New Influence Measures in Polytomous Logistic Regression Models Based on Phi-Divergence Measures |
| |
| Authors: | Nirian MartÍn Leandro Pardo |
| |
| Affiliation: | 1. Department of Statistics, Carlos III University, Madrid, Spainnirian.Martin@uc3m.es;3. Department of Statistics and O.R., Complutense University of Madrid, Madrid, Spain |
| |
| Abstract: | We consider the asymptotic distribution of divergence-based influence measures which are an extension for polytomous logistic regression of an influence measure proposed in Johnson (1985 Johnson, W.O. (1985). Influence measures for logistic regression: Another point of view. Biometrika 72: 5965.[Crossref], [Web of Science ®] , [Google Scholar]), for binary logistic regression. A numerical example compares the classical Cook’s distance with the divergence based influence measures. |
| |
| Keywords: | Divergence measures Cook’s distance Polytomous regression model |
|
|
