Fig. 2From: Evidence of unexplained discrepancies between planned and conducted statistical analyses: a review of randomised trialsNumber of trials with unexplained discrepancies (total N = 89). *Of the n = 22 trials with none; no discrepancies (n = 5), explained discrepancies only (n = 17). **Unclear if discrepancy occurred (n = 11), unclear if discrepancy explained (n = 2). One trial had both a change and an addition for the analysis modelBack to article page