1.3 Data collection

Building a spreadsheet of every queer character ever turned out to be really hard. Since this isn’t an issue that anyone has been tracking yet (aside from GLAAD, who refused to give me their data), the first step was finding a data source. For this, I turned to the next best thing: Wikipedia.

Wikipedia doesn’t have just one list of queer characters. Instead it has six. These are mostly split up by genre (gay characters in sitcoms, gay characters in dramatic series, etc), so before I could even start figuring out who lived and died I had to roll all those pages into one spreadsheet. In total, after I’d taken out the duplicate entries, there were more than 2,000 characters on the list

Not all characters are created equal, however. Other authors decided to cut ones who’d only appeared in a couple of episodes out of their analyses, as these characters were too minor to have any real impact on the show. I did the same, using IMDb to check whether a character had been in three or more episodes (or more than a quarter, if it was a very short show) and dropping them from the data if they hadn’t.

This was the point where I enlisted two of my friends. Turns out checking over 2,000 characters against their IMDb entries is a lot of work.

Tossing out the extremely minor characters brought the list down to about 1,800. Along with figuring out whether they got to stay, we logged some extra data about each character: the show they were on, what year they first appeared, what year they finally left, how many years they were there, the number of episodes they appeared in, the number of episodes the show had in total, which month their last episode was in, and their gender.

All that was left was to figure out who lived and who died. This meant going through the whole spreadsheet again and looking for the characters that appeared on Marie Bernard’s lists of dead queer men and women. These lists may be incomplete, especially since the list of men didn’t get the audience help that the list of women did, but they’re the best we could find. By noting down who had died, the rest was quick to fill in; it’s a pretty safe assumption that if someone isn’t listed as dead, they could be marked as “not dead”.

That’s it. That’s how we got the data. Now here’s what I did with it.