Unawareness and Reverse Bayesianism

Last registered on March 15, 2024

Pre-Trial

Trial Information

General Information

Title
Unawareness and Reverse Bayesianism
RCT ID
AEARCTR-0013143
Initial registration date
March 11, 2024

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
March 15, 2024, 7:01 PM EDT

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

Locations

Region

Primary Investigator

Affiliation
University of Heidelberg

Other Primary Investigator(s)

PI Affiliation
University of Aarhus
PI Affiliation
University of Aarhus

Additional Trial Information

Status
In development
Start date
2024-03-13
End date
2024-06-14
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
This study experimentally investigates behavior under uncertainty when subjects experience growing awareness of states. The theoretical literature on evolution of preferences in response to growing awareness has suggested that subjective beliefs evolve according to reverse Bayesianism. Our experimental design allows us to disentangle behavior across two different dimensions. One dimension is whether or not behavior is consistent with reverse Bayesianism. The other dimension is whether preferences are consistent with subjective expected utility, the more general class of biseparable preferences, or with neither. The former dimension concerns between-awareness-levels behavior, while the latter dimension concerns within-awareness-level behavior. Our experimental design allows us to test connections between the two dimensions, thereby casting light on whether and how behavioral patterns within and between distinct awareness levels are linked.
External Link(s)

Registration Citation

Citation
Dominiak, Adam, Peter Duersch and Marie-Louise Vierø. 2024. "Unawareness and Reverse Bayesianism." AEA RCT Registry. March 15. https://doi.org/10.1257/rct.13143-1.0
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)
Our treatments (refinement and expansion) differ in the information that is given to subjects during the second stage of the experiment. In the first stage of the experiment, subjects face a binary choice between bets on two colors. After the first stage, subjects receive additional information about the colors they can bet on, before facing three additional binary choices.

In the refinement-treatment, subjects discover a new attribute of an existing color (e.g., blue can be either light blue or dark blue). In the expansion-treatment, subjects discover an additional color (e.g., brown exists).
Intervention Start Date
2024-03-13
Intervention End Date
2024-06-14

Primary Outcomes

Primary Outcomes (end points)
Choices of subjects in four binary decision tasks (A or B).
Primary Outcomes (explanation)
Each choice corresponds to a bet on one or two colors.

Secondary Outcomes

Secondary Outcomes (end points)
- Confidence questions for each binary choice (non-incentivized)
- Simple demographics (non-incentivized questionnaire)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
In our experimental design, we implement uncertainty via a sealed urn (a bag), which will be filled with colored balls in unknown proportion. Subjects face four decision tasks. Each decision task is to choose between two binary prospects, called bets on colors. Each bet pays a positive amount of money if the chosen color matches the color of a random and independent draw (with replacement) from the urn. Otherwise, the bet pays nothing. The color of the ball drawn represents the state of the world.

Each subject will face two stages, which are implemented sequentially. In the first stage, subjects choose to bet on one of two colors (e.g., green vs blue). In the second stage, subjects receive additional information about the colors the urn contains. The information represents new information about states of the world. Given this information structure, subjects solve four decision tasks, allowing us to test various decision criteria.

To shed more light on how subjects react to such new information about states, we implement two treatments: the refinement-treatment and the expansion-treatment. In the refinement-treatment, subjects discover a new attribute of an existing color (e.g., blue can be either light blue or dark blue). In the expansion-treatment, subjects discover a new color (e.g., brown).

We follow the standard methodology in behavioral and experimental economics and use subject’s choices to make inferences about preferences. That is, we test adherence of “revealed preferences” to some well-known decision theories such as subjective expected utility, biseparable preferences, and reverse Bayesianism.

Subjects will be informed that they will be compensated with a show-up fee of DKK 75 for their time and that they may earn additional compensation up to a maximum total payment of DKK 225 depending upon the decisions that the subject makes during the experiment and the outcome of the random draws from the urn.

We follow a standard procedure of paying the subjects for one randomly selected decision task out of the four they face. This procedure avoids that subjects hedge their choices across the tasks. The subjects will be informed of this procedure in the instructions.

We will use the subject pool in the COBE Lab’s Sona recruitment system. By doing so, we are following the standard protocol in experimental economics of using mostly student subjects. We do not intend to target any specific populations within this group.

Our experiment is a paper-and-pen experiment. The experiment will take approximately 50 minutes to conduct and will take place in the COBE Lab which is a dedicated facility for human subjects research. Subjects will be randomly assigned to one of our treatments.
Experimental Design Details
Randomization Method
Treatment randomization at signup via SONA.

Outcome randomization via draw from bag.

Payment randomization via roll of dice.
Randomization Unit
experimental session, individual
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
328 subjects or number of subjects reached on June 14 2024, whichever comes first.
Sample size: planned number of observations
328 subjects or number of subjects reached on June 14 2024, whichever comes first.
Sample size (or number of clusters) by treatment arms
50% of subjects per treatment
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Test if the share of subjects adhering to Reverse Bayesianism differs between the two treatments (Hyp1). Test used is two-sample Fisher exact test. Our planned treatment size is 164 subjects. For this group size, power calculation (alpha error prob = 0.05, power = 0.95, p2=0.7) informs us that the minimum detectable effect size would be a difference of 0.1667774 in the p1=>p2 direction.
Supporting Documents and Materials

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

Request Information
IRB

Institutional Review Boards (IRBs)

IRB Name
Aarhus University’s Research Ethics Committee/Research Ethics Committee - BSS (Institutional Review Board)
IRB Approval Date
2024-02-08
IRB Approval Number
BSS-2024-044-L
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