Options as information: Rational reversals of evaluation and preference.

Abstract

This article develops a rational analysis of an important class of apparent preference reversals-jointseparate reversals traditionally explained by the evaluability hypothesis. The "options-as-information" model considers a hypothetical rational actor with limited knowledge about the market distribution of a stimulus attribute. The actor's evaluations are formed via a 2-stage process-an inferential stage in which beliefs are updated on the basis of the sample of options received, followed by an assessment stage in which options are evaluated in light of these updated beliefs. This process generates joint-separate reversals in standard experimental designs. The normative model explains why the evaluability hypothesis works when it does, identifies boundary conditions for the hypothesis, and clarifies some common misconceptions about these effects. In particular, it implies that joint-separate reversals are not irrational; in fact, they are not preference reversals. However, in expanded designs where more than 2 options are jointly evaluated, the model predicts that genuine (and rational) preference reversals will sometimes emerge. Results of 3 experiments suggest an excellent fit between the rational actor model and the judgments of human actors in joint-separate experiments.Keywords: context effect, evaluability hypothesis, preference reversal, rationality Procedure invariance is viewed as a basic requirement of rational choice. This principle states that different methods of eliciting preferences from a decision maker (DM) should yield the same ordering of options. However, several apparent violations of procedure invariance have been documented (reviewed in Hsee, Zhang, & Chen, 2004). This article focuses on one influential class of apparent preference reversals-joint-separate reversals (JSRs) traditionally attributed to the evaluability hypothesis (Hsee, 1996;Hsee, Loewenstein, Blount, & Bazerman, 1999). These effects are often taken to exemplify the divergence between normative and descriptive models of decision making (Hsee et al., 2004;MacLean, 2002;Shafir & LeBoeuf, 2004).In this article, we develop a normative "options-as-information" model of these JSRs. The model considers a hypothetical rational actor in a joint-separate experiment. The actor is rational but not omniscient. His or her beliefs are updated as new options are sampled, and evaluations are updated in light of these inferences. The model implies that JSRs are not preference reversals and are not counter-normative. Instead, when the dynamically evolving belief states of the DM are accounted for, typical JSRs are seen to exemplify the convergence of normative and descriptive analysis, and to illustrate how these levels of analysis can be mutually illuminating (Anderson, 1991;Griffiths, Chater, Kemp, Perfors, & Tenenbaum, 2010;McKenzie, 2003McKenzie, , 2005Oaksford & Chater, 1994). Joint-Separate ReversalsIn a paradigm developed by Hsee and colleagues (Hsee, 1996(Hsee, , 1998(Hsee, , 2000Hsee et al., 1999), two options are defined on two criti...

The Role of Inference in Attribute Framing Effects

“…Modelers drew very different inferences across joint and separate evaluation conditions, and these different inferences were sufficient to reproduce the JSR in recipients, all of whom evaluated only a single option. These results provided support for an “options‐as‐information” model, according to which JSRs occur not because of different attribute weighting in different “evaluation modes” (joint vs. separate) but because of the different inferences that are drawn from different option samples (Sher & McKenzie, ). Sample‐based inference likely also contributes to the asymmetric dominance effect (Prelec, Wernerfelt, & Zettelmeyer, ; Sher, Müller‐Trede, & McKenzie, ) and can lead to intransitive behavior in multi‐attribute choice (Müller‐Trede, Sher, & McKenzie, ).…”

“…JSRs occur when an option that is superior on a difficult‐to‐evaluate attribute receives high ratings when judged jointly with the alternative option and low ratings when judged in isolation (Hsee, ). In a study resembling Experiment 1, Sher and McKenzie () presented modeler participants two options either separately or jointly and then asked them to estimate the mean and range of the difficult‐to‐evaluate attribute. Modelers drew very different inferences across joint and separate evaluation conditions, and these different inferences were sufficient to reproduce the JSR in recipients, all of whom evaluated only a single option.…”

“…To demonstrate the sufficiency of reference point inferences, Experiment 1 employs a yoking procedure recently developed by Sher and McKenzie () to examine how changes in context lead to different beliefs and how these different beliefs subsequently affect evaluations. We used this procedure to break a standard attribute framing design into two parts.…”

Robust, domain‐specific effects of prior context in risk preferences for pension choice

“…We note that, in addition to ALT, DbS, and RFT, Sher and McKenzie (; see also Muller‐Trede, Sher, & McKenzie, ) have proposed that the location of marketplace options with respect to particular attributes (e.g., annual pension income) is relevant to the evaluation of that scale in their “options‐as‐information” model. Decision makers who are somewhat uninformed as to what constitutes a “good” pension income will update their distribution of marketplace options with those presented in a particular choice situation (or prior to it in our case).…”

Improving scope sensitivity in contingent valuation: Joint and separate evaluation of health states

“…In fact, there is some evidence that would support this explanation. Sher and McKenzie (2014) showed one group of subjects (group 1) objects A and B and they were asked to provide their WTP only for object A. They also presented objects A and B to another group (group 2), but they were asked about their WTP only for object B.…”