Do Incentives Change Beliefs? A Field Experiment in Uganda.

Last registered on September 08, 2022


Trial Information

General Information

Do Incentives Change Beliefs? A Field Experiment in Uganda.
Initial registration date
September 04, 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
September 08, 2022, 11:32 AM EDT

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



Primary Investigator

Washington State University

Other Primary Investigator(s)

PI Affiliation
University of California, Riverside
PI Affiliation
University of California, Merced

Additional Trial Information

In development
Start date
End date
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Inaccurate beliefs about the benefits of preventive health products are an important explanation for low demand among the poor. Because credible information about these benefits can be difficult to obtain, financial incentives are widely used to stimulate demand. Aside from a potential price effect, financial incentives may also stimulate demand by credibly signaling the importance of the preventive health investment. Thus, we study whether financial incentives increase demand for health investments through an information channel. We implement a randomized controlled trial in Uganda to study demand for nutritional supplementation, a highly effective health investment to prevent malnourishment. We randomize learning about an incentive, without actually receiving it. We study how this information treatment changes beliefs about the impact of the product on outcomes and willingness-to-pay for the product. We also study how the size of the financial incentive affects willingness-to-pay.
External Link(s)

Registration Citation

Hirshleifer, Sarojini, Shanthi Manian and Ketki Sheth. 2022. "Do Incentives Change Beliefs? A Field Experiment in Uganda.." AEA RCT Registry. September 08.
Experimental Details


There are two randomly assigned “interventions”: 1) learning and participating in a lottery for a free box of MNP, and 2) learning and participating in a lottery for a free box of MNP + additional monetary incentive.
Intervention Start Date
Intervention End Date

Primary Outcomes

Primary Outcomes (end points)
Willingness to Pay for MNP: This is the respondent’s willingness to pay for the MNP elicited by the Becker-DeGroot-Marschak (BDM) method.

Beliefs on effectiveness of MNP

Beliefs on need of MNP
Primary Outcomes (explanation)
Willingness to Pay: The BDM will be implemented as follows: The subject will be asked to state her bid (i.e., willingness-to-pay) for a box of the MixMe MNP and set aside the bid amount. We will then draw a random price for the box of MNP. The possible random prices will range from zero to the at-cost price for the box of MNP. If the realized random price is less than or equal to the subject’s bid, then the subject must purchase the MNP at the randomly drawn price. Instead, if the random price is greater than her bid, then the subject may not purchase the MNP as the random price is greater than their willingness to pay.

Beliefs on effectiveness of MNP: We are examining three different symptoms for which MNP could be effective: small, anemia, frequently sick. We ask about differences of getting healthier with no MNP versus receiving MNP as directed.

Secondary Outcomes

Secondary Outcomes (end points)
Beliefs on your child’s need of MNP: three aspects of beliefs on own child’s symptoms of micronutrient deficiency (small, anemia, frequently sick). We also ask about children’s diet quality.

We will examine treatment effect heterogeneity by an ex ante question on the child’s overall health status

- Ex ante overall health
- Diarrhea/cough/etc
- Highest education of the respondent
- Negative reciprocity
- Decision-making on food
- Food security
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
Eligible households will be randomized into one of three groups: 1) learning and participating in a lottery for a free box of MNP, 2) learning and participating in a lottery for a free box of MNP + additional monetary incentive, 3) not learning about or participation in a lottery. All households will receive information about MNP from a health worker.
Experimental Design Details
Our study uses a household-level randomization. The study sample will consist of households that have an eligible child and the primary caregiver is available at the time of the survey. In one day, the primary caregiver will be visited by a health worker, followed by a survey enumerator. The health worker will inform the primary caregiver about: i) the importance of using MNP regularly using previously piloted materials, and ii) standard price at which the CHPs will sell the MNP.

Eligible households will be randomized into three conditions: high signal, low signal, and control. Those assigned to the high signal or low signal conditions will be informed about and entered into a fair lottery for either a large (high signal) or small (low signal) incentive. The large lottery is a free box of MNP + 60,000 Ugandan shillings; the small lottery is a free box of MNP. They will be informed: “BRAC International is funding this incentive because micronutrient powder is very important for your child’s health." In the control condition, the individuals will not receive any information about an incentive.

We are interested in isolating the information channel of a financial incentive, so only those who lose the lotteries will be included the analysis. Specifically, those who lose the lottery in the high (low) signal condition will become the high (low) signal treatment group. The participants who lose the lotteries are the population of interest, since they receive a signal about the value of the MNP via learning about the existence of an incentive, but they do not directly receive an incentive.

Randomization Method
Randomization is done on the spot by a random number generator in surveyCTO during the intervention and survey. The sample is distributed equally across the three treatment groups in expectation.
Randomization Unit
Randomization is done at the household level. There is only one respondent per household, so this is akin to an individual level randomization.
Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters
Sample size: planned number of observations
1752 households (excluding lottery winners)
Sample size (or number of clusters) by treatment arms
584 subjects per treatment arm, in expectation
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
The minimum detectable effect size for any comparison across two treatment arms is 0.189 standard deviations, at 90% power and a 5% significance level.

Institutional Review Boards (IRBs)

IRB Name
UC Riverside
IRB Approval Date
IRB Approval Number
HS - 20-192


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