Are LGB People More Likely to Be Victims of Crime?

In American survey data, yes, and the gap is large. In British survey data the gap disappears once you account for who LGB people are.

That is not two studies disagreeing. It is the same question answered in two countries with different adjustments, and the difference between them is the most interesting thing in this article.

Those are the headline figures and they are solid. The Bureau of Justice Statistics published them with standard errors, from a survey of hundreds of thousands of households, and later analyses of different years land in the same place.

But the single most useful thing to know about this data is that the word “LGB” is doing damage in that table. Bisexual people are most of the group and carry almost all of the elevation. Lesbian and gay people sit at roughly twice the straight rate. Bisexual people sit at nearly seven times it. Lumping them together produces a number that describes neither.

Every rate here is victims divided by population, and the population is not one number.

It depends on what the survey asked. Identity ("do you identify as gay, lesbian or bisexual") produces the smallest figure. Behaviour ("have you had sexual contact with someone of the same sex") produces a larger one. Attraction produces larger still.

The gap between them is not small. Among American women aged 18 to 49, roughly 10% identify as lesbian or bisexual, about 18% report same-sex sexual contact, and around 22% report some same-sex attraction. Same people, three questions, double the answer.

So a rate built on a count of people who identified as LGB, divided by a population defined by attraction, will be too low. Reverse the mix and it will be too high. Published commentary does both.

Violent crime also concentrates in the young. So comparing an all-ages LGB group against an all-ages straight group compares a young population with an older one, and the raw gap is inflated by that alone.

It is inflated, not invented. The survey publishes age-specific rates, and inside the 18 to 24 band, where everyone’s risk is highest, bisexual respondents were at 198.7 per 1,000 against 32.8 for straight respondents. Six times, among people the same age.

The elevation is not confined to one kind of crime, which is worth establishing early because it rules out some explanations.

Two of those rows are not statistically significant and should not be quoted as though they were: the lesbian and gay figures for aggravated assault, and for intimate partner violence specifically, are within the margin of error of the straight rate.

The stranger-violence row matters. If the gap appeared only in domestic violence you could explain it through relationship patterns. It does not. Bisexual people are attacked by strangers at seven times the straight rate.

There is no homicide figure in this article, because there is no homicide figure.

No national dataset in either country records the sexual orientation of murder victims. The FBI’s homicide reports capture age, sex, race, weapon and the relationship between victim and offender. Not orientation. The UK Home Office homicide index does not either. America’s violent death reporting system has a sexual orientation field, and its own coding manual says it will only pick up people whose orientation was known to friends or family and that the information is not collected systematically.

What people use instead is advocacy organisations counting murders reported in the press as anti-LGBT. That is a count of what was reported and described a particular way. It is not a rate, and this article does not print one.

Here is where it gets interesting, and where most coverage of this subject stops early.

“Are LGB people more often victims” and “does being LGB raise your risk” are different questions. The tables above answer the first. The second needs you to account for everything else that differs between the groups, and LGB samples differ from the general population on nearly every variable that predicts victimisation: they are younger, more urban, less often married, lower income, more likely to report a disability, and far more likely to have experienced homelessness as young adults.

Three countries have tried the adjustment and got three different answers.

The British result deserves more attention than it gets. Same broad question, a national survey of comparable quality, and the association vanishes.

What it does not mean is that LGB people in Britain are not victimised more often. They are, in the raw figures, and British harassment data shows one of the starkest identity gaps anywhere: 35% of bisexual people and 21% of gay and lesbian people reported harassment in a year, against 8.4% of heterosexual people.

What it means is that in Britain, the elevated risk was accounted for by age, disability and the other characteristics in the model. In America it was not. The honest position is that we do not know why the two diverge, and the candidates are all plausible: different covariate lists, different crime definitions, different survey design, or a real difference between the two societies.

That is an unresolved finding, and it stays unresolved here.

Police-recorded hate crime counts are not incidence. They measure what victims reported, what officers recorded, and which agencies sent in returns.

The scale of that problem is easy to miss. In 2023 the FBI received hate crime data from about 16,000 law enforcement agencies. Around 3,100 of them reported a single incident. The rest reported none. Some of those are small departments in quiet years. Some are large cities returning zero.

Against that, the recorded numbers: about 2,400 sexual-orientation bias incidents in the United States in 2023, and 18,702 sexual-orientation hate crimes recorded in England and Wales in the year to March 2025, the third consecutive annual fall.

Now the survey estimate for the same country. Asking people directly, the American victimisation survey found LGBT people experienced 6.6 violent hate crime victimisations per 1,000 people, against 0.6 for everyone else. Scale that to the population and it implies figures far above what police record.

Both numbers are real. They measure different things. A rise in recorded hate crime can mean more attacks, better recording, more agencies participating, or more victims willing to say the attack was motivated by who they are. The published tables cannot separate those.

This one runs against expectation, and it runs in both directions at once.

Lesbian and gay victims reported more often than straight victims. Bisexual victims reported far less, which is the group with the highest victimisation rates.

The practical consequence: police-recorded crime understates what happens to bisexual people and overstates it, relatively, for lesbian and gay people. Any analysis built on police data alone will get the internal ordering of this question wrong.

What is not established is how LGB victims are treated after they report, nationally, against a heterosexual comparison. Community surveys recruited through LGB networks report high rates of perceived disrespect. Those are not national samples and this page does not treat them as rates.

Four serious objections. The weak version, that the surveys are activist, is not one of them: these are government statistical agencies running probability samples.

People who disclose one thing may disclose the other. Someone willing to tell an interviewer they were assaulted may also be more willing to tick “bisexual”. If so the association is partly manufactured. The survey design limits this, because orientation is asked of everyone in the household rather than only of victims, so the denominator includes non-victims who disclosed. It does not eliminate it. Nobody has run a design that separates disclosure from victimisation, and until someone does this stays open.

Some of the cells are tiny. True, and it matters for specific figures rather than the headline. Bisexual men are the clearest case: 64.5 per 1,000 with a standard error of nearly 12. The statistical agency itself flags several subgroup estimates as unreliable. The bisexual women figure is not one of them, resting on a four-year pool with a comparatively small error.

It is really about age and everything else. The strongest objection, and partly correct. It explains about 15% of the lesbian and gay gap, about 41% of the bisexual gap, and in England and Wales it explains all of it. Anyone arguing it explains all of the American gap has to account for the bisexual 18 to 24 rate of 198.7 against 32.8 for straight people the same age.

The group being measured has changed. LGB identification roughly tripled between 2012 and 2025 and shifted young and bisexual. The people answering “bisexual” in 2023 are not the people who answered it in 2012. If those newly identifying have higher baseline risk for other reasons, the rate among identifiers rises without anything happening to anybody. This is a real problem and it means trend claims about this data are unreliable. It does not touch a within-year comparison.

1. In American survey data, LGB people experience violent victimisation at well above the heterosexual rate: 43.5 per 1,000 for lesbian and gay people and 129.1 for bisexual people, against 19.0 for straight people.

2. Bisexual people, and bisexual women above all, carry nearly all of that elevation. A combined “LGB” figure describes neither group accurately.

3. The gap appears in assault, robbery, domestic violence, sexual violence and violence by strangers. It is not confined to intimate relationships or to bias-motivated attacks.

4. Age composition inflates the raw gap, because LGB populations are much younger. It does not create it: within the 18 to 24 age band the bisexual rate is six times the straight rate.

5. Adjustment shrinks the American gap and leaves most of it standing. In England and Wales the same kind of adjustment removes it entirely. Nobody has explained why the two countries differ.

6. There is no homicide figure, in either country, because no national dataset records a murder victim’s sexual orientation.

7. Bisexual victims report to police far less often than anyone else, which means police data gets the internal ordering of this question backwards.

8. Trend claims are unreliable, because the group identifying as LGB has roughly tripled since 2012 and is not the same population across years.