Survey intervention to prevent financial fraud

Last registered on June 06, 2022


Trial Information

General Information

Survey intervention to prevent financial fraud
Initial registration date
June 05, 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, 5:12 AM EDT

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



Primary Investigator

Columbia Business School

Other Primary Investigator(s)

PI Affiliation
Columbia Business School
PI Affiliation

Additional Trial Information

In development
Start date
End date
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
We want to better understand whether people can avoid financial fraud when they are more alert and educated about it.
External Link(s)

Registration Citation

Pagel, Michaela, Shang-Jin Wei and Zhuan Xie. 2022. "Survey intervention to prevent financial fraud." AEA RCT Registry. June 06.
Experimental Details


The purpose is to investigate to what extent people are able to make correct decisions in situations where they are presented with attempts to financial fraud. The study will yield insights about financial fraud and the role of trust in economic decision making.
Intervention Start Date
Intervention End Date

Primary Outcomes

Primary Outcomes (end points)
The outcome variable will be the incidence of financial fraud as measured by users reporting financial fraud to the digital payments platform and by the platform intervening transactions or sending people alerts about transactions that the platform perceives to be likely financial fraud.
Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
We run a survey experiment to prevent financial fraud.
Experimental Design Details
We partner with one of the largest Chinese digital payment platforms. The platform will send users a survey asking questions about financial literacy and financial fraud. Participants are offered to answer the survey as an ad on their app's home screen or in their message inbox. They answer questions for about 6 to 10 minutes. Participants are randomly assigned to 4 different surveys, half of them containing questions about financial fraud.
An invitation ad will be posted on the app's front page or users will find an invitation in their app's inbox. The link will take participants to the survey/questionnaire website.
Randomization Method
Step 1: Use the number of logins of past six month (Dec. 11th, 2021 to May 10th, 2022) to select active Alipay users. If number of logins >=100, the user will be selected.
Step2: Drop a small fraction of users for who the information on gender, province code, or age (if less than 16) is missing.
Step3: Stratifying the population and sampling.
Use class of city (6 classes), gender (2 types), age (4 cohorts, 15-24, 25-39, 40-59, 60+) expenditure rank (5 ranks according to total consumption of Nov. 2021 to Apr. 2022), binary indicator of investment on the platform to stratify the population into groups.
For each user in each group, randomly assign a number one to four to answer our four questionnaires.
Randomization Unit
Individual active user
Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters
Sample size: planned number of observations
We expect to send surveys to 960K users and 36K users in each strata. Then, depending on the click-on-the-survey and complete-the-survey rates, will end up with 100,000 to 500,000 subjects.
Sample size (or number of clusters) by treatment arms
Depending on the click-on-the-survey and complete-the-survey rates, we will end up with 100,000 to 500,000 subjects, half of which will be treated.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
We are interested in the simple difference between the treated-with-financial-fraud-survey groups and treated-with-other-survey groups in terms of the incidence rate of financial fraud (as measured by reported complaints or interrupted or flagged transactions). The type-1 error rate should be less than 5% and the power 80%. If the baseline rate of financial fraud in the control group is 5%, then, in order to detect a 10% difference, i.e., a decrease to 0.45% due to the treatment, we need a sample size of 49,202 subjects (half treated and half control).

Institutional Review Boards (IRBs)

IRB Name
Human Research Protection Office and IRBs at Columbia
IRB Approval Date
IRB Approval Number


Post Trial Information

Study Withdrawal

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

Request Information


Is the intervention completed?
Data Collection Complete
Data Publication

Data Publication

Is public data available?

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials