Documentation: overall.power2 description & multi-arm power

[jamesdalg]

When I do the following simulation using a binary variable, I get outcomes that are a bit difficult to understand:

cluster_size=200
n_sites=50
bin.ma.rct.balanced.4arm.small <- cps.ma.binary(nsim = 12,
nsubjects = list(rep(cluster_size, times=n_sites),
rep(cluster_size, times=n_sites), rep(cluster_size, times=n_sites),
rep(cluster_size, times=n_sites)),
 narms = 4,
 nclusters = n_sites,
 probs = c(0.50, 0.70, 0.01,0.01),
#how to define probs?
 sigma_b_sq = c(0.1, 0.15, 0.01,0.01),
#how to define sigma?
 alpha = 0.05, allSimData = TRUE,
 seed = 123, cores="all")
sapply(bin.ma.rct.balanced.4arm.small$cluster.sizes,sum) %>% sum()
bin.ma.rct.balanced.4arm.small$power
bin.ma.rct.balanced.4arm.small$overall.power2

> bin.ma.rct.balanced.4arm.small$power
          Power Lower.95.CI Upper.95.CI Alpha      Beta
Arm.2 0.3333333   0.1855618   0.5097025  0.05 0.6666667
Arm.3 0.3333333   0.1855618   0.5097025  0.05 0.6666667
Arm.4 0.3333333   0.1855618   0.5097025  0.05 0.6666667
> bin.ma.rct.balanced.4arm.small$overall.power2
                       Power Lower.95.CI Upper.95.CI
probability of success     1   0.7353515           1

Questions:

1. Why are the arm powers 0.33333? I've changed parameters several times. What precisely does this power mean?
2. Why precisely is my power=1? Am I doing something wrong here? What precisely is the power in overall.power2?

[Kenkleinman]

Hi, James--

Happy to address these questions. Maybe via e-mail?

Ken

[jamesdalg]

Thanks for being willing! I've sent you an email, but I should note that there is no description for overall.power2 in the knitr documentation. If you want to update that, that may help others as well.