Establishing the relevance of non-compensatory choice algorithms from stated choice surveys – an exploration
Journal Title: Judgment and Decision Making - Year 2017, Vol 12, Issue 3
Abstract
In applied sciences large-scale surveys are a popular means to acquire insights in the choices that people make in different contexts. In transportation research, for example, tens of thousands stated their choices between alternatives characterized by cost and time attributes. In this study I explore the extent to which the data acquired in such studies may exhibit the co-occurrence of different choice algorithms within such survey populations. For that purpose I propose a novel version of the outcome-oriented approach. It is applied to the outcomes of two Dutch value-of-time surveys. If the recorded choice patterns are viewed apart, most could be the result of several different algorithms, in line with the main criticism of the outcome-oriented approach. The novel version considers the causal relationships between an individual’s personal circumstances, his use of a particular algorithm and the resulting choice pattern. It employs inferential statistics for analyses of the frequencies of the expected and actually recorded choice patterns within groups of respondents. Applied to the Dutch survey results this allowed disentangling, at the aggregate level, the overlap in explaining compensatory and non-compensatory algorithms to a large extent. It revealed that weighted additive (WADD) algorithms incorporating different degrees of loss aversion could explain most recorded choice behaviour while none of the many non-compensatory algorithms that were considered yielded a more than marginal explanation. Replications of this study, preferably by re-analysing other large-scale surveys with more complicated choice sets, is recommended to find out whether or not these findings are incidental.
Authors and Affiliations
Evert Jan van de Kaa
Keywords
Related Articles
- EP ID EP678293
- DOI -
- Views 130
- Downloads 0
TEMAP2.R: True and Error model analysis program in R
True and Error Theory (TET) provides a method to separate the variability of behavior into components due to changing true policy and to random error. TET is a testable theory that can serve as a statistical model, allow...
Glucose promotes controlled processing: Matching, maximizing, and root beer
Participants drank either regular root beer or sugar-free diet root beer before working on a probability-learning task in which they tried to predict which of two events would occur on each of 200 trials. One event (E1)...
Tests of Cumulative Prospect Theory with graphical displays of probability
Recent research reported evidence that contradicts cumulative prospect theory and the priority heuristic. The same body of research also violates two editing principles of original prospect theory: cancellation (the prin...
Posthumous events affect rated quality and happiness of lives
Diener and colleagues (2001) illustrated that individuals rely heavily on endings to evaluate the quality of a life. Two studies investigated the potential for posthumous events to affect rated life quality, calling into...
Two sides of the same coin: Information processing style and reverse biases
This paper examines the effect of information processing styles (indexed by the Rational-Experiential Inventory of Pacini & Epstein, 1999) on adherence to bias judgments, and particularly to reverse biases; i.e., when tw...