Fig. 2From: Should treatment effects be estimated in pilot and feasibility studies?Simulated data for 20 pilot studies (each n = 34) estimating an unknown ‘true’ treatment effect of 10, with a standard deviation of 20, and with an assumed minimum important difference of 4 (dashed horizontal reference line). Error bars are 95% confidence intervals, and the observed underlying standard deviations range from 12.6 to 24.8Back to article page