Who Cares About Character Gender?

As someone who does research on gender and games, I often hear the conventional wisdom that men prefer to play male characters. For example, that’s Mark Rosewater’s explanation for why there are twice as many male planewalkers as female planeswalkers.

Well, guess what? Conventional wisdom is wrong.

A team of researchers from MSU, led by Robby Ratan, looked at game logs from 18,000 League of Legends players. Unlike many studies on character gender, the research team didn’t have to rely on what people said about the kinds of characters they wanted to play. They could look at what players actually did play. This is important because people often present themselves as they’d like to be seen – even when this doesn’t reflect their real behavior.

During the period of the study, around 70% of available characters were portrayed as male, and around 30% were portrayed as female. That makes 70% male and 30% female the “chance rate” – the distribution you’d expect to see if gender weren’t a factor and people were just choosing their characters randomly. In other words, if men prefer to play male characters, you’d expect to see men choose male-gendered characters more than 70% of the time.

It turns out, though, that men play male characters 70% of the time, and female characters 30% of the time. In other words, male League of Legends players don’t seem to care about character gender. You get the same distribution as you would if they were picking characters at random.

Women, on the other hand, played female characters nearly 50% of the time. Only 30% of characters were female, so this rate is significantly more than what you’d expect if female players were choosing randomly.

In other words, women, not men, are the ones who care about playing characters of their own gender.

This implies that designers should make sure women get plenty of opportunities to play female characters; men, on the other hand, will play whatever they’re given. In other words, worry about women, and men will take care of themselves.

Obviously, things are a bit more complex than that. For example, ability types aren’t randomly distributed across character genders. During the data gathering period, there were no female tanks, but plenty of female support characters. That means that the “pure” gender data is probably distorted by the fact that the male and female character pools aren’t equivalent. Character gender choice is going to be partly influenced by the player’s preferred team role.

Another thing to note is that only 4% of the subjects surveyed were women. That’s an unusually small percentage, suggesting that League of Legends is among the most male-dominated games out there. It could be that men only feel secure playing female characters when the activity is so heavily coded male that it doesn’t threaten their gender identity. Alternately, women might care much more about playing female characters when they know they’re in a tiny minority. That might be because they feel they have to work extra hard to maintain their identity, or because only women with strong ties to their gender identity make it into the community in the first place.

Still, game designers can no longer make the same old lazy assumptions about player and character gender – and that’s got to be a good thing all around.

6 comments

  1. Mo says:

    Ayup. I totally will play a female character where it is offered. I might play a male character on a second run through, especially if I think the character options are broader or to explore the gender based difference in story play, but I’ll only go through twice if the game was really worth it (e.g. Mass Effect).

  2. George Locke says:

    If men had no gender preference, then if the background were 50/50, men would choose 50/50. The data do not support that conclusion this is how men would act, which I take to be the gist of your argument.

    Men’s choices mirror the background distribution, which implies effectively randomized choice. However, the background is heavily skewed toward men, so the men are still much more likely to choose to play men. Women correct for the bias in the available characters.

    Considering the data as you describe them, the conclusion I draw is that men are happy to conform with a masculine-dominated character selection whereas women are not.

  3. Daniel Wood says:

    Did the authors correct at all for the fact that in League of Legends, character types are not distributed randomly across gender?

    For example, the majority of ‘support’ characters are female-represented, so if there is an existing trend among female players to play support characters (based on my non-scientific understanding of similar trends in MMOs, etc. this seems likely) then you would expect that to appear as an overall preference of females to play female-represented characters, etc.

    Anyways, I guess as someone who played LoL for awhile I wonder at its suitability for this sort of study (outside of it having the stats available) because the characters themselves are mechanically discrete units — you can only play the character as one gender, and the characters have different power sets. If the assumption is that this will even out over a large enough data set, I am skeptical — because you have to account for the game designers’ own gender-biased tendencies to assign certain character types as female or male.

    And believe me, the people doing character design for LoL are sexist as fuck.)

    • Daniel Wood says:

      Oh how embarrassing, I skipped over the part where you actually talked about this. Anyways consider my question answered.

  4. Mike Holmes says:

    Hi Jessica,

    I’m not all that familiar with League of Legends, but I do see it is a MMORPG of some sort. I think that your analysis may only pertain to this sort of game, therefore. That is, MMORPGs tend to have very little character association by the players.

    I have lots of experience with MMORPGs, having played World of Warcraft for copious hours, and having been a Guild Master for a guild of over 50 active players (something like 250 “toons”). And in studying the behavior of players – this was always a side-project while playing – I’ve found near zero correllation between the behaviors of TTRPG players and MMORPG players.

    I’ve often noted how in WoW the servers are labeled as “Normal,” “PvP” and “Roleplay.” Yes, that’s right, roleplay in WoW is officially considered something other than normal. More to the point, nobody does ANY roleplaying in WoW, other than some few debased examples on the “roleplay” servers. Even there it’s rare, and much of that comes down to toon cyber-sex. Examples of players caring about their characters’ actions are… well I’ve never seen any such play on WoW. And that’s despite looking for it hard.

    Put simply, characters in MMORPGs, labeled “toons” for a reason, are not even avatars. As somebody once said, they are more like “Vehicles.” You’ll note that a player in one of their vehicles will often change to another vehicle, when the vehicle’s powers aren’t right for the job at hand. “We need a healer,” means the player logs off their current character, and on to their healing character. And I am certain about this, because even where a player is referred to in chat by their character’s name, they will usually be referred to by their “main” character (the one that they play most often, or are most associated with), even when playing another character. Often to facilitate this, every character has a variation of the exactly same name. Such as “JonHealer,” JonTank,” JonHunter,” Etc.

    These are simply assets that the players have on hand to solve problems, not at all “characters” in any meaningful sense.

    That said, lots of male players play female toons. It’s kind of a running joke, in fact. And there is, in fact, a meme that people will cite to defend this choice. Which is “If I’m going to be looking at somebody’s ass for 8 hours a day, it may as well be female.”

    Now… I think that this is probably a defensive behavior. To prevent people from thinking that they’re somehow transgendering or something (due to the stigma). I think some of those males get some sort of a kick out of playing a female character of some sort. But I don’t think that kick is “being a female” in a roleplay sense, except perhaps for a tiny percentage. And of those, near zero of them want to share what it is that they’re doing with other players.

    Also note that (even with WoW’s relatively recent move to naming players, and not just toons), often you don’t know the actual gender of the player until you hear their voice in Ventrillo or chat them up. So people don’t have to worry about whether or not they will get “caught” playing a female character; they won’t ever have to explain it to anybody they don’t wish to know. Basically they can do it without fear because they’re hidden from people knowing that they’re doing it.

    Obviously they shouldn’t have to fear it, but they obviously do, hence the meme above used to explain the behavior when it’s discovered.

    All of this contrasts with TTRPGs where the players are face to face, and have to deal with both the gender of the characters and the gender of the players, in a meaningful way. Not that all TTRPG play is full of deep meaning, but even the choice to play a female in the most tactically oriented TTRPG group is still raises people’s eyebrows. To say nothing of how awfully such female characters are… abused… in many such games. Often males in such groups are using the lack of responsibility felt towards theme in play to enact male power fantasies of the worst sort with female characters.

    So obviously this is a nuanced issue, and I think you’ll need more data before you can support some of your conclusions. At least outside of MMORPG play.

    Mike Holmes

  5. Jesse Fuchs says:

    It’s interesting to me that it worked out the same as the random distribution, especially since LoL characters are so distinct in their abilities—there’s probably a better way to put this, but it makes me wonder if the most “male” behavior is to be such a competitive optimizer that gender and other non-gameplay-influencing factors barely register.

    I’d be curious to see the stats in an online game like Monster Hunter, where your character is a (mechanically) featureless cipher and all that matters, male or female, is the equipment you’re currently carrying. (And where the armor, while looking different between male and female characters, tends to be heavy enough that the voyeuristicc factor doesn’t really come into play.) My guess, from highly anecdotal information, is that it would break about 75/25 male-female for guys, and 15/85 male-female for gals, with the difference probably just resulting from female videogamers already having gotten to play as more than enough dudes for one lifetime.

Leave a Reply

Your email address will not be published. Required fields are marked *