User:The Vintage Feminist/Antin et al graphics

"Gender Differences in Wikipedia Editing" by Antin et al, describes the experiences of editors in the first three weeks of editing. This essay converts the results tables in their report into a set of at-a-glance graphics to provide a user-friendly way of considering their findings.

It strongly recommended, therefore, that this essay be read in conjuction with their report.

Background
The report authors were interested in whether gender stereotyping rather than "active sexism" was the reason for the discrepency in the number of female editors compared with male editors on Wikipedia.

Gendered Wiki-work
The report considers the different types of "Wiki-work" that exists, from uploading photos to arbitration. It suggests that it is essential to discover which types of work draw which gender, what "are the real or imagined characteristics of tasks that encourage certain individuals to gravitate towards them", and to what extent stereotypes play a part. In order to grow Wikipedia's user base and encourage diversity, the report authors say that, researching wiki-work from a gender perspective "could allow recruitment and educational efforts to be more focused..." giving "potential contributors more specific ideas of the types of work they might like to do.

Base population
Between 9 September 2010 and 14 February 2011, 256,190 users created an account on the English-language Wikipedia. Of these 13,598 optionally declared a gender in their Wikipedia profile (blue segment, figure 1, right).

The 13,598 gender-declaring Wikipedians were made up of, 11,194 (82%) men, and 2,402 (18%) women (figure 2, left). A ratio that, the report authors' note, is higher than the one quoted in the 2010 UNU-MERIT study (13% women).

Sample 500 editors
A sample of 500 users was then extracted from the gender-declaring Wikipedians (figure 3).

 How the four quartiles were formed: The breakdown of the 500 editors 

The 500 editors were then divided into four quartiles based on how prolific their revisions (edits) were during their first three weeks on Wikipedia. The four quartiles were made up of editors who had made 0, 1, 2, or 4+ revisions (figure 4a).

There were three weeks of data unavailable for 63 editors (figure 4b).

This left an adjusted number of sample editors of 437 (figure 4c).

Gender break down of the top and bottom quartiles

 * Bottom 75% of prolific editors (figures 6 and 7)


 * As this group produces 0 edits, no 'type of revision' data is generated. Therefore, there is no gender breakdown possible for the report's tables (illustrated in figures 9, 10 and 11 below). Put more simply, if someone one makes 0 number of edits in three weeks, there is nothing to analyze.


 * Top 25% of prolific editors (figure 7)

Note: Antin et al do not explain why the gender split in the top group of editors is not even as promised in the method on page 12, where they stated (emphasis added): "We [...] randomly sampled an equal number of men and women from each quartile to build a total sample of 500 editors." The editors for whom data was unavailable only allows for one more editor (see figure 4b).

Coding
The report authors decided that there had to be a consensus of two-thirds of coders on a revision type (e.g. add citation) before the classification was accepted.

Top 25% of prolific editors by gender

 * The gender split for the number of revisions in this quartile was 27% female to 73% male.

Discussion and conclusion

 * Male Wikipedia editors drastically outnumber female editors overall, however the report found that 18% of editors female, compared with the 2010 UNU-MERIT study which stated only 13% of editors were female.
 * In the bottom three quartiles of Wikipedians, men and women made similar numbers of revisions in nearly every category of Wiki-work (fig. 9).
 * In the top 25% of Wikipedians, there is evidence that only 27% of revisions were completed by women (fig. 12).
 * In the bottom 75% of Wikipedians, women appeared to make larger revisions in many categories (fig. 11), however, only 9% of revisions were made by editors in the bottom 75%, so this may not be statistically significant.
 * In the top 25% of Wikipedians, women tended to make more sizeable revisions, particularly in the "Add New Content" and "Rephrase Existing Text" categories (fig. 14).
 * We did not see evidence that men and women are attracted to different types of editing work.
 * Two areas of work, where women made significantly larger revisions, involved creative production, synthesis, and reorganization of text, suggesting that, compared to men, women often develop more successful solutions to R&D problems posed through innovation brokers such as Innocentive.
 * Despite limitations (for example, sample size, and the inability to make claims about patterns over time) this analysis suggests that the story of Wikipedia's "Gender Gap" is perhaps not as straightforward as initial reports have suggested.
 * The results are encouraging in respect that, there was significant gender parity in number of revisions, and women are better represented than the UNU-MERIT study would have led us to believe.
 * Either Wikipedia has been successful in attracting more women, or there may be inaccuracies in prior studies.
 * The results are discouraging in respect that, there are far fewer women editing Wikipedia than men, and there was a particular gender skew in revision quantity among the Wikipedians who do most of the work.
 * Problematically, the most active Wikipedians are also those who largely set policies, arbitrate disputes, and do other high-level tasks into which biases of worldview and temperament can subtly creep. This is the very group among which women may need more representation, but also one which may be more difficult to break into.