First analysis of ‘pre-registered’ studies shows sharp rise in null findingsnature.com415 pointsdanso8 years ago