People You May Know: Friend Recommendation, Network Formation, and Online Community Participation

Friend recommendation has become a standard design in social media platforms over the years, however, its behavioral consequences remain empirically unaddressed. In this paper, we report on a large-scale randomized field experiment carried out on KWAI. Our findings reveal that friend recommendation affects the user’s egocentric network formation. It helps users build larger, more reciprocal, denser, and more diversified egocentric networks, and the effects accumulate over time. On the other hand, friend recommendation motivates users to be more active, engaging more in content contribution, content consumption, and social interaction. We leverage the causal forest method and find younger users, male users, users from big cities, users from urban areas, and users who are at the beginning of building social networks respond more favorably to friend recommendations. Our heterogeneous treatment analysis reveals critical insights for social media platforms in customizing friend recommendation functions for different user groups