ABSTRACT
This chapter discusses the general univariate and multivariate skewed link models for categorical response data. Both independent and correlated categorical (binary or ordinal) responses will be considered. The multivariate skewed link models are particularly useful in modeling correlated categorical response data, since they allow each binary or ordinal component to have different skewness and heaviness of tails of links, and at the same time they capture the correlation among binary and/or ordinal responses.
