Can a questionnaire be feminist?

March 28, 2014

Sari van Anders

So let’s be honest: obviously I’m not going to ask that question if the answer is ‘no’ because that would be pretty boring and random and not very useful. Like, how about we have a post about whether a cactus is feminist (answer = huh?); that would be really useful. So, yes. A questionnaire can be feminist and insert joke here about how now this post is over not really! This post is like a longer version of ‘yes’ divided into different sections: a more abstract/theoretical (epistemological) part, a pragmatic/applied (epistemic) part, a summary part, and some conclusions too. In case you want to skip ahead.

This is the abstracty part.

Some of you might be like: why couldn’t a questionnaire be feminist? Why is this even a question? There are longstanding disagreements in feminist scholarship about whether quantitative work can be feminist. There are some scholars who see any scientific work as inherently non/anti/un-feminist, and scientific work is almost synonymous with quantification (which is largely what questionnaires do). Plus, questionnaires are impersonal, like a lot of science, and meant to distance the scientist/scholar from the subject/participant. Here at Gap Junction Science, we take it for granted that science can be feminist so we’re not going to delve too much into why some folks disagree with that.

I think that there is a useful and important starting point in thinking about whether a questionnaire can be feminist: I would argue that there is no such thing as feminist methods, per se, but only feminist uses of methods (I’m far from the first to say this). WHAT DOES THAT EVEN MEAN?! Well, imagine a method that most people would agree is feminist: the community-based piece of scholarship/activism. This is a way to do scholarship that meets the activism goals of a progressive community (e.g., poverty reduction, sexual health education, etc.). But, what if your community is focused on the best ways to remove women from STEM because they’re the Against-Women-in-Science Community? (And, yes, that is a very contrived example.) Doing a community project with this community to get women out of STEM fields would not be feminist. I hope we agree and you’re not just trying to think of secret ways around this like: WHAT IF ALL THE SCIENTISTS ARE REALLY MONSTERS AND THEN YOU’D WANT WOMEN OUT OF THERE, WOULDN’T YOU?! because: really!! Though that is a great premise for a movie that I now own the rights to SO STOP STEALING MY IDEAS, STEALERS!! Right, so the community-based piece of scholarship/activism may or may not fall in line with feminist goals in some ways (feminists like to think about getting more women into STEM not fewer), even if it falls in line with others (feminists like the idea of engaging with communities).

Methods are neutral, kind of like facts. But neither methods or facts are neutral in an applied way. HUH!? Well, methods – like facts – always have a context or a use. You can’t really have a method without an application (just like you can’t have a fact without an interpretation). An easy way to make sense of this is that it’s the context that makes the fact feminist; and it’s the use that makes the method feminist. An analogy might be digitalis: equally a plant, a very pretty addition to your garden, heart medication, or a good way to kill someone in a murder mystery. Digitalis isn’t intrinsically evil, the use of it in murder is (unless, of course, you’re murdering the monsters who are anti-women-in-STEM in the horror movie I’m now producing).

This is the in-practice part.

The above is kind of abstract, so let’s get pragmatic here. How would you make this questionnaire feminist? Well, feminism involves paying attention to gender inequities and intersecting identities. So, are you asking people to report on their gender or sex? Obviously, you can’t pay attention to gender inequities without gathering data on gender. Wow. That was very insightful; I’m sure you’re impressed. But, really, many people study only males/men, or don’t identify the gender or sex of their participants and, last I checked, you can’t actually analyze data you don’t have. Please let me know if that has changed because that would be awesome.

But if you stop there (including and measuring gender), you’re not really measuring much. For example, are you measuring gender? Or sex? What aspect of either? Both are pretty complex and nuanced. There’s gender role, genitals, identity, upbringing, gender expectations, appearance, hormones &etc. (is there anything better than ‘&etc.’? NO!) (YES: Milkshakes!) So if you find that ‘gender’ matters, is it gender or is it sex? Is it both? And what aspect of either is the aspect that matters? It may be useful to collect other data to help you understand why the difference you found was there. For example, if you find that men respond more quickly on a task than women, are you measuring response time or are you measuring interest in impressing the experimenter? Maybe you’re measuring baseline stress levels (which might differ by gender) that impact response time. Maybe you’re measuring processing time. How will you know if all you have is someone’s self-report of gender/sex? I use ‘gender/sex’ because, like almost all scientists who study human beings, I don’t know whether any differences I find are due to gender or sex, socialization or evolution, and all I’ve measured is someone’s self-report – i.e., gender/sex. Maybe gender/sex is useful when you don’t know whether you’re dealing with gender or sex (which are most cases of human bioscience and many other forms of research too).

A key realization is that gender/sex is not a mechanism – it’s a descriptive factor. If you find differences in some variable by gender/sex, you haven’t found an explanation or a mechanism, you’ve only found a difference. The mechanism is something else. Even if you find a biological difference in brain structure between women and men, that could be due to social experiences and you won’t know the mechanism without measuring possibilities and excluding alternatives. Why am I going on about mechanism? Because we scientists care a lot about mechanism! But correlation (women have this neural pattern, men have that one) is not causation (neural patterns must be hardwired so the differences must be genetic! Um, no.).

Another thing to consider is how you measure gender or sex. Do you give people one of two boxes to check? Is the first box always male? There is some fascinating research showing that majority groups are always presented first. Of course, women aren’t a statistical minority, but they are a minoritized group in terms of power and studies are clear that they’re presented as secondary/lesser. So, thinking about your ordering can be feminist: what does it mean if all scientists always put men first? Oh. What that means is actually pretty clear when I put it like that.

Moreover, are you presenting only two options for gender/sex? For sure, the vast majority of people will feel comfortable fitting into one of those two options – I’m assuming woman and man are the two you’re using though a lot of people use male and female, which don’t really capture human identities and refer to sex more specifically. Sometimes people see using female/male as more scientific because you can use those terms across species (at least those that have male/female) but most human research guidelines advise that you should use women and men because those are whole identities. But also there are some folks who identify as genderqueer, trans, or a host of gender/sex identities that transcend man/woman. Why not make room for people to write in their own answers? That also fits with values around giving people autonomy to self-name. You could also have a third option though calling it ‘other’ might other your participants (make them feel even less valued by science than they already do). I have been guilty of doing this myself, but you could label it ‘nonbinary option’ or something.

Along these lines, are you measuring those intersecting identity variables? Remember: feminism doesn’t assume that all women are women in the same way – it doesn’t (or shouldn’t) essentialize women’s experiences. Race/ethnicity is a key intersecting identity variable and, again, you probably don’t want to give predetermined options unless there’s some a priori reason to do so. Lots of multiracial and biracial folks have argued how alienating questionnaires can be for them when they have to pick one identity over another, as have others who have to choose a label they don’t identify with. Are you sure you know the terms groups are using contemporarily? One option is to just leave blank space for people to write in their identities (on any question) and provide people with the opportunity and autonomy to self-name. You can always categorize people after in the way that is most scientifically useful.

All the same for sexual identity; why not use open-ended questions if you feasibly can? I see a lot of people ask participants to choose ‘heterosexual’, ‘bisexual’, or ‘homosexual’ (or gay/lesbian). But in all my years of research, only one participant has self-identified as homosexual – and I’m a sex researcher so I ask about sexual identity and orientation a lot! Very few women identify as ‘gay’ – instead, terms like lesbian, queer, dyke, etc., are used. Why does this matter? Well, imagine you had to identify yourself as a feminist using the term ‘woman-lover’. That’s not quite it, is it? Imagine you had to identify yourself as a scientist using the term ‘measurer'; that’s also not quite it, eh? Respecting people’s right to tell you what they call themselves seems kind of like a basic human right. And, it seems pretty important for engendering good feeling among your participants.

Another thing to think about is what you’re really wanting to measure. Are you measuring whether someone calls themselves a man or a woman when you’re really interested in how feminine and/or masculine they are? Whether they have a penis or a vulva, ovaries or testes? Whether they menstruate or not? Because, for example, not all women menstruate (e.g., pregnant women, some breastfeeding women, prepubertal women, some adolescent women, postmenopausal women, some perimenopausal women, women on long-term hormonal birth control, some stressed women, some elite athletes, some anorexic women, and women who have transitioned gender/sex) and not all men don’t (including some men who have transitioned gender/sex). And, the whole reason we separate out gender from sex is that not all women are feminine and not all men aren’t feminine. Thinking more carefully about the end goal of why you’re asking something can be useful towards more feminist questionnaires because you can separate out things that are unnecessarily conflated like femaleness and menstruation, or femininity/masculinity from gender/sex. It’s more scientific too. But that’s another post.

[Image with the following text: My questionnaire. Who are you? Why did you do that thing? Will you do it again? What kind of milkshake will you make me? Do you prefer (circle one) “The End”, “Toodleloo!” If you were me, what book would you read next (double unreversed scoring for CanLit)? Please indicate how strongly you agree with this statement on a scale of 1 (strong disagreement) to 7 (strong agreement): windows are good.]

This is definitely the questionnaire I use in all of my studies.

Here is the conclusiony part

Ok, so this post got a wee bit long and we barely got into intersectional identities, much less other demographic factors, much less the rest of your questionnaires did we?! Wow. This feminist questionnaire stuff must be hard, right? WRONG! Most of these things have super quick fixes, actually. For example: instead of giving people preset options, provide space for your participants to self-identify. WOW THAT IS SO ARDUOUS I COULD NEVER DO THAT JUST KIDDING. As long as your sample is under 1000 people, it really won’t take very long to categorize them, especially with practice. I can usually code about 200 participants’ open-ended self-reported gender in probably 3 min! Race/ethnicity takes a bit longer. Maybe 10 min? Surely, though, the gains of doing research in ways that are congruent with your research values are worth these minimal time costs. In terms of figuring out what you’re really asking – well, I suppose that takes a bit of thought and discussion, but it hardly seems like a burden (like, oh shoot! I have to figure out what I’m studying! THIS FEMINISM JUST ASKS TOO MUCH OF MY SCIENCE!). A key tenet of feminism is to address gender inequities and intersecting identities; we don’t want to be marginalizing already marginalized groups further through our questionnaires, or adding to gender and other inequities. That doesn’t ask enough of our science.

Did I cover everything you ever wanted to know about feminist questionnaires? Probably not. Hopefully not. I mean, this is a blog post, hardly the sine qua non of definitive exhaustive scholarship. If anything, this post probably opens up more questions than it answers. Just like a good feminist science blog post should.

Here is the summarized part.

Theory stuff:

• Methods aren’t feminist; applications of methods are.

• Gender/sex is a term that means both gender (social) and sex (bio/evolutionary) aspects and is useful when you’ve got a whole person and have no scientific evidence to partition otherwise.

• Gender/sex differences tell you nothing about mechanism; they’re only descriptive.

• Female/male is seen as ‘scientist’, focusing on non-social aspects of human experience (vs. man/woman)

 Practical stuff:

• Include more than one gender/sex in your research.

• Collect data on gender/sex.

• Collect additional measures on the aspects of gender/sex you’re interested in, e.g., femininity/masculinity, hormones, etc. so you can get to mechanism possibilities.

• Mix up which options come first because people usually put the majority group first.

• Instead of male/female, consider using woman/man.

• Consider using open-ended questions for identities (this allows for more than predetermined binaries/options).

• Measure intersecting identities.

• Identify the variable of interest and measure that, in addition to the identity you think represents it.