Representing LGBT People in Survey Research

In 2008, I jumped at the opportunity to attend the Williams Institute’s primer on quantitative research on sexual orientation. By the time I attended the Fenway Institute‘s Summer Institute on LGBT Population Health in 2011, and thereafter, a lot had changed in survey research on lesbian, gay, bisexual, and trans* people in the US.  The good news is that we have advanced at lightening speed in collecting samples of LGBT people and information about their sexual and non­-sexual lives.  The bad news – reflected by the ease with which one anti-gay study disrupted a Supreme Court case on marriage equality – is that we have not come far enough.

LGBT Inclusion in Survey Research

With the simultaneous decline in anti-LGBT prejudice and growing visibility of LGBT communities in the US, it is unsurprising that research on LGBT people, too, is an evolving process.  In the late 1980s, the beginnings of LGBT inclusion in survey research were the product of attention to the HIV epidemic among gay and bisexual men.  For example, the General Social Survey (GSS) added a question about the sex of their sexual partners in 1989.  Unfortunately, there seemed little hope to investigate the lives of LGBT people otherwise.  For some time, questions about sex (including sexual orientation) remained tucked away in confidential, self-administered questionnaires.

Now, more and more national surveys are asking respondents about their sexual orientation right in the core of the questions.  Self-reported sexual identity has now worked its way to the status of a (sometimes included) sociodemographic characteristic, along with gender, race and ethnicity, education, and so forth.  The most recent surveys, like a 2012 Gallup poll and the 2012 American National Election Survey (ANES), even asked whether adults identify as transgender.  In many ways, we have achieved LGBT inclusion in survey research.  But, in many ways, we are far from accurately representing LGBT adults and youth in social surveys.

“Shades” of Representation

The typical gold standard for survey research is representativeness – ideally, the collection of a large sample that reflects the “true” US population.  Because some segments of the population are small and/or hard-to-reach, additional strategies are sometimes employed to better achieve representativeness.  These strategies include oversampling (i.e., collecting a larger share of a specific subpopulation) and using sample weights to correct for sampling bias.  Social scientists have made great efforts toward achieving representation of LGBT people in quantitative research.  Indeed, several nationally representative surveys now include measures of sexual identity or sexual orientation.

But, what do we mean when we say we have collected a “nationally representative sample of LGB” youth or adults?  I think it is worth teasing out the nuances of “representation.”  Here are some important dimensions, in my opinion:

  • Capturing the “true” size of the LGBT population – LGBT individuals and (not “or”) same-gender couples.
  • Appropriately representing each segment of the diverse LGBT population – trans* people (e.g., transwomen, transmen, genderqueer people), lesbian women, bisexual women, bisexual men, gay men.
  • Effectively representing other axes of diversity in the LGBT population – race, ethnicity, immigrant status, nationality, body size and shape, religion, socioeconomic status, ability, age, marital/partnership status, parental status, geography (e.g., urban, rural, East Coast, South), and political ideology.
  • Representing the experiences, interests, and well-being of LGBT people (e.g., discrimination, political priorities).

Right now, we are just beginning to navigate the challenges of the first of these dimensions of representation – estimating the “true” size of the LGBT population.  In fact, one data collection service, Knowledge Networks, offers sampling procedures to collect representative samples of lesbian, gay, and bisexual (LGB) adults.  However, in doing some digging, I believe their means of means of oversampling LGB individuals and applying inclusive sample weights rely on the American Community Survey (ACS) estimates of “same-sex couples.”  Absent of a Census or similar comprehensive assessment of all LGBT-identified individuals in the US population, our best attempts at nationally representative samples of LGBT people are not necessarily “representative.”

Further, our first achievements of “nationally representative samples” of LGBT people may be underrepresenting certain segments of the LGBT population – namely, older LGB adults, bisexuals (especially men), LGB people of color, and LGB people in rural areas.  And, I must make painfully clear here that we are even further behind on collecting representative samples of trans* adults and youth.  Though we have the very large National Transgender Discrimination Survey, the earliest, very preliminary estimates of the size of the trans* adult population are based on the state of Massachusetts.

Why It Matters

For a number of reasons, it is important to 1) include at least one measure of sexual identity in surveys, 2) include at least one (inclusive) measure of gender identity, 3) include at least one (inclusive) measure of marital/partnership status, and 4) oversample LGBT people and/or apply LGBT-inclusive sample weights.  First and foremost, the evidence that sexual and gender identities are an importance aspect of our social world is undeniable.  The lives of LGBT people are uniquely shaped by their marginalized status – persistent discrimination, invisibility, income inequality, exclusion, relatively worse health and well-being, lack of relationship recognition, etc.  This appears to shape a distinct set of political behaviors and attitudes, one’s social networks, and likely a host of other aspects of the everyday lives of LGBT people.

Second, social scientists must correct the lingering heteronormative and cisnormative biases in survey research.  Surveys that fail to ask respondents’ their sexual identity, and that do not allow respondents to report same-gender partners contribute to the systematic erasure of LGB people and their romantic and familial relationships.  Too few surveys collect information on respondents’ sex-assigned-at-birth, current gender identity, and current gender expression.  Simply asking “female/woman” or “male/man” makes invisible those who are trans* and intersex.  Further, I worry that many researchers using telephone surveys continue to assume the respondent’s gender based on their voice, presumably to avoid offending them by asking.  Even for researchers whose primary interest is not in the lives of LGBT people, these biases are urgent matters for presenting accurate estimates in their research.

Finally, though many scholars do not agree or care to admit it, academic research has great power in defining the population.  In excluding measures that reflect or at least include LGBT people, we send the message that that population is unimportant to social science research.  Or, by slipping in “sex of sex partners” at the end of the survey or in a self-administered questionnaire, we signal the persistent taboo-ness of same-sexuality.  By using one catch-all item for LGBT identity, scholars (unintentionally) erase the diversity within LGBT communities.  But, by treating sexual identity and gender identity as core elements of sociodemographics, we make clear the importance and normalness of these aspects of individuals’ lives.  Why not take the position of having a positive impact on the lives of LGBT people?

We are, indeed, on our way to better representing LGBT people in our research.  In the mean time, there is much room for improvement.  And, it is important to extend and enhance research via other methods (e.g., respondent-driven sampling, qualitative methods), as well, to capture a comprehensive understanding of the identities, well-being, and experiences of LGBT people.

Below, I include a list of resources and additional readings that may be useful for future LGBT research.


Additional Reading

Baulme, Amanda K., and D’Lane R. Compton.  2014.  “Identity Versus Identification: How LGBTQ Parents Identify Their Children on Census Surveys.”  Journal of Marriage and Family 76: 94-104.

Black, Dan, Gary Gates, Seth Sanders, and Lowell Taylor.  2000.  “Demographics of the Gay and Lesbian Population in the United States: Evidence from Available Systematic Data Sources.”  Demography 37: 139-54.

Bradford, Judith B., Sean Cahill, Chriss Grasso, and Harvey J. Makadon.  2012.  Why Gather Data on Sexual Orientation and Gender Identity in Clinical Settings: Boston, MA: The Fenway Institute.

Bradford, Judith B., Sean Cahill, Chriss Grasso, and Harvey J. Makadon.  2012.  How to Gather Data on Sexual Orientation and Gender Identity in Clinical Settings: Boston, MA: The Fenway Institute.

Dilley, Julia A., Katrina Wynkoop Simmons, Michael Boysun, Barbara A. Pizacani, and Mike J. Stark.  2010.  “Demonstrating the Importance and Feasibility of Including Sexual Orientation in Public Health Surveys: Health Disparities in the Pacific Northwest.”  American Journal of Public Health 100: 460-7.

Egan, Patrick J., Murray S. Edelman, and Kenneth Sherrill.  2008.  “Findings from the Hunter College Poll of Lesbians, Gays and Bisexuals: New Discoveries about Identity, Political Attitudes, and Civic Engagement.”  Hunter College, The City University of New York, New York, NY.  Unpublished Manuscript.

Gates, Gary J.  2011.  How Many People are Lesbian, Gay, Bisexual, and Transgender?  Los Angeles: The Williams Institute.

Gates, Gary J.  2013.  “Demographics and LGBT Health.”  Journal of Health and Social Behavior 54: 72-4.

Herek, Gregory M., Aaron T. Norton, Thomas J. Allen, and Charles L. Sims.  2010.  “Demographic, Psychological, and Social Characteristics of Self-Identified Lesbian, Gay, and Bisexual Adults in a US Probability Sample.”  Sexuality Research and Social Policy 7: 176-200.

Herek, Gregory M.  2009.  “Hate Crimes and Stigma-Related Experiences Among Sexual Minority Adults in the United States: Prevalence Estimates From a National Probability Sample.”  Journal of Interpersonal Violence 24: 54-74.

Institute of Medicine.  2011.  The Health of Lesbian, Gay, Bisexual, and Transgender People: Building a Foundation for Better Understanding.  Washington, DC: The National Academies Press.

Meyer, Ilan H., and Patrick A. Wilson.  2009.  “Sampling Lesbian, Gay, and Bisexual Populations.”  Journal of Counseling Psychology 56: 23-31.

Sell, Randall L.  2010.  “Defining and Measuring Sexual Orientation for Research.”  Pp. 355-74 in The Health of Sexual Minorities: Public Health Perspectives on Lesbian, Gay, Bisexual and Transgender Populations, edited by I. H. Meyer and M. E. Northridge.  Springer: New York.

Sexual Minority Assessment Research Team (SMART).  2009.  Best Practices for Asking Questions about Sexual Orientation on Surveys.  Los Angeles: The Williams Institute, UCLA.

Tourangeau, Roger, and Tom W. Smith.  1996.  “Asking Sensitive Questions: The Impact of Data Collection Mode, Question Format, and Question Context.”  Public Opinion Quarterly 60: 275-304.

VanKim, Nicole A., James L. Padilla, Joseph G. L. Lee, and Adam O. Goldstein.  2010.  “Adding Sexual Orientation Questions to Statewide Public Health Surveillance: New Mexico’s Experience.”  American Journal of Public Health 100: 2392-6.