Jackknifing in generalized linear models (Q1260723)

For a generalized linear model, the author proves that the jackknife estimator and the corresponding jackknife studentized statistic of the covariance matrix of the maximum likelihood estimator are consistent and asymptotically normal, respectively. Furthermore, these results are shown to be true even if there exist unequal dispersion parameters in the model. For saving computation the author also discusses how to accelerate computation of the estimate.