摘要
This paper gives the local influence diagnostics for parametric multinomialdistribution, especially contingency tables. Basing on the iterative weighted least squaresprocedure, we obtain the M.L.E. of the parameters and the modified Cook’s distance. Byperturbing the cell probabilities, we obtain Cook’s curvature under the restrictive conditionfor our model. Example is given to demonstrate the power of our methods.
This paper gives the local influence diagnostics for parametric multinomialdistribution, especially contingency tables. Basing on the iterative weighted least squaresprocedure, we obtain the M.L.E. of the parameters and the modified Cook's distance. Byperturbing the cell probabilities, we obtain Cook's curvature under the restrictive conditionfor our model. Example is given to demonstrate the power of our methods.
