing and imprisonment of LGBTQ+ people in the United States, and data from the 2015 United States. Transgender Survey (James et al. Reference James, Herman, Keisling and Mottet2019) saw 58% of respondents reporting some sort of mistreatment in their interaction with law enforcement. Recent scholarship (Girardi Reference Girardi2022) finds that, in the European context, “queer visibility and performativity were associated with higher risks of being victimized and criminalized by law enforcement” ( 859), a finding mirrored by Fileborn (Reference Fileborn2019) when analyzing the context of regional Victoria. Sexual and gender minorities are also disproportionately incarcerated in the United States and face a higher risk of victimization or sanctions (e.g., solitary confinement) when imprisoned (Meyer et al. Reference Meyer, Flores, Stemple, Romero, Wilson and Herman2017). As a result, scholars have found that LGBTQ+ people are more likely to support police reform than cishet people (Jenness and Vogler Reference Jenness and Vogler2025), as well as more likely to perceive law enforcement negatively and as a foe (Owen et al. Reference Owen, Burke, Few-Demo and Natwick2018; Vogler and Jenness Reference Vogler and Jenness2023).

Policing from an Intersectional Perspective

As Shoub (Reference Shoub2025) argues, individual officer biases and the historical perception of some identities as “deviant” can set the scene for police discrimination against them, and this has paved the way for the over policing of LGBTQ+ people, alongside communities of color, across U.S. history.

This is why there is a need, when analyzing police-civilian interactions, to apply an intersectional framework, when possible, as policing has clearly had great implications for the marginalization of both LGBTQ+ people and racial minorities in the United States. “Intersectionality” is a term coined by Kimberlé Crenshaw (Reference Crenshaw1991) to describe how interconnected identities—in Crenshaw’s case, those of women of color—create unique, marginalized positionalities that are shaped by multiple systems of oppression, which intersect and converge to create specific “patterns of subordination” ( 1249) and of marginalization in the socio-economic world. In this regard, marginalized identities and the processes of discrimination leveraged against them (by institutions like the police, too) should be understood as given significance by systems of oppression, such as white supremacy, cisheteronormativity, and the patriarchy, which slot racialized and LGBTQ+ people into subordinate positionalities in the U.S. society and beyond.

Considering the interaction with other marginalized identities—to say, other axes of oppression—when studying police discrimination is hence crucial (Collins Reference Collins2015). In this regard, an intersectional framework can guide us to understand how the structural discrimination and oppression that society leverages against both racial minorities and LGBTQ+ people should be reflected, too, in how the police exercises its power to oppress civilians who belong to a racial minority, the LGBTQ+ community, or both.

Some scholars have recently started to rely on intersectionality to analyze policing at the intersection of race, class, gender, or age, finding that the interplay of these positionalities is indeed a factor in who is targeted by law enforcement and why (Brunson and Miller Reference Brunson and Miller2006; Christiani Reference Christiani, Brown, Block and Stout2020; Cooper Reference Cooper2021; Gonzalez Reference Malone Gonzalez2019; Roach et al. Reference Roach, Baumgartner, Christiani, Epp and Shoub2020). Drawing from an intersectional perspective, a recent study has shown that transgender women of color in New York City are subjected to higher rates of police violence and victimization, compared to both cisgender and transgender White women (Duncan et al. Reference Duncan, Park, Dharma, Torrats-Espinosa, Contreras, Scheinmann, Watson, Herrera, Schneider, Khan, Lim, Trinh-Shevrin and Radix2024). Other authors (Gaynor and Blessett Reference Gaynor and Blessett2022) have delved into what they call the “intersectional subjection” of LGBTQ+ people of color by law enforcement in New Orleans, and how predatory policing furthers the marginalization of these identities. Queerness can hence intersect with race (and class) in making people extremely vulnerable when interfacing with the policing and the carceral system and put them at a higher risk of citation, arrest, or use of force by law enforcement (Morgan Reference Morgan2022, Reference Morgan2021; Nelson Reference Nelson2010). The intersection of queerness and race has also been recently linked to a higher vulnerability to violent crime (Flores et al. Reference Flores, Wilson, Langton and Meyer2023).

This is eliciting calls to broaden the scope of analysis when studying law enforcement profiling. As Morgan (Reference Morgan2022) states, an intersectional approach is fundamental to deepen our understanding of how the police operate in society, as well as how it interacts not only with gender or sexuality but also with race. Mogul et al. (Reference Mogul, Ritchie and Whitlock2011) also correctly note that current scholarship of policing focuses on racial minorities as “monolithic […] when it comes to class, gender, and sexuality” ( 51), while discrimination is probably determined by the intersection of these categories too. With queerness being, by itself, a risk factor in police-civilian interactions, this framework matters greatly both for those who are White but not cisgender or heterosexual, as well as for racial minorities who are LGBTQ+.

This is hence the core argument of the paper, informing its empirical strategy: other marginalized identities like gender and sexual minorities, that have historically been over policed and pushed to the margins through biased policing, matter to study how law enforcement discriminates against civilians. Importantly, these identities matter both by themselves and in their interaction with race. This is what this research aims to explore, understanding policing as an institution that has played a key role in pushing both racialized and LGBTQ+ people to the margins of society.

The Case Study: Los Angeles & San Francisco

Administrative records usually do not record officers’ perception of a civilian’s LGBTQ+ identity: the LAPD and the SFPD data used in this study are hence precious because of their rarity. Most importantly, these cities can be an interesting case study to investigate this phenomenon, especially from an LGBTQ+ perspective, not only because of the unique availability of the data. Despite their reputation as liberal areas, Los Angeles and San Francisco have, more than other urban areas in California, a history of particularly troubled relationships between the LGBTQ+ community and police officers: outside of the cited episode of Compton’s Cafeteria in 1966, the homicide of Harvey Milk in 1978 by a former policeman and the following White Night riots of 1979 are other milestones that signify the tensions that have existed for a long time between the SFPD and LGBTQ+ people (Dwyer Reference Dwyer, Peterson and Panfil2014). Los Angeles has also witnessed similar episodes, such as raids on gay bars, like the Cooper Do-nuts uprising in 1959, considered maybe the first LGBTQ+ uprising in the country, or the 1967 raid and following demonstrations at the Black Cat Tavern, a historic gay bar in the city. Such episodes have spurred clashes between the city’s LGBTQ+ community and law enforcement. Police violence and harassment against the LGBTQ+ communities in the cities have been denounced by many associations in recent years, first amongst them the National Coalition of Anti-Violence Programs (NCAVP). In a 2001 report, the organization extensively documented harassment and violence by law enforcement agents against the LGBTQ+ community of San Francisco in particular (National Coalition of Anti-Violence Programs 2001). Both places are, hence, exemplary sites of conflict between the police and LGBTQ+ communities, marked by the systemic criminalization of sexual and gender minorities by law enforcement.

Despite this, in recent years, the departments have shown some signs of change in their relationship with the LGBTQ+ community. Dwyer (Reference Dwyer, Peterson and Panfil2014) cites, for example, the explicit support of the SFPD to young LGBTQ+ people in an own video for the “It Gets Better” project. The SFPD has also released in March 2018 (so about 4 months before the data collection started) an internal order to establish procedures for interactions with Transgender, Gender-Variant, and Nonbinary individuals (San Francisco Police Department 2018). Guidelines include how to conduct a search, strip-search policies, restroom use, detention in holding cells, and use of correct pronouns. This is particularly relevant as extant research has shown that many police departments in the United States do not have policies regulating interactions with LGBTQ+ people (Ritchie and Jones-Brown Reference Ritchie and Jones-Brown2017). In recent years, the department’s force has also become more diverse in terms of queerness, too, as now the SFPD employs around 200 LGBTQ+ officers, representing about 10% of the force (Borba Reference Borba2022). The LAPD has tackled relationships with the LGBTQ+ community in recent years as well, for example, through training courses (LAPD 2022). It has also addressed the violent harassment towards the LGBTQ+ community in the 50s and 60s (McMillan Reference McMillan2023). The purpose of this research is not to evaluate whether these cultural changes have brought any effect, as it would be impossible due to the lack of data from the period before these took place, but it can still be useful to assess whether disparities are persistent in the departments’ treatment of gender and sexual minorities, despite all these efforts.

These histories and policies do not, however, make San Francisco and Los Angeles unique cases in the U.S. landscape: as mentioned in Section The policing of queerness in the United States, violence and discrimination against LGBTQ+ individuals have been a nationwide phenomenon, rather than one limited to some cities or regions. More than anything else, these two cities can be seen as the epitome of how queerness has shaped police-civilian interactions in urban areas across the country, and their results can be, at least to a degree, generalizable to other contexts.

Describing Disparities in Police Behavior

As a first step, I look at whether any differential treatment in post-stop outcomes can be seen through benchmark tests, as well as by calculating hit rates by group. A benchmark test is simply a ratio of the number of searches, arrests, or uses of force over the total number of stops. A hit rate test, also known as the outcome test (Anwar and Fang Reference Anwar and Fang2006; Ayres Reference Ayres2002), indicates the likelihood of retrieving contraband when performing a search, and it has been used as a measure that can show whether groups are held after different standards of evidence. The simple idea is that if a group (e.g., Blacks) has a significantly lower hit rate than Whites, it would imply that law enforcement is applying different thresholds for when to decide to conduct a search against different groups.

Figure 1 shows the likelihood of being searched, arrested, and of force being used for the interactions between queerness and race (Numeric coefficients are shown in the SI File, Section 9S). A few trends jump to the eye, above all: the likelihood is always the lowest for Whites who are also cishet; being LGBTQ+ significantly increases the risk of being searched, arrested, or a victim of use of force, for racial minorities and Whites alike. Figure S1 in the SI File presents the aggregated estimates for the two main categories (race and queerness) and confirms the trends we observe here.

Figure 1.

Benchmark tests for searches, arrests, and use of force. Points represent the proportions of post-stop outcomes by identity group, with 95% confidence intervals.

Figure 1. Long description

Three bar graphs depict the proportions of post-stop outcomes by identity group, with 95% confidence intervals. Panel A: The first bar graph shows search rates. The x-axis lists four identity groups: White Cishet, Minority Cishet, White LGBTQ+, and Minority LGBTQ+. The y-axis represents the search rate, ranging from 0.15 to 0.35. The graph indicates that Minority LGBTQ+ individuals have the highest search rate, followed by White LGBTQ+, Minority Cishet, and White Cishet individuals. Panel B: The second bar graph displays arrest rates. The x-axis lists the same four identity groups, and the y-axis represents the arrest rate, ranging from 0.10 to 0.25. The graph shows that Minority LGBTQ+ individuals have the highest arrest rate, followed by White LGBTQ+, Minority Cishet, and White Cishet individuals. Panel C: The third bar graph illustrates the use of force rates. The x-axis lists the same four identity groups, and the y-axis represents the use of force rate, ranging from 0.20 to 0.40. The graph indicates that Minority LGBTQ+ individuals experience the highest use of force rate, followed by White LGBTQ+, Minority Cishet, and White Cishet individuals.

Essentially the same can be said when comparing these estimates with hit rates (i.e., the outcome test), as presented in Figure 2: stopped civilians who are LGBTQ+ have a significantly lower standard of evidence applied to them (both when they are White or a racial minority), according to the test, when compared to the reference category of White and cishet civilians. Results also hold when looking at variation between districts (see Figure S4, Online Appendix).

Figure 2.

The outcome test. Points represent the proportion of successful searches (i.e., retrieving contraband) by identity group, with 95% confidence intervals.

I then move on to focus on the interaction of race with queerness, the objective of the second research question. To do so, I run logistic regressions to predict the probability of agents searching, arresting, or using force during a stop, interacting racial status with queerness. For these models, as well as those presented in Section Inferring police bias in post-stop decisions, I include controls for age, type of stop, whether the person had limited or no English, and the stated reason for the stop (RFS in the equation below). Moreover, I add fixed effects for the district, as well as the month, the day of the week, and the 4-hour slice of the day in which the stop happened, which is relevant since police activity is not randomly distributed over weekdays, months, daytime, or area. This allows me to control, as best as the data allows, for a lot of unobserved characteristics of the place and time where the stop happened, while also making the sample of officers making the stop and of people stopped likely more homogeneous and comparable. The model specification, with controls and fixed effects terms, is shown in the equation below.

$\eqalign{&log\left( {{{Pr\left( {outcom{e_i} = 1} \right)} \over {1 - Pr\left( {outcom{e_i} = 1} \right)}}} \right) \cr & = {\beta _0} + {\beta _1}LGB{T_i} + {\beta _2}Minorit{y_i} + {\beta _3}\left( {LGB{T_i} \times Minorit{y_i}} \right) + {\beta _4}Ag{e_i} \cr & + {\beta _5}NoEnglis{h_i} + {\beta _6}Typ{e_i} + {\beta _7}RFS_i^{consensual} + {\beta _8}RFS_i^{parole} \cr & + {\beta _9}RFS_i^{wanted} + {\beta _{10}}RFS_i^{education} + {\beta _{11}}RFS_i^{reasonable} + {\beta _{12}}RFS_i^{probable} \cr & + {\beta _{13}}RFS_i^{traffic} + {\delta _{month}} + {\gamma _{weekday}} + {\theta _{dist}} + {\phi _{timestop}} \cr}$

Results are presented in Figure 3 and again are in line with what was previously found: being LGBTQ+ is positively associated with a higher probability of being searched/arrested/subjected to the use of force. Regression coefficients are presented in the SI File, Table S2.

Figure 3.

The joint effect of race and queerness on police searches, arrests, and uses of force. Lines represent 95% confidence intervals.

The interaction with minority status tends to increase said probabilities, although the effect of being perceived as LGBTQ+ is much stronger for Whites. As can be seen, the effect tends to be bigger when we consider the interaction between queerness and being White, rather than the interaction of queerness with belonging to a racial minority. This is, possibly, because of a lower starting floor for White civilians, and/or a ceiling effect for minority individuals. As a result, we observe that the probability of being subjected to any of the post-stop outcomes associated with being White and LGBTQ+ is almost as high, across all models, as that associated with belonging to a racial minority and being LGBTQ+, and higher than the same probability compared to those civilians who belong to a racial minority but are cishet. Results are robust when using a linear probability model instead of logit.

All the results presented here point in the same direction, signaling that queerness is associated with harsher disparities in police treatment across searches, arrests, and force use. In the next section, I then present the bias-corrected ATT estimator for these same outcomes, with the aim of approximating a better measure of bias in policing and how it affects the groups considered in the study.

Inferring Police Bias in Post-Stop Decisions

In this last section of the empirical analysis, I hence apply the corrective procedure suggested by Knox et al. (Reference Knox, Lowe and Mummolo2020), conditioning the estimates of discrimination in post-outcomes on an approximation of the proportion of biased stops, which I derived from a hit rate test. Figure 4 shows the bias-corrected estimator, i.e., the ATT, among LGBTQ+ civilians, for searches, arrests, and use of force. Basically, the estimands reveal how many searches, arrests, or uses of force would not have occurred had the civilian been White and cishet (when compared with White and LGBTQ+ civilians/minority and LGBTQ+ civilians). A more thorough explanation of how the test works can be found in Section 7S of the Online Appendix.

Figure 4.

ATT estimates of the proportion of discriminatory searches, arrests, and use of force against LGBTQ+ civilians, by race. Lines represent 95% confidence intervals.

As mentioned before, inferring discrimination is a complicated issue, and particularly so when dealing with police data. Hence, and as with the interaction regressions, I include here a wide range of controls that define the police-civilian interaction and civilian behavior (i.e., the same control variables used in the logit interaction models presented in Figure 3). Doing this, I approximate—as best the data allows—an estimand of bias that eliminates as many confounders as possible and isolates the effect of queerness, in its interaction with race, on the police agents’ decisions to search, arrest, or use force.

As can be seen, Figure 4 displays findings in line with what the interaction models of Figure 3 show. A situation of double-disadvantage, where being LGBTQ+ and a racial minority appear to have an additive effect in worsening police discrimination, is apparent across all outcomes. Being identified as LGBTQ+ and a minority significantly increases the risk of being searched, arrested, or subjected to force compared to both white, cishet individuals (i.e., the reference category) and to cishet racial minorities. Hence, while race is definitely an important variable in and of itself, its intersection with being LGBTQ+ has implications that perfectly fit the understanding of society posited by intersectionality: the convergence of marginalized positionalities has a substantial impact in multiplying the threats that the police pose to Black or Hispanic individuals who are also perceived to be LGBTQ+.

Secondarily, but equally importantly, it also appears that White civilians who are also LGBTQ+ face substantially higher discrimination in post-stop decisions compared to the reference categories of White and cishet civilians, on levels often as high as those of racial minorities. The fact that for this subset of White civilians the jump in discrimination is much higher, while for racial minorities the increase resulting by the interaction is smaller, suggests that there may be some ceiling effect at play in determining the incidence of discrimination against minoritized groups, particularly those that are already highly targeted by law enforcement.

These results on aggregate not only confirm the initial hypotheses of the paper relating to intersectional theory and how the interaction of different axes of oppression can multiplicate the incidence of police bias but have also quite relevant implications for our understanding of how policing works in actively furthering the process of marginalization of vulnerable identities. The effect on whiteness is also interesting and signifies how any deviation from the “norm”—being White, cisgender, and heterosexual—can automatically slot people into a marginalized position and increase their risk of suffering discrimination in society at large, and in police-civilian interactions too.

Since this analysis as a proof of bias rests on some assumptions that might not be universally accepted, I also plot the estimates for the ATT assuming that discriminatory stops are the same and null for each group, to see if the results would change. Figure S5 in Section 7S of the SI shows how the ATT changes under this assumption: as it displays, no substantial change is detected in patterns of police bias across the categories. Discrimination towards all other groups is persistent and follows the same motif recognizable in Figure 4, showing that the bias seen in the results is not entirely dependent on conditioning on the assumptions on which these estimands rest.

To conclude, I note that while variation over time is outside of the scope of this paper, I plot the trend over time of stops, including all observations, and then minority and LGBTQ+ stops separately, to check whether any time-variant factors (such as differences in reporting, over time, by police agents particularly, of who is LGBTQ+ or a Black/Latino civilian) could induce any bias in the results. Figure S6 is shown in the SI File and does not show any heterogeneity in the stop trends across the minoritized groups considered in the study. Moreover, Section 10S of the Appendix displays a compositional analysis of stops by type, reason, city, and year, and it shows that, when considering these possible sources of variation (which are also controlled for in the models), patterns of disparity hold for each outcome.