Belief Elicitation under Ambiguity

How to cite

Drobner, Christoph; Coutts, Alexander; Koh, Boon Han; Woolnough, Chris. 2023. "Belief Elicitation under Ambiguity." AEA RCT Registry. AEARCTR-0008197. https://doi.org/10.1257/rct.8197-1.1

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Christoph Drobner* , Alexander Coutts , Boon Han Koh , Chris Woolnough

* Primary investigator

Registry ID
AEARCTR-0008197
DOI
https://doi.org/10.1257/rct.8197-1.1
First registered
Last updated
Status
Completed
Trial dates
2023-09-25 – 2024-05-07
Countries
Germany ()
JEL codes
C91, D81, D83
Registry record
http://www.socialscienceregistry.org/trials/8197

Abstract

Beliefs play a critical role in many economic, political, and life decisions. Hence, researchers have been interested in eliciting individuals' beliefs to better understand their decision-making processes. A fundamental challenge in experimental economics has been creating incentives for experimental participants to truthfully report their beliefs. In recent years, researchers have developed sophisticated methods to elicit the beliefs of experimental participants in an incentive-compatible manner (see Schotter and Trevino, 2014; Schlag et al., 2015; Trautmann and van de Kuilen, 2015; Charness et al., 2021, for reviews). One state-of-the-art method to elicit beliefs is the binarized scoring rule (BSR), which is incentive compatible even under non-neutral risk preferences (Hossain and Okui, 2013). To date, the discussion around the incentive compatibility of belief elicitation methods remains agnostic about the influence of ambiguity. This is surprising in light of the fact that experimenters are often interested in beliefs about ambiguous events. In this project, we investigate how reported beliefs under the BSR deviate from the implied objective probabilities for ambiguous and risky events, and how such deviations depend on the level of detail used to explain the BSR, the stake size, and participants' ambiguity preferences. Finally, we examine the extent to which reported beliefs deviate in line with theoretical predictions based on individuals' ambiguity preferences.

Intervention

Our experiment consists of three individual parts. In Part A, we elicit participants' beliefs about ambiguous and risky events. In Part B and Part C, we elicit participants' ambiguity and risk preferences, respectively. Details of the experiment are provided in the pre-analysis plan.

Experimental design

Our experiment consists of three individual parts. In Part A, we elicit participants' beliefs about ambiguous and risky events. In Part B and Part C, we elicit participants' ambiguity and risk preferences, respectively.

Further details of the experiment are provided in the pre-analysis plan.

Design details

Randomization method
Treatment assignment is done at the individual level by a computer program (oTree).
Randomization unit
Individual level randomization
Clusters
600 participants
Observations
600 participants
Arms
Detailed High: 150 participants Detailed Low: 150 participants Simple High: 150 participants Simple Low: 150 participants
Primary outcomes
The raw data consists of participants' belief reports about risky and ambiguous events for implied objective probabilities p ∈ {0.2, 0.5, 0.8}. We use three outcome variables to measure the accuracy of participants' belief reports: 1) The first outcome variable is a dummy variable that classifies whether the reported belief deviates from the implied objective probability. 2) The second outcome variable is the absolute distance of the reported belief from the implied objective probability. 3) The third outcome variable is the absolute distance of the reported belief from the midpoint 0.5.

Cited by

No citation data available for this registration yet.

Keywords