Title: Regularized Estimation of the Nominal Response Model

Author: Michela Battauz

Affiliation: University of Udine

Abstract:
The nominal response model is an item response theory model that does not
require the ordering of the response categories. However, while providing a
very flexible modelling approach of polytomous responses, it involves the
estimation of many parameters at the risk of numerical instability and
overfitting. In this talk, a penalized likelihood approach to group response
categories and perform regularization is presented. The penalization adopted
is similar to the fused lasso penalty proposed by Tibshirani et al. (2005),
and encourages the slope parameters of the model to assume the same value.
Simulation studies show the good performance of the method, while an
application to TIMSS data illustrates the proposal.

References:
Tibshirani, R., Saunders, M., Rosset, S., Zhu, J., & Knight, K. (2005). 
Sparsity and smoothness via the fused lasso. Journal of the Royal 
Statistical Society: Series B (Statistical Methodology), 67(1), 91-108.