Social Learning, Social Influence, and Projection Bias: A Caution on Inferences Based on Proxy Reporting of Peer Behavior

This study explores the consequences of conflating social learning and social influence concepts and of the widespread use of proxy-reported behavioral data for accurate understanding of learning from others. Our empirical analysis suggests that proxy reporting is more accurate for new innovations, about which social learning is more plausible, than for mature technologies. Furthermore, proxy-reporting errors are correlated with respondent attributes, suggesting projection bias. Self- and proxy-reported variables generate different regression results, raising questions about inferences based on error-prone, proxy-reported peer behaviors. Self-reported peer behavior consistently exhibits statistically insignificant effects on network members' adoption behavior, suggesting an absence of social effects. (c) 2010 by The University of Chicago. All rights reserved..