Replication Studies
This registry tracks our systematic replications of key causal studies from the literature review. We are starting with one replication and hope to add more over time. All replication code and data are publicly archived on Zenodo and GitHub.
Completed Replications
Replication Note: Jackson, Johnson & Persico (2016)
The Effects of School Spending on Educational and Economic Outcomes
Jackson, C. K., Johnson, R. C., & Persico, C. (2016). The Effects of School Spending on Educational and Economic Outcomes: Evidence from School Finance Reforms. The Quarterly Journal of Economics, 131(1), 157–218.
Original Paper: Key Findings
10% increase in per-pupil spending → 7.25% higher adult wages (full sample)
10% increase → 11.6 pp higher high school graduation (low-income children)
10% increase → 6.8 pp reduction in adult poverty (low-income children)
IV first-stage F-statistics: 20–30 (well-identified)
Replication Status by Table
OLS estimates of the effect of school spending on educational attainment
Coefficients and standard errors match within rounding. N = 15,353 individuals.
IV estimates using school finance reforms as instruments
First-stage F-statistics reproduced at state level (F ≈ 2.8). Full district-level IV requires restricted-use PSID geocodes.
Heterogeneous effects by family income
Income-stratified OLS confirmed. IV heterogeneity analysis requires district-level data.
The JJP design requires district-level PSID geocodes (restricted-use file). Without district-level variation, the state-level instrument is near-collinear with required fixed effects (F-statistic: 2.8 vs. JJP's 20–30).
Possible Extensions
Extend the JJP analysis window using NLSY97 and more recent PSID waves to test whether the estimated spending effects persist for cohorts educated under post-NCLB accountability regimes.
Use the Stanford Education Data Archive (SEDA) district-level test score data as an alternative outcome measure to assess whether JJP's long-run earnings effects are mediated by test score gains.
Distinguish between adequacy-based and equity-based finance reforms to test whether the type of reform (not just the amount of spending) moderates the estimated effects.
Code & Data
Future Replication Candidates
The following 17 studies were flagged in the literature review as high-priority replication targets. Each drives significant policy spending, has faced credible methodological challenges, and uses data that is publicly available or accessible via application.
| Citation | Rationale |
|---|---|
| Chetty et al. (2014a,b) | VAM stability across different state testing regimes requires verification; the $250K lifetime earnings figure drives major policy decisions. |
| Rothstein (2010) | Falsification tests for VAMs need updating with newer cohort data and modern VAM specifications. |
| Krueger (1999) | Randomization integrity and attrition concerns remain unresolved; long-run effects are contested. |
| Puma et al. (2012) | Critical to verify fade-out mechanisms vs. non-cognitive mediation with longer follow-up data. |
| Lipsey et al. (2018) | Urgent need to replicate negative findings in other state pre-K expansions. |
| Heckman et al. (2010) | Small sample sizes (n=58) warrant replication with modern intensive programs. |
| Jackson, Johnson & Persico (2016) | Long-run adult outcomes rely on historical rollout variations; needs post-2000 replication. (In progress — see above.) |
| Hanushek (1997) | Needs updating with modern quasi-experimental funding data and SEDA achievement measures. |
| Angrist et al. (2010, 2012) | Boston charter lottery effects need replication in post-pandemic context and other cities. |
| Abdulkadiroğlu et al. (2018) | Louisiana voucher negative effects require multi-state verification. |
| Ehri et al. (2001) | Phonics effect sizes need updating with modern structured literacy trials and pre-registered studies. |
| May et al. (2023) | Reading Recovery long-term fadeout needs replication in other pull-out intervention programs. |
| Cook et al. (2015) | High-school math tutoring effects require replication in non-urban settings and with female students. |
| Nickow et al. (2020) | Requires replication focusing on cost-effectiveness and scaling parameters post-COVID. |
| Durlak et al. (2011) | Needs replication distinguishing universal vs. targeted interventions with pre-registered designs. |
| Borman & Dowling (2010) | Needs replication with modern, post-COVID inequality data and updated multilevel methods. |
| Reardon (2011) | Income-achievement gap trajectory needs post-2020 updating with pandemic-era data. |
Have a suggestion? Contact us to propose a replication target.