Back to History

Fields Changed

Registration

Field Before After
Abstract How do first-time workers make choices about jobs? We partner with the UMass Boston Career Center and the UC Berkeley Career Center to conduct a survey of the senior graduating Classes of 2025 and 2026. Using hypothetical firm-choice experiments, we examine how students trade off first-job salaries with other human capital investments. How do early career workers make choices about jobs? We run a Prolific survey and partner with the UMass Boston Career Center to conduct a study on early-career job choice. Using hypothetical firm-choice experiments, we examine how workers trade off first-job salaries with other human capital investments.
Last Published October 27, 2025 06:15 AM April 07, 2026 12:34 AM
Experimental Design (Public) We partner with the UMass Boston Career Center and the UC Berkeley Career Center to conduct a survey of the senior graduating Classes of 2025 and 2026. Using hypothetical firm-choice experiments, we examine how students trade off first-job salaries with other human capital investments. We conduct a Prolific survey of early-career workers aged 18-26. Using hypothetical firm-choice experiments, we examine how students trade off first-job salaries with other human capital investments. We additionally partner with the UMass Boston Career Center to survey the senior graduating Classes of 2025 and 2026.
Planned Number of Clusters We plan to survey at least 1000 individuals, across 2 schools. If we do not attain enough individuals across the two schools, we are considering expanding the survey to other schools in the University of Massachusetts and University of California system. We plan to survey 2000 participants on Prolific and 500 participants in the University of Massachusetts system.
Planned Number of Observations 1000 individuals 2500 individuals
Sample size (or number of clusters) by treatment arms 1000 individuals, split 500/500 into the AI treatment vs. not. 2500 individuals, split 50/50 into the AI treatment vs. not.
Power calculation: Minimum Detectable Effect Size for Main Outcomes With 1000 individuals, assuming 80% power and 5% statistical significance, our minimum detectable effect size is 0.18 of a standard deviation. With 2500 individuals, assuming 80% power and 5% statistical significance, our minimum detectable effect size is 0.11 of a standard deviation.
Back to top