2011/2/11 Samuel Klein meta.sj@gmail.com:
could this be on ongoing survey, run in a way to target a low randomized % of visitors continuously over time?
Oliver is right that this kind of continuous data-gathering is non-trivial, but there are some great examples to learn from.
My favorite (and I mention it at every opportunity) is the Mozilla Input program. It's built into their Firefox 4 betas, and it gives users a continuous opportunity to report whether Firefox made them happy or sad, in 160 characters that go into a massive text mining database at: http://input.mozilla.com/en-US/beta/
Here's the UI they use in the browser: http://www.mozilla.com/img/firefox/beta/4/feedback-button.png
Imagine this integrated into the Wikimedia UI, and associated with anonymized user info, including gender for those willing to report it. That could be a really useful way to gather tons of data continuously about factors deterring people from contributing, including factors that relate directly to the gender gap. (I personally believe that the value of free-text feedback is greatly underestimated in surveying communities. The Mozilla Input site showcases some great approaches to making sense of tens of thousands of text messages.)