Expertise, Personal Experience, and Algorithm Aversion (new study 1)

Last registered on June 06, 2022

View Trial History

Pre-Trial

Trial Information

General Information

Title

Expertise, Personal Experience, and Algorithm Aversion (new study 1)

RCT ID

AEARCTR-0009542

Initial registration date

June 03, 2022

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 06, 2022, 5:52 AM EDT

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

Locations

Country

China

Region

Primary Investigator

Name

Yu Gao

Affiliation

Contact Primary Investigator

Other Primary Investigator(s)

PI Name

Cong Wang

PI Affiliation

Peking University

Contact Investigator

PI Name

Chong (Alex) Wang

PI Affiliation

Peking University

Contact Investigator

Additional Trial Information

Status

On going

Start date

2022-05-15

End date

2022-06-15

Keywords

Behavior, Lab

Additional Keywords

Algorithm Aversion, Decision from Experience

JEL code(s)

Secondary IDs

Prior work

This trial does not extend or rely on any prior RCTs.

Abstract

Research shows that evidence-based algorithms perform better than humans in predicting the future. Yet people give less weight to AI advice than they should. By exogenously variating personal experience with AI predictions, we explore how personal experience impacts weight on algorithm advice and how the level of expertise moderates this relationship. Our results will help design algorithms that are better adopted by human decision-makers, and mitigate the biases that experts hold on algorithms.

External Link(s)

Registration Citation

Citation

Gao, Yu, Cong Wang and Chong (Alex) Wang. 2022. "Expertise, Personal Experience, and Algorithm Aversion (new study 1)." AEA RCT Registry. June 06. https://doi.org/10.1257/rct.9542-1.0

Sponsors & Partners

Experimental Details

Interventions

Intervention(s)

Intervention Start Date

2022-05-15

Intervention End Date

2022-06-15

Primary Outcomes

Primary Outcomes (end points)

Main outcomes are weight on advice (WOA), the distance between optimal WOA and real WOA, and change in WOA after receiving the information about advisor’s incidental performance.

Primary Outcomes (explanation)

Weight on advice: the difference between the initial and revised judgment divided by the difference between the initial judgment and advice. WOA of 0% occurs when a participant ignores advice and WOA of 100% occurs when a participant abandons his or her prior
judgment to match the advice.

Secondary Outcomes

Secondary Outcomes (end points)

confidence in self, confidence in the advisor, and their tolerance of the advisor

Secondary Outcomes (explanation)

To what extent do you agree with the following statement?
“If the AI underperformed me in its domain, although it rarely happens, I would think that the AI is not good enough”.
“If the expert underperformed me in his/her domain, although it rarely happens, I would think that the expert is not good enough”.
(1-5, from completely agree to completely disagree)

Experimental Design

In Stage 1, each subject will be randomly assigned to the expert or the laypeople group, where they will receive immediate feedback or delayed feedback regarding their performance in a prediction task, respectively. Then, subjects will receive advise from either an AI or a human in new prediction tasks.
In Stage 2, each subject will be exposed to a randomly drawn level of the advisor’s relative performance compared to herself (+24, +12. -12, -24). Then, subjects will receive advise from either an AI or a human in new prediction tasks.

Experimental Design Details

Randomization Method

Randomization will be done by the survey platform.

Randomization Unit

individual

Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters

8 treatments

Sample size: planned number of observations

234 subjects per treatment

Sample size (or number of clusters) by treatment arms

234 * 2 * 2 * 2 = 1872 subjects in total. In theory, a WOA should be between 0 and 1. However, in our pilot study, we observe that about 5%-10% of our subjects report a WOA out of this range. We will oversample about 10% to account for this factor.

Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

pwr.t.test(n = , d = 0.3, sig.level = 0.05 , power = 0.9 , type = c("two.sample"))

Supporting Documents and Materials

IRB