Coming out on Facebook
October 16, 2015 11:31 AM   Subscribe

 
Speaking from cis gay experience, coming out is a more widely-negotiated process than updating a fucking profile status.

I mean, this could also be an indicator of the rate at which people just go "Well, they know everything else about me, and own that data, so screw it."

All this isn't to say that this data isn't interesting - it is. But it's not a meaningful proxy for what "coming out" actually entails in lived experience.
posted by mandolin conspiracy at 11:43 AM on October 16, 2015 [14 favorites]


It was how I came out and it had a pretty damn profound affect on my life. I think you are speaking a little to broadly, mandolin conspiracy?
posted by Annika Cicada at 11:47 AM on October 16, 2015 [11 favorites]


But it's not a meaningful proxy for what "coming out" actually entails in lived experience.

I think it depends on the age and the location of the person, though, and the life circumstances. I think this data is incredibly good and valid, but I just want to see more of it.
posted by roomthreeseventeen at 11:49 AM on October 16, 2015 [2 favorites]


But it's not a meaningful proxy for what "coming out" actually entails in lived experience.

Of course not. And the article doesn't claim that. It's pretty clearly talking *only* about "the rate at which people came out on Facebook."
posted by beagle at 11:51 AM on October 16, 2015


But it's not a meaningful proxy for what "coming out" actually entails in lived experience.

This is the year 2015; it's entirely possible to live one's life with all meaningful interpersonal relationships intermediated by the Internet.
posted by Faint of Butt at 11:56 AM on October 16, 2015 [6 favorites]


This is the year 2015; it's entirely possible to live one's life with all meaningful interpersonal relationships intermediated by the Internet.

My fit of pique is entirely attributable to the fact this makes me feel goddamned old.

But what is pretty amazing is that for some people this can be an all-or-nothing coming out process. At the press of a button, you can disclose with unprecedented speed and reach.

Truly an age of wonders.

Now get off my lawn.
posted by mandolin conspiracy at 11:59 AM on October 16, 2015 [9 favorites]


I suspect that using "specifying a custom gender" is actually going to miss a lot of trans folk--many of whom simply want to be recognized as the binary male/female gender that they identify with and don't want to be publicly labeled as "transgender," especially if they're planning to pass and are capable of such. Unless Facebook can track people who previously identified as one gender and then switched to another (potentially on "Coming Out Day"), transitions are going to go unnoticed.
posted by dlugoczaj at 12:02 PM on October 16, 2015 [2 favorites]


Figure 3 is terrible. I don't know why they didn't plot first differences.

Also there's a really easy first control for growth of Facebook factors: what fraction of cishet interest identifications were added since 2012?

There's interesting things to be done with this dataset but I don't think they're doing it.
posted by PMdixon at 12:05 PM on October 16, 2015


1. Why is Utah so gay?

2. Your profile settings is one way of coming out, but really, it could be just as indicative of "Oh right, Facebook's settings changed... Guess I should update it."
posted by picklenickle at 12:29 PM on October 16, 2015


"Coming out" can have many different levels. I'm not out to my employer, doctor, in-laws, the general public etc. (except crytically and/or anonymously).

Most of my coming out milestones happened online, but updating the gender field on my Facebook profile wasn't a particularly big one for me.
posted by Foosnark at 12:32 PM on October 16, 2015


Shortly after Facebook began allowing custom genders, I set mine to "gender fluid male" and changed my pronoun setting to "they". No one looks at your profile anymore, I figured, and I was right. Nobody ever mentioned it, and over a year later, everyone was still surprised when I actually came out and changed my settings to "female" and "she". But according to Facebook, I went back into the closet that day.

I guess at that point they could have flagged me as gay instead of trans, but I don't get the impression that their methods capture same-sex attraction on change of gender. And anyway, I removed the value for "interested in" about 8 years ago when I got engaged, because I'm only interested in my wife, thanks.

I'm sure their dataset is full of noise like that, a problem which isn't helped by the fact that the analysis isn't normalized or controlled. Even worse, the results aren't particularly surprising or interesting, and the graphs don't tell interesting stories. Figure 1 has a spike in January and a sudden change in the trend in March that they don't bother to examine at all. Boiling down the whole post, I can conclude that people got really enthusiastic when Obergefell was announced, which I already knew.
posted by WCWedin at 12:36 PM on October 16, 2015 [9 favorites]


Wow was that terrible. They should hire the OKCupid blogger.

I'm always skeptical of using percentages on axes of a graph - I don't think it means what readers think it means. This was a great example of this.
posted by freyley at 12:56 PM on October 16, 2015 [1 favorite]


I'm not sure... How would they handle cases where someone abandons their old account and starts a new one? The 'cut-all-ties' scenario seems pretty plausible.
posted by mikurski at 1:12 PM on October 16, 2015


I'm sure their dataset is full of noise like that, a problem which isn't helped by the fact that the analysis isn't normalized or controlled.

The point of statistics is, the noise cancels out. That's not to say there's no art or science to it, but when you're looking at a dataset of millions of events, the ways that one measurement can go wrong fade in significance. To undermine the point of the post, you would have to show a substantial bias towards false positives in the present (but not the past-year) data. "Noise" does not do that. When you have this much data, you can tolerate a lot of noise.

I mean, this is a blog post, and it's very theory-light. There are many things it doesn't say, some of which you might want to see in a more thorough investigation. But the points it does make are overall convincing. The big spikes in that graph, and the overall trend, is not "noise."

I suppose I should disclose that one of the authors is a friend and (hopefully) collaborator of mine. But I feel no conflict in saying that I think this is good work.
posted by grobstein at 1:17 PM on October 16, 2015 [4 favorites]


The point of statistics is, the noise cancels out.

Random noise cancels out. Non-random noise does not.
posted by srboisvert at 1:35 PM on October 16, 2015


Random noise cancels out. Non-random noise does not.

Which non-random noise, in particular, are you worried about, and how does it effect their main conclusions that (1) there's been an increase in people coming out on facebook, (2) there was a spike in both folks coming out on facebook and folks "liking" various LGBT+ paged after various supreme court decisions, and (3) there are state-by-state differences in the amount of people who are "out" on facebook?
posted by damayanti at 1:40 PM on October 16, 2015


Does facebook still require you to specify a gender, or have they gone back to allowing you to not specify one? My account was registered before they required it, so I don't have any gender set. There was some nagging about it when they changed over, but I haven't been prompted in a while now.
posted by vibratory manner of working at 1:46 PM on October 16, 2015


Yeah, not getting the backlash here. The authors, as stated, are not measuring the rate change of coming out; they're measuring the rate change of coming out on Facebook. The latter is far easier to measure and far more of a binary than coming out in real life; that's what allows them to measure it in the first place.

And since Facebook doesn't have an "I'm coming out" button, all they have to measure are the sexual orientation and gender identity fields on one's profile. This is a proxy for coming out, yes, but it seems to be a sound one. (WCWedin, I don't think this study would conclude that you went back into the closet when you switched off of a custom gender, at least for the purposes of Figure 1. They're just counting the number of people who went from cis-het to not each day; they're not counting people who seemed to go the other way.)

It does seem interesting to me that the rate of Americans coming out on Facebook appears to be increasing over time. But I would like to see how this trend correlates to "number of Americans who have Facebook accounts" — i.e., are more Americans coming out of the closet, or are more just signing up for Facebook? Also, their definition of coming out includes "updating one’s profile to express a same-gender attraction" — but it's not clear whether this includes (a) people who indicated they were gay when they signed up for Facebook, or shortly thereafter; or (b) people who changed from "no gender attraction specified" to "same-gender attraction". Group A is clearly not closeted, and Group B may or may not be closeted. These things matter for Figure 1, but not as much for Figure 2.
posted by savetheclocktower at 3:13 PM on October 16, 2015 [2 favorites]


(1) there's been an increase in people coming out on facebook,
Relative to what baseline? Why is this interesting? Should I be surprised? Why?

(2) there was a spike in both folks coming out on facebook and folks "liking" various LGBT+ paged after various supreme court decisions,
This is true but at best falls into the category of confirming the seemingly obvious. And again, how does that compare to spikes around other politically charged news events?

and (3) there are state-by-state differences in the amount of people who are "out" on facebook?
Compared to what? Are the differences large compared to chance noise? Given known demographic relationships? How do they compare to existing state level survey results?

I mean this just reminded me of those two weeks I learned you can just regress 2 data sets onto each other and spend 15 minutes making the scatter plot look pretty, and that will suffice to demonstrate you've done work for the day.
posted by PMdixon at 4:18 PM on October 16, 2015 [2 favorites]


(I guess my actual point is that mucking with time series data is SRS BZNS)
posted by PMdixon at 4:34 PM on October 16, 2015


they're measuring the rate change of coming out on Facebook

I'd say that this is maybe not an accurate measure of "coming out." So, to your point about there not being a "coming out" button, we'll never know.

As far as backlash goes, the data says what it says: the number of people who have updated their profile is this.

The latter is far easier to measure and far more of a binary than coming out in real life; that's what allows them to measure it in the first place.

Potentially, yes.

But I guess the question I have is how many of that cohort have "come out" by saying things on Facebook that is coming out by other means (though posts, chat, etc.) that in turn broadcast to some or all of their full FB network that they're L/G/B/T? So I'd argue it's not necessarily a sound proxy from that standpoint alone. Add to that the way in which (excluding people who live online only) IRL interactions confound it further - it's entirely possible that, IRL: part of, the vast majority, or all of someone's FB network knows their sexual orientation and/or gender identity.

A Rumsfeldian known unknown, if you will.

It's an interesting data set, but one that doesn't capture the multifarious ways in which one comes out that could potentially interact with that data to render the update of someone's gender attraction a boring non-event in someone's network and therefor not constitute a "coming out" event.

But it can also be a way in which one comes out - full stop - too.
posted by mandolin conspiracy at 6:26 PM on October 16, 2015


I never noticed, until reading this thread, that your FB profile includes which gender(s) you are interested in! How come it only gives you two choices?
posted by mittens at 6:29 PM on October 16, 2015 [2 favorites]


I don't know about computer-y things, so maybe I'm overly suspicious, but the fact that your gender and "interested in" fields on Facebook aren't by default fill-in-the-blank (though at least they let you do that for gender now) makes me wonder what they're doing with all that data. Is there a reason why it's more convenient for people to be in a definite category?
posted by thetortoise at 8:43 PM on October 16, 2015


Is there a reason why it's more convenient for people to be in a definite category?

Lots. Memory usage is the simplest one, though probably the least important.
posted by PMdixon at 8:44 PM on October 16, 2015 [2 favorites]


(A not necessarily more useful answer but more complete is that strings are awful to deal with.)
posted by PMdixon at 9:27 PM on October 16, 2015 [1 favorite]


(yeah but come on a couple hundred characters for gender and orientation vs the sheer petabytes worth of confederate flag pictures my relatives post?)

(but i guess it helps with advertising since gay people and straight people need different weird tricks to lose weight)
posted by mittens at 9:29 PM on October 16, 2015 [2 favorites]


I swear about a year ago the facebook algo decided all the queers are besties, because all at once a bunch of people who I barely know and have nothing in common with each other besides being non straight started dominating my timeline. No change in posting behavior on my part as far as I know.
posted by PMdixon at 9:57 PM on October 16, 2015


Is there a reason why it's more convenient for people to be in a definite category?

Monetizing their data about you according to a more complete demographic profile?
posted by mandolin conspiracy at 6:41 AM on October 17, 2015


(1) there's been an increase in people coming out on facebook,
Relative to what baseline? Why is this interesting? Should I be surprised? Why?


"Strikingly, of those who are out on Facebook, approximately 78% made this change to their profile since the beginning of 2012. While this figure may be somewhat conflated by growth in the number of people on Facebook, the sheer magnitude of this increase suggests that the LGBT movement has made significant strides in recent years."

In any case, Facebook is being cautious in saying anything more about this (I don't know, claiming that "Facebook has helped the LGBT+ movement" or "This actually reflects more people coming out in real life") precisely because they know the limitations of their dataset. It's a bit odd to be complaining that, on the one hand, that they don't have the data presented here for X, Y, and Z, therefore, their story is incomplete, but that, on the other hand, they don't try and build a bigger (and more speculative) story with the data they do have.

(2) there was a spike in both folks coming out on facebook and folks "liking" various LGBT+ paged after various supreme court decisions,
This is true but at best falls into the category of confirming the seemingly obvious. And again, how does that compare to spikes around other politically charged news events?


You might think it's "seemingly obvious", but sometimes "seemingly obvious" things are wrong. That's what's data is for! As for comparisons, we don't know-- that's not in this data set. But, future research question!

and (3) there are state-by-state differences in the amount of people who are "out" on facebook?
Compared to what? Are the differences large compared to chance noise? Given known demographic relationships? How do they compare to existing state level survey results?


Compared to other states: "The fraction of out LGBT people on Facebook per state varies widely across the country, with some states having more than double the fraction in others." Note there are no claims for statistical significance, or any claims about, e.g., Mississippi being friendlier to LGBT+ people than NY. again, FB is being rightly cautious about drawing any bigger conclusions. Personally, eyeballing it, I am not surprised that, e.g., the deep south has fewer people out than the north east, so I'm willing to bet that the variation is not random noise. Again, though, future research questions!

The fact that one can think of questions that this data set isn't answering doesn't mean the data set is useless, or an inaccurate representation of what it's saying it's representing ("Number of people who changed their gender and/or orientation on Facebook, which is used as a rough proxy for coming out"). It's a bit like complaining a pie isn't a cake.
posted by damayanti at 6:42 AM on October 17, 2015 [1 favorite]


(Which is to say, it's fine to say that pie is better than cake, or that things would be better if we had pie, instead of cake, but complaining that cake doesn't have crust is rather unfair to the cake.)
posted by damayanti at 6:45 AM on October 17, 2015


In case you want to understand why an organization habitually asks its users' genders and finds it useful to put people in gender binary categories, this guide mentions some potentially legit reasons, e.g., "I need to address my users with pronouns." "I need a user's title for phone or mail communication." And so on. But you can generally ask that question without having to ask for the gender of the person themselves. Wikipedia, for instance, asks, on the user preferences page:
How do you prefer to be described?
[option] (When mentioning you, the software will use gender neutral words whenever possible)
[option] She edits wiki pages
[option] He edits wiki pages

Setting this preference is optional. The software uses its value to address you and to mention you to others using the appropriate grammatical gender. This information will be public.
And then other users can properly use a wiki template to speak with you appropriately -- which is especially helpful in languages where addressing someone or speaking about their actions requires gendering that description.

The gender as a text field approach is being worked on, incidentally, as part of the genderamender open source project, in case any of y'all want input into that.
posted by brainwane at 6:51 AM on October 17, 2015 [1 favorite]


I'd say that this is maybe not an accurate measure of "coming out."

Got a better one?
posted by MisantropicPainforest at 7:20 AM on October 17, 2015


This seems tantamount to suggesting that everyone on Metafilter came out if they marked a user of the same gender as their spouse when folks espoused each other via contacts.

In other words, it doesn't seem very scientific.
posted by terrapin at 7:58 AM on October 17, 2015


The fact that one can think of questions that this data set isn't answering doesn't mean the data set is useless, or an inaccurate representation of what it's saying it's representing

My point is its deeply unclear to me that there's a question they are trying to answer. Basically they've pulled out a couple of time series and a crosstab with no normalization etc and said "look I made a graph."
posted by PMdixon at 9:25 AM on October 17, 2015 [2 favorites]


« Older "The future of war belongs to the bots." [and the...   |   Now you can phone your cat Newer »


This thread has been archived and is closed to new comments