Skip to main content
. 2013 Apr 25;13:60. doi: 10.1186/1471-2288-13-60

Figure 3.

Figure 3

Evolution of the type I error of the Wald test according to the study design. Assuming ϵA = ϵB = 0.999, no limit of detection (“All data”, red line) or a limit of detection at 12 IU/mL (“ML”, blue line). N: number of patients per group of treatment; n: number of viral load measurements per patient; ntot: total numbers of observations per group of treatment.