In this project, we explore how decision-makers interpret informational conflicts and how these conflicts affect decisions. Our study addresses the nuanced differences in decision-making when confronted with conflicting versus ambiguous information, utilizing a behavioral forced-choice paradigm.
- CONSTANTS: Core constants used throughout the analysis.
- Data Folder: Contains all datasets used in the project.
- Helper Functions: Utility functions for data processing and analysis.
- Plotting: Scripts for generating visualizations of the data.
- Subroutines: Core computational functions for the analysis.
- Parameter Fitting: Modules for estimating participant parameters.
- Report Generation: Scripts for creating comprehensive analysis reports.
- Participant Data Processing: Filtering and preparing participant data.
- Risk Choice Analysis: Examining choices under risk.
- Conflict and Ambiguity Data Analysis: Comparative study of choices in ambiguous and conflicted scenarios.
- Statistical Analysis: Utilizing ANOVA and permutation tests to compare conflict and ambiguity attitudes.
- Correlation Analysis: Investigating the relationship between risk, ambiguity, and conflict attitudes.
- Comparative Analysis: Direct comparison between conflicted and ambiguous decision scenarios.
- Regret Analysis and Big 5 Correlation: Exploring the relationship between personality traits, regret, and decision-making attitudes.
A prevalent yet understudied type of uncertainty emerges when several sources provide conflicting information. Here we ask how decision-makers interpret informational conflicts and how these conflicts affect decisions. We test the hypothesis that conflicts, for example one source estimating a 25% success chance of some procedure and another estimating a 75% chance, is interpreted as ambiguity, a success chance ranging between 25% to 75%. Using a behavioral forced-choice paradigm, we presented participants with a series of binary choices involving uncertain monetary outcomes. When choosing between higher uncertain outcomes and lower certain outcomes, we found that participants' attitudes to conflict and ambiguity were indistinguishable. In contrast, when choosing directly between identical conflicted and ambiguous lotteries, participants expressed an overwhelming aversion to conflict. We suggest that these opposing findings reflect context-dependent attitudes to conflict. In isolation, decision-makers reduce conflicted information to general ambiguity. It is only in its comparative form, when contrasted with ambiguity, that conflict is associated with its own, highly aversive, attitudes. Additionally, we tested but did not find evidence that trait agreeableness, regret, and subjective probability mediate conflict attitudes. We discuss the relevance of our findings to advice-giving, and to information communicators who, in our polarizing societies, report more and more conflicting information.
Ohad Dan, Maya Sanghvi, Ifat Levy