Asking gender on surveys

Frequently, when people set up surveys, they ask for gender. This question is often included out of habit, but it still is often a relevant question, even if the only contribution of the question is to show that gender doesn’t have a significant influence on other responses. In modern efforts to be more friendly to gender minorities, the gender question is often updated. And it’s often updated in silly ways.

Here’s the two most common “inclusive” ways I have seen the gender question presented over the last few years:

a) What is your gender?
* Male
* Female
* Transgender
* Prefer not to say

b) What is your gender?
* Male
* Female
* Other

There are minor language variations, such as sometimes using “transgendered” instead of “transgender”, but this is the gist. So, what’s wrong with these?

First, transgender isn’t a gender distinct from male or female. For a long time, however, many people who were not men or women found “trans” to be an inclusive umbrella term which made them visible. In the early 2000s “genderqueer” started gaining popularity, but it wasn’t until this decade that the term “nonbinary” (in contrast with “binary”, someone who is a man or woman) really caught on as an umbrella term for gender identities outside of man and woman. Transgender as an umbrella term still often is used to include nonbinary people, but it also includes a lot of people who are binary. So question a could also be phrased:

a) What is your gender?
* Male
* Female
* Cisgender
* Prefer not to say

If you are cisgender (not transgender), how would you answer that question?

Question b could easily be read such that “other” means the same thing as “nonbinary”. However, we tend to work off of past information in absence of specific information at present. So, often people who offer “other” as a third gender option specifically are referencing trans people: remember that the gender question tends to get included out of habit, and its verbiage is often copied and pasted like a cover sheet from an old homework assignment. To add further unintentional implications, there’s some gender/sex stuff going on here. In the 90s, a big push for trans inclusion and awareness included variations on the line “Gender is between your ears, sex is between your legs.” This idea came with a push to distinguish between man/woman as gender terms, and male/female as sex terms. There are problems with this simplification, notably that neither gender nor sex is that simple, and that it simply shifts gender essentialism to sex essentialism. So, some trans people will look at this question and assume it is implying that they should answer “other” because they are trans. Meanwhile, others will read the sex essentialist interpretation and answer based on whatever they were assigned at birth: trans men may answer “female” and trans women “male”. And others still will answer with their binary gender identity, if they have one, or the gender they are legally. The inclusion of “other” now means that for a given trans person there are <em>three</em> different answers they may give, and you can’t seed out the trans respondents from the cis respondents by simply excluding “other”. At best, this increases your quantity of outlier data.

However, it’s a start, and the language usage has permeated our social transgender awareness even when people don’t know it. So if question b were phrased using “man/woman/other” as the first two options there would be more ambiguity as to whether “other” is implying “transgender” or “nonbinary”, but the usage of “male/female/other” strongly hints at the popular treatment of “transgender” as a third sex, and/or conflation between transgender and intersex, and or vague awareness of intersex people existing and having no idea what that even means. It’s still a better phrasing than question a because it doesn’t explicitly call “transgender” a gender, but what’s your data going to be? For transgender and intersex people who choose a binary gender option with which they identify, that data is just going to get absorbed in the male/female results. Which may be fine, because maybe you didn’t have any specific need to separate out trans and cis people in your data (important thing to consider!). But who’s answering “other”? This may include trans people, nonbinary people, intersex people, as well as…people who prefer not to say. Oops. You can add “prefer not to say” as a fourth option for b, but it doesn’t really address the rest of this mishmash group. So you have data that may contribute to your overall results, and a way to filter it out of your binary gendered results, but it won’t tell you much about results specific to any of the groups in this “other” group.

So, why not go back to explicitly mentioning transgender as an option when you want to know more about the responses from transgender people? You’re still going to mess up your data if you group all trans people together. There are trans men and trans women and trans nonbinary people just as there are cis men and cis women and cis nonbinary people, and just like cis men and women and nonbinary people, trans men and women and nonbinary people have different health concerns and risk factors. For a common example when surveying health concerns, the rate of HIV among trans women is much higher than among trans men, so grouping both together will mess up your data.

One better way to survey with visibility and sensitivity toward transgender, nonbinary, and intersex people, while also collecting usable data, is to do something like this:

What is your gender?
* Man
* Woman
* Nonbinary/Other
* Prefer not to disclose

Are you transgender?
* Yes
* No

Are you intersex?
* Yes
* No

Put all questions on the same page, so that the person can read ahead as they determine how to answer the first question. When gender minorities answer questions about gender, we are always gambling on the expectations of the person asking, and putting all your cards on the table makes us much more willing to play your game in a way that is useful to you.

However, there are still a lot of problems even with this solution. These questions are asking about gender identity and gender assignment history. So, let’s say you are writing a survey about gender and campus safety. If a person is perceived as assigned female at birth or as transfeminine (assigned male at birth while presenting on the female end of some spectrum), they’re at a higher safety risk. But that’s something that these questions won’t address, because they’re not about perception. If the topic you’re addressing is looking at a person’s experience based on how they are perceived in a particular context, another gender question might be something like (using the same example of campus safety):

How is your gender most commonly perceived when you are on campus? “Cisgender” means not transgender. If you’re not familiar with the term, it probably means you.
* Cisgender man
* Cisgender woman
* Transgender man
* Transgender woman
* Other/uncertain

Of course it’s important to avoid a survey that leads its data, so this survey question might be coupled along with the questions above, or might replace it. Are you surveying about safety around strangers or around people the respondents already know, or both? If you’re only surveying about safety around strangers, then this question or something like it may be the only gender question you need to ask. However, you may then be missing critical safety issues such as risk experienced by transgender students who are outed.

On the other hand, let’s say you’re surveying about car purchases. You’re surveying a huge data set to see what kinds of cars people buy, and want to see if the results vary by gender. It’s possible this survey could get down to the nitty-gritty about economic circumstances and community information among white trans men resulting in a higher rate of purchase of second-hand Subarus, in which case it makes sense to ask some very specific questions, but it’s more likely that you’re interested in questions like: are minivan sales up among men? In that case, it makes sense to keep it simple with something like:

What is your gender? Choose whichever you identify with.
* Man
* Woman
* Other
* Prefer not to disclose

This question provides a reasonable option for everyone regardless of gender, allows the option to decline that piece of information from the gender portion of the survey data, and provides useful information for how to answer the question for people who are often stuck gambling.

So as you’re writing survey questions about gender, here are some things to consider:

  • What are you trying to get out of asking about gender? Is it known if your survey topic is dependent on gender identity, social perceptions, legal status, medical status, medical history, gender theory?
  • If it isn’t known, how can you ask for potentially relevant gender information without leading the results?
  • What are you trying to accomplish by explicitly including something acknowledging gender minorities?
  • What information are you giving your gender minority respondents to assist them in answering the question with the information you’re trying to get, so that they don’t second guess
  • What information are you giving your gender normative respondents to assist them in navigating potentially unfamiliar terminology?
Advertisements

Leave a comment

Filed under Paragraphs

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s