2021-03-27

Demographics Done Wrong

The wave of pointlessly cruel discriminatory transphobic hate bills sweeping the nation wherever the party of Trump holds power (not to mention their similarly repugnant voter suppression efforts) is more than I feel up to addressing right now without descending into incoherent expletives, so I won't. Look to people like Chase Strangio or Katelyn Burns or Esther Wang, or organizations like the ACLU or the Trevor Project, or heck, even this satire by Lily Osler for more on those and why they're garbage.

Just reading about all that is bad enough; I'd rather not torment myself by going into enough depth to write about it. Party of personal freedom and small government—up until they decide you're trying to do something they think is icky or that threatens their power, and then suddenly they're all for intruding in everyone's bedrooms, bathrooms, wardrobes, smartphones, and voting booths. And meddling in corporate decisions, too, which the party is ostensibly firmly against, but apparently a publisher isn't even allowed to discontinue printing new copies of a few unprofitable books that are bad for the brand's image without the party trying to make a law about it, now.

Anyway. Let's instead cover something that speaks to the level of cluelessness we're often dealing with: demographics questions.

What prompted this was recently seeing the FDA's current MedWatch form (used to report medical product safety information) and noticing question 3 in particular. "Gender (check one)", the prompt reads. And the options are... "Male", "Female", "Intersex", "Transgender", and "Prefer not to disclose". That's... bad. Painfully bad. So bad I'm not sure it's even really an improvement over the traditional inadequate unnuanced choice between "male" or "female".

At which point, out of curiosity, I tried searching for any professional suggestions that might be out there on how to handle demographics questions about sex and gender properly. Many of them have their own issues, which I'll get into later, but they generally all agree on a few simple points (even if they fail to follow those very well themselves) that the MedWatch form utterly fails to address. Namely, when you're asking about sex or gender, it's important to clearly understand (and unambiguously convey) exactly what information you're actually asking for, why you're asking for it, and how you're going to use it.

So, what information is the MedWatch form trying to collect, and for what purpose? It seems to have no idea. Is it asking about designated sex at birth, or perhaps legally-recognized sex? Those can't be right, since "transgender" wouldn't be recorded at birth and isn't considered a legal sex designation anywhere I'm aware of. Is it asking about gender identity? "Man" and "woman" are generally preferred to "male" and "female" when discussing gender, "intersex" isn't a gender at all, and "transgender" doesn't make sense as a choice, either. You can have a transgender man or a transgender woman or a non-binary transgender person (or a non-binary person who doesn't identify as transgender, for that matter), but transgender isn't a gender on its own.

There simply isn't any single coherent question for which the provided options make any sense, and only allowing one box to be checked just makes it worse. The instructions on the FDA's site aren't any help, either, directing that the patient's "current gender" be selected without providing any clarity about what they mean by "gender", or what the significance of specifying "current" is. And in cases where the choices supplied aren't appropriate, all that's left is "prefer not to disclose". That's a useful option to have for those who genuinely don't want to answer, but it's hardly an adequate substitute for "none of the above" (and if the question is about sex designated at birth, "not recorded" could also be a valid possibility).

That's a common problem with surveys, too. Even if they deign to provide more than two possible responses, the third is all too often some form of "prefer not to answer". That's nice and all, but maybe I would like to answer, and they're not really making it possible to give a meaningful response with the answers I'm being allowed to use. Either I pick an inaccurate response or I pick an inaccurate non-response. Or I just decide it's not worth the bother and abandon the survey entirely, and is that really what you want if you're issuing a survey?

The "best practices" suggestions I found during a brief search weren't necessarily much better. Such guidelines typically recommend separating out questions about gender identity and sex designated at birth, which is a good start. Many of the sample questions I saw took a wrong turn when it came to the options, though. One template I saw repeated in multiple sources includes as options "Male", "Female", "Trans male/trans man", "Trans female/trans woman". Um. Yikes?

Does your form include a question about sexual orientation with options like "normal" and "gay"? Do you ask people whether they're a "woman" or a "black woman"? Or perhaps whether they're "male" or "Latino male"? Do you ask them to pick between "religious" and "religious Muslim"? I certainly hope not.

By the same token, having both "male" and "trans male" and both "female" and "trans female" as options makes no sense. The implication is insulting, and at best, you're likely to compromise your results. Either make sure your categories are properly distinct (such as "cisgender woman" and "transgender woman"), or follow your own guidelines and keep the two concepts in separate questions! Some form of "what is your gender" and "are you transgender", or "what sex were you assigned at birth" and "how do you describe your gender identity", perhaps.

And that's assuming you actually have a good reason to request this information in the first place! If you can't clearly explain what you're asking and what it's for, then why are you asking?

No comments:

Post a Comment