Women and Wikipedia
New research, WikiChix
WikiChix group photo: every year women involved in Wikimedia projects come together for lunch at Wikimania, generating passionate and lively discussion about women's roles within Wikimedia; see "Brief news" below.
A paper titled "WP:Clubhouse? An exploration of Wikipedia’s gender imbalance", to be presented next month at WikiSym 2011 by a team from the University of Minnesota, was posted online on August 11. The team of seven researchers became interested in the imbalance after the January 31 New York Times front-page article on Wikipedia's gender gap (see earlier Signpost coverage: January 31, February 7) and sought a more data-driven analysis of the issue, as opposed to the by now traditional "here is a random 'male'-article, here is a 'female'-article, they are different lengths" approach.
Accompanied by a press release and audio/video summaries from the university, the paper has been widely covered by external media sources—see In the news.
The study confined itself to editors who self-disclosed their gender via a userbox on their user pages or through their user preferences. As the paper notes, this may have introduced a bias, and the gender as self-reported by users (and in particular vandal accounts) may not always reflect the truth.
- 16% of Wikipedia accounts created in 2009 (who state their gender) belong to women.
- 9% of edits from the 2009 cohort are made by accounts belonging to women. A chart plotting gender ration over edit count (similar to one published previously in the Signpost: "Wikipedia's gender gap examined further") further shows that "the gender gap is more pronounced when looking at high-activity editors".
- The gender gap has not changed significantly over the past five years.
- There were significant differences in gender ratio by subject area,
||Percentage of women editing
- Articles where the percentage of female contributors was higher than average tended to be shorter than articles where the percentage of male contributors was higher than average; articles where the gender ratio was closest to the site average were the longest.
- The hypothesis that the "coverage of topics with particular interest to females is inferior to topics with particular interest to males" was also confirmed in a second test – the only one in the paper that did not rely on Wikipedians' self-reported gender: using data on the gender of movie raters on their own MovieLens site, the researchers found that when controlling for some other factors, "articles about 'female' movies are shorter than ones about 'male' movies", and also received a lower quality rating in WikiProject Film's article assessment.
- Women are more likely to edit userpages and user talk pages than men, and men are more likely to edit articles and all other namespaces.
- Women are more likely to seek adoption in the adopt-a-user program.
- Women are more likely to become administrators than men with similar edit counts.
- Women are more likely to vandalize as new users (60% of vandal accounts reporting their gender were women).
- Women are more likely to be reverted when they have very few edits (7% vs 5%); however, in accounts with more than eight edits, the effect disappears.
- Women are no more likely to leave after being reverted than men.
- Women and men were blocked at essentially the same rate (4.39% of female users and 4.52% of male users have ever been blocked).
- WikiChix at Wikimania: A write-up of the WikiChix lunch at Wikimania has been published at outreach:WikiChix Lunch 2011 by Wikimedian Sarah Stierch, who also blogged about an "epiphany" she had at the event: "Sometimes all it takes is an invitation".
- Gender-gap mailing list stagnating: On the Wikimedia Gender gap mailing list inaugurated in February, there were 105 postings in May, 46 in June, 36 in July, and only 5 so far this month. Interested users are encouraged to become involved.
- New tool analyzes article contributors' gender and location: A new tool called "Wiki Trip" displays the gender of the contributors to a specific Wikipedia article, as well as their geographic provenance – restricted to those edits where such information is available, as in the University of Minnesota study. Examples: Friendship bracelets, used in the article in The New York Times mentioned above, as an example of a poorly developed article on a "teenage girls'" topic, has received 100% male edits, whereas in baseball cards, one of the "lengthy articles on something boys might favor", 6% of the editors are estimated to have been female. A few other results: NASCAR 98% male, feminism 91% male, Twilight (series) 55% female.
Check back for the next Signpost on December 1.