![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
eftychia at 02:47pm on 2010-08-25[Oh, forgetful me ...]
So, for the umpteenth time, I got ticked off at a web site that required new registrants to pick a gender of "male" or "female" before leaving a comment, and for the several-th (but less than umpteenth) time, I fired off a note to the contact address for the site, asking why that was a mandatory field, and if it was mandatory then why didn't it include the right answer.
Uh, unlike most other web sites I've prodded about this, instead of getting completely ignored, or getting a patently insincere "to serve you better" with no farther explanation, or a buzzword-based brush-off, I got a polite note from a manager, explaining (acceptably vaguely) why the question is there[1], promising to take it up with the relevant director when he gets back from someplace-else[2], ...
... And asking for suggestions for labels to add to the list for that field, other than "other", which I had already pointed out was, by definition, 'othering' when it's the only option shown beyond M and F.
I want to make sure I'm giving good advice.
I remember reading some useful discussion of exactly this question sometime in -- uh, the last eighteen or twenty months? -- but can't remember where. It was the kind of discussion where folks supported their opinions and tried to take into account data-analysis messiness as well as the feelings of us folks with not quite "standard" gender (and, IIRC, at least part of the conversation looked specifically at "understand our readership better" and "tell advertisers what our demographic is" reasons for collecting the data in the first place, as opposed to medical situations or dating sites, for example.) Do any of y'all happen to remember where that/those discussions took place, or maybe even have them bookmarked? One of the trans-issues sites? Folks trying to get LJ to improve the gender field in user profiles? DW trying to decide how to set up the same field?
I'm mostly looking for a pointer to the discussion I remembered seeing before, rather than wanting to hash it all out from scratch in comments here, but I'll take folks' thoughts here too.
Off the top of my head, I'm inclined to suggest "male", "female", "both", "neither", "other", as a reasonable (though imperfect) starting point, assuming that they want to keep a pull-down list, don't want to try to list every gender-identity label currently in use, and think too many people would pick "decline to answer" if that were offered[3]. But I've got this nagging feeling that there were some problems with that scheme that came out in the last discussion, that I really ought to remember.
[1] "This helps us better understand and define our audience which will in turn define and shape the future of our business. This is primarily a research question." This makes me wonder just what they think the gender info is really telling them -- are they working from a stale (or exaggerated) list of expected behaviours/tastes based on gender, or keeping careful track of how what correlations there are change over time, or just tweaking their content by trial and error to try to keep the male-identified:female-identified ratio in a range that makes their advertisers happy? But ultimately, not my problem. I just want the gender field to be made more inclusive, optional, or both. Whether they're being smart about what they do with the data, I'll probably never know.
[2] But no promise that anything would actually get done, because he doesn't know whether they have the ability to alter that part of the form -- which I'm guessing means that the registration/comments section of the site is a package they bought somewhere else or a setup hosted by someone else, rather than a system developed in-house.
[3] LiveJournal, which has "male", "female", and "unspecified", appears to have about 28% of users listed as "unspecified" (though the stats page makes the male & female numbers add up to 100% and ignores the unspecified precentagewise). InsaneJournal, with the same list of options (and the same way of counting percentages), shows about two-thirds of users picking "unspecified". Dreamwidth, with "male", "female", "other", and "rather not say", shows 39% under "rather not say" and 1.6% under "other". I'm guessing that the folks I'm talking to aren't going to like the idea of having one to two thirds of the answers to the gender question fall into a "myob" category if they do change the registration form.
(no subject)
The only case in which I can see a site caring about your gender are when the site is going to offer medical advice (let's face it -- someone born female isn't going to be worried about her prostate). OTOH, it'd be smarter to have options for transsexuals and folks with both sets of organs (I know at least one person who fits that description) if you're going to go *that* way.
(no subject)
And when they have a good reason for wanting to know, if they only provide two options (either because they've never thought about it, or "because there aren't enough non-binary-gender folks to worry about"), they've got no way of ever finding out whether their assumption about how many of their users are non-binary is wrong.
(no subject)
* or 7, or 9, depending on the desired level of granularity.
(no subject)
Still, it'd be interesting to see what marketing specialists would do with more fine-grained data than just M or F!
(no subject)
(no subject)
What I would like to see on these forms, I think, is at least "female," "male," "other" [with a space to specify further if the person wants], and "rather not say." A fill-in-the-blank without checkboxes would be better, but I suspect someone would then get to massage the data for things like capitalization, the commoner typos, and possibly if they decide (for example) that "female," "woman," "F," "feminine," and "girl" are instances of the same thing.
stating gender accurately
From flaviarassen
But yes, I always have a thought, "& what do you do when you're neither or both...?"
(no subject)
My thought is to make the field free-text. I know that a lot of sites won't like how wide open that is, but in the end I think their objections are misplaced: it's easy enough to map all the synonyms of the genders you have a category for, and anyone who answers outside of those categories is someone who wasn't going to line up well with your targeted ads, cutesy avatar changes, or whatever else is driven by that gender field. What's more, if your interest is truly to do statistical modeling of your users (whether for hypothetical improvement of their experience or merely for demographic targeting) you are better off allowing as much specificity as your users can give you, because you can build clusters of them, which you can't reasonably do with that 25% Other/Unspecified.
If, for example, you wound up with a significant number of users identifying as "m2f", you could track similarities in their aggregate usage much more precisely than if they were lumped into a larger "other" category. Not that you can't have ways that your system recognizes and handles specific cases like "other", "both", "N/A", "etc", but those would no longer be the limit of what you are able to work with.
There is of course the possibility that a free text field will encourage people to enter random inputs. I don't know whether more people would do that than would randomly pick between M/F if forced to choose - when you're intentionally entering spurious info, it doesn't seem too likely that your limited choice of gender category is very statistically useful either. I suppose a Male, Female, Other (Please Specify) ____ setup might cut down on that a little, but then we're back to othering, which the free text alone will never be.
There are two cases where I think this doesn't work terribly well, one a user case and the other an application case. For the user, if they are unwilling to put down in any simple label what they consider their gender to be, this system won't work terribly well for them - I don't think gender should be a required field, or have a minimum number of characters, but if one does force that than it will be tough to accommodate those users.
For the application, I think there could be times when you want a very specific kind of quantification within gender axes. For instance, a survey related to gender issues might want to ask people to choose the closest approximation of their identified gender within the two-dimensional space you describe above. Clearly specific data mapping constraints will have more specific data input needs; in this case I think a Maleness slider and a Femaleness slider would work.
Of course plenty of sites might claim that they need that type of specific categorization for their own purposes, but most of them really just want something easy.
--learnedax