By Laurel Westbrook and Aliya Saperstein
In February, Facebook debuted an open field for gender classification, allowing users to record any identity terminology they like; in April, the White House announced its first “all-gender” restroom; and, over the last several years, colleges across the country – from the University of Vermont to George Washington University – have raised awareness among students, faculty and staff about using preferred gender pronouns (PGPs). These efforts, and many more, call into question two long-standing beliefs about sex and gender: 1) everyone can be classified into one of two categories and 2) it is easy to determine in which category a given person “belongs.”
Despite the recent progress in recognizing gender diversity, common sense, but incorrect, beliefs about sex and gender continue to shape how many institutions operate – including the design and implementation of social surveys. Survey practices are important to consider because they provide a window into who, or what, is deemed to be worth counting. We examined trends in the measurement of sex and gender in the United States by collecting and coding all publicly available documentation for four of the largest and longest-running national surveys: the American National Election Study (ANES), the Panel Study of Income Dynamics (PSID), the General Social Survey (GSS), and the 1979 National Longitudinal Survey of Youth (NLSY). We chose these long running and well-regarded surveys because they are frequently used in academic research, as sources of data to train students in statistical methods, and as models for new surveys.
One surprising finding is that when these surveys are conducted face-to-face or by telephone Americans are not asked to self-identify their sex or gender at all. Instead, the survey interviewer determines the category for the people they interview. The box for “male” or “female” gets checked off based on an unstated set of criteria that could include anything from their name, their voice, their dress or physical appearance, or their relationship to other people in their household. Occasionally, interviewers are instructed to ask a direct question, but only if the person’s sex or gender “is not obvious.” Even then, it is often presumed that asking someone this question will be awkward, likely because of the belief that a person’s sex or gender should be obvious.
Our analysis reveals an array of other survey practices that essentialize, dichotomize, and conflate sex and gender. For example, social scientists typically define gender as normative expectations for behavior, and sex as the placement of people into categories based on biological factors such as genitals and chromosomes, but the surveys we examined treat the two concepts as synonymous. They do this by: only ever asking about a person’s sex or gender not both, providing sex categories as answer options to questions about a person’s “gender” (i.e., responses are “male” and “female” rather than, e.g., “man” or “woman”), and by using the two terms interchangeably throughout their documents. The surveys we examined also do not allow anyone to change sex/gender over time. Instead, they treat all such changes as “errors” in data collection that must be “fixed,” literally erasing any potential transgender experiences from national survey data.
These survey practices are not simply problematic from the standpoint of respecting and representing diversity. They also limit the kinds of research that can be conducted and our ability to pinpoint the mechanisms that perpetuate gender inequality in the United States. For example, without conceptualizing sex and gender as distinct we cannot ask whether health risks are more closely related to chromosomal differences between males and females, to the experience of living as a man or a woman, a combination of the two, or neither. Similarly, by only measuring how survey interviewers perceive respondents, we miss the opportunity to understand how self-identification is related to variations in political attitudes, economic inequality, family formation, self-esteem, or any other social process we care about. Finally, by treating sex/gender as a single binary distinction – rather than spectrums of bodies, identities, and experiences – we will never be able to see when sex and gender and their effects are more a matter of degree than categorical differences in kind.
The belief that people are easily divided into two sexes (and two genders that “match” their sex) has been a fundamental organizing principle underlying how most Americans understand the world. In that sense, our national surveys simply reflect perceptions and practices in society. However, theorists of sex and gender have been questioning our binary language, assumption of natural differences, and fixed categories for several decades. This theoretical tradition, the increasingly evident diversity of lived experiences, and contrasting measurement practices for other characteristics – from race and ethnicity to religion – lead us to question why large-scale surveys and official documents continue to constrain sex and gender classification. The answer is anything but obvious.
Laurel Westbrook is associate professor of sociology at Grand Valley State University. Aliya Saperstein is assistant professor of sociology at Stanford University. Their co-authored article, “New Categories Are Not Enough: Rethinking the Measurement of Sex and Gender in Social Surveys,” is published in the August 2015 issue of Gender & Society. To view the article, click here.