Date: 2026-08-03. Adversarial-presentation format per §0.1.
The polarized core, both sides' preferred specifications
Tennessee VPK (the negative case). RCT (Durkin, Lipsey et al.): achievement gains gone by end of kindergarten; negative effects on state test scores by 3rd grade, larger by 6th; more disciplinary infractions, worse attendance, more special-ed placement (AERA Open RD design, 6th-grade follow-up, Vanderbilt). Preferred read (skeptics): scaled, ordinary-quality public pre-K can harm. Boston (the positive case). Lottery natural experiment (Gray-Lobe–Pathak–Walters): no persistent test-score effects, but +6pp high-school graduation, +5pp college attendance, +5pp college graduation, fewer suspensions — a behavioral/attainment channel that test-score fade-out misses (NBER w28756). Weiland et al. 2025 (oversubscribed-school subset): no lasting test benefits through middle school (EdWorkingPapers). Preferred read (proponents): judge programs on long-run attainment, not 3rd-grade scores.
What would resolve it: long-run attainment follow-up on the TN RCT cohort (does the negative test effect become a negative attainment effect?), and quality-dimension decomposition (Boston: curriculum + coaching + pay near parity; TN: contested implementation quality). Until then: H7.2 supported — scaled-program effects are smaller than demonstration effects, and effect direction at scale is quality-contingent. Extrapolating Perry/Abecedarian to a national program remains indefensible; so does extrapolating Tennessee.
Standing synthesis for the scorecard
Test-score fade-out is the norm everywhere; the live question is whether behavioral/attainment channels survive (Boston: yes; TN: negative). Design implication: quality parameters (curriculum, coaching, workforce stability) are not garnish — they plausibly determine the sign of child effects. This raises the stakes on §11's finding that quality provisions are the most Byrd-vulnerable part of any bill.