Apertus
← Library
sleep experiment

2pm vs 5pm caffeine cutoff

Test how late you can drink coffee without harming your sleep. Compare a 2pm cutoff to a 5pm cutoff over 4 weeks.

A
Treatment A
Last caffeine by 2pm
B
Treatment B
Last caffeine by 5pm
Outcome
Deep sleep
measured in minutes
Design
ABAB crossover
28 days · 14 A/B pairs
Statistical power
79%
With 14 A/B pairs and an expected effect size of d = 0.80, this experiment has a 79% chance of detecting a real effect (α = 0.05, two-sided).
To reach 80% power for this effect you'd need ~15 pairs.
Confounders to control
  • ·alcohol
  • ·exercise timing
  • ·screen use after 9pm
Full protocol document
## Protocol: caffeine cutoff

**Goal:** Find your personal cutoff time for caffeine that doesn't sacrifice deep sleep.

**Design:** 4-week crossover, ABAB. 7 days strict 2pm cutoff, 7 days 5pm cutoff, repeat.

**Constraints:**
- Same total daily caffeine (e.g., 200mg).
- Same caffeine source.
- No naps after 4pm.
- Bedtime within ±15 min of habitual time.

**Outcome:** Deep sleep minutes from sleep tracker.
Start a run
Mode