Gallup just recently released new estimates of the size of the LGBT(+?) population in the U.S. I have been writing about and tracking Gallup’s data for a while. Their new data are useful and allow us to dis-aggregate gender and sexual identity groupings within the larger figures they provide (something not always possible based on how they collected data in the past). I was disappointed that they did not provide any new state-level estimates for the relative sizes of LGBT populations in different states as that is something that deserves more work.
Interestingly, the new data do not rely on the yes/no question Gallup initially used to estimate the proportion of the American public identifying as LGBT. That initial question simply asked respondents, “Do you, personally, identify as lesbian, gay, bisexual, or transgender?” While this did not allow Gallup to separate the L’s, B’s, G’s, and T’s from one another, it was relied upon primarily to reduce false positives. As Gary Gates shared in a Gallup podcast from 2018, Gallup’s initial question “reduced measurement error” by specifically attempting to reduce “false positives”. That’s survey methodologese for worrying that cisgender and straight people may not think of themselves as having gender and/or sexual identities. As such, sometimes they respond incorrectly when asked about their identities. As Gates put it, “With the Gallup question, you’re not asking that group what they are, you’re asking what they aren’t. And they more or less know that. So they may not use terms like heterosexual or straight. But they know they’re not gay [or lesbian, or transgender, etc.].”
Perhaps as a part of education built into the survey design, Gallup now provides questions that allow people to select from a menu of gender and sexual identities, checking as many as they feel apply to them personally, similar to the shift on the U.S. Census question regarding race. This is exciting as it allows us to look into their data with a bit more depth than was previously possible.
As of 2021, Gallup reported that their most recent survey found than 7.1% of Americans now identity as LGBT. It might be more accurate to say that they discovered that 7.1% of Americans identify as “LGBTQ+” because their methodology shows that they incorporated a range of identity options volunteered in interviews, but not formally on their menu of options as “Other LGBT”. The examples they provide in their write-up are queer, pansexual, and same-gender-loving; but it seems like this is how people identifying as genderqueer, gender fluid, nonbinary and more would have also been coded. This is an impressive shift, continuing a shifting trajectory they have been following since 2012. It’s worth noting that they have violated the social science creed: “if you want to measure change, don’t change the measure.” But, it’s also true that when you are measuring identities that transform over time, sometimes old measures simply cease to be as meaningful. As new gender and sexual identities and cultures emerge, how we measure gender and sexual identities must also shift and adapt.
This is exciting and certainly a marked shift. The population of the U.S. in 2021 was approximately 332,915,073 people. 7.1% of that number is about 23 million people. And this is, in all likelihood, a conservative measure of the size of this population as the sexual piece of this figure rests on the dimension of sexuality on which social scientists have routinely received the lowest proportions being identified as members of sexual minority communities (the question is about identity, rather than, for instance, questions about sexual practices and behaviors or attractions and desires).
And yet, as I’ve previously written about with D’Lane Compton (here) and Mignon Moore (here and here), it’s also important to look inside the very diverse collection of identities included in LGBTQ+ to examine among whom these shifts are taking place. Within each of these categories, there is a great deal of variation. I always find looking at these data easier with figures than tables. And I sometimes use these for teaching as well. So, I graphed a few of their tables to consider the shifts they documented and gave us figures for below. I’ll share them here along with some brief commentary I think of when encountering them.
Above, Gallup provided their estimates of the share of the LGBT+ population of the U.S. who selected L, G, B, T, and/or something else (they label the something else’s “other”). Continuing a trend than goes back at least to the early 2000’s, bisexual identities continue to be the lion’s share of the change. For anyone studying sexual demography, that in and of itself is not all that interesting as it has now been the case for a while. But it might be interesting to measure shifts in the the size of that discrepancy in growth between, for instance, lesbian, gay, and bisexual identities over time. Also of note here is the fact that “lesbian” identities were selected less commonly than “gay” (though remember that in the figure above, for each identity category included, people of all genders are represented). Broken down by gender (in the binary way Gallup provided data on gender for this portion) is also interesting.
It is still true that “bisexual” is a more common identity than “lesbian” among women, but a less popular identity than “gay” among men. I have written on this before. We also see here the proportions of women selecting “gay” and “lesbian” (though of course, they were allowed to select both in these data as they were asked to check all that apply). I’m also interested in the slightly higher figure for women among those identifying as “transgender” in these data and the fact that women were also twice as likely as me to have claimed an identity coded as “other”. Finally, here too we see that a larger share of people (women and men) identified as “gay” than “lesbian.”
Catherine Connell and Elliot Chudyk have been interviewing people in the U.S. who identify as “lesbian” recently and I’m interested to see what they discover (see “The Lesbian Project” here if you’re interested). Among the things that stood out to me when I first encountered their call for interviews was the fact that they cast a wide net. Among those they were interested in potentially interviewing were men who identify as lesbian. While a small share of those identifying as men in Gallup’s data also checked “lesbian” (0.1% as Gallup reported), that number stood out to me. I’m happy that Gallup’s method of data collection allowed for that to be discoverable.
Gallup also continues to chart change by generation. But, as I’ve learned more about this from Philip Cohen, the arbitrary nature of generation labels is less productive than simply stating how old people are. Gallup reports their data by generation. Below, I simply graph the same data but simply state the age groupings for people at the time of the interview.
Similar to other data, the youngest adults surveyed have the highest proportions of LGBT(+?) identification in Gallup’s data. 20.8% of people between the ages of 18 and 24 at the time of data collection identified as lesbian, gay, bisexual, transgender, or some other identity other than straight or heterosexual. 10.5% for 25-40 year olds, and only 0.8% of those interviewed over the age of 75. But another fact jumped out at me when looking at the figure above – those not responding to the question. The response rate for this question among 18-24 year olds is the highest of any of the age groupings. Roughly half as many 18-24 year olds chose not to respond to this question when compared with all other four age groups presented here. Reading non-responses can be tricky. But it made me think of a paper published on NBER in 2013 by Katherine B. Coffman, Lucas C. Coffman and Keith M. Marzilli Ericson that I wrote about at the time attempting to provide novel ways of asking about sexuality in particular on surveys in ways that did not trigger social desirability bias or discomfort (even among people on anonymous surveys). Coffman, et al. discovered that directly asking people to identify as something other than heterosexual on a survey resulted in lower proportions of people identifying as something other than heterosexual when compared to an experimental group provided with an indirect method allowing them to identify as something other than heterosexual without having to report that directly on a survey. (I wrote about the study here if you’re interested.) I’m wondering whether a similar bias might explain some of the no responses among older Americans and if it’s possible that younger Americans are less uncomfortable with the direct response approach.
Finally, Gallup also provided data on the proportions of people in different age groupings identifying as LGBT+ (below).
Here, I think there’s less that we learn that is really new. Among the things that stood out for me is the fact that bisexual is the most popular identity listed here among people 56 years old and younger. But over 56 and bisexual is less popular than other sexual identities. The other fact that seemed interesting here to me is the fact that, among 18-24 years olds here, “transgender” was a more popular identity than “lesbian”, and only slightly less popular than “gay.” Now, it’s true that people who selected transgender here could also select gay and/or lesbian as well. But this seemed interesting. I also think that the fact that they had to code progressively more people as “other LGBT” as the ages got younger suggest to me that the menu of options they offered is becoming more limiting and might be less able to capture the diversity of gender and sexual identities moving forward.
Among the things missing from Gallup’s report and something I am in the very early stages of collaborating on with D’Lane Compton and Mignon Moore is an analysis of race. My work with Mignon Moore suggests that some of these exciting trends are particularly pronounced among Americans racialized as Black. It’s always exciting to see Gallup’s new data. But I’m always left wishing they provided more than they did as well.