Power and sample size calculations for Poisson and zero-inflated Poisson regression models

摘要: Although sample size calculations for testing a parameter in the Poisson regression model have been previously done, very little attention has given to effect of correlation structure explanatory covariates on size. A method calculate Wald test is proposed, assuming that may be correlated and multivariate normal distribution. this calculation works with any pre-specified structure, exchangeable AR(1) matrices different values are used illustrate approach. The here based modification methodology already proposed literature. Rather than using discrete approximation distribution which much more problematic higher dimensions, Monte Carlo simulations used. It observed depends number matrix, but so covariates. matrix changes less substantially as dimension increases, it also covariates, lesser extent. extended case zero-inflated order obtain analogous results.