DO NO HARM GUIDE COLLECTING, ANALYZING, AND REPORTING GENDER AND SEXUAL ORIENTATION DATA JONATHAN SCHWABISH, DONOVAN HARVEY, MEL LANGNESS, VINCENT PANCINI, AMY ROGIN, AND GABI VELASCO DECEMBER 2023 GLOSSARY The below sexual orientation and gender identity categorizations and definitions are intended to be broad and capture what these words mean to most people in 2023. Labels are not diagnostic but a way for sexual and gender minorities to find community and communicate aspects of their personhood and experiences in the world. Rather than smoothing over differences in how sexual and gender minorities describe themselves, it is important to recognize that they are not a monolith and have a range of experiences and ways of understanding their identities. By providing this list, we do not mean to imply that there is consensus around what they mean and how they are used. Furthermore, these terms are likely to evolve and change. Asexual. Often called "ace" for short. A person with a Heterosexual. A person who is emotionally, romantically, complete or partial lack of sexual attraction or lack of or sexually attracted only to people who present on the interest in sexual activity with others. Asexuality exists opposite end of the gender spectrum; the term typically on a spectrum, and asexual people may experience no, refers to cisgender men attracted to cisgender women little, or conditional sexual attraction. and vice versa. Bisexual. A person emotionally, romantically, or sexu- Homophobia. Negative and often hateful-and sometimes ally attracted to more than one sex, gender, or gender violent-attitudes toward people who are LGBTQIA+ identity, though not necessarily simultaneously, in the or are thought to be LGBTQIA+ based on misguided same way, or to the same degree. assumptions and beliefs. Cisgender. A term used to refer to people whose Intersex. A term used to refer to people who are born gender identities align with the cultural expectations with a variety of differences in their sex traits and based on the sex they were assigned at birth, or who reproductive anatomy that together exist outside the do not question or disagree with the gender society normative biological sex categories of male or female has expected them to present since they were born. subjectively assigned at birth by doctors based on visible anatomy or X and Y chromosomes alone. Intersex variations Gay. A person who is emotionally, romantically, or sex- include differences in genitalia, chromosomes, gonads, ually attracted to members of the same gender. Men, internal sex organs, hormone production, hormone re- women, and nonbinary people may use this term to sponse, and/or secondary sex traits. describe themselves, but it is most often used by men who are attracted to other men. Lesbian. A woman who is emotionally, romantically, or sexually attracted to other women or nonbinary people. Gender. The social and cultural categorization of peo- Women and nonbinary people may use this term to ple, such as "man" or "woman," based on their identity, describe themselves. behavior, self-expression, and interaction with others. Gender varies across societies and contexts and can LGBTQ+ or LGBTQIA+. An acronym for "lesbian, gay, change over time. bisexual, transgender, and queer," with a "+" sign to recognize the limitless sexual orientations and gender Gender expression. The external manifestations of identities. The LGBTQIA+ includes "intersex and asexual" gender, expressed through a person's name, personal identities. We use the LGBTQIA+ acronym in this guide, pronouns, clothing, behavior, body characteristics, and though we recognize that the identities the term encom- more. Gender expression is not static. passes are separate, and sometimes they are intersectional Gender identity. A person's internal psychological identities that can refer to distinct persons, communities, identification as a man, woman, another gender, or no and contexts. gender. This identification may or may not align with the sex the individual was assigned at birth. 2 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA Nonbinary. Some people may have a gender that Sexual orientation. A term used to refer to a person's blends elements of masculinity and femininity, some pattern of emotional, romantic, or sexual attraction to people do not identify as either male or female, and other people. some people's gender changes over time. People SOGI, SOGIE, or SOGIESC. Variations of an acronym for whose gender is not captured by manhood or woman- "sexual orientation, gender identity, gender expression, hood may use different terms to describe themselves, and sex characteristics." SOGIESC encompasses "sex including nonbinary, but also genderqueer, agender, characteristics" to include intersex people. We use the bigender, genderfluid, and others. SOGI acronym in this guide because our focus is on sexu- Pronouns/personal pronouns. Terms used in place of a al orientation and gender identity, not necessarily on how proper noun, usually a name, when referring to a per- people express their gender. Again, we recognize there son. Examples include they/them/theirs, she/her/hers, are separate and sometimes intersectional identities that he/him/his, and ze/hir/hirs. Pronoun choice is highly can refer to distinct persons, communities, and contexts. personal, and pronouns should not be assigned to Transgender. An umbrella term for people whose gender someone based on assumptions about their perceived identity and/or expression is different from cultural gender. expectations based on the sex they were assigned at Queer. A term people use to express a spectrum of birth. Someone could be a transgender man, a transgen- identities and orientations that are counter to the der woman, transgender nonbinary, or otherwise gender mainstream. It is often used as a catchall for many nonconforming. Being transgender does not imply any people, including those who do not identify as ex- specific sexual orientation, and transgender people may clusively straight and/or those who have nonbinary identify as straight, gay, lesbian, bisexual, and so on. or gender-expansive identities. This term was previ- Two-Spirit. A term used by many Indigenous people in ously used as a slur but has been reclaimed by many North America to describe those who have both a mascu- LGBTQIA+ communities and people. It should not be line and a feminine spirit. It may also be used to describe used to describe someone unless they explicitly indi- a person's sexual, gender, and/or spiritual identity. The cate that they are comfortable being labeled that way. term was coined in 1990 by Myra Laramee (Cree) and is Questioning. A term used to describe people who are intended to be an umbrella term, not to replace tribal- exploring their sexual orientation or gender identity. specific names or traditions nations may have, such as winkte (Lakota) and nadleeh (Navajo).1 Sex or sex assigned at birth. The classification of a person as male, female, or intersex as assigned at birth by doctors based on hormones, chromosomes, and the appearance of external anatomy. Sex, gender, and gender identity are not interchangeable, they are spectrums that have traditionally been divided into binary categories. Sexual and gender minority. A term used to describe people who do not conform to traditional societal norms and expectations regarding sexual orientation and gender identity. It can refer to a diverse group of people who may identify as lesbian, gay, bisexual, transgender, queer, questioning, intersex, asexual, or any other nonheteronormative or noncisnormative identity. The term recognizes the fluidity and complexity of human sexuality and gender beyond the dominant binary narratives. 3 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA TABLE OF 05 PART ONE CONTENTS Introduction Box 1: Opportunities for Future Research 10 PART TWO Historical Context for SOGI Data Collection Early SOGI Data Collection Efforts Federal (and Other Large-Scale) SOGI Data Collection Efforts 15 PART THREE Data Collection Safety and Security Measures to Consider in SOGI Data Collection Data Privacy and Collecting Personally Identifiable Information Building Inclusive Data Collection Methods Box 2: Strengthening Research by Engaging Community 37 PART FOUR Data Analysis Demographic Survey Data Cleaning SOGI Survey Data 41 PART FIVE Communication and Data Visualization Color Terminology in Data Visualization Icons Showing Small Numbers Ordering 46 PART SIX Conclusion 48 PART SEVEN Appendixes 56 Notes 59 References 63 Acknowledgments 4 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA Part One Introduction Every day, millions of people provide data about PART ONE their gender or sexual orientation. They supply Introduction data about themselves or their families, even if they are not fully aware of it, by using their cell phones or shopping online, for example; or JONATHAN SCHWABISH they may answer questions in a formal survey. Urban Institute And with so much (and increasing) technological DONOVAN HARVEY power at our disposal, researchers, analysts, Urban Institute and social scientists have started to think MEL LANGNESS Urban Institute more carefully about how we collect, analyze, VINCENT PANCINI communicate, and respond to data around Urban Institute demographic characteristics and identities. AMY ROGIN Urban Institute These data, when disaggregated by demographics or identities, GABI VELASCO can offer insight into and understanding about disparities across Urban Institute health, income, housing, and other areas-and ultimately inform policy and funding decisions. For example, we know from the Centers for Disease Control and Prevention's (CDC's) Youth Risk Behavior Survey that LGBTQIA+ youth (students in grades 9–12) are at a higher risk of suicide relative to their non-LGBTQIA+ peers. Federal, state, and local governments can use these data to better allocate budget resources and select hospital locations.2 But these data also have a significant capacity to cause harm if misused. The long arc of the fight to recognize and respect the dignity and rights of LGBTQIA+ people has raised the visibility of multiple dimensions of gender and sexual orientation, expanding our conception of these identities beyond the binary definitions of man or woman, straight or gay. But recently, we have seen an increasing and targeted backlash against LGBTQIA+ people, particularly transgender and gender-nonconforming people. As of September 2023, more than 560 anti-LGBTQIA+ bills were introduced across the country.3 These bills-more than 80 of which have already passed-seek to prohibit a number of identity- or gender- affirming practices, including the use of a person's correct pronoun in schools, access to evidence-based and medically necessary care, treatment of gender dysphoria in minors, and the provision of accurate identification documents. Each of these pieces of legislation demonstrates that data regarding gender and sexual orientation can be used for harm-to restrict rights, health care, and freedoms and to further disparage vulnerable, marginalized persons as "other." 6 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA In this fifth guide of the Urban Institute's Do No Harm and transformative. However, data collection for the project, we explore the current state of data around Do No Harm project takes a more general lens, and gender, gender identity, and sexual orientation. The this guide does not attempt to address specific data collection of demographic (or identity-based) data collection considerations that may be prompted by is often complicated by the evolution and nuance of the intersections of gender, sexual orientation, race, language; words or phrases that we used yesterday ethnicity, disability status, age, economic status, and may not be the words or phrases that we use today or other components of individual identity. will use tomorrow. By understanding these changes The findings reported here focus mainly on larger and employing data best practices, researchers, surveys, such as those conducted by the US Census analysts, and other stakeholders can help ensure Bureau, the Bureau of Labor Statistics, and other that such data are used for good-to help address federal agencies. But the lessons can be extended to disparities and inequities faced by LGBTQIA+ people smaller surveys or surveys addressing specific themes and to assess the effects of policies, interventions, and soliciting specific types of data-for example, and societal attitudes on their lives. Making data that are collected by (and for) advocacy groups purposeful and thoughtful decisions about these working on behalf of LGBTQIA+ issues. Similarly, large kinds of data in inclusive and equitable ways can surveys can adopt practices, strategies, and language result in recommendations to policymakers that are from these other types of surveys. In any case, there more likely to be embraced and implemented without is still more work to be done, and we list a set of risking the safety and privacy of the people whose possible future research opportunities in box 1. data are collected. For this guide, we identified and interviewed more Our goal with this guide is to provide a series of than 20 scholars and advocates with expertise in considerations and, in some cases, recommendations collecting, analyzing, and communicating data on regarding collecting, analyzing, and communicating sexual orientation and gender identity, otherwise quantitative data on gender, gender identity, and known as "SOGI" data (see the glossary for sexual orientation. We focus on quantitative (i.e., definitions). Our interview protocol and informed- countable) data, mostly in the context of social consent documents for the interviews are included science research, because that is where our expertise in the appendixes. In addition to the interviews, we lies. But this focus should not suggest that qualitative conducted a literature scan of dozens of reports, data are not important or valuable-in fact, we have academic articles, and books to understand the argued elsewhere that quantitative research needs to current landscape of recommendations. We also add more qualitative dimensions (Schwabish and Feng presented our preliminary findings and sought 2021). We hope this omission leaves the door open feedback in open conversations with representatives for further work on how to best collect and analyze from LGBTQIA+ service and advocacy organizations data relating to gender, gender identity, and sexual across the country. These participants provided orientation. feedback based on their personal and professional LGBTQIA+ experience is not a monolith. Historic and lived experiences, which we have incorporated into current realities faced by LGBTQIA+ people of color our understanding of these issues and the findings include intersecting dynamics of structural racism, reported below. overcriminalization, and violence that impact how We have organized this guide according to each they engage with researchers. Recognizing these step in the data process: collection, analysis, and intersections is essential to ensure that efforts to communication. The following are five key points bring about positive change are comprehensive, just, to keep in mind when working with SOGI data: 7 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA 1. Researchers must tell people why their data are separate, consecutive questions, whereas others say being collected. Survey data collection, especially a single question with a list of options or even an for large government surveys, has traditionally open-ended (i.e., write-in) option is a better approach. reflected the nation's historical structures and But everyone we spoke with agreed that continued institutions that placed certain people-typically research is necessary to understand how to more white cisgender men-in positions of power. As accurately capture people's identities and experiences, a result, researchers have created data that have which echoes existing research literature. long minimized and overlooked people from other 4. Language-to-language translation can be groups, leading to distrust of government and complicated. Most of this guide focuses on the research. Making clear why a person's data will help words and phrases used in US surveys and research answer important research and policy questions is reports to describe LGBTQIA+ people. Making fundamental for building trust, which will ultimately surveys and data collection efforts as well as final result in better-quality data. This task may be easier dissemination of products available to people who for certain data, such as personal health data, or do not speak English, however, can pose additional for surveys aimed at capturing the experiences of concerns. Different language structures and the particular groups or communities. But for large, lack of equivalent terms or phrases are just two of nationally representative surveys-such as those the challenges researchers face. often conducted by the US Census Bureau-it can 5. Privacy and safety are real and serious concerns. be a bigger challenge. Especially with the rise of hate speech and violence 2. Doing SOGI research is just like doing any against transgender people, asking (or requiring) good research. Although existing SOGI data are people to reveal their identity can put them at intermittent and sometimes of dubious quality, the risk for discrimination and harm. Any organization process of collecting and analyzing those data- collecting, storing, and analyzing SOGI data needs both qualitative and quantitative-should be similar to take data privacy seriously to adequately to the process of collecting and analyzing any data. protect research participants. These precautions Before thinking about the best and safest way can extend from safely storing digital files to to conduct an interview about SOGI topics, the using an institutional review board (IRB) to ensure research team should understand how to conduct compliance with applicable regulations, accepted a qualitative interview. Before designing a survey ethical standards, and institutional policies. to collect SOGI information, the research team As with previous Do No Harm Guides, the importance should understand how to implement sound survey of approaching these data efforts with empathy design. Before conducting surveys or interviews and nuance clearly emerged in our interviews. Our with groups that have experienced discrimination interviewees noted how excluding marginalized and violence, the research team should receive groups from data collection and analysis undercuts trauma-informed interview training. No matter the policy solutions and negatively affects the lives type of research, researchers should be clear about of people and communities. Collecting, analyzing, the questions they are asking, what data they need and communicating these data are crucial to to answer those questions, and why. provide evidence, shape public discourse, and guide 3. There is no one way to collect SOGI data. As decisionmaking to protect LGBTQIA+ people's both our interviews and literature review showed, rights and improve their well-being. But none of this there is no universal agreement on the best way data work can be done without considering the to collect SOGI data. Some experts argue that ramifications for people's physical and mental health, asking questions about a person's transgender their ability to live and work, and the threats to their status should be accomplished by asking two personal freedoms. 8 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA Here, we extend the concept of empathy to consider piece of themselves to the work. They grant you how, when people provide their data for research, access into their world. There is therefore an onus they give a part of themselves to the researcher. on the researcher to do something meaningful Researchers incorporating this lens of empathy in their with the data collected. When data shared is not work have an obligation to safeguard and use those used for action, or even worse left unanalyzed on data responsibly. As Kevin Guyan wrote in Queer Data: a hard drive or cloud server, it benefits nobody and When people participate in research related to risks discouraging future participation in research EDI [equity, diversity, and inclusion] they give a projects (Guyan 2022, 62). BOX 1: OPPORTUNITIES FOR FUTURE RESEARCH Many aspects of SOGI data collection, analysis, and communication are not covered in detail in this guide. The following issues should be explored in future work: • Qualitative data collection and communication. Our best practices for or innovative examples of research focus on quantitative data in this guide leaves more to seeking to understand how an individual expresses be said about the importance and value of qualitative their authentic gender identity or how an individual data, which include interviews, observations, case studies, perceives gendered behaviors of others. However, this focus groups, and open-ended text. Qualitative data is a crucial field for future research. collection is important not only as a way to capture a • Intersex people. There are many questions to explore wider range of opinions and experiences but also as around data relating to intersex people, some of which a way to help readers and users better connect with go beyond the scope of our work. For example, what research-but we do not go into great detail about how does someone's sex at birth really tell us, and are re- to conduct qualitative research here. Using qualitative searchers using it as a proxy for how a person is social- data to help open the door to better understand the ized? Is a person's sex a proxy for how the world sees experiences and perspectives of survey respondents them? What are the risks of this kind of assumption in throughout the data collection process can yield more research and analysis? representative and more accurate data. But it can • Asking about pronouns. Most data collection issues we also be a challenge for many researchers, who are not focus on in this guide relate to specific categories of trained in qualitative methods, and for research funding gender and sexual orientation. While pronouns should organizations, which are more likely to fund quantitative never be used to infer a person's gender identity, research than qualitative research (see, for example, sharing pronouns is a common component of personal Carey and Swanson [2003] and Bourgeault [2012]). introductions and can be a sufficient and more appro- Sociologist Tey Meadow, whose work is often centered priate alternative to collecting gender identity data around using qualitative data, told us that the challenge formally, especially in more informal contexts, such as for quantitative researchers is to take a qualitative data a conference registration form or a classroom survey, perspective: "Qualitative researchers have a kind of where the intent is to accurately address people. dense curiosity and an ability to sit with nuance, which makes us question not merely how people in different • Expanding to an international view. While questions categories experience the world, but how the categories regarding the collection and use of SOGI data span the into which they sort them should be constructed. While globe, we have focused on the experience of US data some survey researchers interrogate the assumptions collection agencies, US-based advocates, and other behind their categories, others merely reproduce nor- people conducting research in the US. Important work mative assumptions by using old metrics." is being carried out in this area at organizations around the world, including the World Bank, for example, with • Gender expression and gendered behaviors. Unlike its Equality of Opportunity for Sexual and Gender Minori- race or nationality, gender is a component of one's per- ties report; the International Lesbian, Gay, Bisexual, sonal identity that is both conveyed through and reified Trans and Intersex Association; and the United Nations, by performed physical, social, and aesthetic behavior. to name just a few. Our research did not elicit sufficient information on 9 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA Part Two Historical Context for SOGI Data Collection Collecting data about sexual orientation and PART TWO gender identity and expression allows for a Historical better understanding of sexual and gender Context for minority populations, enabling researchers, policymakers, and advocates to understand SOGI Data differences between these populations Collection and other population groups or the general population across policy areas. Insight from these data highlight important areas for interventions that may improve the lives of members of sexual and gender minority groups. Historically, however, this information has not always been used to improve the lives of LGBTQIA+ people. Throughout US history, public disclosure of one's sexual orientation has often led to disastrous consequences, including jail time; police violence; and discrimination in employment, health, and other areas. For example, many gay men and lesbians served openly during World War II, but as demand for troops declined toward the end of the war, many were dishonorably discharged (Berube 1990). This forcible outing prevented them from receiving benefits under the GI Bill of Rights and often hindered their ability to secure employment. Additionally, in 1953, President Eisenhower issued an executive order barring all homosexuals from federal employment because of fear that communist sympathizers would blackmail them because of their sexuality (Cervini 2020; Johnson 2004). Here, we hope to provide an overview of modern SOGI data collection efforts. Much of this history is summarized in and informed by The Health of Lesbian, Gay, Bisexual, and Transgender People: Building a Foundation for Better Understanding (National Academy of Medicine 2011, chapter 2). We cover some important events in American LGBTQIA+ history, but we do not seek to detail the long and complex history of LGBTQIA+ people or the disparate and discriminatory treatment they have often experienced. Early SOGI Data Collection Efforts After World War II, scientific research that collected sexual orientation information helped improve conditions for LGBTQIA+ people. The publication of Alfred Kinsey's Sexual Behavior in the Human Male (1948) and Sexual Behavior in the Human Female 11 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA (1953)-also called the Kinsey Reports-was a After the Stonewall rebellion, many people became watershed moment in SOGI data collection, because empowered to publicly reveal their sexual orientation. the reports did not treat same-sex attraction and Gay and lesbian communities as well as LGBTQIA+ sexual behavior as categorically different from any organizations, groups, and businesses grew across the other form of sexual attraction or behavior. Compared US throughout the 1970s (Faderman 1991; Levine with previous research, which took for granted that 1979). The LGBTQIA+ activists notched another "homosexual practice" was inherently "deviant," the important victory with the removal of homosexuality Kinsey Reports, and the Kinsey Scale in particular, from the American Psychological Association's "implied that homosexuality was just another form of Diagnostic and Statistical Manual of Mental Disorders sexual activity" (Bullough 1998, 130). Though now in 1973 (Drescher 2015). Still, the social, legal, and considered overestimates, the reports' assessments political backlash against this movement and gay and of the frequency of same-sex attraction and behavior transgender people continued. were among the earliest attempts at measuring the The beginning of the HIV/AIDS epidemic in the 1980s size of the sexual or gender population in the US. marked one of the darkest chapters in LGBTQIA+ Kinsey's work-along with other contemporaneously history. In June 1981, the first cases of the disease, published work from Evelyn Hooker, Clellan Ford, and which would eventually be called AIDS, were reported Frank Beach-presented a new avenue for research to in the CDC's Morbidity and Mortality Weekly Report. address discrimination against LGBTQIA+ people. A By 1995, as many as 1 in 15 gay men in the US had 2011 report from the National Academy of Medicine died of AIDS.5 Driven by overwhelming inaction on said that this era of research "challenged widespread the part of the federal government-exemplified by assumptions that homosexuality was a rare and then president Ronald Reagan's refusal to use the pathological form of sexuality, practiced only by a word AIDS until September 1985, after more than small number of social misfits" (National Academy of 8,000 Americans had died-the HIV/AIDS epidemic Medicine 2011, 37). This scholarship also bolstered marked a turning point in LGBTQ organizing efforts.6 the work of activists, who began to work to end LGBTQ people formed groups, such as Gay Men's harassment and discrimination. Health Crisis and the AIDS Coalition to Unleash Power (commonly known as ACT UP), to provide Despite some success in protecting the rights peer-to-peer support and advocate on behalf of of gay people during the 1960s, anti-LGBTQIA+ those affected by AIDS.7 This infrastructure created persecution was still common.4 Among other forms the foundation for many current LGBTQ services and of discrimination, police still routinely raided spaces groups interested in collecting SOGI information.8 where sexual and gender minorities congregated to arrest and brutalize people (Boyd 2003; Johnson Although these decades brought more visibility 2004). During one such raid on the Stonewall for LGBTQIA+ people, that progress was not felt Inn, a gay bar in New York City, patrons and their equally. Often, white cisgender gay men were at the allies resisted the police for several nights (Adam forefront of these groups-with disproportionate 1995). Known today as the Stonewall rebellion, shares of leadership roles and media attention-to this confrontation-which was planned, led, and the exclusion of Black, Hispanic/Latine/Latinx, and supported by transgender, lesbian, and gay people- Indigenous people; transgender people; and people marked the beginning of the contemporary movement with other intersecting identities. The consequences for LGBTQIA+ rights. of that inequality within the LGBTQIA+ community reverberate to this day, with Black and Latina transgender women experiencing disproportionate rates of violence, homelessness, and discrimination. 12 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA Federal (and Other Large-Scale) orientation questions as part of a suite of changes to SOGI Data Collection Efforts the survey in 2002 (Saperstein and Westbrook 2021). The early 1990s saw a marked increase in the Through the rest of the 1990s and into the 2000s, statistical rigor of sexual behavior research-in large other federal surveys began adding questions about part because of the compelling public health interest sexual orientation. As we discuss later in this guide, created by HIV/AIDS. Not only did the epidemic sexual orientation is a multifaceted construct that reveal the need for robust data on sexual behavior, is typically divided into three dimensions: sexual but "researchers in the United States, many of them identity, sexual behavior, and sexual attraction lesbian, gay, or bisexual themselves … started to argue (National Academy of Medicine 2011). While some for the collection of sexual orientation data in publicly surveys, such as the NHSLS, addressed all three funded data sets" (Sell and Holliday 2014, 967). dimensions of sexual orientation, many others One of the first large-scale SOGI data collection included only one or two of the dimensions. efforts was the 1992 National Health and Social Life This period also marked the first time that same- Survey (NHSLS), which was originally intended to be sex couples were recognized in the US Census. In a pretest survey for the larger Survey of Health and 1990, the Census Bureau added an "unmarried AIDS-Related Practices. Although funding for the partner" category to the "relationship to householder" larger project was blocked in Congress, the NHSLS question to capture the increasing number of couples still proved massively influential. According to a 1995 living together without getting married. However, in retrospective on the NHSLS: "The National Health data files available to the public, the bureau recoded and Social Life Survey is a singular event in the history the gender of individuals who indicated that they of survey research, not because it pioneered new were same-sex couples, treating them as an error. methods, but because it demonstrated that sound, In 2000, the National LGBTQ Task Force and other traditional survey approaches can be applied to the organizations pressed the bureau to stop recoding the study of sexual behavior" (Miller 1995, 418). gender of same-sex partners in public-use data, and The publication of the findings from the NHSLS- the bureau agreed (National LGBTQ Task Force 2017; along with the 1988 introduction of sexual behavior Smith and Gates 2001). questions in the National Opinion Research Center Since these first instances of federal SOGI data (NORC) General Social Survey, one of the longest- collection, best practices started to emerge for the running social surveys in the US-marked a turning burgeoning field. Researchers at the UCLA Williams point in LGBTQIA+ data collection. Institute convened the Sexual Minority Assessment In 1996, four years after the NHSLS was conducted, Research Team, which authored Best Practices for the National Institute of Mental Health funded the Asking Questions about Sexual Orientation on Surveys one-time National Sexual Health Survey, which in 2009 (SMART 2009). This report, along with the was one of the earliest federal surveys to include Williams Institute's 2013 report Best Practices for questions about sexual behavior and sexual identity. Asking Questions to Identify Transgender and Other Other early federal efforts included the National Gender Minority Respondents on Population-Based Longitudinal Study of Adolescent to Adult Health Surveys (GenIUSS Group 2014), would later guide the (Westbrook, Budnick, and Saperstein 2022), which creation of the CDC's optional SOGI module for the incorporated sexual orientation questions into Behavioral Risk Factor Surveillance System (BRFSS) its 2001 panel, and the National Survey of Family (Baker and Hughes 2016). The latter report, also Growth, which added the full spectrum of sexual called the GenIUSS report, was especially timely, as 13 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA states began asking questions about gender identity The Biden administration, however, has resumed in the BRFSS as early as 2007 (Baker and Hughes the federal SOGI data collection efforts. In 2021, 2016). This period also saw the publication of the the Census Bureau added questions about sexual National Academy of Medicine's influential The Health orientation and gender identity to its Household Pulse of Lesbian, Gay, Bisexual, and Transgender People report Survey,11 marking the first time a Census Bureau– (National Academy of Medicine 2011). Although sponsored survey has included SOGI questions.12 not focused on SOGI data collection, the report, In 2022, President Biden signed the Executive Order which is considered one of the first comprehensive on Advancing Equality for Lesbian, Gay, Bisexual, overviews of the field, included guidance for federal Transgender, Queer, and Intersex Individuals. One surveys collecting SOGI data and called for the section of the executive order focused on "promoting collection of these data across federal agencies. These inclusive and responsible federal data collection" to three reports came at a critical time in federal data help policymakers identify, understand, and address collection, and by 2016 "there [were] eleven federal disparities experienced by LGBTQIA+ people. The surveys and one federal study that collect data on executive order also required federal agencies to sexual orientation, including identity, attraction, and submit a "SOGI Data Action Plan" to explain how behavior, and gender identity" (Federal Committee on they will use data to advance equity for LGBTQIA+ Statistical Methodology 2016, 4). people.13 In fiscal year 2023, Congress appropriated $10 million to the Census Bureau to support research But collecting SOGI data remains a contentious on asking SOGI questions in the American Community political issue. In 2017, the Trump administration Survey, which is one of the largest demographic removed questions about sexual orientation and surveys administered by the federal government.14 gender identity from the National Survey of Older Americans Act Participants and the annual program Yet, concerns about data privacy and potential misuse performance report for Centers for Independent of data abound, no matter who is in the Oval Office, Living9 (Cahill and Makadon 2017). The administration given the current charged political environment. In the also curtailed the Census Bureau from adding sexual next section, we discuss these concerns in more detail, orientation and gender as potential question topics including safety and security measures researchers to the American Community Survey.10 should consider when collecting SOGI data. 14 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA Part Three Data Collection There are two sides to survey data collection: the PART THREE experience of the participant and the experience Data Collection of the researcher. Thus, building trust between the participant and the researcher is key to generating high-quality data. Communicating to people why their personal information is necessary and gaining their formal consent can lead to greater collaboration and, ultimately, allow for more inclusive survey methods. No data collection effort is entirely without social implications. Collecting data on LGBTQIA+ people in particular entails special considerations, especially with the recent rise in anti-LGBTQIA+ (most notably anti-transgender) violence and legislation. SOGI data collection efforts are essential for understanding different LGBTQIA+ groups in the US and providing an evidence base for programs and services (Office of Management and Budget 2021). Although aspects of data privacy are encoded in federal law-such as Titles 13 and 26 of the US Code, which lay out regulations and requirements for collecting and storing personally identifiable information (PII)15-anyone collecting, analyzing, and communicating SOGI data needs to be cognizant of the importance of data privacy stewardship. Throughout this section, we try to balance research and survey methodological considerations with the importance of being inclusive of survey participants' identities and protecting their safety and privacy. Unfortunately, these factors often are in tension with one another. Therefore, we recognize that a survey is not meant to capture all the nuances of someone's story. But, when constructed thoughtfully, it can still be a powerful tool for giving visibility and agency to historically marginalized groups. In all cases, researchers have a responsibility to be ethical and empathetic stewards of these data. Safety and Security Measures to Consider in SOGI Data Collection Throughout history, marginalized groups have been repeatedly harmed by research. From the Tuskegee syphilis study, in which 400 Black men with syphilis were purposely untreated; to Henrietta Lacks, whose cancer cells were extracted and used 16 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA without her permission; to the Havasupai Tribe in License Division had emailed colleagues: "Need total Arizona, whose blood samples were used without number of changes from male to female and female to their permission for genetic research-history has no male for the last 24 months, broken down by month. shortage of examples of exploitative and extractive We won't need DL/ID numbers at first but may need research practice. to have them later if we are required to manually look up documents."16 It is not surprising, then, that many people today may feel wary about providing their information With LGBTQIA+ identities so intensely politicized, to institutions or individuals, whether doctors, data on the prevalence and experiences of LGBTQIA+ the federal government, or political pollsters. This people can be used to advance anti-LGBTQIA+ lack of trust can result in lower survey response legislation, which presents safety challenges for rates and less accurate information. Data scientist researchers. As of September 2023, the Trans Kelsey Campbell summarized this challenge in our Legislation Tracker17 had recorded more than 560 conversation: "There is that tension between wanting anti-transgender bills introduced across the country representation and wanting more inclusive measures that seek to limit medical care, bathroom access, on surveys or in systems, but also not having trust in a and accurate identification for transgender people system to use that data appropriately and not use it in (figure 1).18 According to the 2022 Trevor Project an exploitative way." survey, 86 percent of transgender and/or nonbinary young people (ages 13 to 24) reported that the Creating inclusive and representative surveys that political debates around these issues had a negative benefit people and communities can therefore be key effect on their mental health (The Trevor Project 2022). to generating the knowledge needed to pursue better policy and community outcomes. We do not endeavor FIGURE 1 More than 560 Anti-LGBTQ Bills Were Introduced across the to summarize every necessary step of a research US between January and September 2023 project in this guide, but we want to stress that collecting data on LGBTQIA+ identities or experiences must include foundational security measures. No survey effort should put people at risk. Although SOGI data are essential to understanding where LGBTQIA+ people live and work, what benefits and services they receive, and many other types of information, such data can be used to both promote and restrict access to rights and services. Collecting and using SOGI data to locate-and potentially punish-people who identify as LGBTQIA+ is a real concern, especially at the government level. Data on how many or few LGBTQIA+ people live in a given area, for example, can demonstrate the Source: "2023 Anti-Trans Bills Tracker," Trans Legislation Tracker, accessed September 12, 2023, https://translegislation.com. need for services, just as it can be used to target service reductions. In late 2022, the Washington In light of all these challenges, protecting LGBTQIA+ Post reported that the Texas attorney general's office people's identity must be a top priority in data requested a list of individuals who had changed the collection and data analysis efforts, in large-scale gender on their driver's license and other records. The representative surveys and smaller-scale local surveys head of the Texas Department of Public Safety Driver alike (Flores et al. 2021). 17 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA For data collectors, protecting people's privacy and and nonbinary people, much less on how such preserving their safety comes with potential trade- discrimination varies by race and ethnicity (see Sears offs in terms of both data quality and group wishes. et al. [2021] and Badgett, Baumle, and Boutcher In the 2023 Do No Harm Guide: Applying Equity [2018]). Thus, while the balance between data Awareness in Data Privacy Methods, our colleagues collection and data privacy is important on its own at Claire Bowen and Joshua Snoke summarized these an aggregate level, it becomes even more important concerns: as we consider intersections of identities. An underrepresented group might be less Data Privacy and Collecting Personally concerned about their privacy and would rather Identifiable Information have more accurate representation in the data. For researchers who routinely conduct surveys, the For instance, a university might suppress the process of developing security measures for PII may number of transgender students attending their law be familiar. The US Department of Labor defines schools for privacy purposes. Some students might PII as data "(i) that directly identifies an individual support this decision while others may want to be (e.g., name, address, social security number or other accurately represented to allow other transgender identifying number or code, telephone number, students to reach out to them or to know that email address, etc.) or (ii) by which an agency intends the law school is welcoming toward transgender to identify specific individuals in conjunction with students. The outcome depends on the group other data elements, i.e., indirect identification."19 representatives influencing these decisions, and in Demographic information, such as race, ethnicity, either case, the trade-offs and privacy limitations annual income, and sex at birth, is not generally should be acknowledged and communicated to considered PII for large-scale survey efforts. But group representatives. (Bowen and Snoke 2023, 12) the risk of "indirect identification" is particularly As our colleagues note, fears about how data collection high when researchers are working with data from may be weaponized for harm must be weighed against marginalized groups, including LGBTQIA+ people, the importance of having representation. For many and even higher for specific subpopulations, such people, the balance will depend on their specific as transgender people. intersectional identities. As in the above example, a transgender law student may prefer visibility over Survey scope and size can also influence how data privacy to lift up others with their identity. Similarly, may put people at risk of identification. A national a Black transgender woman who has experienced survey is usually too large to put individuals at risk, systemic discrimination may trade some privacy for but a school-based survey asking about gender might more awareness of these harms. However, these generate a dataset that highlights just one respondent circumstances can just as easily be flipped. If the same as nonbinary-a number so low that administrators law student attends school in a state that is stripping and readers may know exactly which student reported transgender protections, they may prefer their data that information. For these reasons, it is a good be suppressed. And if the same Black transgender practice to treat all instruments asking about SOGI woman fears retaliation for their participation in a information like they are collecting PII. survey, they may prefer privacy to visibility. Any US research project seeking to collect data from Ultimately, existing power structures have always marginalized groups, including LGBTQIA+ people, made some identities safer than others in society. should seek approval from an IRB before beginning There is little research-likely because of lack of data collection. Some smaller-scale survey research representative data-for example, on the denial teams-including those at service organizations or of employment opportunities to transgender advocacy groups, which may not have immediate 18 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA access to an in-house IRB-could explore instrument. We derived the following questions and partnering with IRBs at nearby universities or larger considerations from our interviews and review of organizations. existing literature: To secure IRB approval and conduct data collection • Does the research have a legitimate and as ethically as possible, research teams need to defensible purpose for collecting the information develop specific plans for assuring anonymity and (i.e., does the researcher need this information)? storing survey data in ways that minimize the risk Researchers often believe that more data are of linking specific survey responses (e.g., reported better, no matter the research question or final gender identity) to respondents' full names, addresses, product. But if the SOGI data are not going to be or birthdays. Researchers should be prepared to detail used, especially if they are not even relevant to what digital tools they will use for storage, how video or the research question, then they do not need to audio files will be stored safely and securely, how that be collected. Researchers should evaluate whether storage may update or change over time, and what steps the benefits of collecting sensitive information are are needed if the data are combined or merged with worth the risk to participants and the potential other data. Large-scale survey efforts seeking to release unintended consequences of data collection and datasets for future research should be particularly storage. As University of Chicago sociologist Kristin careful about how demographic data are both stored Schilt told us, "You should ask yourself, why am I and published. Obviously, qualitative data collection collecting this data if I am not planning to use it in raises privacy considerations to an entirely new level, my project?" owing to smaller sample sizes and the possibility of • How is the researcher guaranteeing safety and identifying a person, their location, or their occupation confidentiality? Especially for smaller organizations from their quotes or comments. Additional security that may not have official data security protocols measures should be undertaken by researchers and in place, leaving PII lying around on a desk or institutions storing video or audio files, transcriptions, on an unsecure computer risks the privacy and and related qualitative materials. safety of survey respondents. Bobby Jefferson, the global head of diversity, equity, engagement, In addition to building careful measures for data and inclusion at the development firm DAI Global, security, storage, and sharing in the research design told us that he provides survey respondents with phase, organizations that ask SOGI-related questions specific statements regarding data privacy, security, should consider ways to minimize potential harm and storage, which he believes helps increase experienced by respondents. Researchers should survey response rates and the overall accuracy of prepare transparent and detailed informed-consent the data collected. statements outlining who the research team is, where the funding comes from, and what data are • Has the research proposal undergone an IRB being collected and how they will be used. More review? IRBs are designed to protect research information about ethical research principles and study participants and act as a third-party review trauma-responsive survey design can be found in for the research team to ensure that sufficient the "Incorporating Trauma-Informed Care in Survey protections are in place for data collection and data Response" section below. communication. If the researcher does not have access to an IRB within their organization, it may Researchers interested in collecting SOGI data should be worth seeking an outside IRB or partnering with answer four key questions before designing a survey an organization that has an internal IRB. 19 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA • Is the researcher making promises they cannot keep? visibility and remaining cognizant that a survey cannot In our interviews, we heard how important it is to capture every part of a person's identity. tell survey participants why their data matter for Survey Mode the research project. But what happens when the There are many ways to conduct a survey: in person, research strategy, goals, or storage plans change? over the phone, via text, or online. Each survey can For large national surveys, it can be particularly be self-administered (i.e., the respondent answers the hard to ensure that the data will be used for their questions on their own) or administered during an intended purpose. Surveys are often used for many interview, and each method has its own challenges purposes and shared with other organizations. with regard to time, cost, and accuracy. And administration changes mean that the survey • Paper-and-pencil interviewing. This is the conducted under one set of privacy and storage traditional survey method, in which a respondent rules or priorities could be under different rules in a or interviewer fills out a paper form. new administration.20 When conducting surveys, it is important to clearly inform participants of all • Computer-assisted personal interviewing. In the possible ways in which their data may be used. this survey method, the interviewer sits with the For smaller-scale surveys, in particular, researchers respondent and records the answers using a mobile may have more latitude for when certain data fields phone, tablet, or computer. are necessary and how they might revise their data • Audio computer-assisted self-interviewing. A collection to maximize respondent protection. respondent listens to prerecorded questions through headphones and responds to the Building Inclusive Data Collection Methods questions by selecting their answers on a screen or Although several representative surveys in the US touch pad (Morrison-Beedy, Carey, and Tu 2006). currently collect SOGI data, most still collect gender • Self-administered questionnaire. A respondent data in the binary (i.e., man/woman), and if they collect completes the survey on their own, without an sexual orientation data at all, they do so along a binary interviewer present. This method can be efficient division (i.e., straight/gay). Of course, other surveys and inexpensive, but it can also result in missing and data collection efforts have included a wider data because of skipped questions (Morrison- array of identities, such as the National Transgender Beedy, Carey, and Tu 2006). Discrimination Survey (Grant et al. 2011), the 2022 US A large amount of evidence suggests that the number Transgender Survey,21 and the 2023 KFF/Washington Post Trans Survey22 (all of which were offered in English of people who stop taking a survey altogether (typically called survey termination or survey breakoff) and Spanish). These more inclusive surveys provide when they encounter SOGI questions is low (see lessons for how to better frame questions around Atrostic and Kalenkoski [2002] and references identity. therein). The evidence suggests that nonresponse In this section, we draw from existing surveys and our to individual SOGI questions is also low, ranging interviews with SOGI data experts and advocates to from less than 1 percent to about 6 percent (see, outline some principles for inclusive, thoughtful data for example, NASEM [2020]). By comparison, survey collection methods, including survey mode, question questions about income can have nonresponse rates response options, language choices, trauma-informed exceeding 20 percent (Atrostic and Kalenkoski 2002). design, and proxy reporting. Because data collection In a 2019 paper that used the National Health efforts and use cases vary so widely, we hesitate Interview Survey (NHIS), Dahlhamer, Galinsky, and to make blanket recommendations for all SOGI Joestl (2019) found statistically-and meaningfully- data work. With these principles, we try to balance similar rates of survey nonresponse to SOGI questions the tension between offering marginalized groups 20 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA under computer-assisted personal interviewing and The single gender question combines both cisgender audio computer-assisted self-interviewing survey modes. and transgender men/women as men/women and And other research has shown that nonresponse gives decent representation to those outside the rates in self-administered surveys are lower than gender binary through the nonbinary umbrella option. in interview-administered surveys (see Dahlhamer, Transgender status, however, is not explicitly captured Galinsky, and Joestl [2019]; and Jesdale [2021b]). (except for respondents selecting the nonbinary option) with the single gender measure, which may Asking Questions about Gender be insufficient for some use cases. There is no universal agreement on the best way to collect SOGI data. Large-scale survey strategies to The most recent 2023 NHIS (CDC 2023) offers collect gender data are rapidly evolving as research an important view into how large surveys are needs, community understanding, and language experimenting with SOGI questions, particularly in change. At the time of this writing, there are three terms of the single-measure approach. It introduces common approaches to asking questions about a significant change from previous NHIS and gender: (1) directly asking respondents about their other major surveys in the US.23 In the new NHIS, current gender in a single question or measure, (2) respondents are asked how they currently describe using a two-step approach to indirectly capture their gender, and they are able to select one or more gender and transgender status across two consecutive of seven possible answers (see the simplified survey questions, and (3) using an open-ended (i.e., write-in) text in table 1, next page). This is the first time the question. We discuss the first two options, which are NHIS has allowed users to select multiple gender more common in large-scale surveys, in detail below. options, and the first time it has explicitly offered "transgender" and "nonbinary" as answer options. In Single-Measure Approach for Current Categorical Gender follow-up questions, respondents are asked about A single question with limited yet inclusive options can their assigned sex at birth, and to confirm if it differs often be sufficient to record gender for research needs from the previous answer. NHIS data are collected without being taxing on survey respondents. The 2021 throughout the year, so information on how this Australian Census is an example of this approach revised survey worked and how respondents answered (figure 2). In this measure, respondents are asked the questions will not be available until 2024. about their current gender and are provided binary (man/woman), nonbinary, and write-in options. Two-Step-Measure Approach for Transgender Status and Current Categorical Gender FIGURE 2 A two-step approach is commonly used to collect data Sample Gender Question from 2021 Australian Census on gender. Two questions are used in this approach, with the first asking the respondent about their reported sex at birth and the second asking about their current gender. The first question typically limits options to male or female, aligning with how sex is and has been assigned at birth. When this response is combined with the response to the second question- the respondent's current gender identity-the two measures can be used to infer transgender status. For example, the Australian Census uses the gender Source: Australian Bureau of Statistics, Standard for Sex, Gender, measure in conjunction with the sex assigned at birth Variations of Sex Characteristics and Sexual Orientation Variables measure, as shown in figure 3 (next page). The two (Belconnen, AUS: ABS, 2020), https://www.abs.gov.au/statistics/ standards/standard-sex-gender-variations-sex-characteristics-and- measures are combined to derive transgender status, sexual-orientation-variables/latest-release#gender. as illustrated in figure 4 (next page). 21 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA TABLE 1 Summary Reconstruction of the 2023 National Health Interview Survey Questions QUESTION TEXT VA R I A B L E N A M E RESPONSES I N T E RV I E W N OT E S For this next question you may GENDER_A 1 Male select more than one answer. Do you 2 Female currently describe yourself as male, 3 Transgender female, transgender, nonbinary, or 4 Nonbinary 5 Another gender another gender? 7 Refused 9 Don't Know If the respondent selects option 1 (Male) or 2 (Female) only, they are asked the following question: Is [GENDER_A] the sex you were ASATB1_A 1 Yes assigned at birth, on your original 2 No birth certificate? 3 Refused 4 Don't know If the person selects more than one answer or selects option 3 (Transgender), 4 (Nonbinary), 7 (Refused), or 9 (Don't know), they are asked the following question: What sex were you assigned at birth, ASATB2_A 1 Male on your original birth certificate? 2 Female 3 Refused 9 Don't know If the person selects option 5 (Another gender), they are asked to write in their gender: What term do you use to describe GENDSPEC_A Verbatim your gender? 97 Refused 99 Don't know In all three cases, if the answers to the questions do not match, the respondent is asked to confirm the discrepancy, and the data are recorded as the respondent answered. Just to confirm, your sex assigned at GICHECK_A 1 Yes (For brevity, we only include one of the more than birth is [GENDER_A] and [ASATB1_A/ 2 No 25 options included in the NHIS codebook.) ASATB2_A/GENDSPEC_A]. Is that 7 Refused If [GENDER _A] IN (2,3,4), fill "you describe correct? 9 Don't know yourself as female, transgender, and nonbinary" Source: "National Health Interview Survey: 2023 NHIS," Centers for Disease Control and Prevention, National Center for Health Statistics, last reviewed July 5, 2023, https://www.cdc.gov/nchs/nhis/2023nhis.htm. FIGURE 3 FIGURE 4 Sample Sex Recorded at Birth Question from 2021 Australian Census The Cisgender and Trans and Gender Diverse Derivation Matrix, Two-Step Method from the Australian Bureau of Statistics Source: Australian Bureau of Statistics, Standard for Sex, Gender, Source: Australian Bureau of Statistics, Standard for Sex, Gender, Variations of Sex Characteristics and Sexual Orientation Variables Variations of Sex Characteristics and Sexual Orientation Variables (Belconnen, AUS: ABS, 2020), https://www.abs.gov.au/statistics/ (Belconnen, AUS: ABS, 2020), table 8, https://www.abs.gov.au/ standards/standard-sex-gender-variations-sex-characteristics-and- statistics/standards/standard-sex-gender-variations-sex-characteristics- sexual-orientation-variables/latest-release#gender. and-sexual-orientation-variables/latest-release#gender. 22 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA When the two-step approach uses a gender measure language used, the design of the questions, and the that is inclusive of nonbinary people, as in the definitions of identities. In the US, available options in Australian Census example, it can be an efficient the first question are typically male or female. The and accurate way to capture transgender status respondents are then asked their current gender and gender identity. However, the research design identity, with options often limited to male, female, principles discussed above should still be considered. and transgender. If an individual selects male or Some transgender people, for example, consider it female in response to the second question, and this offensive to be asked about sex assigned at birth, differs from the response to the first question on sex resulting in lower response rates or inaccurate data at birth, the interviewer may ask a follow-up question collection. Transparency about data use, research to confirm that the person's sex at birth and current goals, and why transgender status is needed can help gender identity are different. The National Crime encourage representative participation. Victimization Survey (figure 5) and the Census Bureau's Household Pulse Survey (figure 6) are examples of this The two-step approach has been adopted by two-step approach (see also UCLA Center for Health several large-scale survey efforts in the US, but Policy Research [2018] and NIAID [2023]). there is still ample room for improvement in the FIGURE 5 The National Crime Victimization Survey, Two-Step Approach Source: US Department of Justice, Bureau of Justice Statistics, "National Crime Victimization Survey: NCVS-1 Basic Screen Questionnaire," implementation date July 1, 2019, https://bjs.ojp.gov/content/pub/pdf/ncvs20_bsq.pdf. FIGURE 6 Census Bureau's Household Pulse Survey, Two-Step Approach Source: Thom File and Jason-Harold Lee, "Phase 3.2 of Census Bureau Survey Questions Now Include SOGI, Child Tax Credit, COVID Vaccination of Children," US Census Bureau, August 5, 2021, https://www.census.gov/library/stories/2021/08/household-pulse-survey- updates-sex-question-now-asks-sexual-orientation-and-gender-identity.html. 23 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA FIGURE 7 FIGURE 8 National Academies of Sciences, Engineering, and Medicine's Biden Administration's Recommendation for Asking Questions Recommendation for Asking Questions about Sex and Gender about Sex and Gender Identity Identity Source: National Academies of Sciences, Engineering, and Source: White House, "Recommendations on the Best Practices for Medicine, Measuring Sex, Gender Identity, and Sexual Orientation the Collection of Sexual Orientation and Gender Identity Data on (Washington, DC: National Academies Press, 2022). Federal Statistical Surveys" (Washington, DC: White House, 2023), https://www.whitehouse.gov/wp-content/uploads/2023/01/SOGI- Best-Practices.pdf. NASEM (figure 7) and the Biden administration both psychologist Devon Price offers an analogy (which recommend following a two-step approach when they self-described as "flawed") of asking survey asking about gender identity (figure 8). The NASEM respondents about their religious affiliation.26 Price report also recommends breaking out a separate offers the following list of possible responses to such "Two-Spirit" option for people who identify as a question: American Indian or Alaska Native earlier in the survey. 1. Protestant 3. Muslim 5. I converted to 2. Catholic 4. Jewish my religion The Behavioral Risk Factor Surveillance System (BRFSS),24 an annual survey conducted by the CDC, Price notes: "Listing 'I converted to my religion' as a first asks the respondents their sex at birth and religious identity option is like listing 'Transgender' provides four options: "Male," "Female," "Don't know/ as a gender option. In this context, 'Transgender' Not sure," and "Refused." If the respondent chooses is an answer to a question that hasn't been asked."27 the "Don't know/Not sure" or "Refused" option, they Gender questions that offer three options-male, are offered a follow-up question that asks whether female, transgender-are conflating gender the respondent considers themself transgender, assignment and a limited aspect of current gender with six separate options (see table 2, next page).25 identity. Many of these types of survey questions Many of these US examples offer "transgender" as also conflate biological sex and gender terminology. a survey response option to the gender question. The terms male and female refer to sex, while the terms However, it is important to note that the word man and woman refer to gender. But, as NASEM notes, transgender is an adjective, not a noun; in other "most people do not recognize a conceptional distinction words, the word itself does not describe a gender between sex terminology and gender terminology, but a person whose gender assignment at birth does which is likely both a cause and consequence of not match their current gender identity (as illustrated continued conceptual conflation and inconsistent in figure 4). Thus, a transgender woman is a woman, use of terminology in data collection and everyday a transgender man is a man, and some transgender life" (NASEM 2022, 39). In general, data users should people who fall outside this binary may want to be seek to apply the terms as precisely as possible and to identified only as a transgender person. Transgender make clear what data are being collected and why. 24 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA TABLE 2 2021 BRFSS Survey Questions about Gender Identity QUESTION NUMBER QUESTION TEXT VA R I A B L E N A M E RESPONSES I N T E RV I E W N OT E S MSAB.01 What was your sex BIRTHSEX 1 Male This question refers to the at birth? 2 Female original birth certificate of Was it male or 7 Don't know/Not sure the respondent. It does not refer female? 9 Refused to amended birth certificates. If BIRTHSEX does not equal 1 (Male) or 2 (Female), continue to question MSOGI.02 MSOGI.02 Do you consider TRNSGNDR 1 Yes, Transgender, Read if necessary: Some people yourself to be trans- male-to-female describe themselves as transgender gender? 2 Yes, Transgender, when they experience a different female to male gender identity from their sex at birth. 3 Yes, Transgender, For example, a person born into a male gender nonconforming body, but who feels woman would be 4 No transgender. Some transgender people 7 Don't know/Not sure change their physical appearance so 9 Refused that it matches their internal gender identity. Some transgender people take hormones and some have sur- gery. A transgender person may be of any sexual orientation-straight, gay, lesbian, or bisexual. If asked about definition of gender non-conforming: Some people think of themselves as gender non-conforming when they do not identify only as a man or only as a woman. If yes, ask Do you consider yourself to be 1. male-to-female, 2. female-to- male, or 3. gender non-conforming? Please say the number before the text response. Respondent can answer with either the number or the text/word. Source: Centers for Disease Control and Prevention, "2021 BRFSS Questionnaire" (Atlanta: CDC, 2022), https://www.cdc.gov/brfss/ questionnaires/pdf-ques/2021-BRFSS-Questionnaire-1-19-2022-508.pdf. Note: Questionnaires dating back to 1984 can be found on the CDC website at https://www.cdc.gov/brfss/questionnaires/. Also see Centers for Medicare & Medicaid Services, "Behavioral Risk Factor Surveillance Systems (BRFSS)," accessed September 21, 2023, https://www.cms. gov/files/document/sgm-clearinghouse-brfss-updated.pdf. Overall, compared with the Australian Census Even when the two-step approach is implemented measure and the new NHIS, most large-scale US appropriately, several challenges may affect data surveys do not capture a complete and accurate collection in the near and long term. Some people picture of the transgender population. Some may be transitioning (though not necessarily binary transgender people may choose a male or physically), meaning that the gender option they female gender or the transgender option, and some choose today may not be the same one they choose nonbinary people may or may not self-identify as tomorrow. In surveys that use a limited set of transgender. These results illustrate the importance of options-particularly with no write-in option- using terms precisely and correctly. people with genderfluid identities are erased. 25 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA Additionally, none of the methods mentioned may Surveys Specific to Transgender People make sense in the future as states' policies around We close this section by looking at two of the larger gender continues to diverge (figure 9). As of early US surveys directed specifically to people who are 2023, 13 states plus the District of Columbia allowed transgender to see how questions about gender are people to put a third gender category or X marker framed and what lessons they can offer large surveys on their birth certificates,28 which will make the first designed for a general audience. At this time, the question in the two-step approach moot when those 2015 US Transgender Survey (USTS) is the largest children reach survey age. California and New Jersey, survey to examine the experience of transgender for example, adopted a third gender category for birth people in the US, with nearly 28,000 respondents certificates in 2019. By comparison, in April 2022, (figure 10, next page). The 2011 National Transgender Oklahoma passed a law that forbids nonbinary gender Discrimination Survey (NTDS) included more than markers on birth certificates.29 6,000 transgender and gender-nonconforming study FIGURE 9 participants across the country (figure 11, next page). Nonbinary Birth Certificate Laws by State Both surveys started with a straightforward question: "Do you think of yourself as transgender?" (USTS) and "Do you consider yourself to be transgender/gender non-conforming in any way?" (NTDS). The USTS then asked respondents eight additional questions about their gender identity to include anyone who may fit within the survey criteria but may have responded "No" to the initial question. Any respondent who also answered "No" to all the subsequent questions was excluded from the sample. The NTDS took a more straightforward route: respondents who answered "No" to the first question were excluded from the resulting sample, though there were some allowances (based on follow-up questions) for people who did not Source: "Non-Binary Birth Certificates and State IDs: Full Guide," answer the question. Both surveys then implemented US Birth Certificates, accessed September 21, 2023, https://www. some form of a two-step gender question, first asking usbirthcertificates.com/articles/gender-neutral-birth-certificates- states. about sex assigned at birth and then about current gender identity. The USTS offered a long multiple- In any case, Nancy Bates, a retired US Census Bureau selection list of 25 identity labels, spanning gender senior research methodologist, explained that identity, gender expression, sex characteristics, and despite these challenges, data practitioners should other labels, as well as a write-in option.30 The NTDS still go forward with asking questions about gender: opted for a shorter list with binary options, a part- "The reason 'they are not perfect' is not a reason time category, and a free-text option to express not to use them." Multiple people we spoke with nonbinary gender identities. supported the idea that additional survey testing and experimentation are needed to identify the best In these community-specific cases, the direct screening ways to ask SOGI questions, to process the data, and, questions and expanded/write-in identity options ultimately, to communicate the results. provide researchers and other groups unparalleled insight into gender diversity within the transgender community. In seeking to collect data about the 26 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA transgender and gender-nonconforming population, and across the nation (or world)-for a wide variety these survey questions assume a level of familiarity of reasons. While not a panacea for addressing these with terms relating to gender and sexual orientation trends and changes in conventions, consulting and that the general population may lack. Furthermore, working with advocacy groups and specific communities words and phrases used to describe people and more generally can help researchers develop better communities change over time-both in specific areas surveys, resulting in more accurate data. FIGURE 10 FIGURE 11 The 2015 US Transgender Survey The 2011 National Transgender Discrimination Survey Source: Jaime M. Grant, Lisa A. Mottet, Justin Tanis, Jack Harrison, Jody L. Herman, and Mara Keisling, Injustice at Every Turn: A Report of the National Transgender Discrimination Survey (Washington, DC: National Center for Transgender Equality and National Gay and Lesbian Task Force, 2011), https://transequality.org/sites/default/ files/docs/resources/NTDS_Report.pdf. Source: S. E. James, J. L. Herman, S. Rankin, M. Keisling, L. Mottet, and M. Anafi, The Report of the 2015 U.S. Transgender Survey (Washington, DC: National Center for Transgender Equality, 2016). 27 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA Asking Questions about Sexual Orientation of sexual orientation they are seeking to better Sexual orientation consists of three dimensions: understand can result in better data collection and, • sexual identification: the sexual orientation that ultimately, better analysis and recommendations. one identifies with or uses to describe themself Many existing large-scale surveys that ask about • sexual behavior: the sex or gender of one's sexual gender identity also ask about sexual identity. The partners BRFSS, National Crime Victimization Survey, NHIS, • sexual attraction: the sex or gender of the and Census Bureau's Household Pulse Survey all ask individuals one feels emotionally, romantically, the same kind of questions, split separately for people or physically attracted to who identify as men and who identify as women. In the National Crime Victimization Survey, for example, While these dimensions are interconnected, they people who identify as men are provided with three are not always perfectly aligned, and the level of specific sexual orientation options ("Gay"; "Straight, overlap or alignment can vary. Sexual identification, that is, not gay"; and "Bisexual") and three other for example, can evolve over time, and an individual's catchall options ("Something else"; "I don't know sexual behavior can fluctuate or be influenced by the answer"; and "Refused"); people who identify various factors. When conducting a survey (or using as women are asked a very similar question, though survey data more generally), it is important for the first option is listed as "Lesbian or Gay." In both researchers to know which dimension(s) they are cases, straight is the only term explicitly defined in the interested in measuring when asking questions about question, and it is defined as the negation of being sexual orientation. Knowing which dimension(s) gay (figure 12). FIGURE 12 The National Crime Victimization Survey Question about Sexual Identity Source: US Department of Justice, Bureau of Justice Statistics, "National Crime Victimization Survey: NCVS-1 Basic Screen Questionnaire," implementation date July 1, 2019, https://bjs.ojp.gov/content/pub/pdf/ncvs20_bsq.pdf. 28 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA FIGURE 13 FIGURE 14 National Academies of Sciences, Engineering, and Medicine The White House Recommended Sexual Identity Question Recommended Sexual Identity Question Source: National Academies of Sciences, Engineering, and Source: White House, "Recommendations on the Best Practices for Medicine, Measuring Sex, Gender Identity, and Sexual Orientation the Collection of Sexual Orientation and Gender Identity Data on (Washington, DC: National Academies Press, 2022), https://nap. Federal Statistical Surveys" (Washington, DC: White House, 2023), nationalacademies.org/catalog/26424/measuring-sex-gender- https://www.whitehouse.gov/wp-content/uploads/2023/01/SOGI- identity-and-sexual-orientation. Best-Practices.pdf. The official recommendations from the NASEM report Short Lists, Long Lists, and Open-Ended (figure 13) and the White House (figure 14) collapse Survey Questions these two questions into a single question, with The list of response options presented is the core the first option listed as "Lesbian or gay" or "Gay or component of survey design and affects both what lesbian." Both provide a write-in category, but instead data the researcher can collect and how respondents of the diminishing "Other" label, they use a more approach completing the survey. Multiple affirming "I use a different term [free-text]" label for methodological strategies can be used to enable that option. The NASEM report also includes the researchers to capture a wider range of responses, "Two-Spirit" option for those who identify as with the two most common being a long list of American Indian or Alaska Native in an earlier options (e.g., offered as multiple-selection checkboxes question on race. or single-selection radio buttons) and an open-ended option that allows respondents to write out their When we spoke to Nancy Bates about providing response. Another strategy is providing a short list of a recommendation for ordering the options in the options. Consequently, a trade-off exists: a long list NASEM report, she said there is not enough research of identities or an open-ended option can be more to suggest a specific best practice, but a lack of representative, whereas a short list can decrease general knowledge about some of these sexual response time, increase the response rate, and make identities would make alphabetical ordering-where analyzing the data easier. bisexual would appear at the top of the list-confusing for many survey respondents.31 Kevin Guyan Our interviews with experts suggest that, in most reinforces this sentiment in his book Queer Data: cases, shorter lists are preferred. Although longer lists "Several scholars have noted a lack of familiarity can offer more representation, they also generate the or identification with terms such as 'heterosexual' following four primary issues: and 'straight' among heterosexual/straight survey 1. Increased possibility of error. Additional respondents" (Guyan 2022, 56). Again, survey size and subcategories within questions can lead to scope may be a helpful guidepost here; a small-scale misclassification and increased statistical noise. survey for LGBTQIA+ people may not have the same Members of the majority population may not be kind of language considerations, and an alphabetic familiar with the more specific response options ordering of response options may be sufficient. and misinterpret them, resulting in "false positives" 29 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA that bias results (Guyan 2022; Michaels et al. 2017; some survey respondents may choose the larger, Ridolfo, Miller, and Maitland 2012). Sexual identity more "majority" categories to ensure that their questions are particularly susceptible to false responses are not omitted from analysis because of positives when the respondent is unable to select inadequate number of responses. Although a longer multiple categories. In smaller surveys or surveys list of options can provide more representation directed toward specific communities or groups, opportunities, longer surveys mean longer however, longer lists may allow for a wider range of response times, which can result in lower response responses and provide the nuance that often exists rates and smaller datasets32 (see, for example, in sociodemographic survey data. Rolstad, Adler, and Rydén [2011]). For groups 2. Aggregation of results into larger categories. that are already small, losing survey respondents To protect privacy and ensure a large enough reduces the sample size and thus analysts' ability sample size, smaller groups of gender and sexual to draw statistically meaningful conclusions. But orientation identities are often aggregated best practices likely depend on the person or into larger categories. People who select group conducting the survey: a survey from a specific gender identities such as "agender" local LGBTQIA+ advocacy organization may be or "genderqueer" in a survey, for example, are perceived differently than a survey from the US often subsequently collapsed into a "nonbinary/ Census Bureau. gender-nonconforming" or even "other" category 4. Question refusal. In our interview, Marcus in later analysis. In practice, this erases people Berzofsky of RTI International wondered how who originally had the opportunity to select their people perceive lists of different lengths: "From the unique identity from being represented, even if respondent's perspective, is it worse if I had seen a their privacy is not always guaranteed (NASEM list of 20 and still didn't see how I identify? Or if I 2022). That being said, aggregating people into saw a list of four and didn't see it?" With a long list, groups that have been traditionally discriminated Berzofsky noted, a respondent may be offended if against may help support efforts to pursue their identity is not listed, as it suggests the survey equity and equality for those groups and give made an effort but still excluded them. This erasure them a magnified voice. In his book Queer Data, could lead respondents to refuse to answer the in a section based on Weinrich and colleagues questions. Stuart Michael, senior research scientist (1993), Guyan argues, "Depending on the spread at NORC at the University of Chicago, summarized of responses, analysis can find a middle ground this discussion, noting that "in survey research, between the crude binary of heterosexual/ you don't really need to have every single nuanced homosexual and a potentially unwieldy list of category that people use within their everyday life twenty response options" (Guyan 2022, 117). and community." 3. Failure to provide additional insight. Research When implementing a short list, researchers should suggests that when presented with long lists of consider how the response options are ordered. sexual identities, the vast majority of respondents Here again, there are trade-offs, and the best course choose "Straight," "Lesbian," "Gay," or "Bisexual" of action may depend on the survey and the target (see, for example, Bates, García Trejo, and Vines audience. Placing a "straight/heterosexual" option at [2019] and Virgile et al. [2022]). Thus, additional the top of the list of responses to a question about categories can make the survey longer and sexual identity, for example, may suggest a hierarchy more difficult to answer, without resulting in or norm, while alphabetically ordered options may be significantly more variation in responses. We easier to navigate. also heard in our interviews that, anecdotally, 30 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA When adding an "other" category, researchers should (2019) found that of the more than 200 nonblank consider alternative words or phrases that do not write-in answers, only 16 percent represented reinforce the idea that identities not represented sexual minority groups, such as "queer," "pansexual," in the list of options are less accepted or normal. or "asexual." The rest included write-ins, such as Instead, an open-ended text box with a prompt, such "normal," "not your business," or "Christian male." as "I identify as," "I use a different term," or "I am," Similarly, using pooled data from three separate US offers more inclusive options, explicitly eliminating Census surveys, Virgile and colleagues (2022) found the connotation that a person's identity is a choice. that 21 percent of open-ended responses were sexual minority identities; 14 percent were variations on Open-ended questions can sometimes seem like a "straight" or "heterosexual"; 17 percent were blank compromise between a long and a short list because or vague; and 20 percent were considered protest or they provide a write-in option, but this approach hard refusal–type answers, such as "Christian," "child still comes with trade-offs. On the one hand, open- of God," "womanizer," or "human" (see also Otero ended questions offer respondents the ability to be as Class, Meyers, and Berger [2022]). detailed as they wish; on the other hand, they require more time and energy on the part of analysts to parse By comparison, the Pew Research Center found and categorize the responses. that changing a survey question about gender from "Were you born male or female?" to "Do you For specific target audiences who may not be familiar describe yourself as a man, a woman, or in some other with certain terms, open-ended questions can way?" yielded extremely few of these extraneous offer more comfort. Kristin Schilt told us that in her responses-only 4 out of more than 2,600 responses, research with older adults, the team often opts for an or less than 0.2 percent of the sample.33 Similarly, open-ended question on sexual orientation, because the 2023 National LGBTQ+ Women's Community they think that people over 70 have less familiarity Survey,34 which specifically sought LGBTQ+ women with response categories such as queer or asexual. and femmes35 for the survey sample, found very According to Schilt, "Those categories haven't been few of these kinds of responses to the open-ended in circulation on mainstream survey instruments for questions about gender identity. As LGBTQIA+ very long." She then added, "But we don't know if identities face further politicization and vilification in our assumption will be correct-we will find out." The the current political culture, it is possible that protest process of wading through open-ended responses, or hard-refusal answers will increase. however, can be onerous. Amy O'Hara, formerly of the US Census Bureau and now a research professor No matter the type of question-whether a long at the Georgetown University Massive Data Institute, list, a short list, or open ended-how people select expressed some apprehension about the open-ended options should also be considered. A survey might option in our conversation: "As a data person, having enable respondents to select a single option (usually those write-ins means that you're going to have presented as a radio button or an item on a drop- somebody or some algorithm try to code them, and down menu) or multiple options (usually presented that's only going to be as good as what you put into it." as a set of checkboxes). Some LGBTQIA+ people may feel strongly that they fall within a specifically defined Research suggests that answers to open-ended identity, such as a "gay man" or a "bisexual woman." questions-particularly for questions around sexual But for many others, particularly queer-identifying identity-can yield extraneous responses. In their individuals and gender-nonconforming people, analysis of the 2020 Census Barriers, Attitudes, and categorical labels are not a good fit. Motivators Survey, Bates, García Trejo, and Vines 31 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA The very ethos of discrete data collection conflicts as queer in the modern parlance necessarily prefers with the non-normative and nonbinary frames the term. In deciding to use the word queer in its through which many LGBTQIA+ people identify reporting, NPR explicitly noted that older generations and live their lives. Researchers should understand may find the word painful, pejorative, or insulting while that discrete labels cannot capture the full breadth younger generations are more likely to embrace it.38 and variety of LGBTQIA+ identity and experience. Translating survey questions about sexuality from Surveys could provide a multiple-category option; English to other languages also presents unique the Census Bureau, for example, derives a "more than challenges because of the inherent cultural and one race" category by combining answers for people linguistic nuances involved. Terms, expressions, who choose more than one race in its surveys.36 In attitudes, and cultural context related to sexual any case, categorical labels can have consequences, orientation, gender identity, and sexual practices whether intentional or not, so researchers must may differ between English-speaking and non- recognize that the trade-offs between offering a wider English-speaking communities, requiring careful range of response options and optimizing survey consideration to ensure accurate translation.39 Some response rates may depend on the type and purpose concepts may not have direct English equivalents in of the survey. other languages, which necessitates the adaptation There are clear trade-offs between short lists, long and contextualization of questions to capture the lists, and open-ended questions. How people see intended meaning. Additionally, language structures themselves in the survey or final data, how the and syntax can differ, making it crucial to ensure that questions affect survey response time and rates, how questions are linguistically appropriate, clear, and those data will ultimately be used, and what type of comprehensible to respondents. As Chloe Schwenke, survey is conducted are all factors to be considered president and founder of the Center for Values in when designing question formats. In the end, more International Development, said: "We're not trying to research and testing are needed across different impose a Western definition of what is LGBTQ … but groups and communities to better inform survey just to be able to find commonalities is an important practice. piece of the work that needs to be done." Language Shifts and Translation In English, pronouns (e.g., he/him/she/her) are When thinking about large-scale or cross-national associated with different genders. Other languages, research, understanding differences in meaning and such as Spanish and French, classify nouns as semantics both within and across languages can be masculine or feminine (e.g., "el sol" and "la luna") and especially important. Over recent decades, several traditionally indicate people's gender by ending a words related to LGBTQIA+ identities have shifted word in either -a for women or -o for men (e.g., Latina in meaning even within the English language. The and Latino). When describing a mixed-gender group, word queer, for example, was used as a slur to attack these languages tend to ascribe the masculine plural homosexual identity as unusual, wrong, or inferior form, but the use of the -e or -x suffix for nonbinary through much of the 20th century. But when the people (e.g., Latine and Latinx) or to neutrally address word was reclaimed, as "a term of power" in the early a mixed-gender group has increased in the last several 1990s, as linguist Gregory Coles writes, it became a years. Further complicating matters, words that are "category term" rather than a "derogatory term" (Coles masculine in one language may be feminine in another 2016).37 That being said, not everyone who identifies language. 32 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA Other languages similarly have different approaches "heterosexual" option. Following further discussions to gender. In Urdu, for example, gender is indicated with survey participants, the authors found that in the verb, not the pronoun: "he walks" is "who the Spanish-speaking respondents had "difficulties chalta hai" and "she walks" is "who chalti hai." The understanding the term 'heterosexual,' leading to their pronoun "who" is genderless. And in some Indigenous choosing 'something else' or saying that they didn't communities, the term Two Spirit refers to "a person know how to answer."41 who identifies as having both a masculine and a As languages continue to evolve, words and phrases feminine spirit, and it is used by some Indigenous could change to incorporate different identities. people to describe their sexual, gender and/or The increased use of the words Latinx and Latine as spiritual identity." 40 Chrystos, a Two-Spirit poet from alternatives to Latino/Latina, for example, avoids the the Menominee nation, told author Leslie Feinberg: nonbinary gender suffix42 (Schwabish and Feng 2021), "Most of the nations that I know of traditionally had as does the shift in Latin America from the word queer more than two genders. It varies from tribe to tribe. to cuir.43 The concept of Two-Spiritedness is a rather rough translation into English of that idea. I think the English Generally, in large government-led surveys, the terms language is rigid, and the thought patterns that form it used reflect the nation's structures and institutions are rigid, so that gender also becomes rigid" (Feinberg that have placed white people (typically cisgender 1996, 27). men) in positions of power and decisionmaking. Overcoming these challenges in translating may When translating surveys into another language, require using open-ended options to allow survey it is worth taking the time to understand how respondents to self-identify. Federal survey that language treats gender identities and sexual organizations like the Census Bureau are not likely orientations. Translation from English to Spanish, to use separate question structures for different for example, is not necessarily a one-to-one languages, but a smaller survey or a community-based transformation. In their survey, Stuart Michaels and organization conducting a local survey could create colleagues (2017) used different terms for sexual tailored questions in different languages. identity questions in English and Spanish. For English- speaking respondents, they used the phrase "Straight, Language is complicated, nuanced, and forever that is, not (lesbian or) gay." For Spanish-speaking evolving. There are no objective solutions for finding respondents, they translated the word straight to the right words or phrases to describe people heterosexual ("Heterosexual, o sea, no gay o lesbiana"). and their identities. Although we do not focus on While the vast majority of English speakers in their qualitative data collection or research methods here, sample were able to answer this sexual identity engaging with people and communities can help question correctly, almost 60 percent of the non- researchers better understand what questions to ask LGBTQIA+ Spanish speakers did not select the and how (box 2). 33 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA BOX 2: STRENGTHENING RESEARCH BY ENGAGING COMMUNITY Community engagement can take many forms, but at a high level, it is the explicit inclusion of community members as collaborators, reviewers, or participants in the research process (Sankofa, Daly, and Falkenburger 2021). Incorporating community engagement in the research process for SOGI data collection can ground the data in the lived realities of those communities. Community engagement can take place in four main phases: For SOGI data, community engagement can be particularly useful contextualizing the research questions (e.g., understanding in interpreting survey results and ensuring that there is proper the issues and how to engage communities of interest); expertise and cultural competency to understand their nuance. survey design (e.g., knowing what questions to ask and how); As Meghan Maury at the Office of Science and Technology Policy survey implementation (e.g., outreach and survey administra- highlighted in our interview, "Having regular, consistent commu- tion); and data analysis and dissemination (e.g., interpreting nications with community members about their priorities for how and sharing results) (Harrison et al. 2021). data is collected and disseminated" can help make findings acces- Connecting with advocacy groups, nonprofit organizations, or sible for that group and expand the impact of the research. other community-based organizations offers one way to tap The appropriate level of community engagement will depend on into people's knowledge and experience. Community advisory the project's goals, scale, capacity, and limitations. Community boards-groups composed of community members who share engagement requires time and resources to build trust and create an identity, geography, history, or other characteristics or a sustainable partnership. This additional time can pay dividends experiences-can also become an integral part of a project by ensuring that the data collected are robust, accurate, and team (Arnos et al. 2021). Building and promoting a diverse effective. For more information about community engagement, research team or workforce is another way to bring a variety we recommend the Urban Institute report Community-Engaged of perspectives and lived experiences directly into the work. Surveys: From Research Design to Analysis and Dissemination. Incorporating Trauma-Informed Care can help LGBTQIA+ people to more willingly engage in Survey Response with survey instruments and minimize the potential Trauma may be defined as "a response to anything for harm (Jaffe et al. 2015). The Center of Excellence that's overwhelming, that happens too much, too on LGBTQ+ Behavioral Health Equity defines six fast, too soon, or too long-coupled with a lack of key principles of trauma-informed care for LGBTQ+ protection or support."44 Unfortunately, much of people: safety, peer support, empowerment, the general population has experienced some form trustworthiness and transparency, collaboration, of trauma, either as one-time events or through and cultural responsiveness (Levenson, Craig, and cumulative harms over time. In the US, groups that Austin 2021). These principles translate well to the have been subjected to systemic discrimination and four key assumptions of applying trauma-informed marginality have a uniquely traumatizing experience care to research, as defined by Voith and colleagues (Carter, Gibbons, and Beach 2021; Estrada et al. (2020): realizing trauma impacts, recognizing signs 2022; Green, Price, and Dorison 2022; Peterson et al. of or potential for trauma among those involved 2021). As the American Psychiatric Association notes, in a research study, responding by integrating "LGBTQ populations have unique lived experiences trauma knowledge, and resisting the possibility of partly defined by adversity and discrimination,"45 retraumatization. and many of these experiences may be directly For general population surveys that only engage with traumatic, including experiences of familial rejection. LGBTQIA+ topics when collecting demographic data, Incorporating trauma-informed research practices considerations for how to respond to and support 34 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA possible trauma histories among individual 4. culturally competent, representative question respondents are minimal. In these surveys, it is response options and survey administration more important to recognize systemic trauma (e.g., practices institutional discrimination, regular omission of For detailed examples of how to implement these nonbinary identities from demographic forms, etc.) practices, we recommend Trauma-Informed Socially that may affect how LGBTQIA+ people interact with Just Research Framework, developed by Voith and questions about their identities. Targeted surveys colleagues (2022). that ask about specific experiences of harm among LGBTQIA+ people demand more robust application Proxy Reporting of trauma-informed principles. Qualitative and mixed- For many large-scale surveys, proxy response-having methods research incorporating trauma-informed one person respond for all eligible members of the practices are more expansive and emphasize the family, household, or community-offers a useful need for partnership,46 but they are beyond the way to reduce costs and time (Mathiowetz and scope of this guide. Groves 1985). However, proxy responses can impair the quality of the data, depending on what is being Trauma-informed survey research practices can collected and the complexity of the response options respond to participants' fears of submitting data (Fulton et al. 2020). In the context of SOGI questions, about their identity and then not being involved in the there is little research on the effect of proxy reporting, rest of the process. By actively incorporating plans including how sensitive or difficult respondents for community-engaged data dissemination as part of find the questions and whether proxies have the a research approach that prioritizes participant trust knowledge or even the willingness to answer the and researcher transparency, researchers can start to survey questions. address some of these anxieties (Edelman 2023). To ensure that data instruments are as conscientious as In a thorough report on proxy reporting, Holzberg and possible, research teams should participate in trauma- colleagues conducted 132 interviews in four cities in informed research training. Researchers should also the US. They found that most proxy respondents did consider team supports and minimizing trauma for not have difficulty or sensitivity when reporting SOGI researchers, especially if a survey instrument asks information. Respondents were willing to report SOGI about violence or discrimination. information for themselves and other members of their household, with only one respondent refusing Ultimately, maximizing respondent comfort maximizes to answer. There was also a high level of agreement the likelihood of full participation, which strengthens in responses when interviews were conducted in the data (Saleh and Bista 2017). When conducting pairs. Overall, the authors suggest that "asking SOGI a survey instrument with LGBTQIA+ populations, by proxy may be feasible in large-scale, general researchers should keep the following four best population surveys" (Holzberg et al. 2019, 904). practices in mind: Their conclusion, however, comes with an important 1. transparent and robust informed consent caveat. In the qualitative interviews and focus 2. disclaimers for sensitive or potentially upsetting groups, some respondents noted that the questions questions, with reiteration that participation is might be sensitive for other survey respondents voluntary in the household-for example, "[My husband] 3. provision of support resources as needed would find the gay and lesbian, the transgender, at the end of survey administration and the [disability questions] sensitive … He was 35 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA raised in Alabama as a Baptist." Others noted that When working with proxies, researchers should be they were unsure of the correct term or category aware that revealing information to or from the proxy to use-for example, one proxy respondent living may put the safety of another household member in an LGBTQIA+ household reported that "they at risk. Interviewers and interview teams should be [others in the household] see sexuality [as] more vigilant about the level of risk involved in asking fluid. They might answer it 'lesbian,' might answer sensitive identity questions and should use trauma- 'bisexual.'" Altogether, quantitative results suggest informed surveying methods if possible. little difference with using proxy reports, whereas qualitative results suggest that proxies could provide incorrect or misleading information. 36 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA Part Four Data Analysis Analyzing SOGI data is like analyzing any other PART FOUR data. Researchers will need to make some key Data Analysis considerations and judgments, just as they would with any type of data they analyze. Our goal for this section is not to prescribe a set of instructions for different types of analysis but to highlight additional considerations that may be necessary when analyzing SOGI-specific data in different contexts. Although we primarily focus on quantitative data in this guide, qualitative data should be incorporated into research when possible to complement the quantitative analysis. Qualitative data can help researchers understand the "cracks" in the quantitative data-the things that are not captured in the checklist of survey options. Qualitative data can also instill a greater sense of empathy by allowing both the researchers and stakeholders to better connect with the people and communities in question. Especially in the case of SOGI data collection, where existing survey categories may not be sufficient to capture the expansiveness of gender and sexual orientation, qualitative data can humanize the quantitative data and help provide additional perspective and representation. Demographic Survey Data SOGI data can often produce small sample sizes for gender minorities, which can pose many challenges to researchers' data analysis. Researchers may need to combine groups into larger categories to retain an appropriate sample size (e.g., collapsing response options to "cisgender" and "transgender and/or nonbinary" or "heterosexual" and "not heterosexual"), but they should be aware that variation may be lost in the process and should note it in the final analysis. It is incumbent on researchers to read the survey codebook and understand exactly what questions were asked. Shortcutting or summarizing answers to survey questions should be done with care. If a survey asks respondents, "Do you or have you ever had sex with someone of your same sex?" reporting a single answer as "homosexual" may be incorrect, as the measure only asked about sexual behavior and not sexual identity. Researchers should be explicit about what the question asks, acknowledge the specific 38 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA language, and state any summary or proxy words/ Cleaning SOGI Survey Data phrases used. This consideration is especially Data cleaning is the process of fixing or removing important when merging different datasets in which incorrect, corrupted, incorrectly formatted, duplicate, questions about gender or sexual orientation are or incomplete data within a dataset. Several challenges asked differently and may have different implications. can make cleaning SOGI data more difficult than Gender and sex are separate variables, and cleaning other types of data: researchers should not conflate the two. Terms like • Intersecting identities. There are multiple man and woman refer to gender, while terms like components of sexual orientation (e.g., identity, male and female refer to sex. If the survey uses the sexual desire, emotional attraction, and behavior) terms male and female, the analysis results should be and gender (e.g., identity and expression). The reported as such, instead of using terms like man and data should match the actual question-meaning the woman. Surveys often ask about gender but include researcher may need to reframe the analysis or the "male" and "female" as response options. communication of the analysis to accurately reflect Lastly, any question that results in missing data for the identities measured in the data. gender minorities can have a greater effect on these • The "other" category. Many SOGI survey questions groups during the imputation process, because include an open-ended or "other" response option. imputations for small samples can distort aggregate Researchers need to be aware that using such data summary statistics, such as means, medians, and can be time consuming. Additionally, interpretation variances (see, for example, Little and Rubin [2019]). of these responses may be subjective, especially if William Jesdale at the UMass Chan Medical School aggregating them into larger groups. One strategy found that more than half of respondents in the to increase the reliability of the open-ended July–October 2021 Census Household Pulse survey response option is to have multiple people work who identified as transgender were attributable to independently to code the responses, and then the imputation method of sex at birth. Although create a consistent set of outcomes by working nearly 4,000 people out of more than 300,000 together (see, for example, O'Connor and Joffe survey respondents had their sex at birth imputed- [2020] and Braun and Clarke [2012]). accounting for 1 percent of the entire sample- • Splitting and lumping. Splitting (disaggregating) the imputation overly affected estimates of the or lumping (aggregating) identity groups can be transgender population (Jesdale 2021a; Herman and difficult, because the process may require certain O'Neill 2022). The analysis therefore suggests that interpretation of identities. Again, there is no the resulting counts of the transgender population in perfect answer to the question of when to split or the Pulse survey were overestimated, and that more lump responses; it will depend on the individual accurate estimates can be obtained using known data use case. Is the research seeking to make specific, rather than data that relies on imputed sex at birth. statistically meaningful comparisons? If so, it may Until additional research can be conducted on these need larger groups. Is the research seeking to surveys and imputation methodologies for SOGI data, better understand the makeup of a local area or Jesdale recommends restricting imputation to people different gender/sexual orientation populations? If with known sex at birth in analyses (for surveys that so, it may need smaller groups, while recognizing use that approach).47 that, from a purely statistical perspective, drawing conclusions may not be possible. The NASEM report explicitly identifies these practices as 39 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA potentially problematic: "When this [lumping] • Cultural considerations. In some cultures, there occurs, although respondents may have initially had is no distinction between sex and gender, so the opportunity to express their unique identity in respondents may interpret questions on gender data collection, the end result is that their voice is differently. It also may not be possible to fully erased" (NASEM 2022, 67). explain these differences within a survey. It is • Evolution of language. As survey language evolves therefore worth considering how survey options over time to reflect current trends, groups, may be interpreted differently in different and understanding, researchers may need to languages. Most large-scale surveys in the US will consider how a new term relates to a past term. be translated into Spanish, and researchers should As we mentioned earlier, the word queer is a good include considerations to that effect. For smaller- example-although the word was reclaimed from scale surveys-especially those focusing on SOGI being used as a slur in the early 1990s, because of data-researchers should be aware to whom the its history, it is still not a preferred word for many survey is geared and how it may be translated into people. other languages. 40 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA Part Five Communication and Data Visualization PART FIVE With the collection and analysis completed, the final step is to visualize and communicate the Communication work. There are five main considerations when and Data communicating data, all of which are similar to Visualization those outlined in the other Do No Harm Guides. However, unlike the data on race and ethnicity discussed in Do No Harm Guide: Applying Equity Awareness in Data Visualization, there is less evidence on how people are presenting broader SOGI categories-for better or for worse-likely because there is simply not enough data to be visualized. Here, we provide some general guidelines for each of the five considerations. Color Color is one of the most powerful pieces in the data visualization toolkit. It can be used to enhance and clarify while respecting and recognizing the different identities of the people represented in the charts and the people reading them.48 As SOGI data collection improves and more data become available, we will likely see how language and culture coincide with the data, and how the two evolve together through data visualization and other products (Kay and McDaniel 1978; Setlur and Stone 2016). Color perceptions and usage already differ by culture, which will influence how colors assigned to SOGI data are perceived. In Western cultures, for instance, red is often used to show errors or negative values, while in Eastern cultures, red is often used for prosperity and good luck. Since the mid-20th century, the blue- pink color pair is instantly identifiable as representing men and women in Western cultures. But this color scheme is also rooted in sexism, with many societies traditionally assigning less value to feminine-coded colors and products. It does not have to be this way, and other color pairs should be considered.49 42 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA When using a color palette to represent a broader Terminology in Data Visualization range of SOGI data (e.g., men/women, straight/gay), We addressed the evolution of language and any color combinations could be candidates. If we translation earlier in this guide, but when it comes draw color palettes from LGBTQIA+ movement pride to communicating SOGI data, there are a few other flags, then there are essentially endless choices considerations to keep in mind. First, keep language (figure 15). The iconic pride rainbow flag typically consistent with what is asked in the survey, but includes six colors that span the rainbow.50 The balance that continuity with using more equitable transgender flag, which debuted in 1999, consists and inclusive terminology. If the survey collected of baby blue, baby pink, and white. And the progress sex data using the terms man/woman, for example, pride flag uses blue, pink, and white from the a researcher can choose to change them to male/ transgender pride flag, with brown and black to female to be more accurate. A similar change can be represent people of color and the unique challenges made when using race/ethnicity data. If a survey uses that queer people of color face accessing the same the term Latinx, but a researcher recognizes that their rights as white and heterosexual people. audience prefers alternative terms, such as Hispanic Whatever colors are chosen, avoid using sequential or Latino, they can choose to make the change color palettes when presenting data along a continuous (Schwabish and Feng 2021). Whenever terms are spectrum. Sometimes referred to as a color ramp or color changed from the original survey, it is best practice gradient, the sequential palette uses a single hue (e.g., to acknowledge the original language used-the blue) that ranges from a light color for small numbers to respondent did, after all, select the box for "man/ a dark color for large numbers. The sequential blue color woman"-and note that the visualization or text uses a palette used in the pie chart in figure 16, for example, different term, word, or phrase. Endnotes or footnotes suggests a hierarchical ordering-from gay/lesbian/ can also recognize data collection limitations-for same-gender (29 percent) to bisexual (31 percent) to example, a footnote might say, "Nonbinary genders heterosexual (23 percent) and so on-that is not reflected were not available for this analysis and are a in the data (Grant et al. 2011). A sequential color palette limitation." should refer to data values, not to category names. FIGURE 15 FIGURE 16 Flags for the SOGI Community Sequential Color Palette that Does Not Match the Data Source: Authors created the images based on "Flags of the LGBTIQ Source: Jaime M. Grant, Lisa A. Mottet, Justin Tanis, Jack Harrison, Community," Outright International, accessed September 21, 2023. Jody L. Herman, and Mara Keisling, Injustice at Every Turn: A Report of the National Transgender Discrimination Survey (Washington, Note: A. LGBTQIA+ movement pride flag; B. transgender flag; DC: National Center for Transgender Equality and National Gay C. progress pride flag. and Lesbian Task Force, 2011), 29, https://transequality.org/sites/ default/files/docs/resources/NTDS_Report.pdf. 4 3 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA Second, the term other is common in data from the Noun Project, an online resource for icons. visualization, but it literally "others" individuals by Which icon (if any) feels as if it completely captures emphasizing how they are different from some the complexity of being transgender? perceived norm.51 We propose several alternatives FIGURE 17 to the "other" label: Examples of Icons Representing the Term "Transgender" • another gender/sexual orientation • additional groups • all other self-descriptions • people identifying as other or multiple genders/sexual identities • identity not listed • identity not listed in the survey In general, consider how the language could misrepresent the issue, be misconstrued or misunderstood, or be used for harm.52 Are the data being presented proportional to the issue? There are more than 1.3 million adults (ages 18 and older) Source: Noun Project, s.v., "transgender," accessed September 21, and about 300,000 younger people (ages 13 to 17) 2023, https://thenounproject.com/search/icons/?q=transgender. who identify as transgender in the US (Herman, Again, the use case is important here. Using an icon in Flores, and O'Neill 2022). Is it necessary for data a graphic that explores violence against transgender visualization, especially within a larger argument, to people will look and feel very different from using an focus on the transgender community? Does it treat icon in a sign on the bathroom door. In the latter case, all people equally or does it minimize the experiences a reader just needs to know where the restroom is or challenges certain people or groups face? These and whether it is wheelchair accessible, which could questions have answers, and they likely can be be simply symbolized by a wheelchair and toilet. That addressed by including people with lived experience being said, it is worth asking whether icons like those of the topic being discussed or visualized. shown in figure 18 are even necessary. Information that expresses the gender of the people who most often Icons use the bathroom may not be especially important. (It is Instead of using an abstract shape, such as a bar or a worth noting that some may perceive these icons as circle, to represent a data value, it is sometimes better comparing people who do not fit into the binary gender to use icons, which enable readers to see themselves norm with mythical or imaginary things.53) in the data and connect directly with the person or community underlying the data. By offering readers FIGURE 18 Example of Icons Representing Nonbinary Genders a way to see themselves in the data, researchers can help them feel more empathy toward the people the data represent. But icons also can be problematic, especially given the various intersections and complexities of human beings and identities. No single icon will be able to capture all that complexity. Figure 17, for example, shows 4 icons (out of nearly 250 options) found under the search term transgender Source: Photo taken by author. 4 4 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA At their core, icons are just representations of people. the change in a variable that increases from 3 units Although they can do harm, we should be less to 12 units as an increase of 300 percent is factually concerned with their total accuracy in depicting every true, but it can also distort the actual meaning of that characteristic and identity, and instead use them as increase.54 Such distortion can be especially harmful they are intended-as representations of people. In his when reporting about health care for transgender book Joyful Infographics, Nigel Holmes suggests the children, which can be intentionally misleading and following: "When making icons of people for Isotype- risk the health and safety of transgender children and like charts, consider making them all blue. Using a their families. blue [or, presumably, any single color] figure of a Ordering person is not a way to avoid questions of difference or inequality, it simply makes everyone equal, whereas Finally, as with all identities, think carefully about showing people in precisely-color-calibrated, almost- how to order the results in the table, graph, or chart. realistic pictures amplifies the differences-that's Simply because the survey codes man as 1, woman as the intention, of course-but it does not promote 2, and transgender as 3-which we have noted above equality" (Holmes 2023, 106). may be problematic because the word transgender does not describe a gender-does not mean the Showing Small Numbers results need to be shown in that order. As we have demonstrated throughout this guide, As with equity in race and ethnicity data, consider one of the challenges in using SOGI data is the whether sorting the reults by population size, sample issue of small sample sizes. With an estimated 0.5 size (weighted or unweighted), or magnitude makes percent of US adults (ages 18 and older) identifying more sense than the data default (Schwabish and as transgender (Herman, Flores, and O'Neill 2022), Feng 2021). Alphabetical ordering of SOGI data can any sample of transgender respondents, especially be more difficult, because many people may not when disaggregated by race, ethnicity, geography, or be familiar with certain terms like asexual, intersex, something else, will be quite small. or genderqueer. However the results are ordered, These small-sample-size issues require data as long as the decision is made, conscientiously and visualization creators to be conscientious and careful purposefully, the work will be the better for it. about using levels or growth rates. Characterizing 4 5 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA Part Six Conclusion Despite recent efforts to expand and improve PART SIX SOGI data practices, there is still much work Conclusion to be done. And with many trying to strip LGBTQIA+ people of their rights, these data can help improve policy and inform perspectives. The work of collecting, using, analyzing, and communicating SOGI data-and all data that represent people and communities, for that matter-should be carried out carefully, respectfully, and through the lens of equity and inclusivity. We have presented many recommendations and guidelines that people working with data should consider when presenting data relating to gender and sexual orientation. The issues we have highlighted here are not static and do not necessarily have concrete right or wrong approaches. But the principles we have outlined can help people and organizations think more critically about how to work with SOGI data. We urge analysts and researchers to be aware of how decisions made at each step in the pipeline may not only affect data accuracy and representation but also potentially put vulnerable populations at risk. Many issues discussed in this guide and that our interviewees and advisers brought to our attention will continue to evolve and change with society, culture, and norms. By applying SOGI data best practices, experts in data and research fields are in a unique position to improve how people view and understand data relating to gender and sexual orientation, and to help governments, organizations, and other groups implement strategies, programs, and policies that address disparities and inequities-and to use data for good. 47 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA Part Seven Appendix APPENDIX A: PROTOCOL FOR INTERVIEW Confidentiality Statement and Conversation Guide: Interview Introduction and Consent Hello, my name is [insert name] and I'm joined by my colleague • We would like permission share what we learn from you [insert name]. We are researchers with the Urban Institute, a today, but we will take precautions to protect your identity nonprofit research organization based in Washington, D.C. during the data collection process. The Do No Harm Guide - Gender research project, funded by • We will make every effort to protect your identity; the Tableau Foundation, is centered on identifying promising however, there is a small chance your comments and/or practices for collecting data around gender identity and sexual descriptions could be attributed to you in the final report. orientation that would help move the ideas of equity and • We also cannot guarantee the confidentiality of the inclusivity in the research, data science, and data visualization information you provide given the nature of Zoom. We communities forward. By conducting a landscape scan of the believe the risk of sharing information is minimal, but you current state of data collection around sexual orientations are free to decline to respond to any question that you and gender identities/expressions (SOGIEs), interviewing and are not comfortable answering. engaging in thought partnership with LGBTQ+ data leaders, • We will not cite your name in the report unless you grant and developing a series of accessible public products with the us permission to do so. We ask that you participate support of the Tableau Foundation, Urban intends to continue in a private setting away from earshot or viewing by its partnership with the Tableau Foundation in creating data unauthorized persons to include family members. products for all people and communities. The final deliverable will be a report that includes a review of existing best We take all notes on password-protected computers and store practices and interviews with experts around the country. them in folders only accessible to researchers working on this project who have signed a confidentiality pledge. This research will help data communicators create more equitable, inclusive, and accessible data products. While Do you have any questions before we begin? there are increasing conversations around the need to [pause for response] inclusively represent sexual orientations and gender identities/ If you have any questions about this study, you can contact: expressions in data collection and visualization, there does Jon Schwabish, Principal Investigator, the Urban Institute not seem to be much agreement or formal resources around jschwabish@urban.org best practices. This guide will focus on creating guidelines and best practices to advance equity in data products. The Do you agree to participate in this conversation? guide will also help researchers and analysts be more forward [if no] No problem, thank you for your time. thinking as they prepare their data collection, analysis, [if yes] If it's ok with you, we would like to record this and communication efforts. We are conducting a series of conversation to make sure that we get what you have to interviews with experts in LGBTQ+ data collection to help say down correctly. It will also help us move through the guide the development of best-practices as well as to bring interview with fewer pauses for the notes. Once the a first-person perspective to the challenges-and rewards- project is over, we will delete the recording. At any time of creating more equitable and inclusive data products. while we talk, we can also stop the recording if you like. We know that you are busy, so we will be as focused as Do you consent to this interview being recorded? possible. Your participation in this discussion is completely [if yes, start recording.] Thanks. We're recording now. voluntary. That means you may choose to skip any questions [if no, do not record.] No problem, we won't record. you wish, refuse to participate, or stop the interview at any time. In this interview we are interested in learning about your ideas, experiences, and perspectives. There are no right or This interview will last up to 60-90 minutes. My colleague, wrong answers. Do you have any questions before we begin? [name], will be taking notes today and we would like to record If you have any questions during our interview, please do not the interview as well to make sure we capture everything you hesitate to ask. say accurately. 49 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA APPENDIX B: DATA EXPERTS AND AGENCY STAKEHOLDERS INTERVIEW PROTOCOL Interview teams should first ensure the informed consent is covered, which can be found here: [link] Setting the Space Thank you again for speaking with us today. In this conversation, we will use the term "SOGIE" to refer to sexual orientation and/or gender identity and expression data. We want this interview to be only semi-structured and are eager to learn from you as the expert, so please feel free to speak as much as you'd like and we'll probe with questions where needed. SECTION 1: Respondent and Organization Background Thank you for agreeing to speak with us today. We look forward to learning more about your perspectives on and experience with collecting SOGIE data in various research efforts. We'd like to start with some basic background questions about you and your work. 1. Can you tell us about your [institution/agency] and your role? a. What is your title? b. How long have you been in this position? c. What are your primary responsibilities/what is your primary focus? 2. Please briefly provide an overview of your or your agency's work as it relates to demographic data collection broadly. a. [probe] 3. In what capacity do you engage with SOGIE data? a. Is it something you ensure is included in any data collection and analysis effort or is it something that only pops up in dedicated projects? b. Use this question to have the interviewee(s) expand on their SOGIE data work or familiarity with the idea of collecting SOGIE data. i. To consider: does their organization have a specific team(s) that work on these issues? Is everyone in the organization tasked with thinking about these issues? Are there distinctions between the technical and content teams? SECTION 2: Sexual Orientation Data Considerations 4. Have you ever written a survey/form/census that asked respondents about their sexual orientation? a. If no, move to section 3. b. If yes: i. Have you ever designed a survey/project with the explicit goal of learning about sexual orientation of some specific population? 5. What is your opinion on (or experience with) having respondents report their sexual orientation by selecting from a list of options vs. asking via an open-ended question? a. What trade-offs do you consider when designing how respondents report their sexual orientation (i.e., list of options vs. open-ended questions)? 6. In your view, what are the tradeoffs between these different data collection methods and survey response rates? 7. What is your opinion on (or experience with) reporting respondents' sexual orientation, especially in data sets that include a limited number of options? 50 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA SECTION 3: Gender Identity and Expression Data Considerations 8. Have you ever written a survey/form/census that asked respondents about their gender identity? a. If no, move to section 4. b. If yes: i. Have you ever designed a survey/project with the explicit goal of learning about gender identity among some population? 9. What is your opinion on (or experience with) having respondents report their gender identity by selecting from a list of options vs. asking via an open-ended question? 10. In your view, what are the tradeoffs between these different data collection methods and survey response rates? 11. What is your opinion on (or experience with) reporting respondents' gender identity, especially in data sets that include a limited number of options? SECTION 4: Current Scholarship and Landscape of SOGIE Work 12. When working toward collecting SOGIE data, does your institution/agency have its own standard format for these types of questions? Or do you consult any external resources? 13. What, if anything, related to updating SOGIE data collection best practices do you see being discussed in your field of work? a. For example, are you aware of any recent scholarship or conversations around how to ask transgender status or pronouns? 14. Are there any gaps that you see in the current conversation/scholarship around SOGIE data collection? a. What do you see people get wrong most often about collecting SOGIE data? SECTION 5: Trainings 15. Are you aware of any training or technical assistance efforts that exist to support research teams in learning more about SOGIE data and equitable data collection practices? a. What content is covered during this training? b. Who facilitates the training? 16. Are there any gaps that you see in the current conversation/scholarship around SOGIE data collection practices? a. What structure of our work do you think would be most helpful? b. What materials or resources would you find most useful for your own work or to help others at your organization? SECTION 6: Miscellaneous 17. Are there any major institutional or political changes that you anticipate may impact how the field is able to embrace best practices for SOGIE data collection (i.e., elections, legislative changes, lawsuits, system restructuring, university funding, etc.)? 18. Is there anything else you'd like us to know that we did not cover today? SECTION 7: Conclusion + Referrals Thank you so much for speaking with us today. We're looking forward to continuing to learn more about best practices around SOGIE data collection. We have one final question for you: 19. Do you know of anyone else in your field that we should reach out to? If it's a personal connection, could you please share their contact information or otherwise introduce us? 51 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA APPENDIX C: COMMUNITY-BASED PRACTITIONERS AND ADVOCATES INTERVIEW PROTOCOL Interview teams should first ensure the project background + informed consent is covered, which can be found here: [link] Setting the Space Thank you again for speaking with us today. In this conversation, we will use the term "SOGIE" to refer to sexual orientation and/or gender identity and expression data. We want this interview to be only semi-structured and are eager to learn from your expertise in working with LGBTQ+ people, so please feel free to speak as much as you'd like and we'll probe with questions where needed. Warm-up question When you think about the way that data on sexual orientation and/or gender identity is currently collected in most research efforts, what comes to mind? SECTION 1: Respondent and Organization Background We look forward to learning more about your perspectives on how LGBTQ+ people can best be represented in research data collection efforts. We'd like to start with some basic background questions about you and your work. 1. Can you tell us about your organization and your role? a. What is your organization's mission/focus? b. What is your title? c. What are your primary responsibilities/what is your primary focus? d. In what capacity do you work with LGBTQ people? Or on LGBTQ issues? 2. Do you have any experience with thinking about SOGIE data or discussing opinions on how identities are asked about in research with your clients/with the LGBTQ people you work with or in your peer network? SECTION 2: LGBTQ People and SOGIE Data Considerations 3. LGBTQ+ people are by no means a monolith and there can be significant variation in language used for defining sexuality or gender from person to person. In your experience, what do you view as the most representative language to capture this information? i. [probe] 4. What do you view as the tradeoffs between asking people to report their sexual orientation on a survey or form by selecting from a list of options versus asking via an open-ended question? a. Do you have any thoughts on any possible tradeoffs between these survey methods and response rates/sample representativeness? 5. Do you think people feel more included/represented when they report their gender identity on a survey or form by selecting from a list of options, or asking via an open-ended question? a. What about for transgender status? Do you think it's better to ask this in every effort to collect SOGIE data, only if the project has a focus on trans people, etc.? i. If a trans person was taking a survey, do you think they'd prefer to identify as "trans woman" or "woman" and select 'yes I am transgender' in a follow up question? 52 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA SECTION 3: Current Scholarship and Landscape of SOGIE Work 6. What, if anything, related to updating SOGIE data collection best practices do think we should consider or explore in this project? a. For example, are you aware of any recent scholarship or conversations around how to ask transgender status or pronouns? 7. What materials or resources would you find most useful for your own work or to help others at your organization? In other words, how can we make this report as useful as possible? SECTION 4: Miscellaneous 8. Is there anything else you'd like us to know that we did not cover today? SECTION 5: Conclusion + Referrals Thank you so much for speaking with us today. We're looking forward to continuing to learn more about best practices around SOGIE data collection. We have one final question for you: 9. Do you know of anyone else in your field that we should reach out to? If it's a personal connection, could you please share their contact information or otherwise introduce us? 53 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA APPENDIX D: INTERNAL EXPERTS INTERVIEW PROTOCOL Setting the Space Thank you again for speaking with us today. In this conversation, we will use the term "SOGIE" to refer to sexual orientation and/or gender identity and expression data. We want this interview to be only semi-structured and are eager to learn from you as the expert, so please feel free to speak as much as you'd like and we'll probe with questions where needed. 1. In what capacity do you engage with SOGIE data? a. Is it something you ensure is included in any data collection effort, something that only pops up in dedicated projects, etc.? b. Use this question to have the interviewee(s) expand on their SOGIE data work or familiarity with the idea of collecting SOGIE data. 2. What is your opinion on (or experience with) having respondents report their sexual orientation by selecting from a list of options vs. asking via an open-ended question? a. What tradeoffs do you consider when designing how respondents report their sexual orientation (i.e., list of options vs. open-ended questions)? 3. What is your opinion on (or experience with) having respondents report their gender identity by selecting from a list of options vs. asking via an open-ended question? 4. What, if anything, related to updating SOGIE data collection best practices do you see being discussed in your field of work? a. For example, are you aware of any recent scholarship or conversations around how to ask transgender status or pronouns? 5. Are there any gaps that you see in the current conversation/scholarship around SOGIE data collection? a. What do you see people get wrong most often about collecting SOGIE data? 6. Are you aware of any training or technical assistance efforts that exist to support research teams in learning more about SOGIE data and equitable data collection practices? Or that should exist? 7. Is there anything else you'd like us to know that we did not cover today? 8. Do you know of anyone else in your field that we should reach out to? If it's a personal connection, could you please share their contact information or otherwise introduce us? 54 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA Notes, References, and Acknowledgments NOTES 1. Tony Enos, "8 Things You Should Know about Two Spirit People," ICT News, September 13, 2018, https://ictnews.org/archive/8- misconceptions-things-know-two-spirit-people. 2. Caroline Medina and Lindsay Mahowald, "Collecting Data about LGBTQI+ and Other Sexual and Gender-Diverse Communities," Center for American Progress, May 24, 2022, https://www.americanprogress.org/article/collecting-data-about-lgbtqi-and-other-sexual-and- gender-diverse-communities/. 3. The more than 560 anti-LGBTQIA+ bills introduced, as of early September 2023, include 535 bills at the state level and 32 at the federal level. "2023 Anti-Trans Bills Tracker," Trans Legislation Tracker, accessed September 12, 2023, https://translegislation.com. 4. Many transgender people are also part of gay communities and have played a key role in advancing gay rights. However, the history of transgender communities and transgender acceptance is in many ways distinct from the history of gay communities, and we try to make this distinction apparent as necessary in this section. 5. Dana Rosenfeld, "The AIDS Epidemic's Lasting Impact on Gay Men," British Academy, February 19, 2018, https://www.thebritishacademy. ac.uk/blog/aids-epidemic-lasting-impact-gay-men/. 6. Karen Tumulty, "Nancy Reagan's Real Role in the AIDS Crisis," The Atlantic, April 12, 2021, https://www.theatlantic.com/politics/ archive/2021/04/full-story-nancy-reagan-and-aids-crisis/618552/. 7. Nurith Aizenman, "How to Demand a Medical Breakthrough: Lessons from the AIDS Fight," NPR, February 9, 2019, https://www.npr.org/ sections/health-shots/2019/02/09/689924838/how-to-demand-a-medical-breakthrough-lessons-from-the-aids-fight. 8. We use the acronym "LGBTQ" instead of "LGBTQIA+" in this paragraph because, while intersex and asexual individuals were undoubtedly affected by the HIV/AIDS epidemic, neither group was the primary focus of the advocacy or organizing efforts in this period. 9. Sejal Singh, Laura E. Durso, and Aaron Tax, "The Trump Administration Is Rolling Back Data Collection on LGBT Older Adults," Center for American Progress, March 20, 2017, https://www.americanprogress.org/article/trump-administration-rolling-back-data-collection-lgbt- older-adults/. 10. Hansi Lo Wang, "Trump Officials 'Did Not Want' Census Survey to Ask about Sexual Orientation," NPR, September 20, 2018, https://www. npr.org/2018/09/20/649752485/trump-officials-did-not-want-census-survey-to-ask-about-sexual-orientation. 11. "Measuring Household Experiences during the Coronavirus Pandemic," US Census Bureau, accessed July 24, 2023, https://www.census. gov/householdpulsedata. 12. Thom File and Jason-Harold Lee, "Phase 3.2 of Census Bureau Survey Questions Now Include SOGI, Child Tax Credit, COVID Vaccination of Children," US Census Bureau, August 5, 2021, https://www.census.gov/library/stories/2021/08/household-pulse-survey-updates-sex- question-now-asks-sexual-orientation-and-gender-identity.html. 13. "Executive Order on Advancing Equality for Lesbian, Gay, Bisexual, Transgender, Queer, and Intersex Individuals," White House, June 15, 2022, https://www.whitehouse.gov/briefing-room/presidential-actions/2022/06/15/executive-order-on-advancing-equality-for-lesbian- gay-bisexual-transgender-queer-and-intersex-individuals/. 14. Naomi G. Goldberg, "Where Are We with LGBTQI+ Inclusive Health Data? One Year after President Biden's LGBTQI+ Equity EO," National Health Law Program, May 23, 2023, https://healthlaw.org/where-are-we-with-lgbtqi-inclusive-health-data-one-year-after-president- bidens-lgbtqi-equity-eo/. 15. "History: Title 13, U.S. Code," US Census Bureau, accessed July 24, 2023, https://www.census.gov/history/www/reference/privacy_ confidentiality/title_13_us_code.html. 16. Molly Hennessy-Fiske, "Texas Attorney General's Office Sought State Data on Transgender Texans," Washington Post, December 14, 2022, https://www.washingtonpost.com/nation/2022/12/14/texas-transgender-data-paxton/. 17. "2023 Anti-Trans Bills Tracker," Trans Legislation Tracker. 18. See also the ACLU dataset. "Mapping Attacks on LGBTQ Rights in U.S. State Legislatures," ACLU, last updated September 15, 2023, https://www.aclu.org/legislative-attacks-on-lgbtq-rights. 19. "Guidance on the Protection of Personal Identifiable Information," US Department of Labor, accessed September 21, 2023, http://www. dol.gov/general/ppii. 20. In our interview, Bobby Jefferson made an interesting point about changeover in governments that can put federal data and privacy at risk: "The experience at the country level is when a well-meaning international NGO [nongovernmental organization] performs a local survey data collection-women and men and gender and sexual orientation data are collected in the survey. To ensure 'data ownership,' the demographics and the survey results are shared with the local country's officials and ministers for actions. During the next political cycle in the local country the government changes. The policies change regarding the protections of the human rights of individuals' sexual orientation. Now the local country government has access to the demographic data related to gender and sexual orientation. This data can be used to target programs or marginalized groups. It is important to be mindful about 'do no harm' principles and ensuring that the promise of data protection and privacy are kept, no matter where the data are stored." 56 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA 21. "2022 U.S. Trans Survey," accessed July 24, 2023, https://www.ustranssurvey.org. 22. Ashley Kirzinger, Audrey Kearney, Alex Montero, Grace Sparks, Lindsey Dawson, and Mollyann Brodie, "KFF/The Washington Post Trans Survey," Kaiser Family Foundation, March 24, 2023, https://www.kff.org/other/poll-finding/kff-the-washington-post-trans-survey/. 23. The 2023 NHIS Field Representative Manual, a guidebook for NHIS interviewers, notes at the beginning of the gender identity questions that "the purpose of this series of questions is to serve as an experiment to develop and evaluate questions on gender identity." 24. In 2013, the CDC developed a set of SOGI questions to include in the BRFSS, an annual health survey conducted by all 50 states, the District of Columbia, Puerto Rico, and Guam. In 2014, the CDC gave states the option to add the SOGI questions to their questionnaires. In 2014, 16 states included the SOGI questions in the survey, and that number rose to 21 states in 2016 and 33 states in 2022; see https://www.cdc.gov/brfss/questionnaires/modules/state2022.htm. 25. There is a separate question for people who identified as male in the first question and another for people who identified as female. 26. Devon Price, "Bad Gender Measures and How to Avoid Them," Medium, March 14, 2023, https://devonprice.medium.com/bad-gender- measures-how-to-avoid-them-23b8f3a503a6. 27. Price, "Bad Gender Measures and How to Avoid Them." 28. "Non-Binary Birth Certificates and State IDs: Full Guide," US Birth Certificates, accessed September 21, 2023, https://www. usbirthcertificates.com/articles/gender-neutral-birth-certificates-states. 29. Associated Press, "Oklahoma Bans Nonbinary Gender Identities on Birth Certificates," NPR, April 27, 2022, https://www.npr. org/2022/04/27/1095040631/oklahoma-bans-nonbinary-birth-certificates. 30. The results from the 2022 US Transgender Survey have not yet been released, but the questions are reportedly very similar to those in the 2015 survey. 31. Also see the Do No Harm Guide: Applying Equity Awareness in Data Visualization (Schwabish and Feng 2021) for recommendations on ordering results when using race and ethnicity data. 32. Arnold Lau and Courtney Kennedy, "When Online Survey Respondents Only 'Select Some That Apply,'" Pew Research Center, May 9, 2019, https://www.pewresearch.org/methods/2019/05/09/when-online-survey-respondents-only-select-some-that-apply/. 33. Ashley Amaya, "Adapting How We Ask about the Gender of Our Survey Respondents," Decoded (blog), Pew Research Center, September 14, 2020, https://medium.com/pew-research-center-decoded/adapting-how-we-ask-about-the-gender-of-our-survey-respondents- 77b0cb7367c0. 34. The National LGBTQ+ Women's Community Survey, accessed July 24, 2023, https://lgbtqwomensurvey.org/. 35. The word femme is used to describe a lesbian whose appearance and behavior are traditionally seen as feminine. 36. See, for example, Nicholas Jones, Rachel Marks, Roberto Ramirez, and Merarys Ríos-Vargas, "2020 Census Illuminates Racial and Ethnic Composition of the Country," US Census Bureau, August 12, 2021, https://www.census.gov/library/stories/2021/08/improved-race- ethnicity-measures-reveal-united-states-population-much-more-multiracial.html. 37. Erin Blakemore, "From LGBT to LGBTQIA+: The Evolving Recognition of Identity," National Geographic, October 19, 2021, https://www. nationalgeographic.com/history/article/from-lgbt-to-lgbtqia-the-evolving-recognition-of-identity. 38. Juliette Rocheleau, "A Former Slur Is Reclaimed, and Listeners Have Mixed Feelings," NPR, August 21, 2019, https://www.npr.org/ sections/publiceditor/2019/08/21/752330316/a-former-slur-is-reclaimed-and-listeners-have-mixed-feelings. 39. In our interview, Kristen Schilt explained that regional and cultural terminology in the US can also be potentially challenging in survey design. She brought up the work of Ilan Meyer, who found in one of his surveys on sexual and gender identity that "same-gender- loving" was often selected as an identity by Black men who had sexual and romantic relationships with other men, but also by Southern heterosexual white women who imagined this response to mean having close platonic female friendships (Young and Meyer 2005). In a 2004 focus group study of 81 Black men living in New York state and Atlanta, Georgia, researchers found that 10 participants (12 percent of the sample) identified as "same-gender-loving" (Malebranche et al. 2004). 40. "Two-Spirit Community," Re:searching for LGBTQ2S+ Health, accessed September 21, 2023, https://lgbtqhealth.ca/community/two- spirit.php. The term Two-Spirit was coined in 1990 and is a translation of the Anishinaabemowin term niizh manidoowag. It acts as an umbrella term to capture gender identity, dress, and roles in different communities, such as winkt in Lakota, napêw iskwêwisêhot and iskwêw ka napêwayat in Cree, or aakíí'skassi in Siksika (Blackfoot). Michelle Filice, "Two-Spirit," Canadian Encyclopedia, lasted edited September 21, 2023, https://www.thecanadianencyclopedia.ca/en/article/two-spirit. See also Walter and colleagues (2006). 41. Furthermore, recent reporting from the Washington Post shows the debate within the American Sign Language community about disagreements on how to sign identities. Amanda Morris, Alexa Juliana Ard, and Anna Lefkowitz, "How to Sign 'Gay' or 'Queer'? American Sign Language Users Don't Agree," Washington Post, June 29, 2023, https://www.washingtonpost.com/wellness/2023/06/29/ american-sign-language-lgbtq-inclusive/. 42. Irinia Gonzalez, "Most People Don't Know What the Term 'Latinx' Means, a Recent Survey Suggests," Oprah Daily, September 13, 2021, https://www.oprahdaily.com/life/a28056593/latinx-meaning/. 43. Isabel Löfgren, "The 'Cuir' Turn," Eurozine, June 1, 2022, https://www.eurozine.com/the-cuir-turn/. 44. Kristin Moe, "Breaking the Chain: Healing Racial Trauma in the Body; An Interview with Resmaa Menakem," Medium, May 14, 2020, https://medium.com/@kristinmoe3/breaking-the-chain-healing-racial-trauma-in-the-body-af1bffd1d926. 57 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA 45. "Stress and Trauma Toolkit," American Psychiatric Association, accessed July 25, 2023, https://www.psychiatry.org:443/psychiatrists/ diversity/education/stress-and-trauma/lgbtq. 46. Rebecca Wong, "Guidelines to Incorporate Trauma-Informed Care Strategies in Qualitative Research," Urban Wire (blog), Urban Institute, August 30, 2021, https://www.urban.org/urban-wire/guidelines-incorporate-trauma-informed-care-strategies-qualitative-research. 47. Although we do not explore data collection and communication issues for intersex people in this report, we note here that classifying everyone as having "known sex at birth" misses this group, which some estimates suggest accounts for upwards of 2 percent of the US population. See "Intersex Population Figures," Intersex Human Rights Australia, last reviewed September 16, 2019, https://ihra.org. au/16601/intersex-numbers/. 48. With regard to selecting colors for users who may have certain color vision impairments, please see Do No Harm Guide: Centering Accessibility in Data Visualization (Schwabish, Popkin, and Feng 2022). 49. Lisa Charlotte Muth, "An Alternative to Pink and Blue: Colors for Gender Data," Datawrapper (blog), July 10, 2018, https://blog. datawrapper.de/gendercolor/. 50. "Flags of the LGBTIQ Community," Outright International, accessed July 25, 2023, https://outrightinternational.org/insights/flags-lgbtiq- community. 51. See, for example, "Our Story," Othering & Belonging Institute at the University of California, Berkeley, accessed September 21, 2023, https://belonging.berkeley.edu/our-story. 52. See, for example, the recent analysis by Graph Massara, "The Complexities and Nuances of Transgender Coverage," Columbia Journalism Review, May 25, 2023, https://www.cjr.org/analysis/trans-coverage-guide-suggestions.php. 53. See, for example, an interesting discussion on MetaFilter, "Is This Transphobic-Filter," Ask MetaFilter, April 17, 2018, https://ask.metafilter. com/321297/Is-this-transphobic-filter. 54. Jon Schwabish, "10 Ways to Mislead with Data Visualization," PolicyViz (blog), February 7, 2023, https://policyviz.com/2023/02/07/10- ways-to-mislead-with-data-visualization/. 58 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA REFERENCES Adam, Barry. 1995. The Rise of a Gay and Lesbian Movement. New York: Twayne Publishers. Arnos, Diane, Edward Kroll, Emma Jaromin, Hannah Daly, and Elsa Falkenburger. 2021. Tools and Resources for Project-Based Community Advisory Boards. Washington, DC: Urban Institute. Atrostic, Barbara K., and Charlene Kalenkoski. 2002. Item Response Rates: One Indicator of How Well We Measure Income. Washington, DC: Department of Education. Badgett, M. V. Lee, Amanda Baumle, and Steven Boutcher. 2018. Evidence from the Frontlines on Sexual Orientation and Gender Identity Discrimination. Amherst, MA: UMass Center for Employment Equity. Baker, Kellan, and Margaret Hughes. 2016. Sexual Orientation and Gender Identity Data Collection in the Behavioral Risk Factor Surveillance System. Washington, DC: Center for American Progress. Bates, Nancy, Yazmín A. García Trejo, and Monica Vines. 2019. "Are Sexual Minorities Hard-to-Survey? Insights from the 2020 Census Barriers, Attitudes, and Motivators Study (CBAMS) Survey." Journal of Official Statistics 35 (4): 709–29. https://doi.org/10.2478/jos-2019-0030. Berube, Allan. 1990. Coming Out Under Fire: The History of Gay Men and Women in World War Two. New York: Free Press. Bourgeault, Ivy Lynn. 2012. "Critical Issues in the Funding of Qualitative Research." Journal of Ethnographic and Qualitative Research 7 (1): 1–7. Bowen, Claire, and Joshua Snoke. 2023. Do No Harm Guide: Applying Equity Awareness in Data Privacy Methods. Washington, DC: Urban Institute. Boyd, Nan Alamilla. 2003. Wide-Open Town: A History of Queer San Francisco to 1965. Berkeley: University of California Press. Braun, Virginia, and Victoria Clarke. 2012. "Thematic Analysis." In APA Handbook of Research Methods in Psychology, vol. 2, edited by H. Cooper, P. M. Camic, D. L. Long, A. T. Panter, D. Rindskopf, and K. J. Sher, 57–71. Washington, DC: American Psychological Association. https://doi.org/10.1037/13620-004. Bullough, Vern L. 1998. "Alfred Kinsey and the Kinsey Report: Historical Overview and Lasting Contributions." Journal of Sex Research 35 (2): 127–31. https://doi.org/10.1080/00224499809551925. Cahill, Sean R., and Harvey J. Makadon. 2017. "If They Don't Count Us, We Don't Count: Trump Administration Rolls Back Sexual Orientation and Gender Identity Data Collection." LGBT Health 4 (3): 171–73. https://doi.org/10.1089/lgbt.2017.0073. Carey, Martha Ann, and Janice Swanson. 2003. "Funding for Qualitative Research." Qualitative Health Research 13 (6): 852–56. https://doi. org/10.1177/10497323030130060. Carter, Sierra, Frederick Gibbons, and Steven Beach. 2021. "Measuring the Biological Embedding of Racial Trauma among Black Americans Utilizing the RDoC Approach." Development and Psychopathology 33 (5): 1849–63. https://doi.org/10.1017/S0954579421001073. CDC (Centers for Disease Control and Prevention). 2022. "2021 BRFSS Questionnaire." Atlanta: CDC. ---. 2023. "2023 National Health Interview Survey (NHIS) Questionnaire." Atlanta: CDC. Cervini, Eric. 2020. The Deviant's War: The Homosexual vs. the United States of America. New York: Farrar, Straus and Giroux. Coles, Gregory. 2016. "The Exorcism of Language: Reclaimed Derogatory Terms and Their Limits." College English 78 (5): 424–46. Dahlhamer, James, Adena Galinsky, and Sarah Joestl. 2019. "Asking about Sexual Identity on the National Health Interview Survey: Does Mode Matter?" Journal of Official Statistics 35 (4): 807–33. https://doi.org/10.2478/jos-2019-0034. Drescher, Jack. 2015. "Out of DSM: Depathologizing Homosexuality." Behavioral Sciences 5 (4): 565–75. https://doi.org/10.3390/bs5040565. Edelman, Natalie L. 2023. "Trauma and Resilience Informed Research Principles and Practice: A Framework to Improve the Inclusion and Experience of Disadvantaged Populations in Health and Social Care Research." Journal of Health Services Research & Policy 28 (1): 66–75. https://doi.org/10.1177/13558196221124740. Estrada, Fernando, Isaiah Jay Jones, David Rivera, Amaranta Ramirez, and Alison Cerezo. 2022. "Trauma and Latinx Sexual- and Gender- Minority Immigrants in the U.S." Current Opinion in Psychology 47. https://doi.org/10.1016/j.copsyc.2022.101439. Faderman, Lillian. 1991. Odd Girls and Twilight Lovers: A History of Lesbian Life in Twentieth-Century America. New York: Columbia University Press. Federal Committee on Statistical Methodology. 2016. Current Measures of Sexual Orientation and Gender Identity in Federal Surveys. Washington, DC: Federal Committee on Statistical Methodology. Feinberg, Leslie. 1996. Transgender Warriors: Making History from Joan of Arc to Dennis Rodman. Boston: Beacon Press. Flores, Andrew R., Ilan H. Meyer, Lynn Langton, and Jody L. Herman. 2021. "Gender Identity Disparities in Criminal Victimization: National 59 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA Crime Victimization Survey, 2017–2018." American Journal of Public Health 111 (4): 726–29. https://doi.org/10.2105/AJPH.2020.306099. Fulton, Jenna, Jonathan Katz, Jasmine Luck, Amber Henderson, and Mary Davis. 2020. Is a Proxy Response Good Enough? Using Paired Cognitive Interviews to Assess the Accuracy of Proxy Responses. Research Report Series, Survey Methodology #2020-06. Washington, DC: US Census Bureau. GenIUSS (Gender Identity in US Surveillance) Group. 2014. Best Practices for Asking Questions to Identify Transgender and Other Gender Minority Respondents on Population-Based Surveys, edited by J. L. Herman. Los Angeles: Williams Institute. Grant, Jaime M., Lisa A. Mottet, Justin Tanis, Jack Harrison, Jody L. Herman, and Mara Keisling. 2011. Injustice at Every Turn: A Report of the National Transgender Discrimination Survey. Washington, DC: National Center for Transgender Equality and National Gay and Lesbian Task Force. Green, Amy E., Myeshia N. Price, and Sam H. Dorison. 2022. "Cumulative Minority Stress and Suicide Risk among LGBTQ Youth." American Journal of Community Psychology 69 (1–2): 157–68. https://doi.org/10.1002/ajcp.12553. Guyan, Kevin. 2022. Queer Data: Using Gender, Sex and Sexuality Data for Action. Bloomsbury Studies in Digital Cultures. London: Bloomsbury Academic. Harrison, Eona, Matthew Mizota, Hannah Daly, and Elsa Falkenburger. 2021. Community-Engaged Surveys: From Research Design to Analysis and Dissemination. Washington, DC: Urban Institute. Herman, Jody L., Andrew R. Flores, and Kathryn K. O'Neill. 2022. How Many Adults and Youth Identify as Transgender in the United States? Los Angeles: Williams Institute. Herman, Jody L., and Kathryn K. O'Neill. 2022. Well-Being among Transgender People during the COVID-19 Pandemic. Los Angeles: Williams Institute. Holmes, Nigel. 2023. Joyful Infographics: A Friendly, Human Approach to Data. Boca Raton, FL: CRC Press. Holzberg, Jessica, Renee Ellis, Robin Kaplan, Matt Virgile, and Jennifer Edgar. 2019. "Can They and Will They? Exploring Proxy Response of Sexual Orientation and Gender Identity in the Current Population Survey." Journal of Official Statistics 35 (4): 885–911. Jaffe, Anna E., David DiLillo, Lesa Hoffman, Michelle Haikalis, and Rita E. Dykstra. 2015. "Does It Hurt to Ask? A Meta-Analysis of Participant Reactions to Trauma Research." Clinical Psychology Review 40: 40–56. https://doi.org/10.1016/j.cpr.2015.05.004. James, S. E., J. L. Herman, S. Rankin, M. Keisling, L. Mottet, and M. Anafi. 2016. The Report of the 2015 U.S. Transgender Survey. Washington, DC: National Center for Transgender Equality. Jesdale, Bill M. 2021a. "Counting Gender Minority Populations in the Household Pulse Survey (The AGENID=2 Memo)." Providence, RI: National LGBT Cancer Network. ---. 2021b. "Sources of Missing Sexual Orientation and Gender Identity Data in the Behavioral Risk Factor Surveillance System." American Journal of Preventive Medicine 61 (2): 281–90. https://doi.org/10.1016/j.amepre.2021.02.027. Johnson, David K. 2004. The Lavender Scare: The Cold War Persecution of Gays and Lesbians in the Federal Government. Chicago: University of Chicago Press. Kay, Paul, and Chad K. McDaniel. 1978. "The Linguistic Significance of the Meanings of Basic Color Terms." Language 54 (3): 610–46. https://doi.org/10.1353/lan.1978.0035. Kinsey, Alfred C. 1948. Sexual Behavior in the Human Male. Bloomington: Indiana University Press. ---. 1953. Sexual Behavior in the Human Female. Bloomington: Indiana University Press. Levenson, J. S., S. L. Craig, and A. Austin. 2021. "Trauma-Informed and Affirmative Mental Health Practices with LGBTQ+ Clients." Psychological Services 20 (Suppl 1): 134–44. https://doi.org/10.1037/ser0000540. Levine, Martin P. 1979. "Gay Ghetto." Journal of Homosexuality 4 (4): 363–77. https://doi.org/10.1300/J082v04n04_04. Little, Roderick J. A., and Donald B. Rubin. 2019. Statistical Analysis with Missing Data. Hoboken, NJ: John Wiley & Sons. Malebranche, David J., John L. Peterson, Robert E. Fullilove, and Richard W. Stackhouse. 2004. "Race and Sexual Identity: Perceptions about Medical Culture and Healthcare among Black Men Who Have Sex with Men." Journal of the National Medical Association 96 (1): 97–107. Mathiowetz, N. A., and R. M. Groves. 1985. "The Effects of Respondent Rules on Health Survey Reports." American Journal of Public Health 75 (6): 639–44. https://doi.org/10.2105/ajph.75.6.639. Michaels, Stuart, Carolina Milesi, Michael Stern, Melissa Heim Viox, Heather Morrison, Paul Guerino, Christina N. Dragon, and Samuel C. Haffer. 2017. "Improving Measures of Sexual and Gender Identity in English and Spanish to Identify LGBT Older Adults in Surveys." LGBT Health 4 (6): 412–18. https://doi.org/10.1089/lgbt.2016.0168. Miller, Peter V. 1995. "They Said It Couldn't Be Done: The National Health and Social Life Survey." Public Opinion Quarterly 59 (3): 404–19. 60 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA https://doi.org/10.1086/269484. Morrison-Beedy, Dianne, Michael P. Carey, and Xin Tu. 2006. "Accuracy of Audio Computer-Assisted Self-Interviewing (ACASI) and Self- Administered Questionnaires for the Assessment of Sexual Behavior." AIDS and Behavior 10 (5): 541–52. https://doi.org/10.1007/s10461- 006-9081-y. NASEM (National Academies of Sciences, Engineering, and Medicine). 2020. Understanding the Well-Being of LGBTQI+ Populations. Washington, DC: National Academies Press. https://doi.org/10.17226/25877. ---. 2022. Measuring Sex, Gender Identity, and Sexual Orientation. Washington, DC: National Academies Press. https://doi. org/10.17226/26424. National Academy of Medicine. 2011. The Health of Lesbian, Gay, Bisexual, and Transgender People: Building a Foundation for Better Understanding. Washington, DC: National Academies Press. https://doi.org/10.17226/13128. National LGBTQ Task Force. 2017. "LGBTQ Census Advocacy, 1990–2017." Washington, DC: National LGBTQ Task Force. NIAID (National Institute of Allergy and Infectious Diseases). 2023. "Updated Recommendations for Collecting Data on Gender Identity, Sex Assigned at Birth, Intersex Status, and Sexual Orientation." Washington, DC: NIAID. O'Connor, Cliodhna, and Helene Joffe. 2020. "Intercoder Reliability in Qualitative Research: Debates and Practical Guidelines." International Journal of Qualitative Methods 19. https://doi.org/10.1177/1609406919899220. Office of Management and Budget. 2021. Interagency Technical Working Group on Sexual Orientation and Gender Identity Items in the Household Pulse Survey: Report and Recommendations. Washington, DC: Office of Management and Budget. Otero Class, B., M. Meyers, and M. Berger. 2022. "Distribution of Spanish Responses to Open Ended Sexual Orientation Questions." Paper presented at the 77th Annual American Association for Public Opinion Research Conference, Chicago, May 10–13. Peterson, Amanda L., Ansley M. Bender, Benjamin Sullivan, and Marc Karver. 2021. "Ambient Discrimination, Victimization, and Suicidality in a Non-Probability U.S. Sample of LGBTQ Adults." Archives of Sexual Behavior 50:1003–14. https://doi.org/10.1007/s10508-020-01888-4. Ridolfo, Heather, Kristen Miller, and Aaron Maitland. 2012. "Measuring Sexual Identity Using Survey Questionnaires: How Valid Are Our Measures?" Sexuality Research and Social Policy 9 (2): 113–24. https://doi.org/10.1007/s13178-011-0074-x. Rolstad, Sindre, John Adler, and Anna Rydén. 2011. "Response Burden and Questionnaire Length: Is Shorter Better? A Review and Meta-Analysis." Value in Health: The Journal of the International Society for Pharmacoeconomics and Outcomes Research 14 (8): 1101–8. https://doi.org/10.1016/j.jval.2011.06.003. Saleh, A., and K. Bista. 2017. "Examining Factors Impacting Online Survey Response Rates in Educational Research: Perceptions of Graduate Students." Journal of Multidisciplinary Evaluation 13 (29): 64–74. https://doi.org/10.56645/jmde.v13i29.487. Sankofa, John, Hannah Daly, and Elsa Falkenburger. 2021. Community Voice and Power Sharing Guidebook. Washington, DC: Urban Institute. Saperstein, Aliya, and Laurel Westbrook. 2021. "Categorical and Gradational: Alternative Survey Measures of Sex and Gender." European Journal of Politics and Gender 4 (1): 11–30. https://doi.org/10.1332/251510820X15995647280686. Schwabish, Jonathan, and Alice Feng. 2021. Do No Harm Guide: Applying Equity Awareness in Data Visualization. Washington, DC: Urban Institute. Schwabish, Jonathan, Sue Popkin, and Alice Feng. 2022. Do No Harm Guide: Centering Accessibility in Data Visualization. Washington, DC: Urban Institute. Sears, Brad, Christy Mallory, Andrew R. Flores, and Kerith J. Conron. 2021. LGBT People's Experiences of Workplace Discrimination and Harassment. Los Angeles: Williams Institute. Sell, Randall L., and Michelle L. Holliday. 2014. "Sexual Orientation Data Collection Policy in the United States: Public Health Malpractice." American Journal of Public Health 104 (6): 967–69. https://doi.org/10.2105/AJPH.2013.301814. Setlur, Vidya, and Maureen C. Stone. 2016. "A Linguistic Approach to Categorical Color Assignment for Data Visualization." IEEE Transactions on Visualization and Computer Graphics 22 (1): 698–707. https://doi.org/10.1109/TVCG.2015.2467471. SMART (Sexual Minority Assessment Research Team). 2009. Best Practices for Asking Questions about Sexual Orientation on Surveys. Los Angeles: Williams Institute. Smith, David M., and Gary J. Gates. 2001. Gay and Lesbian Families in the United States: Same-Sex Unmarried Partner Households. A Preliminary Analysis of 2000 United States Census Data. Washington, DC: Human Rights Campaign. The Trevor Project. 2022. 2022 National Survey on LGBTQ Youth Mental Health. West Hollywood, CA: The Trevor Project. UCLA Center for Health Policy Research. 2018. California Health Interview Survey Sexual Orientation and Gender Identity Working Group Summary and Final Recommendations. Los Angeles: UCLA Center for Health Policy Research. Virgile, Matthew, Jessica Holzberg, Renee Ellis, and Shelley Feuer. 2022. "Coding Open-Ended Responses to Survey Questions on Sexual Orientation." Paper presented at the Population Association of America Annual Meeting, Atlanta, Georgia, April 7. Voith, Laura A., Tyrone Hamler, Meredith W. Francis, Hyunjune Lee, and Amy Korsch-Williams. 2020. "Using a Trauma-Informed, Socially 61 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA Just Research Framework with Marginalized Populations: Practices and Barriers to Implementation." Social Work Research 44 (3): 169–81. https://doi.org/10.1093/swr/svaa013. ---. 2022. Trauma-Informed Socially Just Research Framework. Cleveland: Case Western Reserve University, Center on Trauma and Adversity. Walters, Karina L., Teresa Evans-Campbell, Jane M. Simoni, Theresa Ronquillo, and Rupaleem Bhuyan. 2006. "'My Spirit in My Heart': Identity Experiences and Challenges among American Indian Two-Spirit Women." Journal of Lesbian Studies 10 (1–2): 125–49. https://doi.org/10.1300/J155v10n01_07. Weinrich, James D., Peter J. Snyder, Richard C. Pillard, Igor Grant, Denise L. Jacobson, S. Renée Robinson, and J. Allen McCutchan. 1993. "A Factor Analysis of the Klein Sexual Orientation Grid in Two Disparate Samples." Archives of Sexual Behavior 22:157–68. https://doi. org/10.1007/BF01542364. Westbrook, Laurel, Jamie Budnick, and Aliya Saperstein. 2022. "Dangerous Data: Seeing Social Surveys through the Sexuality Prism." Sexualities 25 (5–6): 717–49. https://doi.org/10.1177/13634607209869. White House. 2023. "Recommendations on the Best Practices for the Collection of Sexual Orientation and Gender Identity Data on Federal Statistical Surveys." Washington, DC: White House. Young, Rebecca M., and Ilan H. Meyer. 2005. "The Trouble with 'MSM' and 'WSW': Erasure of the Sexual-Minority Person in Public Health Discourse." American Journal of Public Health 95 (7): 1144–49. https://doi.org/10.2105/AJPH.2004.046714. 62 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA ACKNOWLEDGMENTS The authors thank Celina Barrios-Millner, Madeline Brown, Owen Noble, and Sonia Torres Rodriguez for their thoughtful comments and feedback. We also thank the people we interviewed during this project for their invaluable insights and experiences: Nancy Bates, Marcus E. Berzofsky, TJ Billard, Lauren Bouton, Kelsey Campbell, D'Lane Compton, Renee Ellis, Avery Everhart, Shelley Feuer, Kevin Guyan, Jessica Holzberg, Bobby Jefferson, Bill M. Jesdale, Clair Kronk, Meghan Maury, Tey Meadow, Mikelyn Meyers, Stuart Michaels, Amy O'Hara, Kristen Schilt, Chloe Schwenke, Matt Virgile, and Erique Zhang. We are especially indebted to the members of our advisory board, who provided exceptional insight, reflections, and feedback: Kelsey Campbell, Stuart Michaels, and Chloe Schwenke. KELSEY CAMPBELL STUART MICHAELS CHLOE SCHWENKE Pronouns: they/she Pronouns: he/him/his Pronouns: she/her Founder/Data Scientist, Senior Research Scientist, President, Gayta Science: Data NORC at the University Center for Values in Science and Analytics of Chicago International Development with a LGBTQ+ Focus POSITIONALITY STATEMENT This project was led by Jonathan Schwabish (he/him), who directs the Urban Institute's Do No Harm project, in collaboration with Urban Institute researchers holding various LGBTQIA+ identities. Our work draws on our interdisciplinary backgrounds in topics related to economics, public health, public policy, social work, transformative justice, housing justice, and community-based participatory methods. We work at a relatively well-resourced and predominately white research organization based in Washington, DC. We acknowledge the extensive history of intentional and unintentional harm that research organizations have caused to structurally marginalized communities, including LGBTQIA+ people of all identities. 6 3 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA ABOUT THE PROJECT This report was funded by the Tableau Foundation. We are grateful to them and to all our funders, who make it possible for Urban to advance its mission. The views expressed are those of the authors and should not be attributed to the Urban Institute, its trustees, or its funders. Funders do not determine research findings or the insights and recommendations of Urban experts. Further information on the Urban Institute's funding principles is available at urban.org/fundingprinciples ABOUT THE URBAN INSTITUTE The Urban Institute is a nonprofit research organization that provides data and evidence to help advance upward mobility and equity. We are a trusted source for changemakers who seek to strengthen decisionmaking, create inclusive economic growth, and improve the well-being of families and communities. For more than 50 years, Urban has delivered facts that inspire solutions-and this remains our charge today. Copyright © December 2023. Urban Institute. Permission is granted for reproduction of this file, with attribution to the Urban Institute. Cover art by Shutterstock.com. 64 D O N O H A R M G U I D E C O L L E C T I N G , A N A LY Z I N G , A N D R E P O R T I N G G E N D E R A N D S E X U A L O R I E N TAT I O N D ATA 500 L'Enfant Plaza SW | Washington, DC 20024 | www.urban.org