When the Bell Tolls: The Effects of School Starting Times on Academic Achievement

Individual ACT records for all public high school test-takers in the Twin Cities metro area 1993-2002, plus school-level panels for Kansas and Virginia

US public high school students; Minneapolis and several suburban districts made large bell-time changes while St Paul and other suburbs kept early schedules

Difference-in-differences exploiting a genuinely large policy shock — Minneapolis moved its high school start time substantially while neighbouring districts did not — with a decade of individual-level ACT data on the universe of test-takers, then replicated on independent school-level panels from two other states. This is a well-powered, well-identified design on an independent standardized outcome, and it returns a null. Weakness: ACT-taking is a selected sample and composition could shift with the policy; the Kansas and Virginia panels are school-level rather than student-level.

The results do not suggest an effect of school starting times on achievement. The null holds across the Twin Cities individual-level ACT analysis and across the Kansas and Virginia statewide standardized-test panels. This is the cleanest existing test of the distal academic claim on an independent standardized measure, and it fails to find the effect.

Low. The bell-time change was made at district level and applied to whole student populations; the comparison districts absorb secular trends. Residual concern is compositional — families can move between Twin Cities districts, and ACT-taking rates could respond to the policy — but neither mechanism plausibly manufactures a null.