Research talk:Onboarding new Wikipedians/Rollout/Work log/2014-03-12

Wednesday, March 12th[edit]

Initial results from the analysis today.

The proportions of productive new editors before and after the deployment of mw:Extension:GettingStarted are plotted with (normal approx) standard error bars.

Difference in productive editor proportion. The proportions of productive new editors before and after the deployment of mw:Extension:GettingStarted are plotted with (normal approx) standard error bars.

Some wikis get more productive newcomers, some get less. It doesn't look like there's a clear pattern. What if I plot the difference in proportions.

The difference in the proportions of productive new editors before and after the deployment of mw:Extension:GettingStarted is plotted with 95% CI error bars.

Delta productive editor proportion. The difference in the proportions of productive new editors before and after the deployment of mw:Extension:GettingStarted is plotted with 95% CI error bars.

The above plot is sorted by the difference in proportions. Again, no clear trend here. Some wikis went up, other wikis went down. Worse, it doesn't look like the presence of suggestions was meaningful. --Halfak (WMF) (talk) 19:19, 12 March 2014 (UTC)[reply]

I like this strategy for observing the differences cross-wiki. My next job is to generate similar plots for scalar values that lend themselves to geometric means -- like revisions, sessions and sessions seconds. That's going to be difficult using the standard tests since the error around such values are measured in log space. --Halfak (WMF) (talk) 01:56, 13 March 2014 (UTC)[reply]