hayes.power.poisson {clusterPower} | R Documentation |
This function calculates the power for a specified cluster-randomized study based on the methods described by Hayes et al (1999).
hayes.power.poisson(n.clusters, period.effect, btw.clust.var, at.risk.params, cluster.size, effect.size, alpha = 0.05)
n.clusters |
number of clusters |
period.effect |
period effect, on the link scale. See details. |
btw.clust.var |
the between-cluster variance |
at.risk.params |
the expected at-risk time per individual in the study |
cluster.size |
the number of individuals in each cluster |
effect.size |
effect size, specified on the GLM link scale |
alpha |
desired type I error rate |
Calculates, for a specified study design, the power of that study to detect the
specified effect size. The model is specified as a Poisson log-linear random
effects model (period.effect
and btw.clust.var
are parameters from
the model specified in Reich et al (2012)). Based on this model specification, the
coefficient of varation between cluster-level outcomes is calculated using
conditional expectation (see mixed.eff.params()
) and then the formula from Hayes
and Bennett (1999) is implemented.
A numeric vector of length 1, containing the estimated power for the given study specifications.
Reich NG et al. PLoS ONE. Empirical Power and Sample Size Calculations for Cluster-Randomized and Cluster-Randomized Crossover Studies. 2012. http://ow.ly/fEn39
Hayes RJ and Bennett S. Int J Epi. Simple sample size calculation for cluster-randomized trials. 1999. http://www.ncbi.nlm.nih.gov/pubmed/10342698
hayes.power.poisson(n.clusters=36, period.effect=log(.015), btw.clust.var=0, at.risk.params=20, cluster.size=20, effect.size=log(.7))