Test-optional Admissions and Job Market Performance: Experimental Evidence from Japan

How to cite

Uchikoshi, Fumiya; Yamagishi, Atsushi; Gagnon, Etienne. 2026. "Test-optional Admissions and Job Market Performance: Experimental Evidence from Japan." AEA RCT Registry. AEARCTR-0017930. https://doi.org/10.1257/rct.17930-1.1

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

Fumiya Uchikoshi* , Atsushi Yamagishi , Etienne Gagnon

* Primary investigator

Registry ID
AEARCTR-0017930
DOI
https://doi.org/10.1257/rct.17930-1.1
First registered
Last updated
Status
Completed
Trial dates
2025-03-10 – 2025-03-14
Countries
Japan (Greater Tokyo)
JEL codes
C91, J22, J24, J71
Registry record
http://www.socialscienceregistry.org/trials/17930

Abstract

We analyze how test-optional admissions affect students’ job market outcomes. We conduct an experiment that corrects employers’ misperceptions about the prevalence of test-optional admissions in Japan, where both test-optional and test-based admissions co-exist within the same schools and programs. We find that test-optional admissions function as a signal of students’ ability and induce statistical discrimination against test-optional applicants during resume screening. This discount is particularly pronounced at lower-ranked institutions, which tend to enroll students from lower socioeconomic backgrounds. Consequently, the adoption of test-optional admissions may disproportionately harm these students in the labor market.

This experiment is preregistered at OSF, which can be found at http://osf.io/xzrk4/overview

Experimental design

We experimentally correct this school-program misperception on the fraction of test-optional admissions by providing the treatment group with information about the true share of test-optional admissions.

Design details

Randomization method
After answering baseline questions about their views on schooling and entrance examinations, respondents are randomly assigned to either the treatment or control group. Each sample is independently assigned to the treatment group with 50% probability. For both groups, we ask them to evaluate five fictitious randomly-generated job applicant profiles, based on the standard profile Japanese job applicants submit for companies' entry-level hiring for new graduates.
Randomization unit
Individual
Clusters
Our target sample size is 2,500 participants. We continue recruiting the sample until we achieve the 2,500 sample size.
Observations
Our target sample size is 2,500 participants. We continue recruiting the sample until we achieve the 2,500 sample size.
Arms
We initially recruited 2,626 respondents, and the final sample includes the 1,362 who passed the attention check.
Primary outcomes
A 10-point scale question about how likely an applicant is to advance to the next step of respondents' company's recruitment process.

Cited by

No citation data available for this registration yet.

Keywords