One of the key questions about the release of VisualEditor was what effect it would have on newcomer activities. While the visual editor was designed to make editing easier and therefore boost the number of contributions to Wikipedia, it was thought possible that the new feature would also enable a greater number of damaging edits.
In this early study of VE, we found that newcomers with VE enabled by default did not get reverted or blocked more often than those in the control. However, we also saw a significant loss in productivity for newcomers with VE enabled. We concluded that the reason for the loss in productivity was likely due to the early state of the VE software at the time of the experiment.
Given substantial improvements to the VE software, we re-ran a similar experiment. We found that newcomers with VE enabled by default were reverted significantly less and blocked slightly less often than newcomers with VE disabled. Unlike the previous test, we found no significant difference in the productivity or survival of newcomers between conditions. However, we did find that editors spent more time making edits in VE than in wikitext and suspect that this was due to time spent exploring the new editor.
A follow-up to the May 2015 study which used a larger sample size and a longer study period (60 days after registration). It also found that the visual editor had no effect on newcomer survival or productivity. However, it did find that newcomers in the VE group were reverted somewhat less often than those in the control group.
- August 2015 WMF Metrics Meeting report
- At the WMF Metrics and activities meetings/2015-08, there was a presentation by Halfak about VisualEditor AB test and usability results (see also YouTube video):