Analysis of neuroimaging studies shows that close attention to experimental design can increase the statistical robustness of research results.