Hi! I have a lot of Wikimania Scholarship statistics - obviously what is in the report on wiki only highlights a piece of them. As always, comments on meta reports are always helpful (there are none).
There are some gaps in data: I (/WMF) don't have the registration information from year to year, so I can't confirm whether or not someone has registered for and attended Wikimania outside of a scholarship. What I did just quickly look up is our percentage of repeat WMF scholars from 2012 and 2013:
- 85% of 2013 Scholars did not receive scholarship in 2012 or received a scholarship but were unable to attend in 2012
- 5% of 2013 Scholars received a partial scholarship in 2012
- 10% of 2013 scholars were also scholars in 2012
Those are for WMF scholarships. It is important to note that about 40-50% of attendees at Wikimania who arrive on scholarships are actually sponsored directly by chapter rather than WMF. It is possible that this group of people are repeat attendees/scholars (I don't know). It is also true that many chapters send board representatives and/or staff to Wikimania. Again, this may contribute to the feeling that the same people are always attending. (Note: the same is true for WMF and WMF board.)
As is evident in the selection criteria the scholarship committee puts forth, contributions on our wiki projects is the key component to receiving a scholarship. The scores are so close, it is really difficult (impossible?) to receive a scholarship from WMF without having contributions on wiki. The committee also tries to look at someone's contributions in relation to his/her local-wiki context. One specific example of this is a former scholar from the Kyrgyz Wikipedia. On first glance, it looked like her aggregate edit count was low, but on further digging the committee realized she had only been editing for a year, and was already a top 5 contributor on that wiki!
I have lots of comments on the various topics that are getting throw around -- partial scholarships, needs-based scholarships, disclosing of scholars names, data collection ... but I don't feel this is the best forum for discussion. If someone has a wiki page with these topics sectioned off, we should tackle a few together!