It’s Time to Make Surveys More Inclusive

Thanks to intersectionality — a framework coined by legal scholar Kimberlé Crenshaw and used to more accurately consider a person’s experience based on their overlapping, complex identities — I understand how my identities have shaped my experiences, perspectives, and ultimately, who I am. I am a 19-year-old, first-generation, able-bodied, heterosexual, cisgender, middle-class Korean American woman. It would be inaccurate to generalize my identity to only one or two of these markers. However, many surveys do exactly that; I am often minimized to just an Asian woman.

Crenshaw’s original legal discussion of intersectionality focused on how people of intersectional identities are often inaccurately perceived as belonging to only one identity. For example, Black women are often perceived as having the same experiences as Black men or as white women by society and the legal system, when, in reality, the combination of being Black and female is its own unique experience. Contrary to its political opponents, intersectionality does not promote a “hierarchy of victimhood,” nor does it demonize people in positions of privilege, namely heterosexual, white, cisgender males. It rather attempts to help visibilize a more complex understanding of identity by underscoring the “multidimensionality” of lived experiences.

Demographic sections in surveys, however, rarely employ an intersectional framework. Instead, they commonly ask respondents to specify their racial/ethnic identities, gender identities, etc., using multiple-choice or short-answer formats, which frame identities as distinct, separate categories.

There remains great scholarly debate on how to best apply an intersectional framework to survey questions as intersectionality in empirical modes of data collection often complicates data analysis. The multiple social identities and structures that are core to intersectionality are often at odds with principles of scientific inquiry that tend to encourage investigation of one variable, or one facet of identity, at a time. This means that surveys are often designed to neatly categorize respondents’ identities for the purposes of simple and straightforward data analysis and interpretation at the expense of truly representing respondents’ identities. Current practices, for example, often list gender identity options as woman, man, and, sometimes, transgender. This implies that a respondent cannot be both transgender and a woman or transgender and a man.

Researchers who do try to incorporate intersectionality into quantitative modes of research often take an additive approach, viewing conventional race, gender identities, sexual orientation, social class, etc. as factors that add together to create a person’s whole identity. But this approach ignores the experiences that are a result of personal identities that don’t fall into neat categories, such as new and emerging gender identities. For example, it would be inaccurate to frame the experiences of Black, transgender women to simply be the sum of a Black person’s experiences and a White, transgender woman’s experiences.

These challenges in survey design are difficult to address by adopting a singular practice, but one potential solution is including non-uniform answers in responses. Clarifying identity definitions can also be helpful, particularly with gender identity, which many people may confuse with sex assignment at birth and/or sexual orientation. These measures may increase the amount of time needed to clean, analyze, and extrapolate conclusions from collected data, but even so, survey respondents should have the right to express their identities in the way they feel best represents them.

Through these surveys, respondents are being translated into data points. Researchers should strive to make these data points as accurate as possible. Intersectionality is more than an equity-promoting framework — it’s an accuracy-promoting framework, and scholars from all fields must gain greater understandings of methodology that embraces intersectionality.