Introduction and summary
Data collection is an indispensable tool to understand and address challenges facing LGBTQI+ and other sexual and gender-diverse1 communities. Although strides have been made in recent years, a persistent lack of routine data collection on sexual orientation, gender identity, and variations in sex characteristics (SOGISC) is still a substantial roadblock for policymakers, researchers, service providers, and advocates seeking to improve the health and well-being of LGBTQI+ people. More comprehensive and accurate point-in-time and longitudinal demographic data on SOGISC are crucial to:2
- Advance research agendas
- Evaluate population trends
- Identify community-based needs
- Provide high-quality services
- Track and address discrimination
- Equitably distribute funding and other resources
- Shape evidence-based policy solutions to promote equity and reduce disparities faced by LGBTQI+ populations
As the size and diversity of LGBTQI+ populations in the United States continue to expand, particularly among youth and young adults, the importance of collecting data on these communities only continues to grow.3 Failing to collect these data can create harms by hindering the ability of researchers, policymakers, service providers, and advocates to understand the experiences of LGBTQI+ communities, identify disparities, generate policies that promote equity, and evaluate the effectiveness of those policies. Yet currently, the number of federally funded surveys that include questions to identify LGBTQI+ respondents is limited.4
While there are many kinds of data relevant to the experiences of LGBTQI+ communities, this report focuses specifically on data collection to capture SOGISC in two types of settings: general population surveys and surveys of LGBTQI+ communities. General population surveys assess a large sample of the entire population, of which the majority identify as cisgender and heterosexual and will not have intersex traits. In contrast, LGBTQI+ community-based surveys sample a population predominantly comprising sexual and gender minorities. Community-based surveys can provide important opportunities to learn more about even smaller or less-studied sexual and gender minority populations, such as people who are asexual, same-gender-loving, or Two-Spirit. Because of their specific focus on sampling LGBTQI+ populations, community-based surveys may also be more likely to reach larger proportions of groups such as LGBTQI+ people of color, older adults, youth, people with disabilities, people receiving social welfare support services, people whose primary language is not English, and people interacting with the criminal legal system.
When designing questions to accurately and effectively survey either the general population or a predominantly LGBTQI+ population, researchers must carefully weigh a number of key considerations that are explained in this report. Regardless of the type of survey, however, it is imperative to ensure the following:
- Entities collecting demographic data, including data related to SOGISC, do so with a specific and well-defined goal, such as collecting statistics on health experiences or understanding the performance of a government benefit program.
- Data are collected, used, maintained, and shared only with strong privacy, confidentiality, and ethical standards5 in place to minimize the risk of data disclosure and misuse.
- Entities collecting data adopt and post clear nondiscrimination and confidentiality policies. These policies should identify relevant legal nondiscrimination protections; state why the data are being collected and how they will be used; share contact information for resources to enforce protections;6 and provide assurance that confidentiality will be respected and that participation is voluntary, allowing respondents to provide well-informed consent with the knowledge that disclosure is voluntary and that they have the right to opt out of responding.
- Research and researchers comply with and are certified through the Collaborative Institutional Training Initiative and consult community-based resources7 about how to conduct effective and ethical research with LGBTQI+ populations that ensures minority populations have a voice and role in the design of survey questions.
- LGBTQI+ people are meaningfully involved in question development, testing, and the evaluation process.
This report examines best practices and key considerations for collecting data on SOGISC in both general population surveys as well as LGBTQI+ community-based surveys. The first section of the report highlights evidence related to asking these questions in general population surveys and examines critical considerations, as well as avenues of future research that policymakers and researchers should support. The second section of the report compiles both evidence from existing survey designs and interviews with LGBTQI+ individuals to create new suggestions and recommendations for SOGISC question design in surveys of LGBTQI+ communities. Ultimately, improving SOGISC data collection through these kinds of general population and community-based instruments is crucial to identifying disparities and crafting policy solutions that promote more equitable outcomes for LGBTQI+ communities. To view a glossary of the terms that appear throughout the report, expand the text box below.
Asexual: An adjective that refers to a complete or partial lack of sexual attraction or interest in sexual activity with others. Asexual people may experience other forms of attraction and identify as “bisexual,” “gay,” “lesbian,” “pansexual,” “queer,” “straight,” and many other sexual orientation identities.
Bisexual: An adjective that describes individuals who have the potential to be physically, romantically, and/or emotionally attracted to people of more than one gender, although not necessarily to the same degree, in the same way, or at the same time.
Cisgender: A term used to describe people whose gender identity corresponds to that typically associated with the sex they were assigned at birth.
False negatives: A form of measurement error that occurs when a member of a minority population identifies as part of a majority population—for example, if a transgender person reports they are cisgender on a survey.
False positives: A form of measurement error that occurs when a majority population misidentifies as part of a minority population—for example, if a cisgender person incorrectly reports they are transgender on a survey.
Gay: An adjective that describes individuals who are physically, romantically, and/or emotionally attracted to people of the same gender.
Gender: “A multidimensional construct that links gender identity, gender expression, and social and cultural expectations about status, characteristics, and behavior that are associated with sex traits.”8
Gender expression: How an individual expresses their gender to themselves and/or others—including elements such as hair style, clothing, voice and body characteristics, mannerisms, and behavior.
Gender identity: An individual’s internal sense of their own gender. This concept blends both how individuals perceive themselves and what they call themselves.
Gender-nonconforming: An individual with a gender identity/expression that does not conform to the social or cultural norms of the gender they were assigned at birth. A gender-nonconforming person may or may not consider themselves to be transgender.
Genderqueer: An individual with a gender identity/expression that does not conform to the social or cultural norms of the gender they were assigned at birth. A genderqueer person may or may not consider themselves to be transgender.
Intersex: An umbrella term that describes individuals whose sex traits do not fit binary definitions of male or female sexual or reproductive anatomy. People with intersex traits may be born with these differences in sex characteristics or develop them during childhood.
Lesbian: A term generally used to describe women who are primarily physically, romantically, and/or emotionally attracted to people of the same gender. However, some nonbinary people also describe themselves as lesbians, often because they may have some connection to womanhood and are primarily attracted to women.
LGBTQI+: An acronym for “lesbian, gay, bisexual, transgender, queer, intersex” with a “+” sign to recognize the diverse array of sexual orientations and gender identities used by sexual and gender-diverse communities.
Measurement error: A term used to describe the difference between a measured quantity and its true value. This can occur in the form of random errors that are to be expected as well as systematic errors that can seriously affect data quality.
Nonbinary: An umbrella term used to describe people with gender identities that fall outside the gender binary. The term can refer to people who do not identify exclusively as a man or a woman, including those who identify as being both a man and a woman, somewhere in between, or completely outside these categories. Some nonbinary people may identify as transgender, but not all nonbinary people identify as transgender.
Pansexual: An individual who is emotionally/physically attracted to others regardless of their gender identity.
Queer: “An umbrella term used to describe a sexual orientation, gender identity or gender expression that does not conform to dominant societal norms. While it is used as a neutral, or even a positive term among many LGBT people today, historically ‘queer’ was used as a derogatory slur.”9
Same-gender-loving: A term used to describe attraction to the same gender, predominantly used by African Americans as an alternative to what are considered Eurocentric or white terms, such as “gay” and “lesbian.”
Sex: A term used to indicate an individual’s assigned sex, not gender identity, based on a cluster of anatomical and physiological traits, also known as sex characteristics.
Sex characteristics: A term used to describe traits associated with sex. Variations in sex characteristics may appear with respect to an individual’s chromosomes, genitals, or internal organs.
Sexual orientation: An individual’s inner feelings of physical, romantic, and/or emotional attraction with respect to other people.
Transgender: An umbrella term used to describe people whose gender identity differs from cultural expectations based on the sex they were assigned at birth.
Two-Spirit: An umbrella term to describe some Indigenous and First Nations individuals who are not straight and/or cisgender or who possess qualities or fulfill roles of more than one gender. This term acts as an English-language placeholder for tribally specific gender and sexual orientation identities that are centered in tribal worldviews, practices, and knowledges. Many Indigenous communities have specific words in their language to describe these experiences. This term should not be used to describe people who are not Indigenous.
Part 1: The need to collect SOGISC data in general population surveys
General population surveys are used to collect data from representative samples of the population, allowing researchers to create population estimates, describe population demographics, assess disparities, compute statistical weights for survey analysis, and develop policy, program, and funding priorities.10 Federal, state, and local governments rely on data from general population surveys such as the decennial census, the American Community Survey (ACS), and the Current Population Survey (CPS) for everything from allocating budget resources to deciding where to locate hospitals.11
Lack of data collection on LGBTQI+ people in general population surveys
The persistent lack of routine data collection on SOGISC remains a significant challenge for policymakers, researchers, service providers, and advocates seeking to improve the health and well-being of LGBTQI+ people. Currently, most surveys, including those fielded by the U.S. federal government and many state governments, do not collect SOGISC information.12 While some surveys, such as the census and the ACS, now invite people to answer questions about their marital status or living arrangements in ways that allow researchers to identify same-sex couples, this only captures a small segment of the LGBTQI+ population.13 By not asking specific questions about SOGISC, these surveys fail to account for single LGB people or LGB people who are in a relationship but not cohabitating with their partner(s), as well as transgender people and intersex people altogether. For example, according to the latest Gallup data, just 10 percent of LGBT adults in the United States are married to a same-sex spouse, while an additional 6 percent live with a same-sex domestic partner.14 This means that more than 5 in 6 LGBT adults cannot be identified by existing questions in nonexperimental surveys conducted by the U.S. Census Bureau such as the ACS and the census, which only account for cohabitating same-sex couples.15 Failing to collect SOGISC data creates real harms—hindering the ability to fully understand the experiences of LGBTQI+ communities; to craft sound policies that are inclusive of LGBTQI+ people and their needs; and to evaluate the effectiveness of policies to address disparities and promote more equitable outcomes.
More than 5 in 6 LGBT adults cannot be identified by existing questions in nonexperimental surveys conducted by the U.S. Census Bureau such as the ACS and the census.
Adding SOGISC questions to general population surveys is crucial to advancing equity
As recommended by the National Academies of Sciences, Engineering, and Medicine in their 2020 report, U.S. population surveys must routinely collect, analyze, and report demographic data that include SOGISC questions.16 The information gathered by general population surveys shapes major policy decisions and allocations of critical resources related to health care, housing, employment, education, and other services and benefits, affecting the everyday lives of LGBTQI+ people and making the need to adopt SOGISC measures even more urgent. Notably, expanding data collection on SOGISC through government population-based surveys will result in a larger sample size than in privately conducted surveys. This is important because larger samples allow for better, more reliable study and a richer understanding of the diversity of the LGBTQI+ community. Moreover, having larger samples will facilitate analysis and provide greater comprehension of the experiences of populations that are living at the intersection of multiple marginalized identities.
Expanding and enhancing SOGISC data collection also provides critical tools necessary to craft policy solutions to improve outcomes and advance equity. By improving data collection on LGBTQI+ populations, government agencies can take meaningful steps to fulfill the directives set out by executive orders 1398817 and 13985,18 which President Joe Biden signed on day one of his presidency.
Key executive orders to advance equity
Executive order 13988 directs all federal agencies that enforce federal laws prohibiting sex discrimination to also prohibit discrimination based on sexual orientation and gender identity and to take actions necessary to fully implement this policy. This executive order implements the U.S. Supreme Court decision in in Bostock v. Clayton County,19 affirming that Title VII of the Civil Rights Act of 1964’s prohibition on sex discrimination in employment encompasses discrimination based on sexual orientation and gender identity.
Executive order 13985 directs federal agencies to promote equity for LGBTQ+ and other underserved communities through various actions, including but not limited to expanding data collection efforts with the support of the Equitable Data Working Group. This approach includes identifying inadequacies in existing federal data collection, supporting agencies in addressing those deficiencies, and improving the data to help the government assess equity.
Recommendations for the Biden administration on adding SOGISC questions to general population surveys
It is crucial that the Biden-Harris administration adopt a whole-of-government approach and provide meaningful guidance to support routine collection, analysis, and reporting of demographic data on SOGISC in U.S. population surveys. The administration must prioritize actions to add SOGISC measures to general population surveys, which will provide valuable information to identify and address existing and long-standing health and economic disparities20 and to promote more equitable outcomes for LGBTQI+ communities. Specifically, CAP recommends that:
- The U.S. Department of Commerce add SOGISC measures to the ACS,21 which collects information about the social, economic, housing, and demographic characteristics of roughly 3.5 million households each year across the country, and federal agencies use these data to monitor and enforce employment discrimination laws.
- The U.S. Department of Commerce, in partnership with the U.S. Department of Labor, add recommended SOGISC measures to the CPS,22 which gathers crucial information about employment and unemployment; emerging trends in employment status, wages, and earnings; and variables affecting labor force participation.
- The U.S. Department of Commerce add SOGISC measures to the Survey of Income and Program Participation,23 a longitudinal survey that provides important information on income, employment, household composition, and participation in government programs.
- The U.S. Department of Health and Human Services make permanent measures of gender identity and add questions on variations in sex characteristics to the National Health Interview Survey,24 which provides information to monitor trends in health status, determine barriers to accessing care, and evaluate progress to achieving national health objectives. The survey currently includes questions about sexual orientation and gender identity but none about variations in sex characteristics.
- The U.S. Department of Health and Human Services add SOGISC questions to the standardized demographic core questionnaire of the Behavioral Risk Factor Surveillance System, which represents the largest continuously conducted health survey system in the world. Currently, the survey includes questions about sexual orientation and gender identity only as an optional module and contains no measures of variations in sex characteristics.
- The U.S. Department of Health and Human Services add questions on gender identity and intersex traits to the standard Youth Risk Behavior Surveillance System questionnaire, which monitors behaviors and experiences related to the health and well-being of young people.25
It is imperative that the adoption of SOGISC measures in these general population surveys is accompanied by significant investments in future research. Funding and advancing research on these constructs, as well as engaging in periodic assessment of measures, is crucial to ensuring the relevant agencies continue to improve these surveys and update them to reflect the evolving evidence base.
SOGISC questions can function well in general population surveys
LGBTQI+ advocates have long called for the federal government and researchers to add SOGISC questions to major surveys and other data collection efforts as a way to begin addressing the deficiency of data on LGBTQI+ populations and gaps in knowledge and policy affecting LGBTQI+ people.26 Government officials, researchers, and others have questioned whether the general population, composed mainly of non-LGBTQI+ people, would understand these questions, answer them accurately, or even refuse to answer them. The answers to these questions are important because they affect the quality of data gathered about LGBTQI+ people and, by extension, the resources allocated, decisions made, and policies created that affect LGBTQI+ communities.
Importantly, numerous federally supported entities and other expert bodies have issued reports to determine methodological best practices and improve measurement of SOGISC in federal surveys.27 These groups have contributed to a robust and continually growing body of research providing evidence that SOGISC questions can be readily deployed in federally funded and other surveys. For example, evidence indicates that:
- Sexual orientation and gender identity data are not considered especially difficult or sensitive for survey respondents to report,28 meaning that people generally understand what the questions are asking and are willing to answer them.
- Securing participation of sexual minorities in surveys does not require higher levels of effort,29 meaning it will not create added costs or obstacles that would prohibit the government or other researchers from conducting more inclusive data collection.
- People with intersex traits appear willing to disclose their status and support collecting data on this measure in research.30
- People will answer sexual orientation and gender identity questions even across a variety of modes (for example, telephone surveys and paper surveys) and via both self-reporting and proxy reporting (for example, when a single household member responds on behalf of all household members) in federal large-scale general population surveys.31
In other words, existing evidence addresses many of the major hesitations or questions posed by government researchers and others. Put simply, while more research is needed to continue improving SOGISC measures, evidence suggests these questions can function well in major general population surveys, and they should be asked. For example, in 2021, the Census Bureau updated its experimental data collection effort on the COVID-19 pandemic to include questions about sexual orientation and gender identity in the Household Pulse Survey. This historic step marks the first time a Census Bureau-sponsored survey has asked sexual orientation and gender identity questions and highlights the ways in which these questions can—and do—work in large, nationally representative surveys.32 Expanding LGBTQI+-inclusive data collection by asking SOGISC questions on general population surveys is paramount to advancing equity for LGBTQI+ communities.
Key considerations when adding SOGISC questions to general population surveys
Researchers and policymakers interested in improving data collection on LGBTQI+ communities in general population surveys must carefully weigh numerous considerations when designing SOGISC questions.
What is measurement error and why does it matter for LGBTQI+ people?
When conducting surveys, it is imperative to reduce measurement error as much as possible. Measurement error describes the difference between a measured quantity and its true value. This can include false positives—when a majority population misidentifies as part of a minority population, such as cisgender people incorrectly identifying themselves as transgender on a survey—or false negatives—when members of a minority population misidentify as part of a majority population.
Especially for relatively small populations, such as the LGBTQI+ community, it is necessary to consider how measurement error can lead to biased understandings of the size, demographics, and experiences of the minority population. In the case of LGBTQI+ data collection, false positives are a prevalent concern. For example, on the 2010 census, when just a small proportion of different-sex married couples misidentified themselves as same-sex married couples, it resulted in a significant error. In fact, the mistaken census estimate of same-sex households in the country was more than 52 percent higher than in the ACS.33
False negatives occur when minority individuals report majority identities. This can happen when minority respondents do not understand survey terminology; find that the language offensive or that it does not match how they describe themselves; or do not feel comfortable reporting their minority identities due to fear of discrimination or a sense of stigmatization.34 Each of these scenarios can similarly skew results and give an incomplete or inaccurate picture of the demographics and experiences of a minority population.
This highlights the need for standardizing SOGISC questions on general population surveys as well as the importance of pretesting measures through cognitive interviews and experimental studies with members of minority populations in order to detect and reduce the risk of measurement error and improve data quality, usability, and validity.35 Doing so is essential to ensuring the needs of LGBTQI+ communities are reflected in policies, programming, and funding.
Terminology: Balancing community responsiveness and reducing measurement error
In general population surveys, cisgender, heterosexual people without intersex traits are the majority of survey respondents, making up more than 90 percent of the U.S. population.36 When designing general population surveys, researchers must balance the goal of using terminology and questions that reflect LGBTQI+ identities, language, and experiences with the goal of ensuring non-LGBTQI+ people also understand the questions and will answer accurately—in other words, reducing measurement error.
Population-based surveys should use terminology for SOGISC that all respondents, including those who are not LGBTQI+, will understand. In contrast, and as discussed in more detail in the following section, LGBTQI+ community-based surveys may provide an opportunity for researchers to use a broader range of terminology options to capture more of the diversity of LGBTQI+ identities among a sample that is generally more fluent in those terms and concepts.
When administering a general population survey to a sample of people whose primary language is not English, SOGISC questions must also be carefully and accurately translated into languages most relevant to the audience. When surveying communities that may use culturally specific terms—for example, “Two-Spirit” or linguistically or culturally specific terms among many Native American communities,37 and “same-gender-loving” among some African American communities—surveys should adapt response options to employ these terms. In some instances, providing definitions may improve respondents’ understanding of key terms and reduce the risk of measurement errors.
Response options: Balancing community responsiveness and reducing measurement error
Large population-based surveys often employ multiple-choice questions to capture how respondents identify—for example, selecting their race or ethnicity from a list of provided response options. Determining which response options to include requires considering challenges that could arise in both the data collection and the data analysis processes, which may have ethical implications. There are some strategies to balance these concerns, but each has its strengths and weaknesses.
One approach on general population surveys is to prioritize a relatively short, preset list of answers that perform well and, together, capture the majority of respondents’ identities. This may reduce measurement error and make analysis of the data faster and easier, which means researchers can share the findings and help guide policy and programs more quickly. However, this approach may be exclusionary, leaving out the full range of identities and labels with which respondents might identify and rendering small population groups invisible. Alternatively, a second approach is for the question to instead include a much longer list of identities for people to choose from, but this greatly increases the likelihood for measurement error and jeopardizes the usability of the data—especially on a general population survey. A third strategy could be to keep a short list of options but add a free-text or write-in option that allows individuals to describe themselves in their own words.
Both of the latter approaches may add a significant time burden for the researcher and increase the likelihood that some data will be collected and simply not used. This is because an analyst working with the responses must then decide how to tabulate, report, and recode responses into a numeric value necessary for statistical analysis. In such cases, the onus falls on the analyst to decide how best to represent the write-in response, which can lead to them regrouping that individual’s identity into a category that the respondent did not choose or even excluding them from the analysis altogether. Despite these considerations, it is worth emphasizing the significant strengths of the third approach, which, as previously noted, allows respondents to exercise autonomy in self-describing how they identify beyond the limited list of response options. In addition, write-in data allow researchers to monitor and analyze the use of terminology over time, informing decisions about adding new response categories in the future. For these important reasons, the third strategy should be employed whenever possible.
General population survey question designs
In March 2022, the National Academies of Sciences, Engineering, and Medicine published a milestone report summarizing their expert panel’s review of methodological issues and research on measures for collecting data on SOGISC in three settings, including general population surveys.38 That report presents recommendations for these measures based on a set of five guiding principles for data collection that should inform efforts to improve gathering information on LGBTQI+ communities:39
- Inclusiveness: People deserve to count and be counted.
- Precision: Use precise terminology that reflects the constructs of interest.
- Autonomy: Respect identity and autonomy.
- Parsimony: Collect only necessary data.
- Privacy: Use data in a manner that benefits respondents and respects their privacy and confidentiality.
When developing the recommendations consistent with these principles, the panel relied on specific evaluation criteria to ensure they selected measures and response options that met scientific standards for validity and quality, while balancing considerations about community responsiveness and usability by the general population in order to reduce measurement error.40 Accordingly, the panel based their recommendations on testing and research focused on both LGBTQI+ communities and the general population and aimed to ensure that each question captures the right information, that respondents understand the question, that respondents are willing to answer the question, and that they can do so easily and as intended.41 See the Appendix for the specific scientific criteria the panel used to inform their recommendations on measures and response options.
Based on the current state of the evidence base for general population-based surveys, the National Academies report recommends the below question designs for measures of SOGISC. Each explanation of the measure is accompanied by a section on its limitations, as well as CAP’s recommended priorities for ongoing testing and evaluation to continue to improve these measures, which should be supported through funding investments. CAP’s priorities for ongoing testing largely align with avenues for future research in the National Academies report. While more details about each particular measure are provided below, the following set of principles apply to approaches across all measures and can be used to guide approaches to asking these questions on general population surveys:
- Include sexual and gender minorities, as well as people with intersex traits, in question development, testing, and evaluation processes.
- Use questions that clearly specify which component(s) of gender identity, sexual orientation, and sex, including variations in sex characteristics, are being measured. One construct should not be conflated with or used as a proxy for another.
- Sex characteristics, sexual orientation, and gender identity can change over time, so question stems should use time-bound language—for example, use “current gender” in the question stem for measures of gender identity.
- When possible, include a write-in option for questions about sexual orientation and gender identity and analyze responses using a process that is inclusive of sexual and gender minorities, with the intent of improving future question design.
- Avoid “othering” language in response options—for example, use “another gender” instead of “other” or “something else.”
- If a question requires respondents to answer it, include an opt-out component for respondents.
Asking about sexual orientation identity in general population surveys
Sexual orientation can be conceptualized as comprising at least three distinct but related variables: identity, behavior, and attraction.42 However, questions related to each variable can be used individually or together on surveys, depending on the survey goal.43 Because the measure of sexual orientation identity is most routinely used44 and consistently most relevant to assess risk of discrimination and differential outcomes,45 that construct is the focus of this section and of the National Academies report’s recommendations.
It is important to note that questions about sexual orientation identity do not exhaustively describe the complexities of sexual orientation. Some people, particularly youth, may reject labels for their sexual orientation. Others may engage in same-sex sexual behavior or experience same-sex sexual attraction but identify as straight. And still others may use terms that blend concepts of sexual orientation and gender.46 Sexual orientation identity may also change over time, along with the terms people use to describe themselves, so periodic reassessment is necessary on longitudinal, that is over-time, studies.47
The National Academies’ recommended measures include answer options for “lesbian,” “gay,” and “bisexual,” which have remained popular for decades and can be comprehended by respondents, while also providing a free-text option for those who use different terms to describe their sexual orientation identity. The panel also recommends automatically adding “Two-Spirit” as a choice for respondents who have identified themselves as American Indian or Alaska Native. Researchers working with specifically Indigenous communities should also consider including identity terms in tribal languages.
The National Academies panel’s recommended measure for sexual orientation identity
Which of the following best represents how you think of yourself? [Select one]
1. Lesbian or gay
2. Straight, that is, not gay or lesbian
4. [If respondent is American Indian or Alaska Native] Two-Spirit
5. [Free-text option] I use a different term:
(Prefer not to answer)
Sexual orientation identity questions: Limitations
Based on the current state of evidence and evaluation criteria selected by the National Academies panel, this question performs best on general population surveys. However, there are clear weaknesses that should be addressed through future research to improve quality and inclusivity of this question. For example, the set of response options lists the terms “gay,” “lesbian,” and “bisexual” but does not reflect the emergence of newer identities used by many sexual minorities. Additionally, the “straight, that is, not gay or lesbian” wording reinforces heterosexual as the normative is conceptually inaccurate and not applied to the other sexual minority identities listed—that is to say, bisexual or Two-Spirit.
Priorities for ongoing research: Questions about sexual orientation identity
CAP recommends multiple priorities for future research. Future research should examine expansion of response options that better reflect current culture and terms such as “queer,” “pansexual,” and “asexual,” which are popular among younger age groups48 and may therefore continue to grow in prominence in future years. Assessment of free-text responses should inform efforts to determine whether new categories should be added. Additionally, the “straight, that is not gay or lesbian” phrase should be reevaluated to determine if the “that is, not gay or lesbian” wording is still necessary for heterosexual people to answer the question, and, if so, alternative language that is more accurate and less stigmatizing should be assessed. Finally, CAP also supports the National Academies recommendation for additional research to assess performance of this construct in proxy reporting settings, languages other than English, and for major racial and ethnic populations, as well as youth.
Asking about gender identity in general population surveys
The 2022 National Academies report focuses on gender identity questions for general population surveys that allow people to name their gender identity, while also allowing researchers to identify people of transgender experience—that is, any person whose gender identity differs from their sex assigned at birth, regardless of whether or not they explicitly identify with the term “transgender.”49 This is important because, as in any community or demographic group, the language and terms used vary widely. The National Academies panel recommends what is called a two-step approach, which asks about both current gender identity and sex assigned at birth. This approach enables people to be recognized as transgender if they select “transgender” as their current gender identity or if they select a gender that is different from the sex they were assigned at birth, thereby including those who may not use the term “transgender” to describe themselves.50
Importantly, the recommended question for current gender identity includes the option to select “I use a different term” and to write in a different gender identity. This would allow individuals who, for example, feel that the preset list of options does not describe their gender identity, to name themselves in their own terms. For example, respondents who identify as nonbinary,51 or who feel that being transgender is not separate from their gender identity—that is to say, not only “male” or not only “transgender,” but specifically “transgender man”—and others could all use this option to self-identify in their own terms. This free-text option will also allow researchers to observe which identities are most frequently mentioned or which are growing in use, which will help guide future improvements of this question. This is a notable improvement over other variations of the two-step question that do not provide free-text options.52 The recommended approach to gender identity questions attempts to simultaneously allow individual autonomy in self-identifying, while also capturing enough data to provide researchers and LGBTQI+ communities with meaningful information about the large, diverse group of people whose gender identity differs from their sex assigned at birth.
For the two-step question to perform as intended, both components of the question—current gender identity and sex assigned at birth—must be asked sequentially and used in combination to create counts for both transgender and cisgender people. When administering a two-step gender identity question of any kind, it is important to display both components simultaneously, clearly note that both parts will be assessed, and accurately process both responses for analysis and reporting purposes. To minimize false positives and misclassification of a respondent’s gender identity, a third confirmation question is often added to verify differing responses to the first two questions.53
The advantages of this two-step question design are that it is inclusive of people with transgender experience and not just those who identify with the term “transgender” and that it is easy for cisgender respondents to respond to, thus reducing the risk of false positives that may invalidate the data. The two-step measure, with the potential addition of a third confirmation step, can also replace existing stand-alone questions that measure sex or gender in many data collection contexts, as these current nonspecific measures often conflate the two constructs.54 As is the case with the sexual orientation question design above, a “Two-Spirit” or tribal language response option should be available in circumstances where automated data collection is being used and the respondent has identified as American Indian or Alaska Native.
The National Academies panel’s recommended two-step gender identity question
Q1: What sex were you assigned at birth, on your original birth certificate?
(Prefer not to answer)
Q2: What is your current gender? [Mark only one]
4. [If respondent is American Indian or Alaska Native] Two-Spirit
5. [Free-text option] I use a different term:
(Prefer not to answer)
Gender identity questions: Limitations
While evidence finds that this two-step approach performs best and can be asked on general population surveys, there are notable limitations that should be further addressed through research and testing. Some research demonstrates that “current survey measures and terminology for gender identity do not work well for all transgender individuals, resulting in variations in reporting gender identity.”55 One key concern is that some gender minority respondents may find answering questions about sex assigned at birth or disclosure of transgender status through the cross-classification process of the two-step question to be problematic, intrusive, or invalidating.56 It’s also important to note that some gender-diverse respondents may self-identify using multiple terms and that the gender terminology respondents use and recognize may differ based on a variety of factors and evolve over time.57
For example, the current question design on most general population surveys forces respondents to choose between selecting a male or female gender identity option and a transgender identity option, although those are not mutually exclusive; for example, some transgender women and men may want to select female or male, as well as transgender. In addition, there are nonbinary individuals who often do not identify as male or female but may not identify with the term transgender. In terms of terminology, the recommended response options are limited, and although the write-in field will require cleaning and coding for reporting purposes that may be burdensome on the back end, it will also yield valuable data that will allow researchers to assess whether popular new terms that are not currently listed should be added as options. Periodic assessment of these write-in and listed response options is crucial.
Priorities for ongoing research: Questions about gender identity
CAP recommends multiple avenues for future research. Employing a two-step approach to measure gender identity using two questions that interact to provide adequate information about a respondent’s gender and transgender identity, status, or experience is a valuable approach. However, there is additional testing and evaluation that may create a more intuitive set of choices and reduce discomfort or aversion to gender identity questions among transgender and gender-diverse respondents.
For example, additional testing is recommended to evaluate the feasibility of providing respondents the option to select multiple gender identity responses,58 since the term “transgender” by itself—as opposed to in combination, such as “transgender man” or “transgender woman”—is not commonly used by transgender people in the United States to describe their gender. If the gender identity question only allows respondents to select one response option, mutually exclusive response options should be used. However, if response options are not mutually exclusive, respondents should be allowed to select all that apply.
Another priority area of testing is to evaluate the effects of removing the answer option of “transgender,” which is found in most published versions of the two-step question, and add the answer options of “nonbinary” and “another gender: [free-text response].” Removing the “transgender” response option could be valuable because many binary transgender respondents do not select the transgender label by itself to describe themselves;59 in practice select “male” or “female” options; and may find the inclusion of a separate “transgender” option to be confusing or communicate that transgender men and women are not respectively male and female.60 Meanwhile, priority testing should evaluate adding the term “nonbinary,” which is more commonly used by itself among nonbinary people in the United States to describe their gender identity and is increasingly growing in popularity,61 as well as the inclusion of a free-text response option that allows gender-diverse respondents who prefer another term to enter it.
In circumstances where sex assigned at birth is not relevant but it is still important to enumerate transgender people, it may be more appropriate to adopt a modified two-step question that does not rely on sex assigned at birth. This approach would combine a question about current gender, including a free-text option, with a transgender identity question such as “Are you transgender?” or “Do you consider yourself to be transgender?” and provide a free-text field for those who select “Yes” to describe themselves in their own words. This kind of approach is currently under testing and evaluation as part of the New Zealand government’s statistical standard for gender, sex, and variations of sex characteristics, the results of which may be highly valuable to researchers in the United States as they continue to refine and improve question design for general population surveys.62
Additionally, the current evidence base supports asking about sex assigned at birth first, and current gender identity second, but little evidence on the ordering effects of the two-step question is available.63 To reduce discomfort and question aversion among transgender respondents, researchers should test for ordering effects of the two-step question in general population surveys. This may help to address concerns with the current ordering of the two-step question, which some respondents find implies that their sex assigned at birth is their “real” gender and denies the lived experience of transgender individuals.64
Additional testing is also needed to assess the use of the confirmation question following the two-step gender identity question. Currently, the confirmation question is only asked of respondents who indicate that their sex assigned at birth differs from their current gender in order to reduce false negatives that can substantially affect the reliability and validity of population estimates. However, this approach leads to a disproportionate survey burden for transgender populations,65 who may also find it othering to be asked to confirm their responses are accurate. CAP recommends testing the utility of employing the confirmation question for respondents regardless of answers selected for the components of the two-step question in order to capture and reduce both kinds of measurement error.66 It may also be beneficial to include wording that explains the confirmation question is being used to limit measurement error.
Finally, CAP also supports the National Academies recommendation for ongoing testing to evaluate the effects of replacing the use of sex terminology (female/male) with gender terminology (woman/man) and to assess how gender identity measures perform in proxy reporting settings, languages other than English, and for major racial and ethnic populations, as well as for youth.67
Asking about variations in sex characteristics in general population surveys
Variations in sex characteristics refer to a broad range of physical differences in genitals, chromosomes, gonads and other internal reproductive structures, and hormone function that are present at birth, or develop spontaneously later in life, and fall outside what is considered typical within binary medical definitions of “male” or “female.”68 The term “intersex” is often used to describe people whose bodily differences fit this definition.69 There are many types of intersex variations, and estimates based on a partial list of some of the most well-known variations suggest at least 1.7 percent of people are intersex.70
Researchers have identified various challenges related to collecting data on intersex populations. First, some individuals that have intersex traits do not consider themselves to be intersex.71 This may reflect a narrow view of which variations “qualify” as intersex, or an individual may view the term “intersex” as implying a particular experience of gender. As with other sexual and gender minority populations, data collection on intersex populations must take into account the fact that the language used to describe intersex traits and people who have them is complicated and continues to develop.72 Second, some people are not aware of their own intersex traits either because knowledge of their variation has been intentionally concealed from them, or simply because their variation has not yet manifested or been diagnosed.73 Third, and related to the previous points, intersex variations are still stigmatized in many ways,74 especially in medical contexts,75 which may deter individuals from self-identifying as intersex.76 All of these factors pose challenges and may contribute to an undercount of intersex individuals in data collection efforts.
On the other hand, some people who do not have intersex traits may self-identify as intersex on surveys, whether because they view “intersex” as a synonym for transgender or nonbinary or because they otherwise misunderstand what the question is measuring, which may result in false positives.77 Lastly, while some attempts have been made to collect data on intersex traits under measures of sex assigned at birth, few, if any, individuals in the United States are socially or administratively assigned “intersex” rather than male or female at birth.78 For these reasons, a stand-alone measure should be used to identify respondents with intersex traits.79 Based on existing research, as well as guidance from experts and community members, the National Academies panel recommends the following measure out of three versions it considered.
The National Academies panel’s recommended intersex status question
Have you ever been diagnosed by a medical doctor or other health professional with an intersex condition or a difference of sex development (DSD) or were you born with (or developed naturally in puberty) genitals, reproductive organs, or chromosomal patterns that do not fit standard definitions of male or female?
(Prefer not to answer)
Intersex status questions: Limitations
The National Academies panel notes that work remains to be done to test, consistently validate, and implement intersex data measures because research is limited, making the standard measures of validity relied on elsewhere in the report unavailable. The panel endorses the recommended question because it has been tested among intersex populations but notes that it is long and potentially cumbersome. Further, while the two-pronged question recommended by the National Academies panel reflects the fact that individuals may have knowledge of their intersex traits either from personal experience or from a medical diagnosis, this medical framing may also raise sensitivities for a population that has historically experienced discrimination and trauma in medical settings. Use of the terms “condition” and “difference of sex development,” derived from the more overtly pathologizing “disorders of sex development,” may also be alienating to respondents.80
Priorities for ongoing research: Questions about variations in sex characteristics
CAP recommends that policymakers test stand-alone measures to enable data collection on people with variations in sex characteristics/intersex traits. CAP supports the National Academies recommendation to evaluate the use of a stand-alone single-item measure of variations in sex characteristics; the comparative performance of the three measures identified in the National Academies report; the impacts of including definitions and examples in these questions; and the impacts of proxy reporting, particularly among parents who report the intersex traits of their children. Additional testing on the performance of terms such as “intersex trait,” “intersex variation,” and “variation in sex characteristics” suggests that more recent community-driven language should also be assessed.81 People with intersex traits and community-based organizations representing intersex people should be included in question development, testing, and evaluation processes.
Part 2: How to collect SOGISC data in community-based surveys
At this point, most data on LGBTQI+ populations are captured by surveys of LGBTQI+ individuals, largely conducted by academic institutions and universities, think tanks, nonprofits, and advocacy organizations. Many groups with a focus on LGBTQI+ advocacy specifically recruit members of the LGBTQI+ community for surveys intended to capture the diverse range of experiences among these individuals. While general population surveys are critical in better understanding how LGBTQI+ populations compare with non-LGBTQI+ populations in economic and health indicators, surveys of only LGBTQI+ individuals allow researchers to ask more targeted questions about experiences of discrimination and opinions on LGBTQI+ issues. In fact, state and federal government agencies regularly utilize data from these surveys as evidence to support policy changes that protect the rights of LGBTQI+ Americans.82 Some prominent examples of community-specific surveys include:
- TransPop, the first nationally representative survey of transgender Americans83
- Generations: A Study of the Life and Health of LGB People in a Changing Society84
- The Center for American Progress’ nationally representative survey of LGBTQI+ Americans85
- The Trevor Project’s National Survey on LGBTQ Youth Mental Health86
- GLSEN’s National School Climate Survey87
- The National Center for Transgender Equality’s U.S. Transgender Survey88
Despite the fact that surveys within the LGBTQI+ community have long acted as a valuable resource, particularly in the absence of necessary data from large-scale population-based surveys, there is a lack of uniformity in how SOGISC questions are asked on community surveys.89 While this flexibility is important—allowing for a reflection and evolution of particular community norms around language, varying populations, and contexts examined by different surveys—developing general best practices is an important part of ensuring advocacy organizations collect the most usable data possible.
The 2022 National Academies report did not include recommendations specific to surveys within LGBTQI+ communities. Therefore, this section of the CAP report highlights best practices from reports of prior survey findings, qualitative studies, and interviews of LGBTQI+ individuals. In addition to their importance as a source of usable data, community-specific surveys act as an excellent resource for the ongoing testing and adaptation of survey best practices. Many organizations that conduct annual or biannual surveys, including GLSEN, The Trevor Project, and CAP, regularly use responses from prior surveys to better understand preferred terminology and question wording by community members and to make improvements to future survey iterations. As a result, this section lays out suggestions that are on the cutting edge of research on best practices—including significant reference to some designs and concepts that have not yet undergone sufficient testing but that provide a road map for how questions on both community-specific and general population surveys should evolve moving forward.
Key design considerations for LGBTQI+ community-based surveys
Unlike general population surveys, LGBTQI+ community-based surveys provide an opportunity for researchers to use a broader range of terminology and response options to better capture the diversity of LGBTQI+ identities, because they are surveying among a group of people who are more fluent in terms relevant to SOGISC questions.90 As on general population surveys, response options must be adapted to meet the needs of the specific LGBTQI+ population being surveyed. For example, when surveying communities with a sample population largely comprising people whose primary language is not English, SOGISC response options should be carefully and accurately translated into relevant languages. When collecting data among communities that regularly use culturally specific terms—for example, “Two-Spirit” or linguistically specific terms among many Native American communities, and “same-gender-loving” among some African American communities—researchers should tailor the data collection mechanism to employ these terms. For LGBTQI+ youth, researchers must be cognizant of the evolving identities of young people who may still be exploring their sexual orientation and gender identity.91
One important consideration for LGBTQI+-specific surveys is the use of screener questions, either prior to the survey or at the beginning of a survey. If researchers are interested in solely capturing LGBTQI+ individuals, it is necessary to determine via prescreen or opening questions whether respondents fit into these categories. To capture the full range of LGBTQI+ identities, it is necessary to use each of the questions below on SOGISC—at least four screener questions—to determine a respondent’s eligibility. Surveys focusing on transgender individuals must use both components of the recommended two-step question to determine eligibility.
Given the rapid pace at which terminology used by members of the LGBTQI+ community evolves, it is important to include specific definitions for concepts used in questions and response options. This is particularly true if a survey is being fielded among a variety of demographics, such as different age groups, different genders, and different sexualities, who may each use terms differently and may need a clearer understanding of what a question is asking. In addition, it is critical to remember that LGBTQI+ individuals of all ages have identities that can evolve over time. For example, many women have discussed their experiences being legally married to men but socializing as lesbians, with periods of fluctuation in their identities because of social stigmatization and internal uncertainty.92 Transgender individuals may change the way they talk about their sexual orientations as they come to terms with their gender identities and change the way they talk about their gender identities as they understand themselves better, learn more, or receive gender-affirming care.93 Surveys seeking to understand LGBTQI+ identities must be cognizant of the fluid nature of these terms and create as many opportunities as possible to capture these complex dynamics, while also ensuring reliable, usable data.
Asking about sexual orientation identity in community-based surveys
Interviews of LGBTQI+ individuals indicate that they prefer the more straightforward question wording “What is your sexual orientation?” to more nebulous versions such as “Do you think of yourself as…?”.94 It may also be helpful to include the addendum “Regardless of your sexual behavior,” because sexual minorities generally understand sexual orientation to be a broad concept encompassing not just one’s sense of self, but also one’s sexual history or even external presentation.95
Surveys need to include labels that provide a broad range of terms
Share of LGBTQI+ youth 2019 survey respondents who did not identify with the terms “gay,” “lesbian,” or “bisexual” and instead identified as “something else”
The Trevor Project
Share of transgender and genderqueer/gender-variant health care professionals who reported identifying as “something else” to avoid binary-based categories
There is increasing evidence that labels for LGBTQI+ individuals, particularly younger LGBTQI+ individuals, are expanding to include a broader range of terms.96 In The Trevor Project’s assessment of its 2019 survey, it found that 21 percent of LGBTQI+ youth did not identify with the terms “gay,” “lesbian,” or “bisexual” and instead reported that they identified as “something else”—with many of these individuals identifying as queer, pansexual, or on the asexual spectrum.97 As discussed in the 2022 National Academies report, the need for additional labels stems in part from the binary nature of some sexual orientation labels, which describe respondents as experiencing attraction to the “same” gender or the “opposite” gender.98 This language can exclude nonbinary individuals, which was demonstrated in a 2017 review of responses to the National Health Interview Survey among health care professionals, wherein 65 percent of transgender and genderqueer/gender-variant respondents reported identifying as “something else” in order to avoid identifying with binary-based categories.99 While concerns over survey fatigue and respondent confusion limit the capacity to add additional response options to general population surveys, the same considerations are minimized here.
Research on individuals who selected “other” on measures of sexual orientation with limited response options have revealed that “queer,” “pansexual,” and “asexual” are increasingly common identities; these responses are included in CAP’s recommendation on sexual orientation self-identification questions below.100 Studies have also found that many respondents, particularly young respondents, do not identify with any specific labels but do not consider themselves to be straight.101 In addition, many American Indian and Alaska Native individuals identify as Two-Spirit, an umbrella term for people whose sexual orientation or gender identity are not confined to the gender binary.102 Many American Indian and Alaska Native individuals may also use identity terms specific to their linguistic or tribal communities. For example, Native Hawaiians may use the term “māhū,” a term for an individual outside of the gender binary who has both masculine and feminine traits.103 Lastly, the response option “straight, that is, not gay or lesbian” provides a conceptually incorrect understanding of what it is to be straight, reinforces heteronormative ideas, and is not clearly still necessary even in general population survey designs.104 Given these factors, and the generally higher rates of familiarity with LGBTQI+ identities among community members, a community-specific survey could look similar to the example below.
CAP recommendation: Sexual orientation self-identification question
What is your sexual orientation?
- Gay or lesbian
- [If respondent is American Indian or Alaska Native] Two-Spirit
- No specific label, but not straight
- [Free-text option] Something else:
- Prefer not to respond
Sexual orientation identity questions in community-based surveys: Limitations and alternative approaches
As discussed above, one important ethical consideration in survey data collection is the consolidating and reclassifying of survey respondents. Due to the need to accurately and succinctly report survey results, there is an unavoidable degree of response compilation that must be done by organizations collecting these data, particularly in situations where sample sizes are limited. For example, while CAP’s 2020 survey of 1,508 LGBTQI+ individuals provided gender identity response options for nonbinary, genderqueer, agender, and gender-nonconforming individuals, these options netted 75 total respondents, necessitating their combination into one larger “nonbinary” category for data analysis. While researchers must often make such decisions on the back end in the interest of statistically rigorous analysis, it is equally important that survey respondents have the autonomy to identify themselves as accurately as possible, that they are aware of any eventual regrouping that might be done, and that they have the capacity to decide in what way, if at all, they are ultimately grouped.
Survey instruments from The Trevor Project and GLSEN provide essential insights on how to responsibly handle these concerns in community-specific contexts. The Trevor Project uses an open-ended identification question with a free-text option prior to a more targeted question about sexual orientation. This is particularly important for LGBTQI+ individuals of color, who have expressed in cognitive interviews the importance of write-in options to provide communities who may have their own terminology a space to include it—a feature that is also useful for researchers hoping to understand the evolution of terms used to describe sexual orientation.105 Meanwhile, GLSEN provides an extensive list of sexual orientation options and requests that respondents check all the options that apply to them, in addition to providing a free-text option. This check-all approach is important to acknowledge that many sexual minority individuals have multifaceted identities—for example, a person may identify as both asexual and bisexual.106 Both approaches allow survey respondents to self-categorize as specifically as possible, and researchers interested in providing such a question can use whichever option best suits their survey design. Researchers should ensure that a free-text option is still present on a “choose all that apply” question, as individuals may still identify with terms that are not listed.107 These designs are also extremely useful to researchers trying to understand the evolving terminologies used by sexual minority individuals and to make it possible to adjust future surveys based on regularly used responses.108 The follow-up questions included below should precede a response list similar to the above recommendation.
The Trevor Project: Open-ended sexual orientation self-identification
Sexual orientation is a person’s emotional, romantic, and/or sexual attractions to another person. There are many ways a person can describe their sexual orientation and many labels a person can use. How would you describe your current sexual orientation in your own words?
Thank you for telling us about your sexual orientation in your own words. Sometimes we have to create categories to make it easier to understand our findings, and we want to make sure you are represented in the best category. Which of these options best describes your sexual orientation? We understand that there are many different ways you may identify. Please pick the one that best describes you:
1. Straight or heterosexual
8. I am not sure
9. I don’t know what this question means
10. Decline to answer
Note: Survey instrument is on file with the authors.
GLSEN: ‘Choose all that apply’ sexual orientation self-identification
Below is a list of terms that people often use to describe their sexuality or sexual orientation. Please choose all those that apply to you:
9. [Free-text option] None of these apply to me, but I am:
Among the terms you selected, do you identify with one more strongly than the others?
[This question has the same response options as above, but respondents can select only one option.]
Note: Survey instrument is on file with authors.
One additional consideration is that asexuality can be both an identity and a description of an individual’s sexual behavior. Many people who identify as asexual still experience romantic attraction, and people identifying as asexual may have additional self-identification based on these romantic attractions; for example, an individual may identify as both asexual and bisexual. Given the increasing number of LGBTQI+ individuals—particularly among youth—identifying along the asexual spectrum,109 future research should test the efficacy of an additional question with a similar wording, as demonstrated below.
CAP recommendation: Potential add-on asexuality question
Do you consider yourself to be asexual or on the spectrum of asexuality?
- Yes, I am asexual or on the asexuality spectrum
- No, I am not asexual
- Prefer not to respond
Asking about gender identity in community-based surveys
When surveying solely LGBTQI+ individuals, researchers have the capacity to provide a greater number of gender identity options and to place additional focus on varied identities. In particular, the below recommendation includes a nonbinary response option—an essential element in LGBTQI+ specific surveys, given that more than one-third of transgender individuals identify with a nonbinary gender identity.110 The below identities reflect some of the most common nonbinary identities in 2021, as measured by the Gender Census.111 Surveys should include an additional option to reflect respondents who identify with one or more Indigenous gender identity, a consideration that has been tested in at least one prior survey.112
The below is an altered version of the two-step approach. It is based on question sets used in CAP’s 2020 survey of LGBTQI+ Americans and The Trevor Project’s current standards for LGBTQI+ community survey instruments.113 This two-step option does not include a separate “transgender” option among the listed genders. This is because including “transgender” as an option next to “male” and “female” is increasingly esoteric, given that most transgender individuals would not select this to describe themselves, and can create uncertainty among transgender respondents, as the inclusion of a separate option implies that if one is transgender, they are not truly “male” or “female.”114
CAP recommendation: Two-step question for community-specific surveys
What is your current gender?
- Nonbinary, genderqueer, gender-nonconforming, or agender
- [If respondent is American Indian or Alaska Native] Two-Spirit
- [Free-text option] Another gender:
- Decline to answer
What sex were you assigned at birth, meaning on your original birth certificate?
- Decline to answer
Testing from The Trevor Project on a two-step question in a community-specific survey context indicates that nonresponse rates for the question on sex assigned at birth were roughly equal regardless of whether this question was placed before or after the question regarding current gender identity. However, further testing should be done on the ordering effects of these questions.115 In addition, researchers may find it useful to add the term “regardless of your sex assigned at birth” to the question of current gender identity, particularly if this question comes before the question of sex assigned at birth. Doing so clarifies for respondents what surveyors are looking for, especially if they are unaware that there will be a sex assigned at birth follow-up question.116
Gender identity questions in community-based surveys: Limitations and alternative approaches
A two-step question approach that asks about current gender identity and sex assigned at birth is the current standard for general population surveys.117 However, as discussed above, this approach raises several potential concerns. For one, most community-specific surveys are in nonclinical settings where reporting sex assigned at birth is not a necessary variable. The higher rates of transgender and gender-nonconforming respondents in community surveys increase the likelihood that this question will be insensitive and harmful, potentially yielding lower response rates. Anecdotal evidence indicates that some transgender individuals find questions about their sex assigned at birth uncomfortable, implying that the sex listed on their birth certificate is more important than their gender identity.118 Another concern is that this setup requires those analyzing results to make determinations that create the potential for inaccurately representing the transgender community through misclassification and inaccurately characterizing the experiences of survey respondents.
A potential avenue for future research would be to ask about current gender and transgender status, similar to question framing by the Behavioral Risk Factor Surveillance System.119 This approach is suggested by Statistics New Zealand in its 2021 recommendations for data collection and has been adjusted below to include a Two-Spirit option.120 The below option combines a community-specific approach to asking about current gender identity, as shown above, with Statistics New Zealand’s recommendation for a transgender status question.
CAP recommendation: Potential alternative to the two-step question
What is your current gender?
- Nonbinary, genderqueer, gender-nonconforming, or agender
- [If respondent is American Indian or Alaska Native] Two-Spirit
Do you consider yourself to be transgender?
- I don’t know
- Prefer not to respond
As discussed above, the benefit of a question that explicitly asks about transgender status is that it allows researchers to explicitly determine who in their surveys identifies with the term “transgender.” Designers of future research proposals should consider these rationales when testing future sexual and gender minority question and response options.
Just as it is often useful to provide open-ended questions about an individual’s sexual orientation identity, allowing respondents to explain their gender identity in their own words is a valuable effort, especially given the complex and changing ways in which individuals view their own experiences.121 The Trevor Project’s survey instrument provides helpful language for asking these types of questions.
The Trevor Project: Open-ended gender identity self-identification
Gender identity is how someone feels about their own gender. There are many ways a person can describe their gender identity and many labels a person can use. How would you describe your current gender identity in your own words?
Thank you for telling us about your gender identity in your own words. Sometimes we have to create categories to make it easier to present our findings, and we want to make sure you are represented in the best category. Which of the following terms best describes your current gender identity? We understand that there are many different ways you may identify. Please pick the one that best describes you:
1. Girl or woman
2. Boy or man
3. Nonbinary, gender-fluid, or genderqueer
4. I am not sure or questioning
5. I don’t know what this question means
6. Decline to answer
Note: Survey instrument is on file with authors.
The “check all that apply” approach described for questions of sexual orientation is even more relevant for questions of gender identity. This option would allow respondents to select “transgender” or “cisgender” along with options such as “man” or “woman,” which removes the need for separate options such as “transgender man” or “transgender woman.” This may also be useful for respondents who consider themselves nonbinary or gender-nonconforming but still consider themselves to be cisgender. This is supported by research findings that “check-all” options are more applicable and useful on community-specific surveys of gender122 and findings that transgender and nonbinary individuals have a strong preference for multiple options on questions of both sexual orientation and gender identity.123 The “check all that apply” feature is a useful tool for acknowledging the complexity and multifaceted nature of gender identity, and though it provides more difficulty for researchers to code and analyze, it is an important avenue for future research. GLSEN’s current survey instrument provides this option.
GLSEN: “Check-all” gender identity question
Below is a list of some terms that people may use to describe their gender identity. Please choose all those that apply to you:
4. Cisgender (your gender identity is the same as your sex assigned at birth)
5. Transgender (your gender identity is not the same as your sex assigned at birth)
7. Not sure/Questioning
8. [Free-text option] A gender identity not listed here (please describe your gender identity):
Note: Survey instrument is on file with authors.
Asking about variations in sex characteristics on community-based surveys
Questions on variations in sex characteristics are of particular importance on surveys of LGBTQI+ individuals. For one, for a community that has grappled with extremely high levels of stigma,124 questions about experiences of discrimination—which are not typically asked in federal general population surveys—are critical to understanding the struggles facing intersex people. For another, intersex individuals report identifying as LGBTQ+ at higher rates than nonintersex individuals, making community-specific surveys an important resource for gathering information about this population.125 More broadly, there has been a growing and necessary movement to include intersex individuals within the LGBTQI+ community, and intersex individuals should not need to identify as a sexual or gender minority to merit inclusion on a survey of the LGBTQI+ community.
As discussed above, questions about variations in sex characteristics that specify a medical diagnosis will likely lead to underreporting given that 1) intersex individuals have historically experienced high levels of discrimination and trauma from care providers, and 2) some people will be aware of their intersex traits without having received a formal medical diagnosis. As an alternative, GLSEN includes an intersex question on its 2019 survey that does not stipulate a medical diagnosis.
CAP recommendation (question design used by GLSEN): Intersex status question
Some people are born with chromosomes, hormones, internal sex anatomy, or genitalia that do not fit within typical notions of “male” or “female” sex. This is known as intersex. Many learn they are intersex from a medical diagnosis. This is different from being transgender. Are you intersex?
- I’m not sure
- I don’t understand the question
Note: Survey instrument is on file with authors.
As an alternative, the 2020 Pennsylvania LGBTQ Health Needs Assessment, a biannual survey conducted by the Pennsylvania Department of Health to assess the health and wellness needs of LGBTQ individuals, also asks about variations in sex characteristics outside of an official diagnosis, with the wording “Were you born intersex, or with a variation of sex characteristics or sex development?”126 This survey of LGBTQ Americans found that 1.7 percent of respondents reported being born intersex, which the National Academies panel notes matches the highest estimate of intersex prevalence in traditional academic literature. Partially based on the wide range of positive response rates reported by different surveys reviewed by the National Academies panel, the panel’s report emphasizes the need for continued research both on effective intersex question standards and on understanding the needs of intersex individuals.127 To ensure intersex communities and experiences are accounted for in LGBTQI+ surveys, more research on intersex question design and more targeted outreach to intersex individuals for survey participation is needed.
Lastly, it is important to consider the use of a question about variations in sex characteristics as part of the list of prescreen questions on LGBTQI+ specific surveys. To date, no surveys of LGBTQI+ communities use intersex status as a screener question; rather, they include it—if at all—as one of many demographic questions asked of LGBTQ respondents, who are pulled from prescreen questions regarding sexuality orientation and gender identity. The use of a question about variations in sex characteristics as a prescreener would increase the number of intersex individuals present on community surveys.
Expanding and enhancing LGBTQI+-inclusive data collection is paramount to advancing equity for LGBTQI+ communities. Evidence demonstrates that federal general population surveys can and should adopt best-practice SOGISC measures, such as those detailed in this report, that allow for the identification of LGBTQI+ communities. Surveys specific to the LGBTQI+ community create important opportunities for understanding experiences of discrimination, and creating generalized standards for collecting information is critical to ensuring the most reliable and useful data.
While it is imperative to carefully weigh numerous concerns when designing mechanisms to capture data on LGBTQI+ people in general population and community surveys, these considerations should not prevent policymakers and researchers from embracing LGBTQI+-inclusive data collection efforts. Rather, meaningful investments in funding to continue to test, develop, and improve these measures are critical to building more robust data collection systems to gather information on LGBTQI+ people, bring visibility to their experiences, and generate evidence-based policies to effectively meet the needs of LGBTQI+ populations.
The authors wish to thank Kellan Baker, Logan S. Casey, Caitlin Clark, Jonah P. DeChants, Sylvan Fraser, Naomi Goldberg, Shoshana K Goldberg, Sharita Gruberg, Joseph Kosciw, D. Ojeda, Aaron Ridings, Karalin Sprague, Aaron Tax, Harper Jean Tobin, and Bonnie Washick for their valuable feedback on this report. We also wish to thank Carl Chancellor and Shanée Simhoni for their editorial assistance.
Appendix: The National Academies panel’s scientific criteria for recommended measures
- Consistency with the data collection principles discussed above—for example, precision, inclusiveness, autonomy
- Comprehensibility to the general population as well as the LGBTQI+ populations of interest
- Tested in both general population and LGBTQI+ populations
- Requires that respondents select only a single response option in order to simplify enumeration, tabulation, and analysis of the resulting data
- Provides consistent estimates when measured across data collection contexts
- Tested or previously administered with adequate performance using multiple administration modes, such as web-based, interviewer-administered, computer-assisted, and telephone administration
Response options criteria:
- Comprehensibility to the general population
- Consistency with terminology that is currently used in both the general population and LGBTQI+ populations
- Ability to measure current trends
- Ability to measure, assess, and incorporate changes with less well-known terminology
- Balance in providing comprehensive options with minimizing complexity and respondent burden that arises from considering a longer list of response options
- Produces a sufficient number of respondents per category to minimize the need to collapse categories and reclassify respondents
- Considers the effects of response item ordering, including relevant factors such as population prevalence, alphabetical listing, previous testing, and randomization.