Fig. 2

Precision (inverse of the variance of the treatment effect estimator, assuming that the variance of observed outcomes, σ², is 1) against sample size for different designs, in the case of K = 30, M = 100, ρ = 0.05, τ = 0.2. The dashed line shows designs obtained by the algorithm for different sample sizes (with dark-filled circles indicating the designs shown in the second column of Fig. 1). Pale-filled circles denote 100,000 randomly generated designs. Crosses show the performance of staircase designs with varying width of recruitment window