Wikimedia Blog/Drafts/Analyzing user participation with WikiMetrics



Analyzing user participation with WikiMetrics


As a project coordinator on Wiki Loves Monuments projects in France, I'm interested in evaluation of the project on different axis:

  • Content (creation, reuse among Wikimedia projects, etc.)
  • Users (participation, recruitement, motivation, etc.)

We have been tracking the following metrics:Total Images uploaded (Content creation), Total Uploaders (User participation), Number of images used on French Wikipedia (Content reuse), Number of articles on French Wikipedia using images from the content (Content reuse), The percentage of cultural heritage monuments already coverd (Content), Trends between thoses metrics (since early 2013). In order to do that we have had to use different tools such as GLAMorous <>, databases such as <> or even the live participation page <> (based on toolserver tools)

However it was hard to grasp how Wiki Loves Monuments impacted the project in the long run because our metrics were only based on the contributions during the project and not based on the users (participants to the contest). Finally with WikiMetrics we have been able to track a little bit deeper the impact of the contest*. As WikiMetrics is participant centric (based on cohort <>), it allows to go deeper in the analysis.

Before/After Wiki Loves Monuments 2011 in France Study with WikiMetrics

The cohort (or group of user we are going to analyze): The 988 users that have participated to Wiki Loves Monuments 2011 in France.

Time periods: We have chosen two periods of 6 months before and after (Jan-Jun 2011 and 2012) the event in order to compare and get more insight on the impact of the contest (note that the periods chosen may changes the result and introduce a bias, however it is not the point here).

Cohort aggregated results (directly from WikiMetrics): The Wiki Loves Monuments 2011 in France particpants did edit the Wikimedia Commons content ("File:" namespace) 1.7 times more after the event than before. It is interesting to see how an event can impact the engagement of a group.

Individuals analysis: We have noticed that among the users who did not edit during the chosen period before Wiki Loves Monuments 2011, 53 have continued to edit, and 10 are making more than 10 edits per month on Wikimedia Commons. This analysis needed to write a small Python code to analyze the JSON reports <>,