Public knowledge in coordination games: Learning from non-replication

Authors: Roy Chen, Yan Chen, Yohanes E. Riyanto

Presenter

Yan Chen, Daniel Kahneman Collegiate Professor of Information

School of Information, University of Michigan

Homepage: http://yanchen.people.si.umich.edu/

Time and location

North Quad 4330, Thursday (1:00-2:00) pm

Abstract

Recently, social science research replicability has received close examination, with discussions revolving around the degree of success in replicating experimental results. In this report, we lend insight to the replication discussion by examining how even a seemingly minor protocol deviation in the experimental process (Camerer et al. 2016), the removal of public knowledge, can lead to a finding of “non-replication’’ of the results from the original study (Chen and Chen 2011). To examine this possibility, we conduct a meta-analysis of the results obtained from an original study, its replication study, and a series of new experiments in which we vary the condition of public knowledge. Our meta-analysis shows that, with public knowledge, we obtain the original result in (Chen and Chen 2011), whereas without public knowledge, we obtain the null result in Camerer et al. (2016). Together, we use our findings to propose a set of procedure recommendations to increase the validity of replications of laboratory experiments in the social sciences.