The Evidence on Teaching

Teachers, Schools, and Academic Achievement (the teacher-quality magnitude)

Rivkin, S. G., Hanushek, E. A., & Kain, J. F. (with Rockoff 2004 companion) · 2005

grade Bquasi-experimentindependentreplicated
Sample
>500,000 Texas students
Population
Texas elementary students.
Design
Within-school gains models; semiparametric lower bound.
Key findings
Teacher quality is the largest measured within-school factor — and it is invisible on paper. The variance is real; credentials don't predict it.
Genetic confound
Within-school gains net out stable student differences; residual dynamic sorting is the bias the validation literature bounds.
Replication notes
Foundational; replicated across states and specifications.

Effects

OutcomeMetricValueMeasureTimingVsHorizonClass
1 SD of teacher qualitySD/yr>=0.11 math (lower bound; CFR: ~0.14 math / 0.08 ELA) — exceeds a 10-student class-size cutstandardizedannualnonenot-applicabledomain-skill
Observable credentialsprediction~none: certification null, master's null, experience gains concentrated in years 1-5standardizedn/anonenot-applicabledomain-skill

Cited by

Teachers, Schools, and Academic Achievement (the teacher-quality magnitude) · The Evidence on Teaching