The research project concentrates on alternative classes of models for ordinal data generated by rating surveys with the aim to compare them from methodological and empirical point of views.
The main objective is to offer a real support to decisions based on advanced statistical models, flexible to use and easy to interpret, which expresses intentions of respondents with higher fidelity.
Some data sets (of interest for economic, psychological, social, medical and sensometric sciences) will be identified to check the performance of the methods, their relative merits and drawbacks, and the possible validity of further extensions.
A focal point of the research is the awareness that at least two latent components move the evaluation process: attractiveness towards the item (feeling) and indecision in the response (uncertainty).
This consideration leads to a mixture model which will be the benchmark of the Start-up phase. This model presents several advantages with respect to alternative approaches and the project aims to deepen these aspects from theoretical and empirical viewpoints.
