Improving Social Mobility by Helping Rural Students Make Informed College Choices Country Proposed Start Date

How to cite

Wang, Ao; Ye, Xiaoyang; Wang, Shaoda. 2019. "Improving Social Mobility by Helping Rural Students Make Informed College Choices Country Proposed Start Date." AEA RCT Registry. AEARCTR-0004952. https://doi.org/10.1257/rct.4952-1.1

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Ao Wang* , Xiaoyang Ye , Shaoda Wang

* Primary investigator

Registry ID
AEARCTR-0004952
DOI
https://doi.org/10.1257/rct.4952-1.1
First registered
Last updated
Status
On going
Trial dates
2019-06-01 – 2020-09-01
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/4952

Abstract

Centralized admission mechanisms, especially the “Deferred Acceptance (DA)” algorithm, have been
increasingly adopted in college admission among developing countries. For fairness, exam score is often the only
determinant for admission in these systems. Using administrative data from China, which uses a constrained DA
mechanism for college admission, we find that conditional on exam score, poor and disadvantaged students frequently
use dominated strategies (put unambiguously more selective colleges in lower-ranked positions) in college application,
resulting in substantially worse admission outcomes. In collaboration with an influential research center at Peking
University, we propose a randomized controlled trial to understand why disadvantaged students make dominated
choices in college application, and how these suboptimal high-stakes decisions could be improved. The interventions
consist in tutoring on the DA mechanism, customized presentation of college list, and providing admission probability
with the aid of machine learning techniques. The general equilibrium implications of such interventions will be
evaluated.

Experimental design

Based on students' basic information, declared preferences and their score, our algorithm will generate a customized list of colleges for each student who participate in the survey experiment. We will randomly pick half of the colleges on the customized list and present them to students to facilitate their application. This variation is cross-randomization across all arms. In addition to the customized list, we will provide students with useful information such as telling them the admission probability of relevant colleges, or teach them to correctly order the colleges in their rank-order lists.

Design details

Randomization method
Randomize by a computer
Randomization unit
Inidividual
Clusters
Roughly 50 schools
Observations
364 in the first round (this summer), expect to get more (up to roughly 2,000) in the next round (next summer)
Arms
Arm 1 (control): 84 completed this summer, expect to get more in the next round. Arm 2 (information about probability only): 97 completed this summer, expect to get more in the next round. Arm 3 (information about probability and tutorial): 183 completed this summer, expect to get more in the next round.
Primary outcomes
Whether students make more informed choice in college application

Cited by

No citation data available for this registration yet.

Keywords