Measuring femininity and masculinity in social surveys

Interest in measuring femininity and masculinity through surveys has increased along with more attention to inclusion and nonbinary gender identities. In a recent talk to the Clayman Institute Faculty Research Fellows titled “Gender Beyond Categories: Femininity, Masculinity, and Gender Expression,” Aliya Saperstein offered a critical review of gender scales deployed in national surveys to capture gender expressions across the United States population. Saperstein is a professor of sociology and the Benjamin Scott Crocker Professor in Human Biology.   

Saperstein highlighted that the genealogy of such scales runs through Stanford University, beginning in the 1930s with the work of Lewis Terman and Catherine Cox Miles. Though unwieldy by contemporary standards, their 456-item “masculinity-femininity test” set the precedent for many subsequent scales in psychology. Saperstein and colleagues, however, have dedicated their research to updating these measures, to better capture gender diversity and non-conforming gender expressions in their complexity and make them more accessible in a wider range of research settings.  

Terman was interested in IQ testing and quantifying human traits and used similar approaches to measuring gender. Saperstein emphasized how these earlier scales treated gender as a “bipolar measure,” which means that “they treated femininity and masculinity as opposite ends of the same scale.” The construction of the scales themselves revealed another unstated assumption, as “masculinity was assigned the positive or higher end of the scale, and femininity was assigned the negative or lower end of the scale.” 

In the 1970s, feminist psychologists, including Stanford’s Sandra Bem, began to critique the Terman and Miles approach. Instead of treating masculinity and femininity as opposites, Bem created scales that treated them as separate or independent sets of traits. In what became known as the Bem Sex Role Inventory (BSRI), respondents rated themselves on a battery of 60 traits – 20 feminine, 20 masculine, and 20 neutral. Some characteristics used for rating on the BSRI included “gullible” and “loves children” (for feminine) and “self-sufficient” and “makes decisions easily” (for masculine).  

Contrary to the Terman and Cox measurement scale, in which higher femininity came at the expense of higher masculinity ratings, an advance in the BRSI was that respondents could score high on both scales, low on both scales, or anywhere in between. Though the BSRI continues to be used today, Saperstein noted that it has come under criticism for not really measuring the extent to which people identify as feminine or masculine. Instead, it measures “cultural expectations or stereotypes about men and women, and how they should or shouldn’t behave.” 

In contrast, the approach Saperstein advocates asks people to report how they see themselves in terms of femininity or masculinity directly, on a scale ranging from “not at all” to “very,” along with how they think they are perceived by others. The rating scales are similar to how surveys ask people to describe themselves as “liberal” or “conservative,” without defining the concepts for them. “We tried labeling the top end of the scales ‘extremely,’” Saperstein said, “but then we found people avoided the top category because saying they were ‘extremely masculine’ or feminine must have seemed too extreme.”

Saperstein and colleagues use these gender scales in conjunction with a two-step approach to measuring categorical gender identity. The idea is to first ask a participant their sex assigned at birth, and next, to ask their gender identity with more than two response options. Not only does the two-step approach create more flexibility and diversity, compared to a single binary “female” and “male” distinction, it also does not confound gender and sex. The gender identity question also includes a category with an open-ended response, which allows researchers to be attentive to emergent and changing gender labels. Together, these measures provide researchers with a broader range of tools to study contemporary gender diversity and inequality.