In last month’s column, I worried about whether encouraging us to preregister our hypotheses and analysis plan before running studies would stifle discovery. I came to the conclusion that it needn’t — but that we need to guard against letting the …