Journal of Official Statistics, Vol. 35, No. 4, 2019, pp. 835–858, http://dx.doi.org/10.

2478/JOS-2019-0035

Measuring Sexual Orientation and Gender Identity in

the National Crime Victimization Survey
Jennifer L. Truman1, Rachel E. Morgan1, Timothy Gilbert 2, and Preeti Vaghela2

The National Crime Victimization Survey (NCVS) collects information on nonfatal personal and
property crimes both reported and not reported to police. As part of the ongoing redesign efforts for
the NCVS, the Bureau of Justice Statistics (BJS) added sexual orientation and gender identity
(SOGI) questions to the survey’s demographic section in July 2016. The inclusion of these
measures will provide important national-level estimates of victimization among lesbian, gay,
bisexual, and transgender (LGBT) people and allow researchers to understand victimization risk
and access to victim services. This article includes a discussion of the sexual orientation and
gender identity measures that were added to the NCVS, and findings from the monitoring activities
conducted during the first six months of data collection. In addition, population counts by sexual
orientation and gender identity are estimated using July through December 2016 NCVS data.

Key words: National crime victimization survey; sexual orientation; sexual identity; gender
identity; victimization.

1. Introduction
There is a growing interest in understanding the national status of the lesbian, gay,
bisexual, and transgender (LGBT) population across key indicators of social, health, and
economic well-being (IOM 2011; SOGI Federal Working Group 2016a, 2016b, 2016c).
Historically, few national surveys have collected data on sexual orientation and gender
identity. Research on LGBT persons is developing in the health and social fields,
specifically in the US Federal Statistical System, with the addition of sexual orientation
measures to the National Health Interview Survey (NHIS), both sexual orientation and
gender identity measures on the Survey of Prison Inmates (SPI), Behavioral Risk Factor
Surveillance System (BRFSS), and Population Assessment of Tobacco and Health
(PATH), and the potential inclusion of sexual orientation and gender identity (SOGI)
measures to the Current Population Survey (CPS) (Dahlhamer et al. 2014; Ellis et al. 2017;
SOGI Federal Working Group 2016a, 2016b). However, sexual orientation and gender
identity have been identified in other research as correlates of victimization, and national-
level data are needed on the criminal victimization experiences of LGBT people.
1
Bureau of Justice Statistics, 810 7th Street, N.W., Washington, D.C. 20531, U.S.A. Emails:
[email protected], and [email protected]
2
U.S. Census Bureau, 4600 Silver Hill Road, Suitland, MD 20746, U.S.A. Emails:
[email protected], and [email protected]
Acknowledgments: The authors would like to thank Lynn Langton, Meagan Meuchel, and Meg Ruhnke for their
thoughtful review of this article. The findings and conclusions in this article are those of the authors and do not
necessarily represent the official position and policies of the Bureau of Justice Statistics (BJS) or the U.S. Census
Bureau.

q Statistics Sweden
836 Journal of Official Statistics

As one of two primary sources of information on the nature of criminal victimization

incidents in the United States, the National Crime Victimization Survey (NCVS) is a perfect
vehicle for collecting information on the victimization experiences of LGBT persons. (The
other primary source of information on criminal victimization in the United States is the
Federal Bureau of Investigation’s Uniform Crime Reporting program.) In 2016, after
substantial research and testing, measures of SOGI were added to the NCVS (Martinez et al.
2017). The estimates produced through the survey will provide researchers and policy
makers with information on the types of victimization experienced by the LGBT population
and their access to victim services. Measuring sexual orientation and gender identity on the
NCVS also provides data on other types of victimization experienced by LGBT people,
including identity theft and stalking; and their interactions with law enforcement using
data from the NCVS supplemental surveys (i.e., NCVS Identity Theft Supplement,
Supplemental Victimization Survey, and Police Public Contact Survey). Overall, the
inclusion of sexual orientation and gender identity in the NCVS provides more accurate and
detailed data that can be used to inform public policy regarding this vulnerable population.
LGBT persons are at risk of experiencing certain types of victimization at a
disproportionately higher rate or the same rate as their heterosexual peers. In particular, for
both women and men, LGBT persons report intimate partner violence and sexual violence
at rates equal to or higher than heterosexual women and men (Krebs et al. 2016; Walters
et al. 2013). Transgender persons experience intimate partner and sexual violence at higher
rates than those who do not identify as transgender (Krebs et al. 2016; National Coalition
of Anti-Violence Programs 2016). Sexual minority youth disproportionately experience
health risks, including violence; and are at risk of peer victimization related to their sexual
orientation and gender identity or expression (Collier et al. 2013; Kann et al. 2011).
Important changes in federal laws related to protecting LGBT survivors of violence have
occurred in recent years, including the Violence Against Women Reauthorization Act
(VAWA) of 2013. VAWA sought to improve care and access to victim services for LGBT
victims and explicitly prohibited discrimination of victims or survivors of violence based on
actual or perceived SOGI status, which works to ensure access to key services (VAWA 2013,
Pub. L. No. 113-4, Stat. 47). VAWA also identified LGBT victims as an underserved
community, which allowed organizations to receive more funding to focus on LGBT
domestic violence, dating violence, sexual assault, and stalking victims. In addition, The
Matthew Shepard and James Byrd Jr. Hate Crimes Prevention Act of 2009 (HCPA) included
language to allow for prosecution of hate crimes committed against persons based on their
actual or perceived sexual orientation or gender identity (HCPA, 18 U.S.C. § 249). Measuring
SOGI on the NCVS provides important data on victimization, access to victim services, and
experiences of hate crimes to support these laws.
As research continues to develop in this area, federal data are needed to expand the
knowledge on criminal victimization of the LGBT population. The addition of these
measures to the NCVS provides important national-level estimates of victimization among
LGBT people and allows researchers to understand victimization risk and access to victim
services. This article will address the following research questions: (1) How did
interviewers and respondents react to SOGI questions asked in the context of a crime
survey; (2) What was item nonresponse for the SOGI questions, and how did this vary by
demographic characteristics; and (3) How do SOGI population estimates collected on a
Truman et al.: Measuring SOGI in the NCVS 837

crime survey differ from other types of population surveys, specifically health-related
surveys? It includes a discussion of the sexual orientation and gender identity measures
that were added to the NCVS, and findings from the monitoring of data collection
activities. In addition, population counts by sexual orientation and gender identity are
estimated using July through December 2016 NCVS data.

2. Methodology

2.1. Timeline of Pretesting, Implementation, and Monitoring of SOGI Data Collection

In the fall of 2015, the Center for Survey Measurement (CSM) at the U.S. Census Bureau
conducted cognitive testing of the proposed sexual orientation and gender identity questions
for the NCVS (Figure 1). For more information about the cognitive interviews, please
review the report by the U.S. Census Bureau (Martinez et al. 2017). Data collection of these
questions began in July 2016. The U.S. Census Bureau conducted a debriefing questionnaire
(August 2016 to September 2016), focus groups (September 2016), and targeted interviews
(October 2016) with NCVS interviewers who collected SOGI data. Additional monitoring
of the SOGI data has been conducted from November 2016 to the present.

2.2. SOGI Data Collection Production Interviews in the 2016 NCVS

The inclusion of the SOGI questions in the NCVS began in July 2016. Before
administering these new items, interviewers completed a self-study training to introduce
them to the new items in the NCVS instrument and allow them to practice with the items
before their first interview. In addition to training, the NCVS computer-assisted personal
interviewing (CAPI) instrument includes a Frequently Asked Questions (FAQs) section
for interviewers if respondents have questions about the SOGI items, including why
the questions are important and relevant to a crime survey and definitions of each of the
concepts (see Appendix 1, Section 5). The SOGI questions were placed at the end of
the interview in a section with questions on disability, citizenship, veteran status, and
household income. Once the questions were in the field, the U.S. Census Bureau and BJS
began monitoring responses, refusal rates, and any information reported by interviewers.

Data Collection Additional Monitoring

Focus Groups

Debriefing Questionnaire Targeted Interviews

Cognitive Testing

Fall 2015 July 2016 August 2016 to September 2016 October 2016 November
September 2016 2016-Present

Fig. 1. Timeline of pretesting, implementation, and monitoring of SOGI data collection.

838 Journal of Official Statistics

2.3. Defining Sexual Orientation and Gender Identity

Sexual orientation is defined by three dimensions: sexual identification (identity), sexual
attraction, and sexual behavior. The measure used in the NCVS focuses on sexual identity.
Sexual identification (identity) refers to the way a person identifies with a given sexual
orientation (SMART 2009; SOGI Federal Working Group 2016a). The most commonly used
terms to describe sexual orientation include lesbian, gay, bisexual, and heterosexual/straight
(SOGI Federal Working Group 2016a). Sexual attraction refers to the relationship between a
person’s gender and the gender of the person(s) that they feel attracted to. Sexual behavior
refers to the relationship between a person’s gender and the gender of the individual(s) with
whom they engage in sexual activity. Some measures of sexual orientation include all three
dimensions of the concept and some only focus on sexual identity.
Like sexual orientation, gender identity is comprised of several dimensions, specifically
differences between the concepts of sex and gender. Sex is an individual’s biological
classification at birth as either male or female (IOM 2011; SOGI Federal Working Group
2016a). Gender is socially constructed and based on how the individual presents to
society, as either male or female, and encompasses the concepts of gender identity and
gender expression (SOGI Working Group 2016a). Gender identity refers to a person’s
internal sense of gender, while gender expression is the way one sees themselves or how
they present their gender to society (SOGI Working Group 2016b). An individual’s sex
and gender may be consistent (cisgender) or may be different (transgender).

2.4. Measures of Sexual Orientation and Gender Identity

2.4.1. Sexual Orientation

The sexual orientation questions that were administered in the NCVS came from the
National Health Interview Survey (NHIS), conducted by the National Center for Health
Statistics (NCHS 2018), and measure sexual identity. These questions are also consistent
with recommendations on measuring sexual orientation made by the Sexual Minority
Assessment Research Team (SMART 2009). The NHIS questions had been previously
tested using cognitive interviews and have performed well with persons age 18 or older.
The question used in the 2016 NCVS had also been tested using cognitive interviews
(Martinez et al. 2017), and is as follows:

Sexual orientation question

1. Which of the following best represents how you think of yourself?
W [Lesbian or] Gay
W Straight, that is, not [lesbian or] gay
W Bisexual
W Something else
W I don’t know the answer
W REFUSED
Truman et al.: Measuring SOGI in the NCVS 839

The phrase “lesbian or” is only displayed and read if the respondent had been assigned
female on the household roster, and answer categories displayed in all capital letters are
not read aloud.

2.4.2. Gender Identity

The gender identity questions that were administered in the NCVS come from
recommendations from the Gender Identity in U.S. Surveillance (GenIUSS) group and the
California Health Interview Survey (CHIS 2018; GenIUSS Group 2014). The current
recommendation for measuring gender identity is to use a two-step approach that asks
about assigned sex at birth and current gender identity (GenIUSS Group 2014). Two
questions are used to classify respondents as transgender or cisgender; and this method has
been successful in identifying transgender individuals compared to single questions
(Reisner et al. 2014; Tate et al. 2013; Xavier 2000; Xavier et al. 2007). In addition, using
the two-step approach allows those transgender individuals who identify their current
gender as male or female and not as transgender to identify as such, but still be classified
as transgender using the two-step approach. In the NCVS, persons were identified as
transgender if their responses to sex at birth and current gender identity were different, but
not if they said “none of these” on the gender identity question. These gender identity
questions had been previously tested using cognitive interviews and have performed well
with persons age 18 or older, and were cognitively tested again for the NCVS (Martinez
et al. 2017). The questions used in the 2016 NCVS are as follows:

Gender identity – assigned sex at birth question

1. What sex were you assigned at birth, on your original birth certificate?
W Male
W Female
W REFUSED
W DON’T KNOW
Gender identity – current gender identity question
2. Do you currently describe yourself as male, female, or transgender?
W Male
W Female
W Transgender
W None of these
W REFUSED
W DON’T KNOW

The respondent is asked the following confirmation question if they answer “male” and then
“female,” “transgender,” or “none of these” to the gender identity questions or if they answer
“female” and then “male,” “transgender,” or “none of these” to the gender identity questions.
840 Journal of Official Statistics

Gender identity – confirmation question

3. Just to confirm, you were assigned {FILL} at birth and now describe
yourself as {FILL}. Is that correct?
W Yes
W No
W REFUSED
W DON’T KNOW

The first fill in the confirmation question is populated with the answer given by the
respondent to the first gender identity question (assigned sex at birth), and the fill options
are “male” or “female.” The second fill is populated with the answer given by the
respondent to the second gender identity question (current gender identity). The fill
options are “male,” “female,” or “transgender.” Additionally, if the respondent answered
“None of these” to the second gender identity question, then the phrase “describe yourself
as {FILL}” is replaced with “do not describe yourself as male, female, or transgender.”
If the respondent answers “No” to the confirmation question, then the survey instrument
forces the interviewer to have the respondent change their answer to either the first or
second gender identity question. The confirmation question will be asked again until the
answer is “Yes” or the answers to the gender identity questions are not discordant.

2.5. Analytical Strategy

We used a mixed methods approach to answer our three research questions. To understand
interviewer and respondent reactions to SOGI questions asked in the context of a crime
survey (research question 1), we solicited feedback from interviewers in three ways: a
debriefing questionnaire, focus groups, and targeted interviews. The online debriefing
questionnaire was sent to all Census Bureau interviewers working on the NCVS, and they
responded to the questionnaire in August and September 2016. Approximately, 899
interviewers (77.3% response rate) reported completing at least one NCVS interview
between July 1, 2016 and the time of the debriefing questionnaire, and completed the full
debriefing questionnaire.
The debriefing questionnaire collected quantitative data about interviewer perceptions
of instrument problems, and respondents’ experience in and reactions to answering SOGI
questions. Interviewers were also allowed to elaborate about their experiences with
administrating the questions using a write-in response for comments. All write-in
comments for each question were analyzed using grounded theory (for more information,
see Charmaz 2006). Initial codes were first created from line by line coding by two
independent coders. Then the independent coders developed theoretical memos, which are
reports written by the researcher that document their thoughts about the individual codes
and how codes can be related to each other. From these memos, the coders were able to
generate themes. Then the coders independently coded each response for the agreed-upon
themes, met to review any inconsistencies, and recoded responses as necessary. For
Truman et al.: Measuring SOGI in the NCVS 841

each response that had a difference in coding, the coders discussed why they coded each
response with their specific theme until they both agreed on the same theme for
the response. The researchers then conducted theoretical sampling. Theoretical sampling
refers to a sample that is selected to gather more information or to develop a theme. This
theoretical sampling took the form of focus groups and targeted interviews with
interviewers, and was used to collect more data and help refine themes.
Six different focus groups, each one hour in length, with select interviewers from across the
country were conducted by conference call in August 2016, after a full month of data collection
with the new questions. Each focus group included two interviewers from one of the six U.S.
Census Bureau regional offices (ROs). Interviewers from each RO were eligible to participate
if they had conducted more than the average number of NCVS interviews within their region
(number of interviews ranged from 13 to 22 across region) in July 2016. At least one
interviewer selected per RO worked in a state that had legislative actions regarding transgender
issues. The topics discussed in the focus groups included the ease of administering the
questions, respondent reactions to the questions, and any other comments the interviewers
had about the questions. As the researchers facilitated the focus groups, they wrote notes
when interviewers mentioned new or relevant themes. After each focus group session, the
researchers typed up their notes about the focus groups. These notes were reviewed again by
the coders for themes. The coders also developed memos to refine the themes.
Targeted interviews were also conducted with interviewers who had collected specific
answers to the SOGI questions. Twenty interviews were conducted in October 2016 with
individual interviewers who had collected answers of lesbian, gay, bisexual, something
else, or I do not know the answer to the sexual orientation question; or transgender, none
of these, or differing male and female responses to the gender identity questions. The focus
of these interviews was understanding how the SOGI questions worked with LGBT
respondents. The researchers took notes during the interviews, created a detailed write-up
about the interview and the interview responses, and then these interview notes were
reviewed by the coders for themes.
We examined item nonresponse by respondent sociodemographic characteristics and
breakoffs for each item to understand item nonresponse for the SOGI questions in the
NCVS (research question 2). Logistic regression models were used to examine the
likelihood of nonresponse to the SOGI questions across various sociodemographic
characteristics (i.e., age, sex, race and Hispanic origin, educational attainment, household
income, interview language, and mode of interview). These models were also stratified
by Hispanic origin to determine if there were subgroup differences in likelihood of
nonresponse to the SOGI questions. Many of these demographic characteristics are used to
create post-stratification weights of NCVS data. Therefore, it was important to analyze
their effect on nonresponse to the SOGI questions, because these are variables known to
account for nonresponse in the data. We also stratified our results by race and Hispanic
origin because previous research has found that racial minorities have higher rates of
nonresponse to SOGI questions compared to non-Hispanic Whites (Jans et al. 2015; SOGI
Federal Working Group 2016b). Finally, population totals were estimated to compare
sexual orientation and gender identity estimates administered in the setting of a crime
survey to estimates from other types of population surveys, specifically health-related
surveys (research question 3).
842 Journal of Official Statistics

3. Results

3.1. Debriefing Questionnaire, Focus Groups, and Targeted Interviews

During the debriefing questionnaire, interviewers were asked to report any type of issue
experienced while administering the SOGI questions to respondents. Interviewers were
able to select all types of issues they encountered. The results from this debriefing
questionnaire are not meant to make inferences about all interviewers, but merely to report
the experiences of interviewers who responded to the debriefing questionnaire. Although
rare, some interviewers reported both experiencing no issues, as well as a general option
for other types of issues. The results indicate that interviewers felt the gender identity
questions were easier to read than the sexual orientation question. About half of
interviewers (52%) reported no issues with the gender identity questions, while a smaller
percentage (37%) reported no issues when administering the sexual orientation question
(Figure 2). About 50% of interviewers reported at least one respondent having a negative
reaction to the sexual orientation question itself, while 39% of interviewers reported a
negative reaction the gender identity questions. These findings should be interpreted with
caution as it is unknown whether the problems were with a single respondent or a larger
proportion of respondents interviewed by each interviewer. Interviewers may also
misremember how often respondents reported a concern or may only focus on the
concerns that were important or stood out to them. Additionally, a smaller percentage of
interviewers reported that respondents had difficulty answering the gender identity
questions (4%) than the sexual orientation question (9%).
A major theme that emerged from the qualitative analysis was that respondents
questioned the relevance of sexual orientation and gender identity to crime. This theme
included any mention that respondents questioned the reason for asking the questions,

60.0
51.5 52.3
50.0
Percentage of interviewers

38.8
40.0 37.0 Sexual orientation
Gender identity
30.0

20.0
13.9
9.7 9.0
10.0
3.7
0.1 0.3
0.0
Respondent negative No issues Other issues Respondent difficulty Instrument problems
reaction to the answering
questions

Fig. 2. Issues experienced by interviewers for sexual orientation and gender identity items.
Note: These are response options to the question: Have you experienced any of the following issues? Mark all that
apply. Percentages will not add to 100% because respondents were allowed to select more than one answer.
Interviewers reported these issues from at least one respondent. An example of other issues reported is
respondents questioning the relevancy of the question to crime. Instrument problems refer to issues with the
software an interviewer uses to administer the survey. N ¼ 899.
Source: U.S. Census Bureau internal debriefing questionnaire for NCVS interviewers.
Truman et al.: Measuring SOGI in the NCVS 843

including simply questioning the relevance of these questions to experiencing crime and
wanting to understand why the government was interested in this information. FAQs were
included in the CAPI instrument to aid interviewers in addressing this concern with
respondents. The FAQs specify that sexual orientation and gender identity are correlated
with victimization, and the questions are included to better understand this relationship.
In addition, the FAQs mention that discrimination against persons because of their sexual
orientation or gender identity is prohibited by federal hate crime statutes and the 2013
reauthorized VAWA; and the inclusion of these items allows researchers to better address
policy-relevant questions about hate crime victimization and victim services. Overall, the
inclusion of the FAQs resulted in positive reactions from the respondents and assisted
interviewers in being able to address any concerns. Upon hearing these questions, some
respondents also tried to answer the relevancy of these questions themselves. They
attributed the asking of these questions to current events involving transgender
individuals, to politics, or to the change in the cultural discourse around LGBT issues.
A second theme that emerged was some respondents felt the interviewers should have
been able to tell their sexual orientation or gender identity by just looking at them, and
expressed discomfort at the perceived suggestion of not being straight. These respondents
mentioned that the answer to these questions should be obvious without having to ask, and
that asking these questions indicated that they might not present as straight or cisgender.
Although the majority of respondents did not have difficulty understanding the question,
interviewers remarked that some respondents had negative or emphatic reactions to the
content of the question, which indicates that some respondents may have been
uncomfortable answering. In particular, some older adults had negative reactions to the
gender identity questions, and expressed discomfort at answering the sexual orientation
question or did not know how to answer because they were confused by the terms.
Additionally, some men (across all age ranges) answered emphatically that they were
straight and male. A smaller portion of respondents felt some level of discomfort about
answering as lesbian, gay, or bisexual, and hesitated slightly before answering. Some
interviewers themselves expressed concerns that the sexual orientation question would
impact response rates to future interviews, but these comments were only given by about
1% of all interviewers who answered the full debriefing questionnaire.
Interviewers did state that some respondents had positive reactions to the questions. The
sexual orientation question was easy for LGB respondents to understand and answer; some
respondents had already divulged their sexual orientation earlier in the interview, and
many just answered the question matter-of-factly. Many of these respondents had positive
reactions to the question, stating that they were “thrilled” and “appreciated that the
question was included” in the survey.
The focus groups also gathered information from interviewers about the something else
response category for the sexual orientation question. About 0.22% of all respondents age
16 or older selected this response category and the NCVS instrument did not collect any
additional information when a respondent chose this answer. In the targeted interviews,
interviewers clarified that respondents who chose something else tended to move on with
the interview without voluntarily providing additional information about why they chose
that response. However, some interviewers perceived that English-speaking respondents
who identified as something else may not have wanted to disclose their sexual orientation
844 Journal of Official Statistics

or felt that their sexual orientation was not captured by the categories presented (queer,
pansexual, asexual, etc.). During Spanish-speaking interviews, respondents who selected
an answer of something else generally needed the interviewer to repeat the question and
usually responded that they were “normal.” This indicates that respondents who answered
something else in Spanish might have experienced confusion about the terminology used.
Therefore, when some straight respondents were asked about their sexual orientation in
a Spanish-speaking interview, those not familiar with the term “straight” selected the
something else response option.
Interviewers also stated that there were concerns related to the none of these response
category in the current gender identity question. About 0.17% of all respondents age 16 or
older selected this response category and the NCVS instrument did not collect any
additional information when a respondent chose this answer. In the targeted interviews,
interviewers reported mixed reasons for the use of the none of these response category. In
some instances, interviewers sensed that this answer was the result of respondents being
generally offended and not wanting to answer the gender identity questions, rather than
describing themselves as something other than male, female, or transgender. The data from
the focus groups and targeted interviews with interviewers suggested that the none of these
response category may have been marked by interviewers in instances when respondents
did not really want to answer the question but did not outright refuse. This issue was
discovered early in data collection, and messages were sent to interviewers to reinforce the
proper use of the none of these response category. However, in other instances interviewers
did believe that respondents used this category because their gender identity was not
represented in the gender described question (i.e., bigender or genderqueer).
In summation, while the interviewers reported that some respondents had negative
reactions or sensitivity to the SOGI questions, overall it appeared that respondents were
able to understand and answer the questions. Nonetheless, because many interviewers
experienced at least some pushback from respondents, it is important to examine
nonresponse patterns for respondents.

3.2. Item Nonresponse and Breakoffs

Item nonresponse to the SOGI questions was low compared to other questions in the
NCVS. About 2.77% of respondents refused to answer the sexual orientation question.
About 0.41% of respondents answered don’t know to the question. These two nonresponse
categories combined with other missing responses (i.e., respondents that were eligible, or
in-universe, but had a missing response due to changes to variables used to define the
universe of the question, such as age or sex, in postdata collection processing) to the sexual
orientation question account for 3.51% of all respondents age 16 or older (Table 1). Less
than one percent (0.97%) of respondents refused to answer the gender identity questions.
Only 0.01% of respondents answered don’t know to the questions. These two nonresponse
categories combined with other missing responses to the gender identity questions account
for 1.33% of all respondents age 16 or older. Comparatively, about 25% of respondents
answered don’t know or refused to answer a question about household income.
These nonresponse rates varied by certain demographic characteristics. A similar
percentage of male and female respondents refused to answer the SOGI questions.
Truman et al.: Measuring SOGI in the NCVS 845

Table 1. Unweighted percent of nonresponse to sexual orientation and gender identity (combined responses to
sex at birth and current gender identity questions) items.

Sexual orientation Gender identity

Refusal Total Refusal Total
nonresponse nonresponsea nonresponse nonresponsea
Demographics Percent Percent Percent Percent

Total 2.77% 3.51% 0.97% 1.33%

Sexb
Male 2.71% 3.48% 0.95% 1.33%
Female 2.82% 3.54% 0.99% 1.33%
Race/Hispanic originb
Non-Hispanic white 2.77% 3.45% 1.00% 1.32%
Non-Hispanic black 2.92% 3.73% 1.12% 1.59%
Hispanic 2.53% 3.33% 0.71% 1.09%
Ageb
16 –17 1.97% 3.05% 0.54% 0.88%
18 –24 1.97% 2.84% 0.41% 0.73%
25 –34 2.63% 3.33% 0.68% 1.04%
35 –49 2.49% 3.20% 0.93% 1.35%
50 –64 3.22% 3.87% 1.17% 1.49%
65 or older 2.96% 3.80% 1.21% 1.54%
Region
Northeast 3.66% 4.82% 1.10% 1.76%
Midwest 2.76% 3.39% 0.91% 1.14%
South 2.44% 3.07% 0.95% 1.34%
West 2.78% 3.62% 1.03% 1.29%
Location of residence
Urban 3.50% 4.38% 0.97% 1.37%
Suburban 2.65% 3.41% 1.07% 1.42%
Rural 1.78% 2.25% 0.68% 0.96%
a
Total nonresponse includes all nonresponse, including refusals, don’t knows, and in-universe missing.
b
Based on data collected on the household roster for the sampled household, including the name, age, sex, race,
Hispanic origin, marital status, and education level of each person living in the household.
Source: Bureau of Justice Statistics, based on internal data from the U.S. Census Bureau, National Crime
Victimization Survey, July–December 2016.

Nonresponse to the sexual orientation question was similar across the race and Hispanic
origin groups, while a slightly lower percentage of Hispanics (0.71%) refused to answer
the gender identity questions compared to non-Hispanic Whites (1.00%) and non-Hispanic
Blacks (1.12%). Respondents age 16 to 24 generally had lower refusal rates to the sexual
orientation question than persons age 25 or older. About one percent of respondents age 50
to 64 and 65 or older refused to answer the gender identity questions, compared to only
0.54% of respondents age 16 to 17 and 0.41% of respondents age 18 to 24. This difference
in nonresponse by age supports the qualitative result that found that older respondents had
a harder time understanding or answering the sexual orientation and gender identity
questions. Differences in nonresponse were also observed by region for the sexual
orientation question. More respondents in the Northeast refused to answer the sexual
orientation question than respondents in the Midwest, South, and West. For gender
846 Journal of Official Statistics

identity, nonresponse was similar across region. A greater percentage of respondents

living in urban areas refused to answer the sexual orientation question than respondents
in suburban and rural areas. Less than one percent of respondents living in rural areas
(0.68%) refused to answer the gender identity questions, compared to 0.97% of
respondents in urban areas and 1.07% of respondents in suburban areas.
Breakoffs from the survey at the SOGI questions were also low. Breakoffs include
respondents that stopped participating in the survey at this question. Of all breakoffs,
0.24% happened at the sexual orientation question. About 0.10% of all breakoffs happened
at the sex at birth question, 0.04% of all breakoffs happened at the current gender identity
question, and no respondents broke off from the survey at the gender confirmation
question. For comparison, about 13% of all breakoffs occurred when respondents were
asked if their house was rented or owned, which is very early on in the instrument, and
about 0.29% of all breakoffs occurred at the hearing-based disability question, which is at
the end of the instrument with the SOGI items.
The analysis examined various predictors of nonresponse to the sexual orientation question
and found that total household income, age, mode of interview, educational attainment, and
race predicted nonresponse of sexual orientation (i.e., a refusal). The effect of every one
category increase in income was to decrease the odds of nonresponse to the sexual orientation
question by a factor of 0.97 holding sex, race and Hispanic origin, age, educational
attainment, interview language, and mode of interview constant (p , 0.001, Table 2). The
effect of every one-year increase in age was to increase the odds of nonresponse by a factor of
1.01 (p , 0.001). The effect of interviews conducted over the phone, compared to interviews
conducted in person, was to increase the odds of nonresponse by a factor of 1.34 (p , 0.001).
The effect of every one year increase in educational attainment was to increase the odds of
nonresponse by a factor of 1.01 (p , .05). The effect of being a race other than Hispanic, non-
Hispanic White, or non-Hispanic Black decreased the odds of nonresponse by a factor of 0.76
compared to being non-Hispanic White (p , .05).
The analysis also examined various predictors of nonresponse to the combined gender
identity questions (i.e., a refusal or I don’t know the answer) and found that nonresponse
is closely related to age and mode of interview. The effect of every one-year increase in
age was to increase the odds of nonresponse to the gender identity question by a factor of
1.01 (p , .001), holding sex, race and Hispanic origin, income, educational attainment,
interview language, and mode of interview constant (Table 2). The effect of interviews
conducted over the phone, compared to interviews conducted in person, was to increase the
odds of nonresponse by a factor of 1.77 (p , .001). The effect of mode of interview on
nonresponse to gender identity differed by Hispanic origin. Among Hispanics, the effect of
interviews conducted over the phone was to decrease the odds of nonresponse by a factor of
0.46 compared to interviews conducted in person (p , .05, Table 3). Among non-Hispanic
respondents, the effect of interviews conducted over the phone, compared to interviews
conducted in person, was to increase the odds of nonresponse by a factor of 2.10 (p , .001).

3.3. SOGI Population Estimates

Using NCVS data from July through December 2016, population estimates and percent
distribution of SOGI among all persons age 16 or older were estimated. The NCVS data
Truman et al.: Measuring SOGI in the NCVS 847

Table 2. Unweighted logistic regression of nonresponse to sexual orientation and gender identity questions by
sociodemographics.

Sexual orientation Gender identity

Respondent characteristic Logged odds Odds ratios Logged odds Odds ratios

Female 2 0.03 0.93 0.04 1.08

(0.03) (0.05)
Race (ref ¼ non-Hispanic
white)
Non-Hispanic black 2 0.01 0.99 2 0.11 0.80
(0.05) (0.08)
Hispanic 2 0.09 0.83 2 0.01 0.98
(0.05) (0.01)
Non-Hispanic other 2 0.14* 0.76* 2 0.13 0.77
(0.06) (0.10)
Age 0.01*** 1.01*** 0.01*** 1.01***
(0.01) (0.01)
Household income 2 0.03*** 0.97*** 2 0.02 0.98
(0.01) (0.02)
Years of education 0.08* 1.01* 0.01 1.00
(0.01) (0.01)
Language (ref ¼ English)
Spanish 0.09 1.19 0.29 1.77
(0.10) (0.22)
Other language 0.17 1.40 2 0.04 0.92
(0.26) (0.36)
Mode of interview
(ref ¼ face to face
interviews)
Telephone 0.14*** 1.34*** 0.29*** 1.77***
(0.03) (0.06)
Intercept 2 4.50*** 2 6.09***
(0.30) (0.47)
Note: Standard errors in parentheses.
***p,0.001, **p,0.01, *p,0.05.
Source: Bureau of Justice Statistics, based on internal data from the U.S. Census Bureau, National Crime
Victimization Survey, July-December 2016.

were weighted to the US population age 16 or older in order to produce estimates. Overall,
1.26% of all persons age 16 or older identified as gay or lesbian, 0.60% identified as
bisexual, and 0.11% identified as transgender (Figure 3; Appendix Tables 1 and 2). Among
transgender respondents, 51.7% identified as transgender on the current gender identity
question and 48.3% reported discordant sex at birth and current gender identity. These data
indicate that it is important to collect gender identity using the two-step method to provide an
accurate measure of the transgender population. It is possible that if only current gender
identity was collected, about half of transgender respondents may not be identified as such if
they only selected their currently identified gender (i.e., male or female) and not transgender.
Looking at demographic characteristics, the majority of persons age 16 or older
regardless of sex, age, race and Hispanic origin, or where they lived identified as straight.
The percent distributions of those persons who identified as gay or lesbian were similar for
848 Journal of Official Statistics

Table 3. Unweighted logistic regression of nonresponse to gender identity by sociodemographics by Hispanic

and non-Hispanic respondents.
Among Hispanic Among non-Hispanic
respondents respondents
Respondent characteristic Logged odds Odds ratios Logged odds Odds ratios

Female 0.05 1.11 0.03 1.07

(0.16) (0.05)
Age 0.01 1.00 0.02*** 1.02***
(0.01) (0.01)
Household income 2 0.05 0.95 2 0.02 0.98
(0.04) (0.02)
Years of education 2 0.02 0.98 0.01 1.01
(0.01) (0.01)
Language (ref ¼ English)
Spanish 0.45 2.47 4.94 . 999.9
(0.23) (165.0)
Other language 5.12 . 999.8 2 0.15 0.75
(478.80) (0.36)
Mode of Interview (ref ¼ face
to face interviews)
Telephone 2 0.39* 0.46* 0.37*** 2.10***
(0.18) (0.06)
Intercept 2 10.09 2 11.12
(478.80) (165.0)
Note: Standard errors in parentheses.
***p,0.001, **p,0.01, *p,0.05.
Source: Bureau of Justice Statistics, based on internal data from the U.S. Census Bureau, National Crime
Victimization Survey, July-December 2016.

males and females, 1.41% of males age 16 or older identified as gay compared with 1.13%
of females who identified as gay or lesbian (Figure 4). A higher percentage of females
identified as bisexual (0.85%) than compared to males (0.32%). Looking at distributions
by race and Hispanic origin, 1.38% of non-Hispanic Whites age 16 or older, 1.12% of

1.40
1.26
Percent of persons age 16 or older

1.20

1.00

0.80
0.60
0.60

0.40

0.20 0.11

0.00
Gay/lesbian Bisexual Transgender

Fig. 3. Percent of persons age 16 or older who identified as LGBT.

Source: Bureau of Justice Statistics, based on internal data from the U.S. Census Bureau, National Crime
Victimization Survey, July– December 2016.
Truman et al.: Measuring SOGI in the NCVS 849

4.00

3.50
Percent of persons age 16 or older

Bisexual Gay/lesbian
3.00

2.50 1.67

1.06
2.00
0.60 1.66
0.32 0.85
1.50 0.37
0.52 0.64 0.24

1.00
1.68 1.65
1.41 1.38 1.33 1.34
1.13 1.12 0.11
0.50 1.03
0.76
0.55
0.00
Male Female Non-Hispanic Non-Hispanic Hispanic 16–17 18–24 25–34 35–49 50–64 65 or older
White Black

Demographic characteristics

Fig. 4. Percent of persons age 16 or older who identified as LGB, by demographic characteristics.
Source: Bureau of Justice Statistics, based on internal data from the U.S. Census Bureau, National Crime
Victimization Survey, July–December 2016.

non-Hispanic Blacks, and 1.03% of Hispanics identified as gay or lesbian. About 0.60% of
non-Hispanic Whites, 0.52% of non-Hispanic Blacks, and 0.64% of Hispanics identified as
bisexual. Percent distributions of those who identified as gay or lesbian varied by age. A
higher percentage of persons ages 18 to 24 (1.68%) and ages 25 to 34 (1.65%) identified as
gay or lesbian compared to persons ages 16 to 17 (0.76%), 35 to 49 (1.33%), 50 to 64
(1.34%), and 65 or older (0.55%). A larger percentage of younger persons ages 16 to 17
(1.66%) and 18 to 24 (1.67%) identified as bisexual than persons ages 25 to 34 (1.06%), 35
to 49 (0.37%), 50 to 64 (0.24%), and 65 or older (0.11%).
The percent distributions varied among region and location of residence as well. About
1.38% of persons who live in the Northeast and 1.42% of persons who live in the West
identified as gay or lesbian, compared to 1.06% of persons who live in the Midwest and
1.23% of persons who live in the South (Figure 5). A higher percentage of persons who

3.00
Percent of persons age 16 or older

2.50
0.74
Bisexual Gay/lesbian
2.00
0.93
0.53
1.50 0.43
0.57 0.49
1.00 0.64
1.80
1.38 1.23 1.42
0.50 1.06 1.06
0.78

0.00 Northeast Midwest South West Urban Suburban Rural

Region and location of residence

Fig. 5. Percent of persons age 16 or older who identified as LGB, by region and location of residence.
Source: Bureau of Justice Statistics, based on internal data from the U.S. Census Bureau, National Crime
Victimization Survey, July–December 2016.
850 Journal of Official Statistics

live in the West (0.93%) identified as bisexual than persons who live in the Northeast
(0.53%), Midwest (0.57%), or South (0.43%). A larger percentage of persons who live in
urban areas (1.80%) identified as gay or lesbian compared to persons who live in suburban
areas (1.06%) or rural areas (0.78%). Similarly, a higher percentage of persons who live in
urban areas (0.74%) identified as bisexual compared to persons who live in suburban areas
(0.49%) or rural areas (0.64%).
The majority of persons age 16 or older regardless of sex, age, race and Hispanic origin,
or where they lived, identified as either male or female. The distribution of persons age 16
or older who identified as transgender was similar across demographic characteristics. The
distribution was similar among race and Hispanic origin, about 0.10% of all non-Hispanic
Whites, non-Hispanic Blacks, and Hispanics age 16 or older identified as transgender.
Among age groups, 0.22% of persons ages 16 to 17 and 0.21% of persons ages 18 to 24
identified as transgender compared to about 0.10% of persons age 25 or older.
Distributions were similar among regions and location of residence. About 0.18% of
persons who live in the West identified as transgender, compared to about 0.10% of
persons who live in the Northeast (0.08%), Midwest (0.07%), or South (0.11%). About
0.10% each of persons who live in urban (0.13%), suburban (0.11%), and rural (0.06%)
areas identified as transgender.
Based on 2015 NHIS data, 97.6% of persons identified as straight, 1.6% identified as
gay or lesbian, and 0.8% identified as bisexual (Figure 6) (NCHS 2015b). This compares
to 2016 NCVS data where 97.5% of all persons age 18 or older identified as straight,
1.3% identified as gay or lesbian, and 0.6% identified as bisexual. In general, given the
differences in methodology the two surveys found reasonably comparable population
estimates. It should also be noted that while the National Survey on Drug Use and Health
(NSDUH) and National Survey of Family Growth (NSFG) use slightly different question
wording, the NCVS estimate of all persons age 18 or older that identified as gay or lesbian
was also comparable to those findings (Medley et al. 2016; NCHS 2015a). However, the
percentages of adults who identified as bisexual in the NSDUH and NSFG appeared to be
higher than the estimate from the NCVS (Medley et al. 2016; NCHS 2015a).

3.00
Bisexual Gay/lesbian
Percent of persons age 18 or older

2.50

2.00 0.80

1.50 0.56

1.00
1.60
1.28
0.50

0.00
NCVS, 2016 NHIS, 2015

Fig. 6. Percent of adults who identified as LGB.

Source: Bureau of Justice Statistics, based on internal data from the U.S. Census Bureau, National Crime
Victimization Survey, July–December 2016; and National Center for Health Statistics, National Health Interview
Survey, 2015.
Truman et al.: Measuring SOGI in the NCVS 851

The NCVS was the first national household-based survey to include a gender identity
measure for all respondents ages 16 or older. Currently there are limited national,
population-based data collections that the NCVS can be compared to; however, some
researchers have estimated the transgender population using the CDC’s Behavioral Risk
Factor Surveillance System (BRFSS) (Flores et al. 2016; Herman et al. 2017). Although
the data collection, methodology, question wording, and sampled population vary by each
of these surveys, it is useful to compare estimates to assess data quality. The BRFSS uses
a one-step measure to identify transgender persons (Do you consider yourself to be
transgender?), while the NCVS uses the two-step measure as previously discussed. The
research using BRFSS found that 0.6% of U.S. adults identified as transgender. The NCVS
estimate for adults who identified as transgender was 0.1%, which is lower than the
estimate from the research using the BRFSS data.

4. Discussion
This article addressed three research questions about sexual orientation and gender
identity measurement in a large-scale population-based federal survey. A multi-method
approach was taken to address these questions including conducting a debriefing
questionnaire, focus groups, and targeted interviewers with interviewers, analyzing
nonresponse, breakoffs, and estimating the populations. Additionally, the sexual
orientation estimates generated from the NCVS data were compared to existing estimates
from another household-based survey. Overall, interviewers indicated that both they and
the respondents did not have difficulty understanding or comprehending the SOGI items.
Nonresponse and breakoffs were low for both sexual orientation and gender identity.
Similar LGB population totals using NCVS data were estimated compared to another
population-based federal survey.
Interviewers and respondents generally reacted positively to the addition of the SOGI
items. Including SOGI items in a crime survey did not present any major problems related
to the collection of these data or other demographic data or victimization data. Based on
the debriefing interviews, focus groups, and targeted interviews, interviewers indicated
that the most common issue respondents had was related to the relevance of these items on
a crime survey. However, interviewers were able to address these concerns by using the
FAQs that were provided in the CAPI instrument and explain their importance in better
understanding the relationship between these characteristics and experiences with criminal
victimization. Respondents were understanding and more positive about the questions
once this information from the FAQs was provided and explained. During the focus
groups, interviewers indicated that English-speaking respondents may have chosen the
something else response category to sexual orientation because they may not have wanted
to disclose their sexual orientation or felt that their sexual orientation was not captured
by the categories presented. Whereas, Spanish-speaking respondents may have selected
something else when there was confusion about the terminology used. This finding is
consistent with other research in the field (Stern et al. 2016). Given this, BJS and the
Census Bureau continue to monitor the something else response category and consider
approaches to improving the measurement of sexual orientation among non-English
speakers, in particular. In general, interviewers reported fewer issues with the gender
852 Journal of Official Statistics

identity questions than the sexual orientation question. Interviewers reported mixed
reasons for the use of the none of these response category in the second step question for
gender identity. In some instances, interviewers felt that respondents chose that response
category when they were offended and did not want to answer the question; whereas, in
other cases interviewers indicated that respondents used this response category because
their gender identity was not represented in the existing response categories. While the
interviewers reported that some respondents exhibited sensitivity and other issues toward
the questions, overall respondents were able to understand and answer the questions.
Item nonresponse and breakoffs were low for both the sexual orientation and gender
identity items. Older and more educated persons were more likely to refuse to answer the
sexual orientation question. Older persons were also more likely to refuse to answer the
gender identity questions. Respondents who were interviewed over the phone were more
likely to refuse both the sexual orientation and gender identity questions compared to those
who were interviewed in person.
A direct comparison could be made since the NCVS used the same measure as the
NHIS. In general, the NCVS found reasonably comparable population estimates given the
methodological differences between the two surveys. These are both household-based
surveys using the same question wording, and may be why the results are similar. The
NCVS estimates were also compared to other research estimating the transgender
population using the BRFSS. The NCVS estimates of the transgender population were less
than the population estimated using BRFSS data. These differences are likely a result of
differences in the two surveys, including question design and wording, survey context
(crime vs. health), and data collection methods.
As with any research, there were some limitations to the collection of SOGI items in the
NCVS. The NCVS did not collect data on the something else response category for sexual
orientation; therefore, it is difficult to fully assess whether persons identified as something
else or whether they did not want to respond to the item. In addition, the NCVS did not
collect information on the none of these response category for the second step question for
gender identity. Again, this makes it difficult to determine whether these respondents may
be identifying as gender non-conforming, or if they were choosing the response category
for another reason. The findings on respondent reactions and any negative reactions came
from the interviewers and not directly from the respondents. This is a limitation as the data
received was from the perspective of the interviewers whose experience or interpretation
of the situation could have been different than the respondents. Future research should
seek to address these issues.
SOGI questions can be successfully administered on a large-scale population-based
survey. Respondents are able to answer the questions with general ease and have
minimally negative reactions to the items. It is recommended that these items be added
to other surveys that may be considering including them. However, there are a few
considerations to keep in mind. Placement of the questions should be considered; in the
NCVS they were placed at the end of the instrument so that they would not have any effect
on crime rates. One should also consider why the questions are being added to the survey.
It was clear from this research that respondents questioned the relevance of these items on
a crime survey, but were willing to answer them once they understood their purpose.
Providing FAQs for interviewers to be able to reference when providing additional
Truman et al.: Measuring SOGI in the NCVS 853

clarification to respondents was effective, and therefore is a recommended approach

for other surveys. Finally, if gender identity is used, using the two-step method is
recommended. Based on data from the NCVS, about half of transgender persons age 16 or
older did not identify as transgender in the current gender identity question. Instead, they
reported different responses to their sex at birth and current gender identity.
The BJS and the Census Bureau continue to monitor performance, data quality, and field
interviewers’ concerns as the data are collected. Based on experiences from the NCVS
data collection of sexual orientation and gender identity, some key issues emerged as areas
to focus future research. These areas are also consistent with those identified in the SOGI
Federal Working Group’s Toward a Research Agenda for Measuring Sexual Orientation
and Gender Identity in Federal Surveys report (2016c). These areas include monitoring
response categories, specifically something else for sexual orientation and none of these
for gender identity. Current terminology is continually evolving. In addition, more
research is needed administering these items in languages other than English. Prior
research has indicated that terms related to sexual orientation and gender identity may
have different meaning or lack a direct translation in other languages. Finally, as data
collection in the NCVS continues, analyses on victimization experiences, including types
of victimization and incident characteristics, can be examined. The field of research
around sexual orientation and gender identity continues to grow. Despite concerns that
may exist related to adding these questions to surveys, the NCVS has shown that they can
be successfully administered in a household-based crime survey.
5. Appendix
854

Table A1. Weighted estimates of persons age 16 or older by sexual orientation.

Number of persons age 16 or older Percent of persons age 16 or older
Something Don’t Something Don’t
Demographics Gay/lesbian Straight Bisexual else know Gay/lesbian Straight Bisexual else know

Total 2,937,090 226,389,420 1,384,380 545,910 1,021,175 1.26% 97.46% 0.60% 0.24% 0.44%
Sexa
Male 1,580,010 109,099,895 357,990 206,580 534,920 1.41% 97.60% 0.32% 0.18% 0.48%
Female 1,357,080 117,289,525 1,026,390 339,330 486,255 1.13% 97.34% 0.85% 0.28% 0.40%
Race/Hispanic origina
Non-Hispanic White 2,060,130 145,625,345 902,920 387,775 581,885 1.38% 97.37% 0.60% 0.26% 0.39%
Non-Hispanic Black 315,070 27,541,315 146,235 47,995 99,515 1.12% 97.84% 0.52% 0.17% 0.35%
Hispanic/ 380,315 36,040,060 237,225 50,610 197,175 1.03% 97.66% 0.64% 0.14% 0.53%
Agea
16 – 17 54,385 6,879,450 117,780 34,960 29,455 ! 0.76% 96.68% 1.66% 0.49% 0.41%
18 – 24 446,655 25,406,200 444,175 115,320 151,420 1.68% 95.64% 1.67% 0.43% 0.57%
25 – 34 666,605 39,119,015 428,790 115,645 150,150 1.65% 96.64% 1.06% 0.29% 0.37%
35 – 49 756,280 55,478,330 207,295 111,045 220,765 1.33% 97.72% 0.37% 0.20% 0.39%
50 – 64 773,525 56,561,360 139,435 111,960 214,400 1.34% 97.86% 0.24% 0.19% 0.37%
65 or older 239,635 42,945,065 46,905 56,985 254,990 0.55% 98.63% 0.11% 0.13% 0.59%
Region
Northeast 547,815 38,541,700 211,480 100,715 264,070 1.38% 97.17% 0.53% 0.25% 0.67%
Midwest 539,575 49,692,250 291,160 129,155 208,905 1.06% 97.70% 0.57% 0.25% 0.41%
South 1,083,410 85,985,440 378,465 125,960 241,835 1.23% 97.92% 0.43% 0.14% 0.28%
West 766,290 52,170,030 503,275 190,085 306,365 1.42% 96.73% 0.93% 0.35% 0.57%
Location of residence
Urban 1,387,665 74,434,300 573,265 263,430 408,800 1.80% 96.58% 0.74% 0.34% 0.53%
Suburban 1,284,360 118,401,820 593,855 218,425 531,135 1.06% 97.83% 0.49% 0.18% 0.44%
Rural 265,065 33,553,300 217,260 64,055 81,245 0.78% 98.16% 0.64% 0.19% 0.24%
Note: Numbers rounded to nearest 5.
! Interpret with caution. Estimate based on 10 or fewer sample cases, or coefficient of variation is greater than 50%.
a
Based on data collected on the household roster for the sampled household, including the name, age, sex, race, Hispanic origin, marital status, and education level of each person
living in the household.
Journal of Official Statistics

Source: Bureau of Justice Statistics, based on internal data from the U.S. Census Bureau, National Crime Victimization Survey, July-December 2016.
Table A2. Weighted estimates of 16 or orlder by gender identity.
Number of persons age 16 or older Percent of persons age 16 or older
None of Don’t None of Don’t
Demographics Male Female Transgendera these know Male Female Transgendera these know

Total 113,568,135 122,754,055 265,290 438,135 210,430 47.87% 51.74% 0.11% 0.18% 0.09
Race/Hispanic originb
Non-Hispanic White 74,150,450 78,093,575 160,180 276,075 90,355 48.54% 51.12% 0.10% 0.18% 0.06
Non-HispanicBlack 12,741,565 15,913,260 35,610 39,415 39,640 44.29% 55.31% 0.12% 0.14% 0.14
Hispanic 17,826,680 19,634,245 40,440 77,155 44,395 47.38% 52.19% 0.11% 0.21% 0.12
Ageb
16 – 17 3,660,205 3,565,610 16,185 ! 18,555 ! 2,095 ! 50.40% 49.10% 0.22% ! 0.26% ! 0.03
18 – 24 13,366,840 13,567,930 56,545 69,160 24,130 ! 49.35% 50.09% 0.21% 0.26% 0.09
25 – 34 20,417,715 20,937,435 52,680 83,480 15,255 ! 49.19% 50.44% 0.13% 0.20% 0.04
Truman et al.: Measuring SOGI in the NCVS

35 – 49 28,266,940 29,319,320 71,785 79,630 60,850 48.91% 50.73% 0.12% 0.14% 0.11
50 – 64 28,371,520 30,612,310 37,565 107,570 54,590 47.94% 51.72% 0.06% 0.18% 0.09
65 or older 19,484,920 24,751,455 30,530 79,740 53,505 43.88% 55.75% 0.07% 0.18% 0.12
Region
Northeast 19,328,775 21,304,905 31,405 93,370 86,080 47.32% 52.16% 0.08% 0.23% 0.21
Midwest 25,098,465 26,851,795 38,715 87,620 36,660 48.16% 51.53% 0.07% 0.17% 0.07
South 42,505,110 46,537,240 96,730 129,615 61,815 47.58% 52.10% 0.11% 0.15% 0.07
West 26,635,780 28,060,115 98,445 127,535 25,875 48.47% 51.07% 0.18% 0.23% 0.05
Location of residence
Urban 37,866,975 41,179,230 106,905 156,960 61,220 47.71% 51.88% 0.13% 0.20% 0.08
Suburban 59,198,250 63,627,320 138,385 249,660 111,660 48.00% 51.59% 0.11% 0.20% 0.09
Rural 16,502,910 17,947,505 20,000 ! 31,515 37,550 47.78% 51.96% 0.06% ! 0.09% 0.11
Note: Numbers rounded to nearest 5.
! Interpret with caution. Estimate based on 10 or fewer sample cases, or coefficient of variation is greater than 50%.
a
Includes respondents with discordant responses to the sex at birth and currently described gender questions. That is, respondents may have identified using the transgender response
category or identified with a different gender than their assigned sex at birth.
b
Based on data collected on the household roster for the sampled household, including the name, age, race, Hispanic origin, marital status, and education level of each person living in
the household.
c
White and black race categories exclude persons of Hispanic or Latino origin.
Source: Bureau of Justice Statistics, based on internal data from the U.S. Census Bureau, National Crime Victimization Survey, July-December 2016.
855
856 Journal of Official Statistics

6. References
Charmaz, Kathy. 2006. Constructing Grounded Theory: A Practical Guide Through
Qualitative Analysis. Los Angeles: Sage. Available at: http://www.sxf.uevora.pt/
wp-content/uploads/2013/03/Charmaz_2006.pdf (accessed July 2019).
CHIS (California Health Interview Survey). 2018. CHIS Questionnaires. Los Angeles:
UCLA Center for Health Policy Research. Available at: http://healthpolicy.ucla.edu/
chis/design/Pages/questionnairesEnglish.aspx (accessed November 2018).
Collier, K.L., G. van Beusekom, H.M.W. Bos, and T.G.M. Sandfort. 2013. “Sexual
orientation and gender identity/expression related peer victimization in adolescence:
A systematic review of associated psychosocial and health outcomes.” Journal of Sex
Research 50: 299 –317. DOI: https://doi.org/10.1080/00224499.2012.750639.
Dahlhamer, J.M., A.M. Galinsky, S.S. Joestl, and B.W. Ward. 2014. “Sexual orientation
in the 2013 National Health Interview Survey: A quality assessment.” Vital Health
Statistics 2(169): 1 – 24. Available at: https://www.cdc.gov/nchs/data/series/sr_02/
sr02_169.pdf (accessed July 2019).
Ellis, R., M. Virgile, J. Holzberg, D.V. Nelson, J. Edgar, P. Phipps, and R. Kaplan. 2017.
Assessing the Feasibility of Asking Sexual Orientation and Gender Identity in the
Current Population Survey: Results from Cognitive Interviews. Washington, DC:
Center for Survey Measurement, U.S. Census Bureau and Office of Survey Methods
Research, Bureau of Labor Statistics. Available at: https://www.bls.gov/osmr/research-
papers/2017/html/st170210.htm (accessed July 2019).
Flores, A.R., J.L. Herman, G.J. Gates, and T.N. Brown. 2016. How Many Adults Identify
as Transgender in the United States? Los Angeles, CA: The Williams Institute.
Available at: https://williamsinstitute.law.ucla.edu/wp-content/uploads/How-Many-
Adults-Identify-as-Transgender-in-the-United-States.pdf (accessed July 2019).
GenIUSS Group. 2014. Best Practices for Asking Questions to Identify Transgender and
Other Gender Minority Respondents on Population-Based Surveys. Los Angeles, CA:
The Williams Institute. Available at: https://williamsinstitute.law.ucla.edu/wp-content/
uploads/geniuss-report-sep-2014.pdf (accessed July 2019).
Herman, J.L., A.R. Flores, T.N.T. Brown, B.D.M. Wilson, and K.J. Conron. 2017. Age of
Individuals who Identify as Transgender in the United States. Los Angeles, CA: The
Williams Institute. Available at: https://williamsinstitute.law.ucla.edu/wp-content/
uploads/TransAgeReport.pdf (accessed July 2019).
IOM (Institute of Medicine). 2011. The Health of Lesbian, Gay, Bisexual, and
Transgender People: Building a Foundation for Better Understanding. Washington,
DC: The National Academies Press. Available at: http://www.nationalacademies.org/
hmd/Reports/2011/The-Health-of-Lesbian-Gay-Bisexual-and-Transgender-People.
aspx (accessed March 2018).
Jans, M., J. Viana, D. Grant, S.D. Cochran, A.C. Lee, and N.A. Ponce. 2015. “Trends in
sexual orientation missing data over a decade of the California Health Interview
Survey.” American Journal of Public Health 105(5): e43 – e50. DOI: https://doi.org/
10.2105/AJPH.2014.302514.
Kann, L., E.O. Olsen, T. McManus, S. Kinchen, D. Chyen, W.A. Harris, and H. Wechsler.
2011. Sexual Identity, Sex of Sexual Contacts, and Health-Risk Behaviors Among
Truman et al.: Measuring SOGI in the NCVS 857

Students in Grades 9 – 12 –– Youth Risk Behavior Surveillance, Selected Sites, United

States, 2001 – 2009. Washington, DC: Centers for Disease Control and Prevention.
(MMWR 60(SS07):1 – 133). Available at: https://www.cdc.gov/mmwr/preview/
mmwrhtml/ss6007a1.htm (accessed July 2019).
Krebs, C., C. Lindquist, M. Berzofsky, B. Shook-Sa, K. Peterson, M. Planty, L. Langton,
and J. Stroop. 2016. Campus Climate Survey Validation Study Final Technical Report.
Washington, DC: Bureau of Justice Statistics. (NCJ 249545). Available at:
https://www.bjs.gov/content/pub/pdf/ccsvsftr.pdf (accessed July 2019).
Martinez, M., A. Henderson, J. Luck, and M.C. Davis. 2017. Cognitive Pretesting of the
National Crime Victimization Survey Supplemental Victimization Survey. Suitland, MD:
U.S. Census Bureau, Center for Survey Measurement. Available at: https://www.census.
gov/library/working-papers/2017/adrm/rsm2017-03.html (accessed November 2018).
Medley, G., R.N. Lipari, J. Bose, D.S. Cribb, L.A. Kroutil, and G. McHenry. 2016. Sexual
Orientation and Estimates of Adult Substance Use and Mental Health: Results from the
2015 National Survey on Drug Use and Health. Washington, DC: Substance Abuse and
Mental Health Services Administration. Available at: https://www.samhsa.gov/data/
sites/default/files/NSDUH-SexualOrientation-2015/NSDUH-SexualOrientation-2015/
NSDUH-SexualOrientation-2015.htm (accessed July 2019).
National Coalition of Anti-Violence Programs. 2016. Lesbian, Gay, Bisexual,
Transgender, Queer, and HIV-Affected Intimate Partner Violence in 2015. New
York, NY: Emily Waters. Available at: https://avp.org/wp-content/uploads/
2017/04/2015_ncavp_lgbtqipvreport.pdf (accessed March 2018).
NCHS (National Center for Health Statistics). 2015a. Sexual identity for men and women
18– 44 years of age. Washington DC: NCHS. Available at: https://www.cdc.gov/nchs/
nsfg/key_statistics/s.htm#sexualorientation (accessed March 2018).
NCHS (National Center for Health Statistics). 2015b. Sexual Orientation and Health
among U.S. Adults: National Health Interview Survey, 2015. Washington DC: NCHS.
Available at: https://www.cdc.gov/nchs/data/nhis/sexual_orientation/asi_2015_stweb-
site_tables.pdf (accessed March 2018).
NCHS (National Center for Health Statistics). 2018. NHIS Data, Questionnaires and
Related Documentation. Washington DC: NCHS. Available at: https://www.cdc.gov/
nchs/nhis/data-questionnaires-documentation.htm (accessed November 2018).
Reisner, S.L., K.J. Conron, L.A. Tardiff, S. Jarvi, A.R. Gordon, and S.B. Austin. 2014.
“Monitoring the health of transgender and other gender minority populations: validity
of natal sex and gender identity survey items in a U.S. national cohort of young adults.”
BMC Public Health 14(1): 1224. Available at: https://bmcpublichealth.biomedcen-
tral.com/articles/10.1186/1471-2458-14-1224 (accessed July 2019).
SMART, The Sexual Minority Assessment Research Team. 2009. Best Practices for
Asking Questions about Sexual Orientation on Surveys. Los Angeles, CA: The Williams
Institute. Available at: https://williamsinstitute.law.ucla.edu/wp-content/uploads/
SMART-FINAL-Nov-2009.pdf (accessed July 2019).
Stern, M.J., S. Michaels, C. Milesi, M.H. Viox, and H. Morrison. 2016. Task 1.32.a:
Develop and Test Sexual and Gender Minority Status (LGBT) Items Final Summary
Report. Washington, DC: Centers for Medicare and Medicaid Services, Office of
Minority Health. (HHSM-500-2014-00035I). Available at: https://www.cms.gov/
858 Journal of Official Statistics

About-CMS/Agency-Information/OMH/Downloads/Research-Report-sexual-gender-
minority-status-items.pdf (accessed July 2019).
SOGI Federal Working Group (Federal Interagency Working Group on Improving
Measurement of Sexual Orientation and Gender Identity in Federal Surveys). 2016a.
Current Measures of Sexual Orientation and Gender Identity in Federal Surveys.
Available at: https://s3.amazonaws.com/sitesusa/wp-content/uploads/sites/242/2014/
04/WorkingGroupPaper1_CurrentMeasures_08-16.pdf (accessed March 2018).
SOGI Federal Working Group (Federal Interagency Working Group on Improving
Measurement of Sexual Orientation and Gender Identity in Federal Surveys). 2016b.
Evaluations of Sexual Orientation and Gender Identity Survey Measures: What Have
We Learned? Available at: https://s3.amazonaws.com/sitesusa/wp-content/uploads/sites/
242/2014/04/Evaluations_of_SOGI_Questions_20160923.pdf (accessed March 2018).
SOGI Federal Working Group (Federal Interagency Working Group on Improving
Measurement of Sexual Orientation and Gender Identity in Federal Surveys). 2016c.
Toward a Research Agenda for Measuring Sexual Orientation and Gender Identity
in Federal Surveys: Findings, Recommendations, and Next Steps. Available at:
https://s3.amazonaws.com/sitesusa/wp-content/uploads/sites/242/2014/04/SOGI_Re-
search_Agenda_Final_Report_20161020.pdf (accessed March 2018).
Tate, C.C., J.N. Ledbetter, and C.P. Youssef. 2013. “A two-question method for assessing
gender categories in the social and medical sciences.” Journal of Sex Research 50(8):
767– 776. DOI: https://doi.org/10.1080/00224499.2012.690110.
The Matthew Shepard and James Byrd Jr. Hate Crimes Prevention Act of 2009 (HCPA),
18 U.S.C. § 249 (2009). Available at: https://www.congress.gov/111/bills/s909/BILLS-
111s909is.pdf (accessed July 2019).
VAWA, Violence Against Women Reauthorization Act of 2013, Pub. L. No. 113-4, Stat.
47 (2013). Available at: https://www.congress.gov/113/plaws/publ4/PLAW-
113publ4.pdf (accessed July 2019).
Walters, M.L., J. Chen, and M.J. Breiding. 2013. The National Intimate Partner and
Sexual Violence Survey (NISVS): 2010 Findings on Victimization by Sexual
Orientation. Atlanta, GA: National Center for Injury Prevention and Control, Centers
for Disease Control and Prevention. Available at: https://www.cdc.gov/violencepre-
vention/pdf/nisvs_sofindings.pdf (accessed July 2019).
Xavier, J.M. 2000. The Washington Transgender Needs Assessment Survey: Final report
for phase two. Washington, DC: Administration for HIV/AIDS of the District of
Columbia. DOI: https://doi.org/10.13140/RG.2.2.23988.65924.
Xavier, J., J.A. Honnold, and J. Bradford. 2007. The Health, Health-related Needs, and
Lifecourse Experiences of Transgender Virginians. Richmond, VA: Virginia Department
of Health, Division of Disease Prevention. Available at: http://www.vdh.virginia.gov/
content/uploads/sites/10/2016/01/THISFINALREPORTVol1.pdf (accessed July 2019).

Received March 2018

Revised April 2019
Accepted August 2019