Technion, IEM faculty - Statistics Seminar
Speaker: Dotan Tzor, Technion
Title: Statistical methods of estimation and prediction in repeated choice games
Date: 25/03/2012
Time: 14:30
Place: Bloomfield-527
Abstract:  <>
Or read it here:
Comparisons of learning models in repeated choice games that focus on
the effect of experience on decisions is a major research interest of
experimental and behavioral economics. Current studies reveal that
models which assume reliance on small sets of experiences appear to fit
the summarized data very well (Erev, Ert and Roth, 2010a, 2010b; Nevo
and Erev, 2012). However, the parameters of those models were not
estimated by traditional econometric methods, but by wild grid-search
simulation techniques, which did not use all the available information.
The main goal of the current research is to try to improve the
estimation procedure of those models by using the maximum likelihood
(ML) approach instead of a grid-search simulation based approach. The
models consider in the work have features that are not as of a standard
parametric representation, therefore places several challenges. We
formulate the likelihood function to several modifications of those
small decisions making models. We present extensive simulation study to
assess the finite sample properties of the proposed methodology, and
show that our proposed MLE performs very well in terms of bias, variance
and computation time, per se, and in compare to the wild grid search
technique, as long as the model is well-specified.
We also consider possible miss-specifications in the model, and analyze,
based on simulations, the robustness of each of the two estimation
methods to a certain miss-specification in terms of estimation and
prediction accuracy. As a demonstration, we analyze a real data set,
taken from a choice prediction competition (Erev et al., 2010a).
Technion Math. Net (TECHMATH)
Editor: Michael Cwikel   <> 
Announcement from:  <>