How Does Your Kindergarten Classroom Affect Your Earnings? Evidence from Project Star
Chetty R, Friedman JN, Hilger N, Saez E, Schanzenbach DW, Yagan D · 2011
grade Arctindependentunreplicatednumbers spot-checked
Sample
11,571 Project STAR students, ~95% matched to tax records
Population
Tennessee STAR participants followed into their late twenties.
Design
Long-run follow-up of the STAR class-size experiment using administrative tax records. The experimental contrasts (small vs regular class, teacher assignment) are clean; the much-quoted kindergarten-score-to-earnings gradient is correlational and should not be read as an experimental result.
Key findings
The canonical fadeout-with-persistence result: STAR's test-score effects vanish by grade 8, yet college attendance rises and adult outcomes move. Few levers in education have grade-A long-run validation like this — and few illustrate so clearly that test-score fadeout is not the same as effect fadeout.
Genetic confound
Experimental contrasts are clean: randomization within school means genes cannot differ between arms. The kindergarten-score-to-earnings gradient is family-and-genetics loaded and should be used only as description.
Replication notes
Verified including the earnings-window correction (ages 25-27). Complemented by Dynarski, Hyman & Schanzenbach's Clearinghouse-based attainment estimates.
DOI / URL
Effects
| Outcome | Metric | Value | Measure | Timing | Vs | Horizon | Class |
|---|---|---|---|---|---|---|---|
| College attendance at age 20 | pp | +1.8 (base 26.4%) | standardized | age 20 | business-as-usual | adulthood | attainment |
| Earnings at ages 25-27 | effect | UNDERPOWERED, not null: the point estimate is negative and imprecise, but the CI upper bound (+3.4%) contains the +2.7% predicted from the score effects | standardized | ages 25-27 | business-as-usual | adulthood | attainment |
| Test-score persistence | pattern | score effects fade to insignificance by grade 8 while attainment and noncognitive gains persist | standardized | grades 4-8 | business-as-usual | over-2yr | domain-skill |
| Kindergarten teacher with >10 years experience | earnings | +$1,093 mean earnings at ages 25-27 (much reduced and non-significant for grade 1-3 entrants — the effect is front-loaded) | standardized | ages 25-27 | none | adulthood | attainment |
Cited by
- Class-size reductionmixedconf: highgc: low
- Teacher quality — selection over credentials and workshopsstrong supportconf: highgc: low