The math behind that little statistic was so terrible I had to write a blog post about it.
http://blog.mvolz.com/2014/08/what-percentage-of-wikipedia-editors-are-mums/ First off, in their blog post, Andreas & Collida multiply the percentage of contributor respondents who were women (12.64%) by the percentage of all respondents (contributor and reader, male and female) who were parents- 14.72%- while seemingly missing that the study in fact provided a breakdown of this: 13.7% of all female respondents were parents. (15.1% of the male respondents were). Secondly, Andreas & Collida cherry pick a lower bound number for women contributors (8.5%) (source unkown) and presented the number from the survey (12.64%) as an upper bound. A literature search gave me an upper bound of 16.1% from Hill & Shaw. Furthermore, the source Andreas & Collida used contained biased statistics. The original WMF/UNU-MERIT report had no methods section and didn’t control for sampling bias. The Hill & Shaw paper controls for sample bias based on a survey by Pew, which used better sampling methods. Hill & Shaw tried to control for the survey’s selection bias and found that they “estimate that females, married people, and individuals with children were underrepresented in the WMF/UNU-MERIT sample while immigrants and students were overrepresented.” This means that the two statistics Andreas & Collida chose to multiply together; female editors/contributors and males and females with children- were *both* underestimates in the WMF/UNU-MERIT survey. Hill & Shaw provide the adjusted numbers for these accordingly; they estimate that 16.1% of contributors (as opposed to 12.64%) are female, and that 25.3% have children. We can perform a similar analysis as Andreas & Collida using those adjusted numbers by multiplying them, a result of about 4.1%- more than double their highest estimate. Of course, this number is also flawed; we don’t have the actual breakdown of what percentage of female contributors have children, and instead are multiplying aggregate numbers. A better estimate could be obtained by redoing Hill & Shaw‘s analysis on the raw dataset. On Tue, Aug 26, 2014 at 9:14 PM, Tim Davenport <shoehu...@gmail.com> wrote: > There is a new blog post up on Wikipedia-criticism site Wikipediocracy that > should be of interest to this list. > > Andreas Kolbe with Nathalie Collida, "Why Women Have No Time For Wikipedia: > Thoughts on the Online Encyclopedia's Gender Imbalance." > > http://wikipediocracy.com/2014/08/26/why-women-have-no-time-for-wikipedia/ > > One interesting assertion made by the authors in their lengthy essay is that > fewer than 1 in 50 WP contributors is a mother: > > "It is sometimes argued that women simply have less time to contribute to > Wikipedia, due to family commitments. This is a fallacy. Firstly, the United > Nations University survey found that only 33.29% of respondents had a > partner, and only 14.72% had children. The difference between readers and > contributors was negligible here, and the survey report did not indicate any > difference in these percentages for male and female respondents. It is > patently obvious that girls and women in the age groups that are most > strongly represented in Wikipedia’s demographics typically do not yet have > families of their own. Their lack of participation is unrelated to their > being bogged down by family responsibilities. > > "Of course, these figures also tell us something else: if only 14.72% of > contributors have children, and the percentage of female contributors lies > somewhere between 8.5% and 12.64%, then it looks like only 1.25%–1.86% of > Wikipedia contributors are mothers. > > "That is less than 1 in 50." > > > Tim Davenport > "Carrite" on WP /// "Randy from Boise" on WPO > > > > _______________________________________________ > Gendergap mailing list > Gendergap@lists.wikimedia.org > https://lists.wikimedia.org/mailman/listinfo/gendergap > _______________________________________________ Gendergap mailing list Gendergap@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/gendergap