Dubious statistical results are becoming less common in psychology