Do online communities benefit from appointing volunteer moderators? Evidence from a regression discontinuity design

We study whether online communities can benefit from appointing users to "volunteer moderators", in terms of community members or leaders who are supposed to encourage participation and plant the seeds of community. We exploit a quasi-experiment on Stack Exchange, a network of more than 160 online communities, over the period from 2010 to 2017. These communities regularly hold democratic elections of moderators. By focusing on elections decided by a narrow vote margin, we can exploit a regression discontinuity that yields a quasi-random assignment of moderatorship. We find that online communities can significantly benefit from volunteer moderatorship: closely elected moderators contribute not only more content, they also provide more extensive content to the community. Furthermore, the direct outcomes of moderatorship appear to have positive spillover effects on community discourse: moderators spark more extensive discussions, which are rated higher and bookmarked more often.