Narratives, Memory Recall, and Optimistic Beliefs in Gambling and Artificial Intelligence

Last registered on August 27, 2026
View Trial History

Pre-Trial

Trial Information

General Information

Title
Narratives, Memory Recall, and Optimistic Beliefs in Gambling and Artificial Intelligence
RCT ID
AEARCTR-0019323
Initial registration date
August 19, 2026

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
August 27, 2026, 11:43 AM EDT

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

Locations

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

Request Information

Primary Investigator

Name
Kim Ristolainen
Affiliation
University of Turku
Contact Primary Investigator

Other Primary Investigator(s)

Additional Trial Information

Status
In development
Start date
2026-09-07
End date
2026-12-22
Keywords
Behavior, Finance & Microfinance, Health, Labor
Additional Keywords
Belief formation, AI expectations, Gambling behavior, Narratives, Memory
JEL code(s)
D91, D84, G41, C93
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
Individuals often form beliefs about uncertain future events using limited information, personal experiences, and commonly encountered narratives. This study examines how making different types of experiences and narratives salient affects subjective beliefs under uncertainty.

We conduct a randomized survey experiment using a large random sample of Finnish adults. Respondents are randomly assigned to different prompts before reporting subjective expectations in two distinct domains: lottery outcomes and the future capabilities of artificial intelligence. The experimental prompts vary the salience of personal experiences, observed events, and commonly encountered narratives.

The study examines whether these prompts causally affect subjective beliefs and whether their effects differ across types of prompts and respondents. The survey also measures characteristics related to memory and belief formation, perceptions of similar experiences, and respondents' financial and investment characteristics.
External Link(s)

Registration Citation

Citation
Ristolainen, Kim. 2026. "Narratives, Memory Recall, and Optimistic Beliefs in Gambling and Artificial Intelligence." AEA RCT Registry. August 27. https://doi.org/10.1257/rct.19323-1.0
Sponsors & Partners
Experimental Details

Interventions

Intervention(s)
Participants are randomly assigned to survey prompts designed to vary the salience of different experiences, observations, and commonly encountered narratives before reporting subjective expectations about uncertain future events. The experiment contains separate randomized interventions in two domains.
Intervention Start Date
2026-09-07
Intervention End Date
2026-12-22

Primary Outcomes

Primary Outcomes (end points)
1. Subjective probability that the respondent would obtain at least five correct numbers in the Finnish Lotto at least once over a three-year period when playing one Lotto line per week.
2. Subjective probability that the respondent would obtain at least five correct numbers in the Finnish Lotto at least once over a fifteen-year period when playing one Lotto line per week.
3. Subjective probability that artificial intelligence will be able to perform the respondent's own work, or work corresponding to the respondent's education, as well as or better than the respondent within three years.
4. Subjective probability that artificial intelligence will be able to perform the respondent's own work, or work corresponding to the respondent's education, as well as or better than the respondent within fifteen years.

All primary outcomes are elicited on a 0–100 percent scale.
Primary Outcomes (explanation)
The four primary outcomes will be analyzed separately. No composite primary outcome is required for the main treatment-effect analysis.

For each domain, treatment effects will primarily be estimated relative to the corresponding no-prompt control group. We will also examine prespecified contrasts between narrative/statement-based prompts and experience- or exposure-based prompts. Raw percentage-point effects will remain the primary scale of interpretation.

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
The experiment is embedded as the second part of a larger survey administered by the Finnish Institute for Health and Welfare (THL). THL administers the survey, including distribution of survey invitations and collection of responses. The study is a randomized online survey experiment conducted using a large random sample of Finnish adults. Participants report subjective expectations in two domains involving uncertain future events.

Within each domain, respondents are randomly assigned at the individual level to one of several survey conditions that vary the salience of different experiences, observations, or narratives. Each domain also contains a no-prompt control condition.

The primary analysis compares subjective beliefs across randomly assigned survey conditions. The study additionally examines whether treatment effects differ across types of prompts and across prespecified respondent characteristics.
Experimental Design Details
Not available
Randomization Method
Randomization will be implemented electronically by the survey software using computer-generated random assignment. Respondents will be assigned independently at the individual level to experimental conditions with approximately equal probabilities within each experimental block.
Randomization Unit
Individual survey respondent.
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
1
Sample size: planned number of observations
Up to 30,000 Finnish adults will be invited to participate in the survey. The final number of observations will equal the number of invited individuals who complete the relevant survey outcomes. For ex ante power calculations, we assume approximately 4,500 completed responses, corresponding to a 15 percent response rate among 30,000 invitees.
Sample size (or number of clusters) by treatment arms
Conditional on approximately 4,500 completed responses and equal randomization probabilities, each five-arm experimental block is expected to contain approximately:

900 respondents in the no-prompt control condition;
900 respondents in treatment arm 1;
900 respondents in treatment arm 2;
900 respondents in treatment arm 3; and
900 respondents in treatment arm 4.

The exact realized numbers may differ because the final sample size depends on survey participation and random assignment.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Power calculations assume approximately 4,500 completed survey responses, individual-level randomization, five approximately equally sized experimental arms, a two-sided significance level of 5%, and 80% statistical power. With approximately 900 respondents per experimental arm, the study is powered to detect a minimum effect of approximately 0.13 standard deviations in a comparison between any single treatment arm and the no-prompt control group. Because treatment is randomized at the individual level and is not clustered, no adjustment for clustering is required. Comparisons that pool multiple treatment arms will have greater statistical power.
Supporting Documents and Materials
IRB

Institutional Review Boards (IRBs)

IRB Name
The Ethics Commitee of the THL (Finnish Institute for Health and Welfare)
IRB Approval Date
2026-08-18
IRB Approval Number
THL/2084/6.02.01/2026
Analysis Plan