Journal-initiated retractions can reduce the number of citations of the author's earlier publications, a study finds (S. F. Lu et al. Sci. Rep. 3, 3146; 2013). The authors analysed the effects of 667 retractions — mostly in the sciences and dating mainly from 2000 onwards — on citations of the author's earlier work. When a journal initiated the retraction, the number of annual citations of earlier papers fell by 6.9% on average. But author-initiated retractions had no such effect. The scientific community rewards honesty, says study co-author Ben Jones, an economist at Northwestern University in Evanston, Illinois. Self-reporting indicates that “you really care about getting it right”, he says.