New Editor Experiences/Community selection

We chose the communities by first:

Selecting projects with between 300 and 3 000 monthly active editors.
Removing Meta and the English Wiktionary as out of scope.
Removing the Persian and Turkish Wikipedias because of the political situations in Iran and Turkey.
Removing projects with less than 1 000 interview prospects (editors who (1) had registered on that project in the past 90 days, (2) had given an email address, and (3) had made at least 5 edits (4) all of which geolocated to the country of interest.)

This left, in no particular orders, the following Wikipedia–country pairs:

Korean (South Korea)
Hebrew (Israel)
Ukrainian (Ukraine)
Dutch (Netherlands)
Swedish (Sweden)
Polish (Poland)
Czech (Czech Republic)
Chinese (Taiwan)
Portuguese (Brasil)
Italian (Italy)
Indonesian (Indonesia)

We then investigated these communities on a number of criteria, including the new editor retention rate, internet penentration, level of offline programmatic activity, cost of research, receptivity to WMF initiatives, and our pre-existing knowledge of those communities (the less we already knew, the more interesting a community would be).

Final selection

Taking those criteria into account, we selected the Korean (second choice: Chinese) and Czech (second choice: Ukrainian) Wikipedias.

Korean

Korean and Chinese were both similar in many respects: they had very high levels of internet access and economic development, low levels of offline programmatic activity, and high rates of new editor retention. In addition, we felt that, like other East Asian communities, these communities participated relatively little in the global movement, so we knew particularly little about them.

We chose Korean over Chinese because the Chinese Wikipedia seemed particularly unrepresentative of Wikipedias in general. In particular, its block in China is an unusual level of geopolitical tension, while its use of the language variants system to support a mix of Simplified and Traditional characters creates unusual technical complexity for new users and has unfortunately prevented the wiki from getting the visual editor for the foreseeable future.

Czech

Czech and Ukrainian were also quite similar: both had lower rates of editor retention, but high rates of offline programmatic activity. Moreover, there are many active communities in Eastern Europe which are relatively strongly linked to each other, so results from one such community may be unusually applicable. We chose the Czech over Ukrainian mainly because of its unusually low rate of editor retention (less than half that of Korean).

Other wikis

The reasons we decided against other communities were as follows:

Indonesian was a particularly interesting choice because of Indonesia's low internet penetration (22%), but access itself (as opposed to things like awareness, motivation, or experience) is not a central focus of this project. Moreover, its penetration is the lowest of the geographic base of any large or mid-sized wiki (including ones like Spanish, Arabic, Thai, and Vietnamese), so the lessons learned might not be generally applicable.
Portuguese was an early favorite, but the WMF has historically made Brasil the focus of a number of major projects (such as the Catalyst program), with disappointing results. We decided this history made it worth pursuing other options.
We didn't have strong objection to Italian, Polish, or Hebrew, but at the same time, we weren't particularly drawn to them either.