Research talk:HTTPS Transition and Article Censorship

Consider using pageview ratios instead of absolute numbers

Great research project! We already did a parallel investigation in one case in August - I briefly outlined it in an email to Ellery in early October, but he didn't have access to the internal Phabricator ticket with the details, so I'm now posting the charts used back then here for illustration.

To mitigate confounding factors, i.e. other causes that increased traffic to these articles (which were particularly relevant in the Turkish case, because the local community had highlighted the blocking of those articles in a site banner, cf. [1] and [2]), I used the ratio of pageviews from the blocking country to the overall pageviews, instead of the overall number. This helped to verify that there was indeed some *partial* blocking (some ISPs but not others) going on in that case.

Regards, Tbayer (WMF) (talk) 21:56, 19 November 2015 (UTC)

Related research project

Research:Analyzing Accessibility of Wikipedia Projects Around the World will presumably focus on project-level censorship, but I understand the team is interested in looking at article-level data too. Regards, Tbayer (WMF) (talk) 18:23, 19 May 2016 (UTC)