COLLECTIVE DELUSION IN THE SOCIAL SCIENCES: PUBLISHING INCENTIVES FOR EMPIRICAL ABUSE

The implications o f the editorial bias of academic journals for the selection of articles with apparently statistically significant findings are widely recognized but largely ignored. Few worry about the incentives the publishing bias presents to researchers for empirical abuse that brings into question the basis of social science knowledge. One possible solution, desirable but probably impractical, is to review articles with statistical results and conclusions omitted. Another, more practical, approach is to guarantee journal space for replication of previously published research. Finally, editors should take greater care in warning readers about findings that implicitly make unfounded statistical claims.