A robust Bayesian test for identifying context effects in multiattribute decision-making

Bibliographic Data

ID21640685
AuthorsDimitris Katsimpokis (0000-0002-4073-0206, University of Basel, corresponding author), Laura Fontanesi (0000-0001-5067-9451, University of Basel), Jörg Rieskamp (0000-0003-2632-8015, University of Basel)
Year2023
Volume30
Issue2
Pages498-515
Publication date2023-04-01
Open AccessYes
TypeARTICLE
VenuePsychonomic Bulletin & Review (JOURNAL)
Journal identifiersISSN: 1069-9384 • E-ISSN: 1531-5320
PublisherSpringer Science and Business Media LLC (PUBLISHER)
DOI10.3758/s13423-022-02157-2
PMID36167914
OpenAlexW4297338249
LanguageEN
Citations received2
References cited72

Research on multiattribute decision-making has repeatedly shown that people’s preferences for options depend on the set of other options they are presented with, that is, the choice context. As a result, recent years have seen the development of a number of psychological theories explaining context effects. However, much less attention has been given to the statistical analyses of context effects. Traditionally, context effects are measured as a change in preference for a target option across two different choice sets (the so-called relative choice share of the target, or RST). We first show that the frequently used definition of the RST measure has some weaknesses and should be replaced by a more appropriate definition that we provide. We then show through a large-scale simulation that the RST measure as previously defined can lead to biased inferences. As an alternative, we suggest a Bayesian approach to estimating an accurate RST measure that is robust to various circumstances. We applied the two approaches to the data of five published studies (total participants, N = 738), some of which used the biased approach. Additionally, we introduce the absolute choice share of the target (or AST) as the appropriate measure for the attraction effect. Our approach is an example of evaluating and proposing proper statistical tests for axiomatic principles of decision-making. After applying the AST and the robust RST to published studies, we found qualitatively different results in at least one-fourth of the cases. These results highlight the importance of utilizing robust statistical tests as a foundation for the development of new psychological theories

Axiom · Bayesian probability · Data mining · Econometrics · Machine learning · Preference · Statistical hypothesis testing · Statistics · Computer Science · Decision-Making and Behavioral Economics · Economic and Environmental Valuation · Forecasting Techniques and Applications · Mathematics · Psychology · Artificial Intelligence

Tools

Open DOIOpen Access