The Evidence on Teaching

Teaching Higher / textbook-effects high-water mark (and its non-survival)

Kane, T. J., et al. · 2016

grade Dquasi-experimentindependentfailed
Sample
~1,500 teachers, 5 states
Population
US grades 4-5 math.
Design
Cross-sectional textbook-school matching.
Key findings
The 'textbooks rival teacher quality' claim at its peak — instructive precisely because it didn't survive better data.
Genetic confound
Correlational adoption; schools choosing texts differ on unobservables.
Replication notes
A caution against building policy on single cross-sectional results.

Effects

OutcomeMetricValueMeasureTimingVsHorizonClass
1 SD of textbook effectiveness (2016 estimate)SD~0.10 — did NOT survive the same team's 2019-20 six-state replication (~no differences)standardizedannualactive-alternativeend-of-treatmentdomain-skill

Cited by

Teaching Higher / textbook-effects high-water mark (and its non-survival) · The Evidence on Teaching