Estimating equations for association structures.

摘要: This paper investigates generalized estimating equations for association parameters, which are frequently of interest in family studies, with emphasis on covariance estimation. Separate link functions used to connect the mean, scale, and correlation linear predictors involving possibly different sets covariates, separate proposed three parameters. Simulations show that robust 'sandwich' variance estimator jackknife parameters generally close empirical sample size 50 clusters. The results contradict Ziegler et al. Kastner Ziegler, where obtained from software MAREG was shown be unsuitable practical usage. problem appears arise because does not account variability estimation scale but may valid fixed scale. We also find formula approximate is deficient, resulting systematic deviations fully iterated estimator. A general provided performs well numerical studies. Data a study genetics alcoholism illustrate importance reliable biomedical applications.