The Effect of Evaluation on Teacher Performance
Taylor ES, Tyler JH · 2012
grade Bquasi-experimentindependentunreplicatednumbers spot-checked
Sample
Cincinnati mid-career mathematics teachers cycling through the district's evaluation system
Population
Cincinnati Public Schools teachers and their students.
Design
Exploits the district's staggered assignment of teachers to a year-long, high-intensity classroom-observation and feedback process. Because the year in which a teacher is evaluated is effectively arbitrary, teachers can be compared before, during and after their own evaluation year.
Key findings
Serious classroom observation with feedback produced a lasting improvement in teacher performance (+0.11 SD in math) that persisted after the evaluation year, largest for weaker teachers. Evaluation worked here as development, not as a sorting device.
Genetic confound
Within-teacher comparison; student sorting is not the identifying variation.
Replication notes
The strongest single result for evaluation-with-feedback as a development mechanism rather than an accountability mechanism.
Effects
| Outcome | Metric | Value | Measure | Timing | Vs | Horizon | Class |
|---|---|---|---|---|---|---|---|
| Student mathematics achievement in post-evaluation years | SD | +0.11, largest for initially weaker teachers | standardized | years following evaluation | business-as-usual | unclear | domain-skill |
| Cost | cost | ~$7,500 per teacher for a persistent human-capital gain | standardized | one-off | unclear | not-applicable | unclear |
Cited by
- Teacher quality — selection over credentials and workshopsstrong supportconf: highgc: low