Decision making under conflictive uncertainty

Abstract

Living in a post-truth era, individuals are increasingly exposed to conflicting information. Such conflict can induce "conflictive uncertainty", a state of uncertainty about which information and source to trust. Conflictive uncertainty can impact emotions (e.g., frustration, distress), cognition (e.g., suspicion, distrust), and behaviours (e.g., reluctance to follow health-promoting advice). Yet, the psychological mechanisms driving these impacts are not well understood. Particularly, there are limited studies exploring why people are averse to conflicting information and conflictive uncertainty. Previous studies have suggested that decreased perceived source credibility may be a factor in conflict aversion. Sources who provide conflicting information are usually perceived as less credible compared to sources who provide other types of information, such as ambiguous or risky information. The decreased perceived source credibility may further diminish people's preference for options involving conflictive uncertainty, causing conflict aversion. Therefore, the aim of this thesis is to examine the mediation relationship between conflictive uncertainty, decreased perceived source credibility, and conflict aversion. It addresses three main questions: (1) validation of a cross-domain measure of conflict aversion; (2) identification of an appropriate method to study causality in a mediation relationship; and (3) empirical testing of the cause of conflict aversion. To validate a cross-domain measure of conflict aversion, this thesis assessed the convergent validity and test-retest reliability of three uncertainty preference measures: forced binary choice, certainty equivalent, and matching probability. The results showed that these measures exhibited unsatisfactory convergent validity and test-retest reliability in one-off assessments but achieved relatively satisfactory validity in repeated measurement conditions. Despite the limited reliability, matching probability consistently captured average preferences over two measurement periods. Thus, matching probability was selected for further experimental studies due to its consistency and cross-domain applicability. To identify an appropriate method for studying causality in a mediation relationship, this thesis first reviews the advantages and disadvantages of commonly used regression-based mediation approaches in psychology, such as Baron and Kenny's criteria, Sobel test, and structural equation modelling. It also discussed the challenges associated with using psychological states as mediators. To address these challenges, this thesis introduces some alternative causal approaches from recent causal inference literature, including the counterfactual framework, encouragement design, blockage design and so on. Using these advanced approaches, the thesis conducted four experiments to examine the mediation relationship between conflictive uncertainty, decreased perceived source credibility, and conflict aversion. Experiment 1 found a significant indirect effect from conflictive uncertainty to conflict aversion via decreased source credibility. Experiment 2 and 3 demonstrated that interventions on perceived source credibility significantly altered participants' degree of conflict aversion. Experiment 4 showed that by blocking the decrease in perceived source credibility, conflict aversion can be weakened or even eliminated. These experiments support the hypothesis that decreased perceived source credibility under conflictive uncertainty is a cause for conflict aversion. The findings of this thesis have significant implications for managing the impact of conflicting information. Additionally, the application of advanced causal approaches provides valuable guidance for studies aiming to explore different types of evidence supporting causality within a psychological mechanism. Show more