EXPRESS: Multi-Self Presentation and Knowledge Contribution in Platform Operations: Evidence from a Natural Experiment

Operations of online user-generated content (UGC) platforms often involve tackling under-provision , wherein users lack motivation to contribute sufficient content. We study a novel design deployed to address this issue: the integration of a proprietary social feature that allows users to post casual free-form content (e.g., casual social content) alongside standard formal content on the platform. Leveraging a quasi-experiment on a large Chinese online knowledge-sharing platform, we assess the effects of integrating such a social feature on users' formal knowledge contribution (i.e., writing answers to others' questions ). Using difference-in-differences (DiD) analyses paired with various matching strategies, such as propensity score matching and look-ahead propensity score matching, we find that users who adopt and frequently use the feature (i.e., high adopters) show a stronger incentive to contribute answers without compromising quality. These results are further validated by a set of robustness checks. Further mediation analysis suggests that such a complementary effect might be because of two mechanisms related to the integrated social feature's multi-self presentation function: First, the new feature increases high adopters' engagement on the platform and leads them to consume more knowledge content, which in turn facilitates knowledge contribution; Second, high adopters receive more social recognition benefits by building a more complete self through the integrated social feature, and such increased social recognition benefits further motivate more subsequent content contribution. Moreover, employing a text-mining approach to assess the semantic similarity between users' casual social content and their formal knowledge contributions, we find that the complementary effect is amplified for high adopters whose casual posts remain more closely connected to their formal knowledge domain (i.e., are more semantically aligned with their formal knowledge). Our results have novel implications for platform design and operations to motivate users to contribute valuable content.