YOUR FRIENDS LIKE IT, DO YOU? THE EFFECTS OF SOCIAL RECOMMENDATION SYSTEMS ON CONSUMER PREFERENCE HOMOGENEITY Seth Li, Elena Karahanna University of Georgia
Social Recommendation System My Facebook Friends 2
Research Question � What are the effects of social recommendation systems on consumer preference homogeneity? Social Recommendation Systems Consumer Preference Homogeneity Network Structural Characteristics 3
Methodology � Dataset: Last.fm, a music recommendation provider ◦ 1,892 users,12,717 user-friend relations � Independent variables: social network characteristics (generated using UCINET) � Dependent variable: Euclidean distance (measure of preference dissimilarity) 4
5
Hypotheses � Consumers differ in preferences � Heterogeneous influences Less similar to others (individual level) & less homogeneity (cluster level) Hypothesis Depiction Description Result Degree centrality (-) H1a Supported homogeneity Betweenness centrality (-) H2 Supported homogeneity Closeness centrality (-) H3 Supported homogeneity 6
Hypotheses � Consumer has limited cognitive capability: Too many friends influenced by only a few � More connections information redundancy Hypothesis Depiction Description Result Degree centrality (+) H1b Supported influence heterogeneity A C Cluster density (+) H4 Supported homogeneity B D 7
Summary of Findings Group Level Individual Level Heterogenize Heterogenize Social - As centrality increases Recommendation - As centrality increases Homogenize - Effects get weaker System - Consumers selectively choose - As density increases recommendations 8 8
Recommend
More recommend
