Disclaimer: I agree there is a real social problem here. The rest of this message is solely about statistics (inspired by the language of statistics used in the article, like "A/B Test," "statistical significance," and "99.9% confidence level"). The article opens with an analysis of messages to one male and one (self-selected) female from a community which has a large male majority. Most people in the world are heter…
I think the question here is "will gamers harass men or women more". In this case you want your community to be representative of the community at large, so this gender imbalance makes sense. Your point makes sense, but seems irrelevant to the discussion here.
In fact, the last time we had a discussion about sexism here, someone posted a study to me claiming that it's often confounding community values that create a hotbed of apparent sexism.[1]
So if we're talking about the situation where what we see is a normal (and consistent at large with society) rate of men harassing women, we should be asking what we can do to fix the other parts of the community that enable these people to be unduly felt, such as not properly spam filtering messages between users. Why is it, for instance, I can tell my email client to never display a message with the word 'cunt' in it, except for whitelisted users, but I can't do this when trying to interact with people in public, such as reddit or Kongregate?
I believe what we see time and again when we analyze the numbers is not that we're seeing an inherently higher rate of sexism, to which we need to do something to fix the attitudes of men in general, but rather that we're seeing the undue influence of the usual rate of problematic actors - that is, it's way easier to bother a lot of people as a troll and dominate the conversation online than in person, and its this difference that is the cause of the problems, not the background rate of problematic actors.
[1] http://firstmonday.org/ojs/index.php/fm/article/view/4291/33...