Ordinal regression models allow for parsimonious modeling of ordinal response data, which is preferable for any type of categorical data because the information content in the response is always low. In this thesis, the multinomial logistic regression model as well as ordinal regression models are introduced and the relevant theory is laid out. In the second part the proportional odds model, one of the most popular ordinal logistic models, is used to analyze a customer satisfaction survey. The proportional odds model as multicategorical extension of the binary logistic regression model for ordinal response variables assumes that all regression parameters with the exception of the intercepts are independent of the response category and models only so called “global effects”. This makes the model simple and allows for easy interpretation of the regression parameters. To facilitate the application of the model, a graphical method to investigate if the relevant assumptions are fulfilled by the data set is developed.     «

Ordinal regression models allow for parsimonious modeling of ordinal response data, which is preferable for any type of categorical data because the information content in the response is always low. In this thesis, the multinomial logistic regression model as well as ordinal regression models are introduced and the relevant theory is laid out. In the second part the proportional odds model, one of the most popular ordinal logistic models, is used to analyze a customer satisfaction survey. The...     »