Excerpts from the project presentation:

- Large literature on how interventions such as better teacher quality or smaller classes affect achievement as measured by test scores

- Much less evidence on whether interventions that increase test scores improve outcomes such as earnings

- Problem: few datasets link information on early childhood test scores with data on adult outcomesWe link data from the STAR experiment to US tax records to evaluate the long-term impacts of education interventions

- Question: Are higher test scores a good proxy for improvements in adult outcomes? Do small classes and better teachers/peers improve adult outcomes to the extent they improve test scores?

....

Conclusions

1. Early childhood class effects fade out in test scores but reappear in adult outcomes

2. Contemporaneous test scores are a valid proxy for the benefits of early childhood interventions

- 1 SD higher in test scores increases earnings by 14.8%
- Intervention-based estimates similar to OLS with controls