Sociologists Andrew M. Penner and Aliya Saperstein have published yet another study that demonstrates how we categorize others in terms of race — not just racial stereotypes, but even racial identity — is dependent upon their other characteristics. In their most recent, published in the June 2013 issue of Gender & Society, the researchers found that individuals’ socioeconomic position and gender predicted whether their race would be recorded by interviewers as Black, white, or other:
Researchers study what shapes racial classification. In a novel study that looked back at how survey interviewers racially classify people over the course of their adult lives, sociologists Andrew Penner and Aliya Saperstein discovered that from one year to the next some people’s race appeared to change. This change occurred when the interviewer in one year wrote down one race, but in the next year the interviewer wrote down a different race. Penner and Saperstein call these changes in classification “racial fluidity,” and the researchers wanted to know what affected how a person’s race was perceived.
Though they found general factors that seemed to determine respondents’ racial classification, some were gender-specific:
The study found that men and women had similar levels of racial fluidity overall, and some factors, such as where the people lived, resulted in similar changes for both women and men. All else being equal, people were more likely to be classified as white and less likely to be classified as black if they lived in the suburbs, while the opposite was true for people living in the inner city.
However, other factors that triggered changes in racial classification differed by gender. In particular, poverty made men and women less likely to be classified as white, but the effect was stronger for men. Penner explains, “This is consistent with traditional gender roles that emphasize men’s responsibility as breadwinners, so that poverty changes how men are seen more than how women are seen.”
On the other hand, women, but not men, who have received welfare benefits are less likely to be seen as white and more likely to be seen as black, even though the U.S. Department of Health and Human Services estimated that in 2010 70% of welfare recipients are not black. Penner continues, “This result speaks to deeply entrenched stereotypes of ‘welfare queens’ originally made popular by Ronald Reagan in the 1980s. Knowing that a women is on welfare triggers a racial stereotype that isn’t triggered for men.”
Consistent with other widespread stereotypes, being a single parent affected a woman’s likelihood of being classified as white more than a man’s, while having been in prison affected whether men were classified as white but not women.
Some Additional Thoughts
This study, and their larger research project on racial fluidity, is a major contribution to the sociological understanding of race. Put bluntly, their work provides further evidence that race is socially constructed. It is not fixed (i.e., unchanging) nor universal. Rather, race is contextual, fluid, and, most importantly, an arbitrary way of classifying people.
It is also a commendable extension of intersectionality, wherein the researchers highlight the intersections of gender and social class in racial classifications. How we view others in terms of race is contingent upon their socioeconomic standing and gender. To study race separate from other important social characteristics is to paint an incomplete picture. I particularly appreciate their detailed discussion of doing intersectionality (i.e., applying an intersectional framework) in quantitative research — a practice that remains contentious among (and even antithetical to some) intersectionality scholars.
One question that lingers in my mind is the perceivers’ background. That is, do these dynamics play out the same way for all interviewers? Are they unique to interviewers of a particular background? Heck, let me just say what I really mean — is it just white interviewers whose racial classifications appear to be contingent on classed and gendered notions of whiteness and Blackness? The researchers accounted for various characteristics of the interviewers, including gender, level of education, and age — none of which effected racial categorization. But, interviewers’ self-identified race did.
In particular, respondents were significantly more likely to be coded as white if the interviewer was white (at least compared to Black interviewers); the reverse was true for coding respondents as Black. (Maybe these dynamics would reflect other race interviewers’ racial classifications if the survey was more racially inclusive than Black, white, and “other.”) Since this was merely background noise for the researchers’ primary analyses, they did not dig deeper into this. Why are white interviewers more likely to see respondents as white, and Black interviewers more likely to see respondents as Black?
I suppose from my own experience — notably, as someone who is racially ambiguous — there tends to be just as much racial inclusion as there is racial “Othering.” Some whites and Blacks have seen me as “one of their own,” while others see me as belonging to some other racial group. So, I am surprised by this bias of categorizing others as one’s own race. Certainly more research is needed to better understand these dynamics.
A point that seems lost in the academic press releases, commenting on “how others see your race,” is that those “others” are NLYS interviewers. Certainly, interviewers and researchers are mere humans; thus, it would be inappropriate to expect them to be totally free of society’s influences (including stereotypes and biases). I could make an issue of the supposed generalizability of the study — that we cannot assume trained interviewers’ racial classifications reflect those of laypeople. But, their other work makes this concern unnecessary.
One article about the study noted:
These changes were not random, as one might expect if the interviewers were just hurrying to finish up or if the data-entry clerks were making mistakes. The racial classifications changed systematically, in response to what had happened to the respondent since the previous interview.
Interviewer error is inevitable. But, this kind of systematic racial misclassification raises some cause for concern. These “mistakes,” to some unknown degree, biased research based on the NLYS data. In particular, it may have produced inaccurate estimates of racial differences on some outcome (e.g., health).
Fortunately, NLYS along with many other widely used surveys (and the US Census) have ceased interviewer-imposed race and ethnicity. Now, respondents themselves provide their self-identified race and ethnicity. While this eliminates interviewer bias, this approach is still imperfect for the fluidity and complexity of race. In another study by Saperstein and Penner, individuals’ racial self-identification depended upon prior incarceration. While this may appear to be evidence that respondents lie about their race (which is possible), it actually suggests that even how individuals see their own race depends upon their experiences and status. Arguably, these contingent self-identified racial categorizations may reflect how others see them.
In other ways, researchers and interviewers may continue to impose their perceptions on respondents. I have witnessed first hand the imputing of respondents’ gender. The rationale given against explicitly asking respondents their gender was to avoid offending them: “can’t you tell by my voice that I’m a man!” I am confident that most people were accurately classified by their self-identified gender. But, I worry about the unknowable number of people who were misclassified. I wondered why, when asking about personal opinions and intimate details of strangers’ lives, there was fear of offending them by asking about something so readily volunteered, constantly provided on official forms.
Although our openness as researchers introduces messiness and complexity, I feel we owe it to the people we study to willingly capture the messy, complicated details of their lives and identities. I fear we too often choose the convenience of easily contained categories and quantifiable experiences over the rich complexity and diversity of our social world. Though barely mentioned in the press for the article, Penner and Saperstein’s study reminds us just how complicated and messy that world is.