"Edward Castronova":http://business.fullerton.edu/ecastronova/ has written a new paper about the economics of massively multiplayer online games: "The Price of 'Man' and 'Woman': A Hedonic Pricing Model of Avatar Attributes in a Synthethic World":http://papers.ssrn.com/sol3/papers.cfm?abstract_id=415043# uses the online auction market for gaming characters to explore the characteristics that drive real-world prices for fantasy creations.
There's some interesting stuff in Catronova's paper, the key finding of which is that avatar level is the strongest driver of auction price. This makes sense -- after all, who wants to buy a level 5 hobbit? (As an aside, it's rather fun to see RPG-speak rendered in full-on academic language: "Hacking monsters, winning loot, and leveling up are fun," becomes, "By interacting with others, and the environment, the avatar can attain significant new skills, abilities, cash, and items. All of these things help the player enjoy the world and improve its contribution to the player's sense of wellbeing.")
Some notion of character gender plays a role too, however: "All else equal, a female avatar sells for about $41 less than a male avatar, which is about 12 percent of the mean price of $333. It is also comparable to the effect of an avatar level: all else equal, each increase in level adds about $33 to the avatar price" (33). Castronova suggests very reasonably that this difference could be attributed to the male majority of players simply preferring male avatars. While cross-gender playing does take place, 87% to 89% of players' _main_ characters match their gender -- this would underscore the suggestion that when players _do_ switch genders, they do it for experimentation, not for long-term play, so they chose highly sexualized classes and races to make the most of their transgendered experience. Overall, however, Castronova's approach to gender seems kind of under-socialized, hinting at the importing of gender stereotypes from the real world but not getting into the implications of that possibility.
The paper leaves a question hanging: The marketplace for online avatars makes buyer and seller information really important. While Castronova cannot access detailed information, he could access information on the volume and type of objects bought and sold by each. That information would greatly enhance his ability to characterize auctions: Auctions could be identified according to the overall activity of their buyers and sellers. I have a feeling that would reveal something interesting on the subject of avatar gender.