Search
Program Calendar
Browse By Day
Browse By Time
Browse By Person
Browse By Room
Browse By Division
Browse By Session Type
Search Tips
Personal Schedule
Sign In
X (Twitter)
Theoretical and empirical approaches to the design of effective messages to increase healthy and reduce risky behavior have shown only incremental progress. Recommendation systems as employed by companies like Amazon and Netflix, utilize user-based collaborative filtering (UBCF) which holds the potential to leapfrog conventional models of message research and achieve individual-specific tailoring at scale. This paper reports the effectiveness of statistical “cloning” and “contra-cloning”, an innovative test of the applicability of UBCF, in the context of anti-smoking PSAs. With a nationally representative sample of current smokers and an archive of professionally produced PSAs, “clones”/“contra-clones”—other participants sharing “similar”/“dissimilar” rating patterns with the target individual—were found to positively/negatively predict the target individual’s evaluation of a PSA unseen, even after controlling for individual characteristics and the confounding of mere aggregation. The UBCF approach, and recommendation systems broadly, have important implications for health communication research in general and message tailoring in specific.
Sijia Yang, University of Pennsylvania
Poong Oh, U of Southern California - Annenberg School for Communication
Hyun Suk Kim, University of Pennsylvania
Minji Kim, U of Pennsylvania
Joseph N. Cappella, U of Pennsylvania