Skip to main content
Figure 2 | BMC Medical Research Methodology

Figure 2

From: Sample size calculations for skewed distributions

Figure 2

Negative binomial sample size. Each vertical axis shows power estimated by simulation for comparing two means and 90% nominal power. Each data point is based on 10,000 simulated datasets with two arms of equal size (Q0 = Q1 = 0.5), the ° symbol shows results for logarithmic link, equation (7), and × for identity link, equation (9). The vertical lines show the 95% confidence interval for the proportion of datasets reaching 5% two-sided significance level when analysed by GLM with logarithmic link, i.e. a measure of Monte Carlo error. The number beside each plot symbol is the sample size calculated by the relevant method. In the upper panel, μ0 = 71.4, μ1 = 50 (30% vaccine efficacy ) and various values of k (assumed equal in both arms) on the horizontal axis. In the lower panel, μ0 is as before, k is 0.33 and various values of efficacy are used (horizontal axis) with μ1 = μ0 × (1-efficacy).

Back to article page