Category Archives: Quantitative Methods

“Which of the following best describes you?”

I tried something different on the second to last day of class. We spent the last week on the larger ethical and philosophical questions of the work that sociologists do, and I told them about something that’s been troubling me for the last few years, and asked for their advice about what to do.

I asked them to talk about my use of the gender binary in survey research and in my class examples.

I’ve explored this issue in the past in terms of how we think about race/ethnicity as a concept used in class. On the one hand, using current bureaucratic or institutional understandings of race (such as those used by the census) are important ways of assuring the reliability and comparability of your data. On the other hand, using those examples reifies race as a biological, immutable, or relatively unchanging concept when in fact it is none of those things. On the third hand (no digit binary for me), what I do right now is present the socially constructed elements and problematize race as a measurement – and then spend the rest of the semester treating race in quantitative analyses as if the measurements we use are unproblematic (collapsing race categories with a recode, for example, or only looking at the larger racial groups because the n is too small to analyze the nuances of the smaller racial categories). An inelegant solution, to be sure.

Obviously official statistics are still replete with the gender binary, and it’s official statistics we have to deal with. Sometimes you’ll see a “prefer not to answer” but that creates missing data problems as well as validity issues (it may not really be that someone prefers not to answer, but that the valid response for that person isn’t there). “Other” (with or without fill in the blank) is even more problematic, as it further marginalizes those with a non-traditional gender identity.

This is what they had to say.

First off, avoiding talking about gender at all (or race) is not an option in sociology. Moreover, we agreed that variation in your population is a factor. Since we are conducting a survey of undergraduates, and gender is a variable, the issue is pressing. We talked adding a few more categories and then discussed what those categories might be. We looked at some examples:

In the end, they preferred phrasing the question as “the gender I identify with is…” with a blank for all respondents. This means that coding is bit more difficult — rather than having an easy 1 for female, 2 for male, you have more work to do; you may get 8 or 10 or even more different possibilities, just as you do when you ask an open-ended question about racial categories. However, they agreed that the extra work required was worth it.

The more difficult question I had for them, however, had to do with the examples I use in class. Do I use too many examples with a gender binary? There are just lots of useful — and sociologically relevant! — examples. One of the labs they complete looks the relationship between gender and income, using a t-test, which examines the difference between average male income in a sample and average female income, testing for a statistically significant difference. We also look at gender and political preference (a lab which undermines the notion of the “gender gap,” at least in a sample of Latino respondents). Dichotomous variables are easy to work with and don’t complicate things very much. I often use these examples because they need to be simple to run and understand — in essence, stereotypical. For the same reason, we might examine black and white samples and leave off the others race — theoretically problematic but allows for “cleaner” examples.

They decided that I should continue to use those examples because they are important ones, but also to find at least one example of a data set or research question that expanded the notion of gender. It’s not too confusing, I asked, to present gender in all its lovely socially constructed glory but then to back off with “But we’re going to do this anyway?” Nope. We can handle it, they said. They pointed out that discussing gender in this way is a new development, and still evolving and so I shouldn’t feel bad that I hadn’t discovered the solution to this problem! But I should show the class what the possibility would look like, even if institutionally we hadn’t come up with any alternatives.

Productive session.