Choosing between Causal Interpretations: An Experimental Study

Last registered on June 20, 2025
View Trial History

Pre-Trial

Trial Information

General Information

Title
Choosing between Causal Interpretations: An Experimental Study
RCT ID
AEARCTR-0016173
Initial registration date
June 05, 2025

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
June 20, 2025, 11:34 AM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

Country
Switzerland
Region

Primary Investigator

Name
Heidi Christina Thysen
Affiliation
Norwegian School of Economics
Contact Primary Investigator

Other Primary Investigator(s)

PI Name
Sandro Ambuehl
PI Affiliation
University of Zurich
Contact Investigator

Additional Trial Information

Status
In development
Start date
2025-06-06
End date
2025-10-31
Keywords
Behavior
Additional Keywords
JEL code(s)
C91, D01, D83
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
We often find ourselves compelled to make choices without fully knowing the causal mechanism through which our action influences the outcome we care about. In this paper, we study people’s choices when faced with exogenous interpretations about how variables are connected. Do they exclude inaccurate interpretations (fact-based choices), or do they follow interpretations that sound appealing at face value (utility-based choices)? What part of the narrative determines its appeal? What kind of empirical implications from the narratives do subjects check against the facts? A recent theory literature makes various assumptions about how such choices could be made, but empirical evidence is scarce.
External Link(s)

Registration Citation

Citation
Ambuehl, Sandro and Heidi Christina Thysen. 2025. "Choosing between Causal Interpretations: An Experimental Study." AEA RCT Registry. June 20. https://doi.org/10.1257/rct.16173-1.0
Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information
Experimental Details

Interventions

Intervention(s)
In the main part of the experiment subjects choice an action by following the advice of one of two advisors. The experiment uses variations across menus to reveal what information subjects use to guide their decisions. The set of menus are created in a way such that different choice strategies leads to different choice patterns or fingerprints. This is within subject.

In the second part of the experiment, subjects face additional choices. There two different ways of framing the choices and two different ways of organizing the information provided. The framing is either with advisors (similar to the first part), or where the choices are framed as bets on the color of the balls that are drawn. The advisors (or the descriptions of the bets) either contain information about what happens in both states given that this action is chosen, or information about what happens in a given state for both available actions.

Intervention (Hidden)
The across-subject variation is a three-fold increase in the stake size to investigate whether choice strategies varies with the stakes. A subject faces either regular stakes in both parts of the experiment or high stakes in both parts of the experiment.
Intervention Start Date
2025-06-06
Intervention End Date
2025-10-31

Primary Outcomes

Primary Outcomes (end points)
The distribution over choice strategies or types by stake size
Primary Outcomes (explanation)
The distribution over choice strategies or what in the PDF is estimated using a finite mixture model. Every distribution over types give an implied distribution over the choices for each menu (first moments), and across every pair of menus (second moments). The distribution over types are estimated by minimizing the distance between observed and implied moments. This is done separately for the high and low stakes treatments.

Secondary Outcomes

Secondary Outcomes (end points)
Effect of framing on revealed ambiguity attitudes.
Secondary Outcomes (explanation)
If there are behavioral differences by presentation mode, this raises the question of which presentation mode reveals preferences, and which is affected by confusion. To answer this question, we change the payoff matrix such that one alternative stochastically dominates the other. We will interpret the choice of dominated option as a mistake, and the frame in which people choose the dominated option less frequently as the one that better reveals the subjects’ true preferences. Specifically, if dominated option is chosen more frequently in the state-constant frame, then we interpret the choices of the high promise as more of a cognitive mistake rather than the reflection of an underlying preference.

Experimental Design

Experimental Design
In the main part of the experiment, subjects choice an action by following the advice of one of two advisors. The experiment use variations across menus to reveal what subjects use to guide their decisions. The set of menus are created in a way such that different choice strategies leads to different choice patterns or fingerprints. This is within subject. The across subject there is variation in the stakes to investigate whether choice strategies varies with the stakes.

In the second part of the experiment the subjects are faced with additional choices. Here all the variation is within subject. There two different ways of framing the choices and two different ways of organizing the information provided. The framing is either with advisors (similar to the first part), or where the choices are framed as bets on the color of the balls that are drawn. The advisors (or the descriptions of the bets) either contain information about what happens in both states given that this action is chosen, or information about what happens in a given state for both available actions.

See PDF for more details, and data generating processes as well as parameters.
Experimental Design Details
Randomization Method
Randomization is done on the computers. In particular, the subjects within a session are randomized into the high and low stakes treatment so that every session includes both stakes conditions.
Randomization Unit
For reduced form analyses, we cluster standard errors at the subject level.
Was the treatment clustered?
Yes

Experiment Characteristics

Sample size: planned number of clusters
400 subjects
Sample size: planned number of observations
For identification and testing of the types distributions there are 14*400 choices.
Sample size (or number of clusters) by treatment arms
200 subjects in the low stakes treatment and 200 subjects in the high stakes treatment.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Supporting Documents and Materials
IRB

Institutional Review Boards (IRBs)

IRB Name
Human Subjects Committee of the Faculty of Economics, Business Administration and Information Technology
IRB Approval Date
2025-03-20
IRB Approval Number
2025-020
Analysis Plan

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Post-Trial

Post Trial Information

Study Withdrawal

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Intervention

Is the intervention completed?
No
Data Collection Complete
Data Publication

Data Publication

Is public data available?
No

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials