Biased Beliefs and Search in Education Markets
Last registered on November 04, 2018


Trial Information
General Information
Biased Beliefs and Search in Education Markets
Initial registration date
November 01, 2018
Last updated
November 04, 2018 11:30 PM EST
Primary Investigator
Princeton University
Other Primary Investigator(s)
PI Affiliation
Princeton University
PI Affiliation
Columbia University
Additional Trial Information
On going
Start date
End date
Secondary IDs
This study examines the role of biased beliefs about the distribution of school characteristics and search costs in the school choice process. To disentangle these two mechanisms, we conduct a randomized control trial in the Dominican Republic in which we vary the nature of the information that is given to parents. In the first treatment arm, we attempt to change parents’ beliefs by providing information on the price and quality distribution of schools in their neighborhood. In the second treatment arm, we try to also reduce search costs by including information on the attributes of each individual school in addition to the distributions. The experimental results are used to inform a structural model of search and school choice.
External Link(s)
Registration Citation
Agte, Patrick, Christopher Neilson and Claudia SC. 2018. "Biased Beliefs and Search in Education Markets." AEA RCT Registry. November 04.
Former Citation
Agte, Patrick et al. 2018. "Biased Beliefs and Search in Education Markets." AEA RCT Registry. November 04.
Experimental Details
Intervention Start Date
Intervention End Date
Primary Outcomes
Primary Outcomes (end points)
The primary outcomes of this study are the results from the parent’s search decision: the consideration set and the school choice. The standard search model predicts that parents search less if they underestimate the utility of schools that are not in their consideration set. Thus, we hypothesize that correcting their beliefs may encourage parents to search more and result in a larger consideration set and a better school choice. We also hypothesize that households in the second treatment may send their children to better schools than households in the control group. However, we are ambivalent on the effect of the second treatment on the consideration set. While the additional information in the brochure should increase the parent's passive knowledge of schools, we are only able to collect the names of the schools that parents actively report in the survey. This might be different from their passive knowledge if parents only remember to report the schools for which they have actively searched.

The measures of knowledge and school choice may include the following:

- Number of schools that the parent knows
- Number of schools in the neighborhood that the parent knows
- Average quality, price, school day duration, and distance of the schools that the parent knows

- The attributes of the school in which the parent enrolls the child. In particular, we will focus on quality, price, school day duration, and distance of the school.

These main outcomes might be used as moments in the model estimation. We are still working on the empirical version and estimation procedure, but the main idea is that the model should replicate the difference between knowledge outcomes of families in the first treatment arm and the control group and the difference between the choices of families in the first and the second treatment arm.
Primary Outcomes (explanation)
Secondary Outcomes
Secondary Outcomes (end points)
Our secondary outcomes are different factors in the school choice process: effort, beliefs, and perceived value of search. We hypothesize that the first treatment increased the effort that parents put into the search process. They should also have more correct beliefs and should perceive the value of search more than parents in the control group. By contrast, parents in the second correct treatment arm should put less effort in search than parents in the first treatment arm, but they should still have more correct beliefs and value search more than the control group.

The measures of knowledge and school choice may include the following:

- Number of schools visited by the parent
- Number of schools called by the parent
- Number of friends consulted by the parent
- Time invested in searching for schools

- Difference between the believed quality ranking of the enrolled school and the true ranking

Perceived value of search
- Probability of finding better school given initial awareness set
- Whether the parent recommends continuing searching given an initial awareness set
Secondary Outcomes (explanation)
Experimental Design
Experimental Design
To disentangle the relative importance of search costs and biased beliefs, we conduct a field experiment with 724 households in 19 neighborhoods in Santo Domingo, Dominican Republic. To be eligible, households needed to have a child in the relevant age for choosing schools and state that they plan to enroll them for the first time in a primary school in the academic year 2018/19.

Initial beliefs and awareness sets were collected during a baseline and an intervention survey. Beliefs on the distributions were obtained by asking the respondent about the perceived quality and price of private and public schools at different terciles. The surveys also collected information on the number of children in the household, the respondent's network, and the valuation of specific school attributes. Information on schools in the neighborhoods was obtained from administrative data and school surveys. This includes information on prices, student/teacher ratios, and students’ performance on a national exam in 3rd grade.

Stratified by neighborhood, parents were randomly assigned to either the control group or to one of two treatment groups:

- Parents in the first treatment arm received a brochure with information on the joint distributions of quality and price of schools in their neighborhood. All the information is aggregated at the neighborhood level.

- Parents in the second treatment arm received a brochure that contains the same information as the brochure in the first treatment arm plus information on individual school attributes. Additional information on the school’s performance on a national exam in 3rd grade was given verbally since we were not allowed to include it in official documents.

- Parents in the control group received a placebo letter that contained general information on schooling.

Surveyors carefully explained the content to all parents to make sure that they understood the information. Endline data are collected in a phone survey that is conducted two months after the households receive the treatments.
Experimental Design Details
Randomization Method
Randomization is done in office by a computer.
Randomization Unit
Was the treatment clustered?
Experiment Characteristics
Sample size: planned number of clusters
Sample size: planned number of observations
724 households
Sample size (or number of clusters) by treatment arms
Control: 241 households
Treatment 1: 239 households
Treatment 2: 244 households
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
IRB Name
Institutional Review Board, Princeton University
IRB Approval Date
IRB Approval Number
Analysis Plan

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

Request Information
Post Trial Information
Study Withdrawal
Is the intervention completed?
Is data collection complete?
Data Publication
Data Publication
Is public data available?
Program Files
Program Files
Reports, Papers & Other Materials
Relevant Paper(s)