A SAS-macro %CANLOG performing maximum likelihood estimation and testing in a hierarchy of ordinal response models contained in the generalized logistic model for multinomial response is available from the author Jochem König upon request. Models of this type are discussed by Anderson (stereotype model) and Feldmann.
It may be used in combination with a second SAS-macro %CLASS that performs logistic discrimination based on fitted submodels of the multinomial logistic or cumulative odds model. The decision theoretic approach to discrimant analysis is used for construction of allocation rules. Cost structures are provided that take account of the ordinal outcome scale.
Online documentation
References:
Feldmann U. Partial distributional canonical discriminant analysis. Biometrical Journal, 1993; 35:427-443.
Anderson JA. Regression and ordered categorical variables (with discussion). Journal of the Royal Statistical Society, Series B, 1984; 45:1-30.
Anderson JA. Logistic discrimination. Handbook of Statistics, North-Holland, 1982; 2:169-191.
McCullagh P. Regression models for ordinal data (with discussion). Journal of the Royal Statistical Society, 1980; B42:109-142.
Panel on Discriminant Analysis, Classification and Clustering. Discriminant analysis and clustering, Statistical Science, 1989; 4:34-69.
Feldmann U. Strategies of ordinal classification. 1999. Paper presented at the ISCB 1999 in Heidelberg.
Feldmann U, Steudel I (2000), Methods of ordinal classification applied to medical scoring systems. Statistics in Medicine 19, 575-586.
