On the Asymptotic Variances of û Terms in Loglinear Models of Multidimensional Contingency Tables
多维列联表对数线性模型中 û 项的渐近方差
Loglinear models are classified as direct or indirect depending on whether the maximum likelihood estimates of cell values are explicit functions of the sufficient statistics or not. For saturated (hence, direct) models, Goodman (1970) and Bishop, Fienberg, and Holland (1975) used the δ method to calculate the asymptotic variances of various û terms in the loglinear models. In the present paper, this approach has been generalized to direct unsaturated hierarchical loglinear…