On the asymptotic moments of Pearson type statistics based on resampling procedures (Q1966028)

The author presents a discussion of overdispersed linear regression models. Dispersion estimates constructed with the help of the bootstrap technique based on resampling from some observed centered Pearson residuals are proposed. This technique is an alternative to those based on the generalized Pearson statistic. The asymptotic properties of the bootstrap dispersion estimate are obtained. Approximations to the first two unconditional moments are calculated.