This is an in-progress meta-analysis of randomized evaluations of job training programs in the U.S., currently with 144 estimates from 56 studies.
Most data extraction done by Claude, under supervision. “Claude is AI and can make mistakes.”
Github repo