Job Seekers’ Preferences for Educational Requirements and Their Perception of AI Cognitive Ability

How to cite

Chen, Zeyang; Lin, Lin; Liu, Kai; Wang, Fei; Zhao, Zhong; Zou, Xianqiang. 2025. "Job Seekers’ Preferences for Educational Requirements and Their Perception of AI Cognitive Ability." AEA RCT Registry. AEARCTR-0017409. https://doi.org/10.1257/rct.17409-1.0

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

Zeyang Chen* , Lin Lin , Kai Liu , Fei Wang , Zhong Zhao , Xianqiang Zou

* Primary investigator

Registry ID
AEARCTR-0017409
DOI
https://doi.org/10.1257/rct.17409-1.0
First registered
Last updated
Status
On going
Trial dates
2025-12-02 – 2026-03-31
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/17409

Abstract

This study uses an online information-intervention experiment to examine whether bachelor’s degree holders exhibit a systematic preference for jobs with higher educational requirements, and whether perceptions of AI cognitive ability can alter such preferences. Participants are randomly assigned to treatment and control groups, with the treatment group receiving information about the high cognitive performance of a state-of-the-art AI model. All participants then complete a series of job-choice tasks in which they evaluate paired job positions using a seven-point preference scale. By comparing choices across conditions, the study identifies how perceived AI capability and non-educational job attributes influence graduates’ willingness to consider positions requiring only an associate degree. The findings shed light on the behavioral roots of educational-requirement mismatch in the labor market and the potential of information interventions to mitigate it.

Experimental design

The experiment consists of two core modules—AI cognition intervention and job-preference elicitation—supplemented by additional survey items on participant characteristics.

(1) AI Cognition Intervention (Between-subjects randomization): Participants are randomly assigned to treatment or control groups based on the odd–even status of their birth date. Initial AI cognition is measured using “AI model college entrance examination score predictions.” The treatment group receives information stating that “the Doubao large language model achieved a Gaokao-equivalent score close to 700 points,” while the control group receives no such information.

(2) Job Preference Measurement (Within-subjects randomization): Baseline comparison condition: Job A requires “associate degree or above,” Job B requires “bachelor’s degree or above,” with all other attributes held constant. Non-educational attribute treatments: Additional attributes are added only to Job A—such as career development potential, skill-training opportunities, wage premiums, job stability, and employee feedback—while Job B remains unchanged. Participants rate their preferences on a 7-point scale (1 = strongly prefer A; 4 = no preference; 7 = strongly prefer B).

Design details

Randomization method
Randomization done by the server of the survey platform
Randomization unit
The AI Cognition Intervention is a between-subject randomization at individual level. The Job Preference Measurement is a within-subject randomization where the sequence of choice questions are randomized.
Clusters
4800 individuals
Observations
4800 individuals
Arms
2400 individuals receive the AI Cognition Intervention, and 2400 do not.
Primary outcomes
Job Preference Score: A 7-point scale measuring participants’ preference between two job postings (1 = strongly prefer Job A; 7 = strongly prefer Job B; 4 = no preference). Captures the central behavioral outcome—avoidance or acceptance of jobs requiring only an associate degree.

Cited by

No citation data available for this registration yet.

Keywords