Biased Beliefs and the Dynamic Role of Information in College Choice
Last registered on March 23, 2019

Pre-Trial

Trial Information
General Information
Title
Biased Beliefs and the Dynamic Role of Information in College Choice
RCT ID
AEARCTR-0003239
Initial registration date
March 15, 2019
Last updated
March 23, 2019 8:05 PM EDT
Location(s)
Region
Primary Investigator
Affiliation
Columbia University
Other Primary Investigator(s)
PI Affiliation
Columbia University
Additional Trial Information
Status
In development
Start date
2018-12-01
End date
2020-05-31
Secondary IDs
Abstract
Even though information frictions have been widely studied regarding decisions involving higher education, most research has studied these frictions as static, overlooking the dynamic component that might be embedded in the process. In this study, we will focus on both components, static and dynamic, analyzing not only the potential frictions in the discrete decision of college-degree choice, but also the dynamic decision in investment on college admission preparation itself. We will test this by conducting a randomized control trial (RCT) for senior high school students in Chile attending a large college admission test preparation institution, and providing students with information about institution-degree characteristics, at (i) the beginning of the academic year (dynamic treatment) or (ii) shortly before the college application window (static treatment). By leveraging the results of the RCT, we will be able to recover the key parameters of the decision model and their interaction.
External Link(s)
Registration Citation
Citation
Allende Santa Cruz, Claudia and Magdalena Bennett. 2019. "Biased Beliefs and the Dynamic Role of Information in College Choice." AEA RCT Registry. March 23. https://doi.org/10.1257/rct.3239-1.0.
Former Citation
Allende Santa Cruz, Claudia, Claudia Allende Santa Cruz and Magdalena Bennett. 2019. "Biased Beliefs and the Dynamic Role of Information in College Choice." AEA RCT Registry. March 23. http://www.socialscienceregistry.org/trials/3239/history/43927.
Sponsors & Partners

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

Request Information
Experimental Details
Interventions
Intervention(s)
Intervention Start Date
2019-03-01
Intervention End Date
2020-01-31
Primary Outcomes
Primary Outcomes (end points)
beliefs about the return of higher education programs, student test scores in the higher education admission test, tests taken by the student, rank order list of programs, enrollment data.
Primary Outcomes (explanation)
Secondary Outcomes
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
Experimental Design
Experimental Design
This project tries to understand information frictions in college choice by combining: (i) an information intervention, (ii) detailed administrative and survey data, (iii) a model of college choice and investments in human capital, and (iv) counter-factual analysis of the scaled-up policy. The
project will be conducted between March 2019 and July 2020. The objective of the experimental interventions is to exogenously shock students' beliefs about the returns of different programs. This variation will allow us to distinguish between the e ects of biased beliefs and other unobservables that can potentially affect students' scores and choices that are also correlated with their information set. The treatment will provide them with information about these returns. The speci c design of the treatment will depend on the results of our baseline survey and pilots, but the plan is to tailor information based on their stated preferences. We will conduct follow up surveys after the treatment to see the effectiveness of the interventions in terms of correcting beliefs.
Experimental Design Details
Randomization Method
Randomization done in office by a computer
Randomization Unit
In order to mitigate potential spillovers from the treatment groups on to the control group, we will randomize at the center-time slot level to minimize potential information sharing between students. Each center has di erent time slots in which students can attend classes
(e.g. afternoon or evening, each day of the week), so students in di erent slots do not share classes.
Was the treatment clustered?
Yes
Experiment Characteristics
Sample size: planned number of clusters
We expect to have 640 clusters in total. Randomization will be done at the "center–class cluster"
level. This means that for each center, we will build the class cluster for students (identified
by the classes students share), and randomize these structures to the different treatments
and control groups. Given that each center has on average just under 1,000 students, and
classes are capped to 25-30 students, many students that share the same time slot do not
actually attend the same classes, which gives us the advantage of randomizing at a lower
level without losing the safeguards against spillovers that the "center–time slot"
randomization had. Class clusters will be constructed by identifying all the class IDs for
each student, and mapping their classmates to identify all students they share classes with.
Given the number of students, for the same time slot there are different sections for each course.
Then, we estimate a conservative average of 20 class clusters per center, for the 32 centers.
Sample size: planned number of observations
12,000 students
Sample size (or number of clusters) by treatment arms
213 treatment 1, 213 in treatment 2, 213 in control.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Given that will randomize at a cluster level for two treatments and a control group, we calculate the minimum detectable effect (MDE) using a standard power calculation for cluster-level randomization (Duflo et al., 2007). We obtain the following minimum detectable effects (MDE) for different scenarios of intra-cluster correlation. Data from the 2016 admission process at the school and SES group-school level would suggest that the intra-cluster correlation would be between 0.15 and 0.2, in which case we would have the statistical power to detect effect sizes of 0.10-0.11 SD.
IRB
INSTITUTIONAL REVIEW BOARDS (IRBs)
IRB Name
Teachers College IRB
IRB Approval Date
Details not available
IRB Approval Number
19-036 Protocol
Post-Trial
Post Trial Information
Study Withdrawal
Intervention
Is the intervention completed?
No
Is data collection complete?
Data Publication
Data Publication
Is public data available?
No
Program Files
Program Files
Reports and Papers
Preliminary Reports
Relevant Papers