Williams and his team found that male characters are “vastly more likely to appear” in games than females. They made up 85 per cent of characters, compared to 51 per cent of the real population.
The paper notes that this gap is to some extent driven by the popularity of sports games (none of the sports titles sampled had female athletes), and adds that the figures are closer to the demographics of game developers (a little over 10% female) than game players (a little under 40%). Player characters are even more likely to be male (89.55%) – almost identical to the proportion of male game developers.
Compared to the real population, African Americans were under-represented by 13 per cent and Hispanic/Latino people by 78 per cent. Asians were over-represented by 25 per cent and white people by 7 per cent.
The incidence of black characters, on the other hand, would be much lower if not for sports games. Native Americans and biracial people are also underrepresented, although I’m not sure if that’s statistically significant. When only player-characters are considered, every group declines (compared to all characters) except whites and biracial people.
Elderly people and pre-teen children are also underrepresented compared to population – I’ll leave that one as an exercise for the reader.
I note that the sampling is heavily skewed towards console-based games, and covers 2005-2006, thus predating both the Wii and much of the growth of casual gaming; of MMORPGs, only WoW was sampled. (Those 150 games commanded 95% of total game sales, however, so it’s not going to be a big skew. Also, the most popular games are less representative than the typical game produced by developers, indicating that players also play a role in the cycle of creation and consumption.)