Gamasutra: The Art & Business of Making Gamesspacer
View All     RSS
October 20, 2014
arrowPress Releases
October 20, 2014
PR Newswire
View All
View All     Submit Event

If you enjoy reading this site, you might also want to check out these UBM Tech sites:

The Gender Cocktail. Part II: Big chests and little waists?
by Dmitri Williams on 04/01/14 05:52:00 pm   Expert Blogs

The following blog post, unless otherwise noted, was written by a member of Gamasutra’s community.
The thoughts and opinions expressed are those of the writer and not Gamasutra or its parent company.


Part I of this looked at gender representations and found that they were heavily skewed male. In this installment, I’ll share the visuals for some context. We now know that females are underrepresented, but when they do show up, how are they represented?

The expected answer would be that they are hypersexualized. Big breasts, tiny waists. Well, let’s see. And while we’re at it, why not also take a look at the male characters? The research on body shapes is large and consistent and I’ll summarize: if you look at hyper-sexualized versions of your gender, you feel like crap, and it doesn’t matter what gender you are.

OK, so how can we compare game body shapes to real ones? This is pioneering work done by my colleague Nicole Martins at Indiana University. The full papers are here for females and here for males. It turns out that there are databases of “real” body shapes out there, primarily made for the clothing industry. We used the CAESAR dataset. These have 3D data on hips, busts, leg lengths and dozens of other body parts. Using these data we used Poser to create a “typical” female and male shape as our baseline.

Then, using the same sample of characters from the first post, we took screen captures of the characters, imported them into Photoshop, measured their dimensions and then created a composite Poser figure for them.

What we found for female characters surprised us. First, no big breasts. No, not a typo and not an error. The numbers do not lie. The CAESAR dataset shows an average chest circumference of 38 inches and the composite game character had an average of 32 inches.

One other thing we didn’t expect were giant heads, but in retrospect it makes a lot of sense. Games don’t all have a zillion pixels of detail and to show human expression the heads often end up disproportionately large.

We dug a little deeper and split the results into high-res and low-res games to see if this head thing continued, and to see if maybe in the more detailed versions we’d find our digital vixens. We found the opposite. When we looked at high res, the average bust size was 29 inches and low res, 38 inches. The latter is “realistic.”

Then we took a look at the men in the same approach, and this time with slightly better rendering software.

And again, freakishly large heads, likely for the same reason. What would be a test of hypermasculinity? Big chests apply here, too. And we found the same pattern. The real-world baseline is 40 inches of chest circumference, and the game characters had an average of about 43 inches. Bigger, sure, but not herculean. When we split this into high-res and low-res, the differences dropped out entirely for the high res versions. The low-res was 45 inches, making us wonder if this was a graphical issue or really about large, caveman dudes. We dug further into the ratings and found that for the children’s games (by ESRB rating), the average chest size was 46 inches. It’s possible that this is hypersexualization, but also possible that in low-res situations this is a simple technique for making clear the character’s maleness.

What’s the bottom line? For male characters, everyone looks pretty realistic except for in children’s games where they are he-men. That’s potentially a damaging stereotype for young males who, after all, are going to grow up to be smaller-chested on average than these characters. (I’m letting my son stick with Stevie in Minecraft…).

What do we make of all of these stats? First, it’s clear that females have been under-represented. Second, at least for the year we looked at, females were not hypersexualized as we expected. Third, it’s male characters--in children’s games--that are actually hypersexualized.

There’s an argument to be made here that rather than there being anything sinister at play, games are simply made by and for a group (males). Give any creator license to choose their creation and they pick content that’s of interest to them. Dudes make games for dudes. Thus, the cycle of female representation can only change if women make games in greater numbers. Here are the data on women in the industry from IGDA numbers (2004 and 2009) and Game Developer (2012):

2004: 93% male/7% female

2009: 86% male/14% female

2012: No overall numbers, but by job function

Producer: 77% male/23% female

Programmer: 96% male/4% female

QA (the gateway job): 93% male/7% female

Given these numbers, we’d expect to see little change in characters, but maybe there’s some hope in customizability? Or maybe that’s spitting in the wind. Still, given how many females play, you don’t need to be socially progressive to consider adding more balanced character content. It makes good business sense.

And, truth be told, as a father of a daughter, I’d like to see my kid controlling some badass chics blowing away aliens on the way to save her hapless boyfriend.


Related Jobs

Filament Games LLC
Filament Games LLC — Madison, Wisconsin, United States

Quality Assurance Associate
Filament Games LLC
Filament Games LLC — Madison, Wisconsin, United States

Game Engineer II
Filament Games LLC
Filament Games LLC — Madison, Wisconsin, United States

Game Engineer I
InnoGames GmbH
InnoGames GmbH — Hamburg, Germany

Team Lead Online Marketing - TV (m/f)


Matthew Fundaun
profile image
It's very interesting to read through this, Dimitri. Right off the bat, I'd like to thank you for bringing methodical study and analysis into the video game community, and helping us get an objective a viewpoint as possible.

To ask a few questions, to get a better read on these results: This study purely covers body type and dimensions, yes? Not portrayal, clothing, or anything else?

The datapoints on the ratio of male to female programmers are unnerving, yes. The teacher in my programming class has actually brought it up before, and she showed us a graph stating that the ratio has only been getting worse for the past years.

(On another note: The link to the male study isn't working.)

Ian Richard
profile image
That was my thought too.

I'd hesitate to draw any conclusions of hypersexualization from any data that was only counting approximate body shape.

While this is an interesting read and fairly surprising to me... I think we'd need to consider more variables before I'd buy into any conclusions.

Christian Nutt
profile image
Portrayal and context is incredibly important, because as has been widely reported, the physique of male and female game characters is often exaggerated, but the implied message (which can be boiled down incredibly simply in many cases to something like "she is sexy" vs. "you are this badass") is very different.

Nicole Martins
profile image
Hi everyone! Sorry I am late to this party. You all are correct in that we only looked at body types in this study. Thus, we cannot make claims about hypersexualization. We actually measured some aspects of hypersexualization, but did not report the findings in this paper because I wanted the focus to remain on the body.

But for those of you who are interested in this topic, check out the work by Paul Downs and Stacy Smith. They have a content analysis published in Sex Roles that examined the hypersexuality of video game characters. As you might have guessed, female characters were significantly underrepresented as compared to their male counterparts. When females were shown, they were more likely to be shown partially nude, have unattainable body proportions, and highly sexualized.

Thanks for bringing attention to our work, Dmitri!

Dmitri Williams
profile image
I noticed that the link to the male paper isn't working. Here's a direct link from DropBox:

Matthew, Ian and Christian, I agree that context is important and a large sample study like doesn't offer context. It's really just a broad representation of the dimensions, and all of the breakdowns we did we for high vs. low res and by ESRB rating. My takeaway point was really just the surprise at the overall patterns, which were not what I expected.

Prof. Martins has done more work on this area and may have some more contextual data.