Thirst for confirmation in multi-attribute choice: Does search for consistency impair decision performance?

Details

Serval ID
serval:BIB_A02B4A91D495
Type
Article: article from journal or magazin.
Collection
Publications
Title
Thirst for confirmation in multi-attribute choice: Does search for consistency impair decision performance?
Journal
Organizational Behavior and Human Decision Processes, 100, 128-143.
Author(s)
Karelaia  N.
Publication state
Published
Issued date
2006
Abstract
Experimental evidence suggests that people often do not feel comfortable with making decisions based on a single piece of evidence and that they systematically look for confirming evidence before choosing. The goal of this paper is to investigate whether such behavior is appropriate for multi-attribute binary choice. We model the experimentally observed "thirst for confirming redundancy" (Bruner, Goodnow, & Austin, 1956) through a simple heuristic strategy (CONF) that needs two consistent cues to make a binary choice. Analytical expressions for the probabilities that CONF chooses correctly between two alternatives and takes a decision after considering fewer than all pieces of evidence are presented. Importantly, CONF is advantageously insensitive to cue ordering. The model performs equally well in both structured environments, where cues are ordered by validity, and unstructured environments, where the cues are not consulted in the order of their validity. We further compare the performance of CONF with the performance of other heuristics in a series of simulated three-cue environments where the cues are continuous and vary in both predictive ability and inter-correlation. We show that across environments, CONF balances the advantages and disadvantages of other simple models. We conclude that confirmation-seeking in multi-attribute decision making is a simple, fast, and robust "ignorance" strategy that hedges your bets when you know that you might not know.
Create date
19/11/2007 11:42
Last modification date
03/03/2018 20:04
Usage data