Hinrichs 2011.

Methods	Non‐randomized cross‐over trial
Participants	Eligibility: public school students in grades 10 to 12 from 48 districts (73 schools) in the USA who took the ACT between 1993‐2002 Participants: 196,617 students
Interventions	Delaying school start time in Minneapolis and its suburban districts from 07:30 to 08:40, and keeping start time at St. Paul and surrounding suburbs schools constant
Outcomes	Test scores: Minnesota composite ACT test score data (1993 to 2002) for grade 10 to 12, statewide standardized test scores from Kansas and Virginia Average daily attendance rates Outcome measurement: Student survey data (demographic), school data (start time, length of school day, attendance rate, % of students on free lunch) obtained through direct contact with schools and the National Center for Education Statistics's Common Core of Data ACT data
Notes	Funding sources: Georgetown University
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Non‐randomized trial; high risk by definition
Allocation concealment (selection bias)	High risk	Non‐randomized trial; high risk by definition
Blinding of participants and personnel (performance bias) All outcomes	Low risk	ACT and standardized achievement data used
Incomplete outcome data (attrition bias) All outcomes	Low risk	School‐level participation rate controlled
Selective reporting (reporting bias)	Low risk	No evidence that some outcomes were not reported
Other bias	Low risk	No evidence of other biases
Similarity of baseline measures (RCT/CBA only)	Low risk	Matched school districts
Similarity of baseline characteristics (RCT/CBA only)	Low risk	Baseline characteristics listed; analyses done by subgroup to control for baseline differences in characteristics
Diffusion of treatment effects (RCT/CBA only)	Low risk	Students were assigned to different start times by school; start times are not subject to diffusion as a treatment