Transgender issues and images are increasingly present in popular media, literature, journalism, and the United States legal system. Few data sets, however, include information on the transgender community.

In a new working paper, I evaluate first-name and sex-coding changes in administrative files from the Social Security Administration (SSA) to model which individuals are likely to be transgender. I then match these likely-transgender individuals to their responses in the 2010 Census to create a preliminary dataset for studying the characteristics of likely-transgender individuals.

It is crucial to note that the number of people I identify as “likely-transgender” is not an estimate of the number of transgender individuals in the population, or even in the Social Security data. Gender identity is too complex to be captured only by name changes and sex-coding changes in a particular file. Further, some transgender individuals may not have a Social Security number, may not change their names with the SSA, or may have gender-neutral names. While this approach does not provide exact estimates of who is and is not transgender, it does allow researchers to learn about relative changes in trends over time and cross-sectional differences in characteristics for a subset of likely-transgender individuals.

Beginning in 1936 (the first year Social Security numbers were issued), I find claims for male-to-female and female-to-male name changes, many of which include sex-coding changes in the same directions. Figure 1 shows these “transgender-consistent” claims make up just under 0.02 percent of all claims and remained roughly proportional as the number of people with Social Security numbers grew.

Next, I match likely-transgender individuals to their records in the 2010 Census to explore residential patterns and responses to the question on sex. Figure 2 shows that likely-transgender individuals in the Social Security files are most concentrated in western and northeastern states that had legal protections against discrimination based on gender identity or expression. They are least concentrated in states without such protections.

I also find that likely-transgender individuals answer questions about sex differently than nontransgender individuals. Figure 3 shows likely-transgender respondents report both sexes, or leave the question blank, more often than nontransgender respondents. This is interesting because this question appears everywhere, from surveys to credit card applications to the forms we all fill out at the doctor’s office.

Together, these results help illuminate U.S. transgender history from the 1930s onward, and they help address questions about the characteristics and residential patterns of likely-transgender individuals. This work also demonstrates the potential for integrating data from existing sources, including administrative records, to produce new information products that provide deeper insights into the U.S. population. Despite several limitations to this approach, it is an important step toward learning more about transgender individuals’ experiences in the U.S.

The paper is available at:

Note: Please also see the Census Bureau statement on transgender data collection:

Census Bureau Statement on Transgender Data Collection

June 3, 2015 — Content changes in the decennial census and surveys are managed through an interagency process by the Office of Management and Budget (OMB). Decisions on new content are reached through careful consideration and public input and linked to a federal, legislative or programmatic need for the data. At this time the Census Bureau does not have plans to test questions about gender identity or sexual orientation for the 2020 Census or American Community Survey.

The Census Bureau, however, is a member of the Office of Management and Budget’s Federal Interagency Group on Improving Measurement of Lesbian, Gay, Bisexual Transgender (LGBT) populations in Federal Surveys, and the Census Bureau continues to work with the Office of Management and Budget and other federal agencies to examine the changing requirements and data recommended for program implementation. Additionally, a measure of sexual orientation was recently included in the National Health Interview Survey (NHIS) conducted by the Census Bureau and sponsored by the National Center for Health Statistics (NCHS).

The Census Bureau is committed to reflecting the information needs of our changing society. The Census Bureau is constantly examining the effectiveness of census and survey questions to collect accurate data on families and people and works with the National Advisory Committee on Racial, Ethnic and Other Populations and the Census Scientific Advisory Committee on recommendations for changing population needs.

Benjamin Cerf Harris, Ph.D. is an Economist for the Center for Administrative Records Research and Applications.