Can the Prediction Market Improve Predictions of COVID-19?

The goal of this study is to better understand how people predict the future risks of the novel Coronavirus (COVID-19).

Specifically, the investigators will ask the following research questions:

• How well do participants predict the future risks of COVID-19?
• Can the predictions be improved by using a prediction market mechanism?
• Does the prediction market reduce people's fear of COVID-19?

Study Overview

• Status: Completed
• Conditions: Health Knowledge, Attitudes, Practice
• Intervention / Treatment: Other: Prediction Market

Detailed Description

The proposed study is an online experiment. Students enrolled at National University of Singapore are recruited to participate in the study.

Participants will first complete a pre-experiment survey, which contains basic demographic questions. Then, participants will be randomly assigned to one of two conditions: "Survey" and "Prediction Market".

"SURVEY" CONDITION:

Participants in the "Survey" condition are asked 16 prediction questions in a survey format. The questions are of the following format:

"What do you think will be the total cumulative number of cases in Singapore on 8th of June, at 12pm?"

Each question has 5 answer options. Each answer option is a range of outcomes, e.g. "< 28,900", "between 28,900 and 33,899", "between 33,900 and 38,899", "between 38,900 and 43,899", and "> 43,899". Participants are required to enter their perceived likelihood of each answer option in %.

The 16 prediction questions come from the following variations: 4 countries (Mexico, Singapore, Turkey, USA) x 2 outcome measures (cases, deaths) x 2 time periods (8th of June, 6th of July).

Participants have 24 hours to submit their predictions.

After the 24-hour period, participants are requested to fill out a post-experiment survey, which includes questions about their subjective attitudes and fears towards COVID-19.

"PREDICTION MARKET" CONDITION:

For participants in the "Prediction Market" condition, the same 16 prediction questions are presented in the form of prediction markets. The prediction market is a well-established method of eliciting people's predictions. The method is briefly described below.

There are 16 prediction markets, one for each question. Participants are given 100 tokens per market, which can be used to buy "stocks" on possible outcomes. There are 5 possible outcomes per market (identical to the 5 answer options per question in the "Survey" condition).

Each stock (i.e., possible outcome) will have a price that is dynamically determined by the central marketplace, which is a function of real-time demand and supply of the option. If the option is popular, its price will become higher, and vice versa.

Participants can trade at any time, and as many times as they want, during a 24-hour period. Upon closure of the prediction market, participants will be rewarded proportional to the number of shares that they hold on options that later turn out to be true.

The final prices of stocks correspond to the group's predictions of COVID-19.

After the 24-hour period, participants are requested to fill out a post-experiment survey, which includes questions about their subjective attitudes and fears towards COVID-19.

=====

HYPOTHESES

The prediction market leads to better predictions about COVID-19. The investigators will compare the survey predictions and the prediction-market predictions with the actual realized outcome. The investigators hypothesize that the prediction-market predictions are more accurate than the survey predictions through information aggregation.

The prediction market reduces fear. Fear is measured by participants' responses to subjective attitude questions in the post-experiment survey.

• Study Type: Interventional
• Enrollment (Actual): 560
• Phase: Not Applicable

Contacts and Locations

Study Locations

• Singapore
  • Singapore, Singapore
    • National University of Singapore

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: 18 years and older (Adult, Older Adult)
• Accepts Healthy Volunteers: Yes
• Genders Eligible for Study: All

Description

Inclusion Criteria:

• National University of Singapore students

Exclusion Criteria:

• N/A

Study Plan

How is the study designed?

Design Details

• Primary Purpose: Other
• Allocation: Randomized
• Interventional Model: Parallel Assignment
• Masking: None (Open Label)

Number of Arms

2

Arms and Interventions

Participant Group / Arm	Intervention / Treatment
No Intervention: Control; Participants' COVID-19 predictions are elicited via a survey
Experimental: Treatment; Participants' COVID-19 predictions are elicited via a prediction market	Other: Prediction Market; Participants "bet" on likely future outcomes using a prediction market

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Predictions of COVID-19 Cases and Deaths	Participants are asked 16 questions of the following format: "What do you think will be the total cumulative number of cases in Singapore on 8th of June, at 12pm?" Each question has 5 answer options. Each answer option is a range of possible outcomes. The primary outcome measure is participants' perceived likelihood of each answer option. The 16 questions come from the following variations: 4 countries (Mexico, Singapore, Turkey, USA) x 2 outcome measures (cases, deaths) x 2 time periods (8th of June, 6th of July).	24 hours

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Fear	Fear is measured by participants' responses to subjective attitude questions in the post-experiment survey. The questions are on a 5-point Likert scale.	24 hours (participants are required to submit post-experiment survey within 24 hours of completion of the main experiment)

Collaborators and Investigators

• Sponsor: National University of Singapore
• Investigators: Principal Investigator: Teck Ho, PhD, National University of Singapore

Publications and helpful links

General Publications

• Camerer CF, Dreber A, Holzmeister F, Ho TH, Huber J, Johannesson M, Kirchler M, Nave G, Nosek BA, Pfeiffer T, Altmejd A, Buttrick N, Chan T, Chen Y, Forsell E, Gampa A, Heikensten E, Hummer L, Imai T, Isaksson S, Manfredi D, Rose J, Wagenmakers EJ, Wu H. Evaluating the replicability of social science experiments in Nature and Science between 2010 and 2015. Nat Hum Behav. 2018 Sep;2(9):637-644. doi: 10.1038/s41562-018-0399-z. Epub 2018 Aug 27.

Study record dates

Study Major Dates

• Study Start (Actual): May 15, 2020
• Primary Completion (Actual): May 16, 2020
• Study Completion (Actual): May 17, 2020

Study Registration Dates

• First Submitted: May 27, 2020
• First Submitted That Met QC Criteria: May 29, 2020
• First Posted (Actual): June 1, 2020

Study Record Updates

• Last Update Posted (Actual): June 2, 2020
• Last Update Submitted That Met QC Criteria: May 30, 2020
• Last Verified: May 1, 2020

More Information

Terms related to this study

• Other Study ID Numbers: SG-COVID

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: YES
• IPD Plan Description: Investigators will not be storing or sharing any personal identifiers. All individual level data will be anonymized, and only anonymized data will be shared with other researchers, upon request.
• IPD Sharing Time Frame: After completion of all analysis. It will be made available in the supporting documentation.
• IPD Sharing Access Criteria: It will be made available in the supporting documentation.
• IPD Sharing Supporting Information Type: STUDY_PROTOCOL; ICF; ANALYTIC_CODE

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No