First analysis of ‘pre-registered’ studies shows sharp rise in null findings Logging hypotheses and protocols before performing research seems to work as intended: to reduce publication bias for positive results. Matthew Warren 30 Oct 2018
