Models of stochastic choice and decision theories: why both are important for analyzing decisions

Abstract

We select a menu of seven popular decision theories and embed each theory in five models of stochastic choice, including tremble, Fechner and random utility model. We find that the estimated parameters of decision theories differ significantly when theories are combined with different models. Depending on the selected model of stochastic choice we obtain different rankings of decision theories with regard to their goodness of fit to the data. The fit of all analyzed decision theories improves significantly when they are embedded in a Fechner model of heteroscedastic truncated errors or a random utility model. Copyright © 2009 John Wiley & Sons, Ltd.

Do Preference Reversals Disappear When We Allow for Probabilistic Choice?

“…However, as Stott (2006) and Blavatskyy and Pogrebna (2010) have shown, different assumptions about the way in which the stochastic component is specified can produce quite different estimates of underlying parameters. To avoid becoming embroiled in debates about the sensitivity of our results to particular functional forms, our strategy in this paper is to try to investigate the issues raised above within a framework of probabilistic choice so general that it encompasses a very broad range of more specific stochastic models and relies on a bare minimum of assumptions.…”

Random expected utility and certainty equivalents: mimicry of probability weighting functions

“…Whether the same can be said of the standard model is unclear to me. However, meaningful preference measurement may not be possible without strong assumptions concerning the random part of decision behavior (Wilcox 2008;Blavatskyy and Pogrebna 2010;Wilcox 2011;Apesteguia and Ballester 2016). Many scholars say that elicited certainty equivalents, or quantities that are argued to be estimated certainty equivalents, permit "nonparametric" identification and estimation of preferences (Gonzales and Wu 1999;Abdellaoui 2000;Abdellaoui, Bleichrodt and Paraschiv 2007).…”

Testing for independence while allowing for probabilistic choice

“…This presents us with a challenge, because there are a number of ways in which the stochastic component might be specified; and the parameters we estimate and the conclusions we draw from our tests may vary considerably depending upon the particular specifications used (for a discussion of these issues, see Stott 2006;Blavatskyy and Pogrebna 2010;and Chapter 6 in Bardsley et al 2010).…”