School start times and academic achievement - a systematic review on grades and test scores

21 studies of middle and high school students, PRISMA systematic review

Middle and high school students; outcomes restricted to grades and standardized test scores

PRISMA systematic review with a pre-defined risk-of-bias tool built from modified GRADE items and ROBINS-I. The authors explicitly declined to meta-analyse because heterogeneity in outcome type, delay dose, exposure length, and design was too great and risk of bias too high. They then grouped studies by outcome, dose, evidence quality, and design to hunt for a hidden effect, and found none. This is the single most disciplined appraisal of the achievement question and it is the deflationary one.

Of 21 studies, 9 reported no association between later starts and achievement, 5 mixed, 5 positive, 1 negative, 1 unclear. No generalisable improvement in grades or test scores emerges beyond the level of single studies, under any grouping by outcome, dose, quality, or design. The authors are careful about what this does and does not mean: improved sleep very likely improves cognition, but that has not been shown to translate into measured grades and scores across students, and grades/test scores may be poor instruments for detecting it. Their bottom line — schools could start later "while achievement is likely maintained" — is a null-with-no-harm, not an achievement endorsement.

Not directly addressed. Most pooled studies are observational comparisons between schools with different bell times, where family sorting into districts and schools is a live confound; the review's risk-of-bias tool captures this indirectly as a high risk of confounding across most included studies.