Working Paper: NBER ID: w29990
Authors: James J. Heckman; Jin Zhou
Abstract: Empirical studies in the economics of education, the measurement of skill gaps, the impacts of interventions on skill formation, and the value-added literature rely on psychometrically validated test scores. Test scores are taken as measures of an invariant scale of human capital compared over time and people. We examine if conventional skill measures are comparable on mastery of specific task knowledge. An unusually rich dataset from an early childhood intervention is used to test the assumption of scale invariance. We reject the scale invariance hypothesis for multiple skills and cast doubt on the uncritical use of test scores in research.
Keywords: No keywords provided
JEL Codes: C81; I21; J71
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| conventional skill measures (J24) | unreliable invariant scale of human capital (J24) |
| traditional test scores (C12) | not comparable across individuals and over time (J78) |
| mastery of tasks within defined skill levels (J24) | more accurate measure of knowledge (D80) |
| traditional measures (C52) | fail to capture nuanced development of skills (J24) |
| alternative skill measures (C52) | impact assessment of human capital (J24) |