Until recently, researchers have struggled to answer what should be a simple question: How many Americans are lesbian, gay, bisexual or transgender (LGBT)?
Alfred Kinsey's prominent study of sexuality in men from the 1950s led to the often-quoted estimate that 10% of the U.S. population is gay or lesbian. Other significant milestones in estimating the population of gay Americans include Edward Laumann and colleagues' The Social Organization of Sexuality, published in 1994; Gary Gates' The Gay & Lesbian Atlas, published in 2004; and The Williams Institute at UCLA School of Law's 2011 estimate (also authored by Gates).
Since 2012, Â鶹´«Ã½AV research has shown that the proportion of U.S. adults identifying as LGBT has increased from 3.5% in 2012 to 4.5% in 2017. The overall uptick is related to an increase in young people who identify as LGBT. Â鶹´«Ã½AV asks, "Do you personally identify as lesbian, gay, bisexual or transgender?" People who answer yes are classified as LGBT.
But the matter of how to best inquire about sexual orientation and gender identity is far from settled. In 2016, together with a group of researchers, I started working with Â鶹´«Ã½AV on two National Institutes of Health-funded studies: Generations, studying sexual minorities, and TransPop, studying gender minorities. To recruit separate sexual and gender minority groups, we had to begin to unpack the LGBT category that Â鶹´«Ã½AV had been collecting data on. Typically, sexual and gender minority statuses are assessed separately, and we wondered how Â鶹´«Ã½AV's single LGBT question compared.
First Experiment
In 2016, we analyzed Â鶹´«Ã½AV's LGBT question with two questions that assess gender identity. Using the so-called two-step method, we first asked, "What sex were you assigned at birth, on your original birth certificate?" with the response options of female and male. And second, "Which of the following terms best describes your current gender identity?" with the response options of woman, man, trans woman (male-to-female), trans man (female-to-male) and non-binary/genderqueer.
People are categorized as transgender if their sex assigned at birth is different from their gender identity, even if they don't use the label transgender to identify themselves. For sexual orientation, we asked, "Which of the following do you consider yourself to be?" with the response options of straight or heterosexual, lesbian, gay, bisexual, queer or same-gender loving. We found that the proportion of the population classified as LGBT using the Â鶹´«Ã½AV LGBT question was 3.83% and the proportion classified using our sexual orientation question was 4.35%, which is not statistically different. The proportion of transgender people using the two-step questions is 0.49% (of course, the Â鶹´«Ã½AV item cannot differentiate between LGB and transgender folks).
With the new set of questions, we found that among transgender people, a majority (66%) identified as straight -- and the people who identified as straight were not picked up by Â鶹´«Ã½AV's LGBT question. This is not hard to explain. People who we classified as transgender but were not identified with that label (e.g., a woman who was assigned male at birth) and who were not LGB would correctly answer the LGBT question with a "no," but would be misclassified based on our definition of transgender.
Second Experiment
In another experiment, we went back to test the sexual orientation question. One issue that has been occupying researchers has been whether a question of sexual orientation should only include the more commonly understood categories of lesbian, gay, bisexual and straight, or a more elaborate list of identities that are used by LGBT people, such as queer and pansexual. The advantage of the former is that it would presumably not confuse people who are straight and less familiar with sexual orientation terms (or even with the notion of sexual orientation and sexual identity, which is why those words do not appear in the question).
Research by a federal working group, among others, has shown that given the opportunity to choose among lesbian, gay, bisexual and straight, straight people would correctly report their sexual orientation. The advantage of the second format, which includes more identity labels, is that it allows sexual minorities to choose the identity label that best matches their identity.
For example, the argument goes, a person who identifies as queer might hesitate or even refuse to answer if given the more restricted options -- leading to an underestimate of the sexual minority population. It should be noted that there is concern about misclassifications in both formats, but that misclassifying straight people can lead to greater overall error because of their overwhelming numbers (according to Â鶹´«Ã½AV, straight people make up 95.5% of the U.S. population).
Misclassification can occur if a straight person misunderstands a term meant to convey a sexual minority identity and responds to such an item positively. For example, same-gender loving is a term used by a minority of black LGBT people in the U.S., but could be interpreted literally as something different; asexual, which a small number of people use as a sexual identity, could be misunderstood as referring to disinterest in sex; and queer could be used by LGBT allies who are not themselves sexual minorities.
Because of their vastly different representation in the general population, even if a small proportion of straight people are misclassified, this would lead to a gross overestimation of the sexual minority population (these comments do not apply in the same way to studies that sample within LGBT communities). We had some cause to worry about this because in one experiment we found a larger number of white people than would be expected using the same-gender loving label.
To test this, we asked people in one form, "Which of the following do you consider yourself to be?" with the response items lesbian or gay, straight (that is, not lesbian or gay), bisexual, something else or don't know. In a second form, people who said "something else" to the same question were given the option of identifying themselves as queer, pansexual, asexual, demisexual or same-gender loving; 0.6% of the total population indicated one of these labels. What we found surprising, however, is that a slight majority of people (53%) chose none of these options after choosing "something else."
It is possible that other sexual minority options are necessary, but anecdotal data warn us to the possibility that among those who choose none of the sexual minority labels, there may be many straight people who do not grasp the sexual identity question at all and, in other surveys, have written entries like "normal." Further, we found that more people in the second form than in the simpler first form (4.5% vs. 3.1%, respectively) gave unusable responses (don't know, refuse, something else).
Our conclusions are mixed. The second form provides an opportunity for 0.6% of the population to indicate their sexual minority status -- they make up 10% of the sexual minority group -- which would otherwise be registered as "something else" and handled by researchers in inconsistent ways. But we have also learned that 53% of "something else" respondents may not be classified as sexual minorities at all. On the other hand, the first form was preferable both in terms of its lower cost (one less question) and better response rates.
Deciding which question to use when asking about sexual orientation may vary depending on whether the general population or a select group (e.g., LGBT community members) is the target sampling population and whether specific age groups are targeted (adults vs. high school students). Other considerations include the interview modality (interviewer- vs. self-administered) and language and acculturation.
We will continue to study these issues -- but it is important to look back and note the tremendous progress already made in studying LGBT populations.
Ilan H. Meyer, Ph.D., is a Distinguished Senior Scholar of Public Policy at The Williams Institute at UCLA School of Law.
The work described was performed by Â鶹´«Ã½AV for the Generations and TransPop studies, funded by a grant from the Eunice Kennedy Shriver National Institute of Child Health and Human Development and through supplemental grants from the National Institutes of Health, the Office of Behavioral and Social Sciences Research, and the Office of Research on Women's Health.
The investigators are: Ilan H. Meyer, Ph.D., Walter O. Bockting, Ph.D., David M. Frost, Ph.D., Phillip L. Hammack, Ph.D., Jody L. Herman, Ph.D., Marguerita Lightfoot, Ph.D., Sari L. Reisner, Sc.D., Stephen T. Russell, Ph.D., and Bianca D.M. Wilson, Ph.D.
