The Later Start Time Debate

The biological change. At puberty the circadian system undergoes a phase delay. Melatonin secretion, which in a 10-year-old begins in the early evening and produces sleepiness around 8.30-9 pm, shifts later by roughly one to two hours, so that a 15-year-old may not be able to fall asleep much before 11 pm. The shift tracks pubertal stage and is associated with gonadal hormone changes rather than age in years, and it is measured under controlled conditions - dim light, fixed schedules, no phones - so it is not simply an artefact of social life. A second change runs alongside it: sleep pressure appears to accumulate more slowly in adolescents, so even a tired teenager takes longer to fall asleep than a younger child would.

Why this matters for the start time. Adolescents still need around eight to ten hours. If sleep onset cannot occur before 11 pm, an 8.30 am start with a 7.15 am wake-up produces chronic restriction during the week regardless of effort. A 9.15 am start gives back roughly the amount of sleep that the phase delay took away.

Evaluating the 'lazy' claim. It is not supported as a primary explanation, because the phase delay is present in adolescents who have no access to screens, jobs or late socialising, and the same young people often sleep much later on free days - which is a marker of a shifted clock catching up on debt, not of indiscipline. However, the claim is not entirely wrong as a partial account. Bedtime is also pushed later by homework, devices, part-time work and socialising, and these are genuinely modifiable. The accurate statement is that biology sets the earliest feasible sleep time and social factors determine how far past it the adolescent goes. Telling a 15-year-old to go to bed earlier does not work if they are lying awake - that is the one intervention the biology rules out.

Evidence 1 is the weakest, despite looking most impressive. A 68% versus 41% gap is correlational. Students who get eight hours differ systematically from those who get under seven: household income, housing stability, part-time work hours, mental health, motivation, caring responsibilities and chronotype all plausibly affect both sleep and achievement. Reverse causation is also live - disengagement from school and depression both reduce sleep and reduce attainment. The gap is almost certainly real but substantially smaller than 27 percentage points once confounders are accounted for. It also cannot tell us whether changing the start time would change sleep, because it is about students who already differ.

Evidence 2 is the strongest but still weak. It is the only evidence involving an actual change. Its problems are fatal to a causal claim as it stands: only the Year 12 cohort was involved, so any difference could be a cohort effect; the timetable changed at the same time (no Monday periods, shortened lunch), so the start time is confounded with scheduling and with lesson time; there was no comparison group; the outcome was self-reported alertness, not achievement; and one term is far too short for any achievement outcome to emerge.

Evidence 3 is not evidence about effects at all. It measures acceptability, not impact. It is important for implementation but says nothing about whether achievement would rise.

What would establish causation. A cluster-randomised design, or a stepped-wedge design in which different year levels or schools adopt the change in a randomised order, with objective sleep measured by actigraphy, pre-registered achievement outcomes, and the timetable held constant. Existing meta-analyses of school start time changes report small positive effects on sleep duration and on some outcomes, with achievement effects modest - typically well under a tenth of a standard deviation. The board should expect a real but small effect, not a transformation.

Advice: do not implement the 9.15 am start in the current year. Start collecting proper baseline data now, and implement in 2025 when transport can be aligned.

The reasoning is about which claim the board is entitled to act on. The evidence that a later start increases sleep duration in adolescents is reasonably strong and consistent. The evidence that it raises NCEA attainment is weak, and the board's own data does not support it. The board should therefore act on the sleep claim and not promise the achievement claim.

Why not implement now anyway. Without the bus change, students who rely on buses would either arrive late or be dropped at school an hour early, which would cancel out the sleep gain for exactly the students with the least flexibility - a classic case of an intervention that widens inequality. Announcing a 9.15 am start before transport is available, and before the 2023 pilot confound is resolved, risks a public failure that would make the 2025 change harder.

What to do this year. (i) Collect baseline data: actigraphy or a validated sleep measure on a sample of Years 11-13, plus attendance, lateness, and a wellbeing measure, so that a proper evaluation is possible. (ii) Negotiate the 2025 bus contract now, and design the evaluation with a comparison group - for example, stagger the change across year levels. (iii) Address the modifiable social factors immediately: later scheduling of sport and part-time work where possible, no assessments scheduled before 9 am, and evidence-based sleep education rather than lectures about laziness. (iv) Take the 54% opposition seriously as an implementation problem, not as disproof - parental concerns about work and transport are legitimate and are exactly what the 2025 plan needs to solve. This is a low-regret strategy: reversible, informative, and it does not stake the board's credibility on an effect that has not been demonstrated.