Video Game Demographics – I finally read the article

Earlier this week I posted about an article that was reported (vaguely) in New Scientist. Some helpful Twitter folks were kind enough to forward me the original article so I could comment more thoroughly…eventually. Here we go. [Edit: I completely forgot to post this before I left for the long weekend. Kimbo fail.]Background

Basically this is a study of in-game demographics. There were other studies suggesting that video games are similar to TV in influencing the perception of social reality and subsequent social identify formation. In normal person speak: video games may be giving people a false impression of real life diversity.

It turns out they weren’t so much interested in who is playing the games (I may have been biased to consider this because I play World of Warcraft). They were interested in whether the population of game characters [more on this below] in the top 150 games (as per US sales, apparently irrespective of where the game was developed) was representative of the population of the US (as per the 2000 census).

When a game was duplicated on multiple systems, they used the system with the highest graphic rating. They weighted the games so that the most bought were worth more in the analysis to better approximate what people were actually exposed to (given my initial impression of the experiment from the NS article, I had initially misunderstood this part, I see what they’re going for now). Digression: a lot of people play more than one game, so although more people will be exposed to the popular games, each individual may have more diverse game exposure.

Anyway, “expert” gamers played each game for 30 minutes on the easiest setting (if that was a choice). This was recorded and coded later. When the characters were able to be chosen by the gamers, they choose randomly so that there was an equal chance for representation of males and females and different ethnicities. I had wondered about this very thing in my last post (because of World of Warcraft, etc).

1) Characters
I don’t think it was a good move to include player-chosen characters at all now that I understand the purpose of their analysis. A “random selection” of these characters does not necessarily represent real life. Real life players are choosing characters non-randomly (i.e., because they are creating a representation of themselves or they are creating a persona – which are purposeful choices), so this selection method does not necessarily represent real life exposure to characters in those games. Also the whole point was to examine whether these representations were representative of the population – the population has an uneven distribution so randomizing for evenness is inappropriate to the analysis anyway.

Furthermore, they make a distinction between primary and secondary characters as if the salience of these representations makes a difference regarding perception. If that is the case and I am playing with my female character that I non-randomly chose, then my impression of in-game representation may be different than if I had chosen a male. Watching a strong female character progress through a game may have a great impact on the general perception of female game character representation for me. So again, I think it was inappropriate to include these particular characters in this specific study.

They made a distinction between primary (player-controlled) versus secondary (game-controlled, non-playable) characters. I see some problems with this. For example, the female voice (GLaDOS) is a non-playable character, but she appears throughout the entire game. She would have counted as a secondary character if she were human. But actually according to their system, she would not have counted at all because she is a non-human character, even though she’s present through the entire game. The reasoning behind excluding non-human characters was that people don’t connect with them as well. This eliminated almost half of their data set. So maybe we should be concerned about human representation in general in games or else people might get the impression that the world is overrun with aliens and robots. :)

They did not indicate what they consider a “child” in game: For example, Link in the popular Legend of Zelda games is a child or teenager depending on the story. Would they count him as a child? A teenager? I don’t know. How did they determine the ages of the characters? I assume they used the same categories as the census, but what did they do when there were ambiguous ages, such as Link’s?

Did they also consider that the reason children and the elderly are less represented is because the average gamer is neither? Although games may not represent the overall US population, they may represent the overall gamer population (something they did mention in their discussion). That is not necessarily good or bad – it depends on how you look at it. On the one hand, it may be part of the reason fewer children and elderly play. On he other hand, it may be that the gaming industry is catering to it’s most cash-giving demographic. Also, babies in general can’t do much so it might not occur to game developers to have many game heroes that are young children, reducing the available range of characters even though they apparently consider the entire category from the census in their comparison. They cite that television has similar representation, but I see the same issues with TV (children are usually genre-specific, Kid’s channel, etc – I don’t know what TV’s excuse for elderly representation is).

2) Games
They mention the game ratings system, but don’t really provide any analysis on how that would affect game exposure (and therefore their interpretation of demographic exposure and social perception). They do weight for sales so I guess it would even out, but not necessarily (as an adult in one household could play any game but they may limit what their children play). They do compare diversity according to game rating though, and it turns out that in the games they looked at, games rated E were more racially diverse. This could mean that certain demographics are exposed to…well…certain demographics.

They also didn’t provide much context for how a demographic was included in the games. Like I said, the main distinction was “primary” vs “secondary”. But if I’m playing a game with adorable babies doing adorable things that’s a little different than if the game is zombie babies that I’m shooting in the face. Not exactly representative of real babies. Now, Poop Hero, that would be another story… What if the characters are more often that not stereotypically represented. For example, if there are many women in games, but they are always big-breasted cartoons, then what does that mean for “diversity” and “representativeness”? I don’t even want to think how Natives are “represented”…

They didn’t mention if the games were created/developed only in the US. I’m not saying that makes in-game demographics a non-issue, but it might give a false impression of US game developers. It’s possible that games produced in other countries are more representative of that country or, if they were developed in another country specifically for the US, they might have based the demographics on what they assumed about the US. I think it would have been an interesting analysis to see if game demographics differed depending on country of origin. But that would probably have been too unwieldy.

The most popular games aren’t necessarily representative of all games, so where does the “problem” lie (if there is one) and what are the ways to fix things? It’s possible the medium as a whole is more representative of the population, but gamers in the US are drawn for some reason to games that are less diverse. By only considering US sales and US population, but not where the game was produced they haven’t convincingly demonstrated that games are misrepresentative of the general population.

A good try with some interesting information, but still lost of noise and some questionable method choices. Not that I necessarily fault them to that – you’d be amazed what you think is a good idea at the time and then after the study, you’re like “oh balls I should have done X” or “we should have eliminated Y”, etc. And no study is the perfect study, so I’m not convinced that this study alone is sufficient evidence for underlying social inequality in video games but I don’t deny that it’s possible. And I’m not clear on the significance of demographic representation in the grand scheme of things now that many popular games allow you to choose, and in some cases custom-develop (as with Miis), your own main characters.

Mostly my beef is with the rather craptacular summary in New Scientist which was typically too vague. A few clarifying sentences would have made all the difference. Ironically their vagueness made me go find out more about the study, but I don’t have time to do that all the time. I barely had time to do it this time (hence it coming way later than I promised).


One response to “Video Game Demographics – I finally read the article

  1. Pingback: g a m i n g     s t a t i s t i c s | about gamers