Syntes: Breaking the Early Bell: Lessons from the First Statewide Mandate on School Start Times (2026)

Källa: https://doi.org/10.3386/w35184 (NBER Working Paper 35184, May 2026). Earlier version: IZA DP 17930 (May 2025), https://hdl.handle.net/10419/320524

Verified content

• [full text, May 2026] SB 328 required middle schools to start no earlier than 8:00 and high schools no earlier than 8:30; binding for roughly 80% of California districts (authors' survey); rural districts exempt (p. 3).
• [full text, May 2026] Sleep: YRBS (high-school students, self-report) share sleeping at least 8 hours up 13%, larger among boys; ATUS time diaries, ages 15–18, weekdays: sleep hours +0.712 (SE 0.241), share sleeping at least 8 hours +0.122 (SE 0.058); children under 13: +0.762 hours (SE 0.175) (pp. 3–4; Table 3, p. 27).
• [full text, May 2026] Test scores: SEDA grade 8, 2015–2024 excluding 2020–21; two-way fixed effects maths +0.045 SD (SE 0.012) and English +0.033 SD (SE 0.011); matched DiD against similar districts in other states maths +0.096 SD (SE 0.015) and English +0.077 SD (SE 0.013), all p < .01 (Table 5, p. 33). Pre-period event-study coefficients small and not significant; positive in 2023 and 2024 (p. 32). Urban–rural triple difference: matched maths +0.09, English +0.14 SD (p. 33). Gains larger for Hispanic and economically disadvantaged students.
• [full text, May 2026] Within-state design (teachers' commute arrival times as a proxy for earlier start times; California Department of Education records): maths, English and science gains accumulating over 2023–2025; no effect on chronic absenteeism (p. 4).
• [full text, May 2025 IZA version] Earlier estimates on SEDA grades 3–8 were larger (matched DiD maths 0.169, English 0.149 SD; Table 4) and SAT totals +25 points (Table 5); the May 2026 version restricts SEDA to grade 8.

Limits

• [assessment] Working paper, not peer reviewed; estimates changed between the 2025 and 2026 versions.
• [assessment] Sleep is self-reported (survey or time diary), not measured by device; sleep (high-school students, teenagers) and test scores (grade 8) come from different data and partly different age groups, so the study does not show that the same students who slept more scored higher.
• [assessment] Post-policy years 2023–2024 follow the COVID-19 disruptions; California-specific post-pandemic recovery is a possible confound the authors address with the rural-exemption triple difference and the within-state design.

Added 2026-10-22 by the project (nightly attack run). [full text, May 2026] = read in the NBER May 2026 PDF (page images); [full text, May 2025 IZA version] = read in the stored IZA text; [assessment] = the project's own assessment.

October 2026 revision (read 2026-11-12)

• [full text, Oct 2026] Title page "May 2026, Revised October 2026". YRBS share sleeping ≥ 8 h +0.036 (SE 0.017; 13% of the mean), ≥ 7 h +0.070 (Table 2). ATUS ages 15–18, weekdays, metro areas: sleep hours +0.464 (SE 0.207; "about 28 minutes"), share ≥ 8 h +0.097 (SE 0.044) (Table 3). The May 2026 figures above (+0.712 h; +0.122) are superseded.
• [full text, Oct 2026] Academic outcomes now come only from a within-California design: schools in counties with earlier pre-policy start times (proxied by teachers' commute arrival times in the ACS) against the top quartile plus rural counties; CDE records 2014–15 to 2024–25 without 2019–20 and 2020–21; county-clustered SEs. Share meeting or exceeding standards: high school maths +1.82 (SE 0.69), English +1.75 (SE 0.59), science +1.73 (SE 0.63); grade 8 maths +1.70 (SE 0.48), English +1.70 (SE 0.54), science +0.69 (SE 0.40, p < .10) (Table 5). Suggestive declines in dropout (−0.97 points) and suspensions (−0.32 points), both p < .10.
• [full text, Oct 2026] The SEDA cross-state grade-8 analysis (matched DiD +0.096 / +0.077 SD; TWFE +0.045 / +0.033) and the urban–rural test-score triple difference are gone. The authors: "the uneven recovery from the pandemic across states means that cross-state comparisons of test scores after 2022 cannot isolate the policy"; the within-state design "does not rule out differences in recovery across counties that happen to be correlated with pre-policy start times". State-level SAT (+25 points) is reported as descriptive only (Appendix Table A.5).
• [assessment] The kill-condition of C14 still fails on the revised paper (more sleep, higher test scores), but the test-score evidence now rests on a proxy for treatment intensity inside one state. C14 Bounds corrected 2026-11-12.