Based on results of the recent elections in the US, I thought I would try some comparisons with our own elections, to get a better idea of the effective size of the Wikimedia electorate. What I did was to take the elections results for cities in my general area, since I can find those very quickly. I chose those cities where the number of ballots cast in a contested local race (usually for city council) was closest to the number of ballots cast in two of our elections (the Board of Trustees election earlier this year, and the Arbitration Committee election last December on the English Wikipedia). I figured that the population of these cities would be a reasonable estimate of the effective size of our community. Naturally, this is an impressionistic rather than a rigorous statistical analysis.
At the time of the Arbitration Committee election, the number of ballots cast on the English Wikipedia was about the same as several cities with populations ranging roughly from 3,000-6,000 people. The upper end of this range is quite close to the number of people who edited five times or more last December, according to Erik Zachte's statistical reports. So it seems this may be the number that is closest to a real-world population for our community. Since those statistics have just been updated (thanks, Erik) we can see that the English Wikipedia is now nearly equivalent to a town of 15,000 people.
The turnout for the Board of Trustees election was comparable to a real-world population of about 7,000-11,000 people. It's a little harder to find an equivalent number in our ongoing statistics, in part because project-wide participation is difficult to capture (<cough>single login</cough>). Even the number of users with five or more edits that month on just the Wikipedias collectively was significantly larger than this.
--Michael Snow