Impact of Community Feedback on User Generated Content: Empirical Evidence from an Online Community and Experiments

Abstract

How do contributors in online knowledge communities respond to peer feedback from the community? We study this question by empirically analyzing users’ contributions with data from StackOverflow.com, coupled with an experiment on Amazon Mechanical Turk (mturk.com). Based on an analysis of about 1.7 million data points (all relevant activity during a two-year period on this website), we find that users’ responses are positively linked to community feedback: more positive community feedback for a focal user in the previous period leads to higher contribution by that focal user in the subsequent period. The experiment on mTurk further helps establish a causal link between positive evaluations by the community and user contribution and then explores the underlying mechanisms driving this phenomenon. Our experimental evidence suggests that users who receive peer feedback (a) make internal and external attributions, which in turn (b) impact their emotional affects in terms of excitement, hope, and worry, which then (c) have an impact on their willingness to contribute. Interestingly, the experimental evidence suggests that users who receive more positive feedback, attribute this in equal measure internally to perceived self-efficacy and externally to perceived fairness, whereas users who receive negative feedback attribute it somewhat to lack of perceived self-efficacy, but much more to the lack of perceived fairness of community feedback. These findings have important implications for online platforms trying to leverage knowledge sharing by community members, as well as for researchers trying to better understand the drivers of online knowledge sharing.

Keywords: peer feedback, knowledge management, user-generated content, randomized experiment, panel data analysis