Fig. 2

Bias distribution. Distribution of bias in the 1000 simulated samples under random (panel a), case-control (panel b), stratified (panel c) and nested case control sampling (m=1, panel d). For the first three scenarios a sample size of 100 was used, while in the nested case-control sampling a mean of 180 subjects was considered. Data were subject to a random censoring of 15 % (over administrative censoring). The box represent the first and third quartile, the black line the median and the empty dots represent outliers defined as bias more than 1.5 times the interquartile range above the third quartile (or more than 1.5 times the interquartile range below the first quartile). Dotted lines represent the reference for bias (bias equal to 0)