Abstract
I discuss reasons for why false-positive results are produced and published and report the results from large-scale replication projects where laboratory experiments are redone with new data to test the same hypotheses as in the original studies. In some of this work my collaborators and I have also added prediction markets and decision markets to study whether researchers can predict replication outcomes. The results suggest some but not perfect “wisdom of crowds.” I also discuss potential solutions to replication problems, including potential credibility problems and solutions for nonexperimental results.