What is an n-of-1 experiment?
By the Claritips research team·Updated July 26, 2026·7 min read·4 sources THE SHORT ANSWER
An n-of-1 experiment is a structured trial with a sample size of one: a single person changes one variable for a set period, measures the same outcomes daily, and compares the results against their own baseline. It answers the question population studies can't: not "does caffeine affect sleep on average," but "does it affect mine."
Where it comes from
N-of-1 trials are a real clinical method, not a wellness invention - used since the 1980s to personalize treatment when patients respond differently to the same drug [1]. Reviews rank well-conducted n-of-1 trials among the strongest evidence for decisions about that specific patient [2] - which is exactly the decision you're making about your own habits. The wellness version trades clinical rigor for practicality, but the logic is identical: your own baseline is the control group.
The five rules of an honest self-experiment
1
Baseline first
Log at least a week of normal life before changing anything. Without a baseline there’s nothing to compare against - you’re just vibing with extra steps.
2
One variable
Change exactly one thing. Cut caffeine AND start running AND go to bed early, and a good result tells you nothing about which one worked - or which two canceled out.
3
Same measures, daily
Pick your outcomes up front (restfulness, energy, mood on a 1-5 scale) and rate them every day, ideally at the same time. Memory is not a measurement instrument.
4
Flag the weird days
Sick, traveling, celebrations - mark them and exclude them from the math. One wedding shouldn’t decide your caffeine policy.
5
Two weeks, then two averages
Long enough to outlast withdrawal effects and bad-luck streaks; short enough to actually finish. Verdict = average during minus average before. That’s it.
The honest limitations
A self-experiment isn't blinded: you know you changed something, and expecting improvement produces some improvement (placebo). Bad weeks also tend to be followed by better ones regardless of what you did (regression to the mean) [3]. Three defenses: demand a decent effect size (half a star, not a decimal), run the full two weeks, and be genuinely willing to record "no effect" - a crossed-off suspect is a real result.
And the boundary that matters: self-experiments are for habits - caffeine timing, bedtime, walks, screens. Never for prescribed medication, and never as a substitute for a clinician when something is actually wrong.
Run one in 14 days
Claritips is an n-of-1 engine wearing a calm app: 60-second daily check-ins build the baseline, Experiments hold you to one variable with compliance dots, flagged days sit out, and the verdict is two averages compared - "sleep during experiment: 4.1★ vs 2.9★ before." Or start on paper with the free Pattern Finder kit.
Start your first experiment - free → FAQ
Is n-of-1 the same as "quantified self"?
Related but stricter. Quantified self is tracking anything about yourself; an n-of-1 experiment adds structure - baseline, single variable, fixed duration, predeclared outcomes - so the numbers can support a conclusion rather than a hunch.
Do I need a wearable?
No. For habits, a 1-5 subjective rating logged daily is a legitimate outcome measure - clinical n-of-1 trials use patient-reported scales constantly. How rested you feel is the thing you actually care about anyway.
What effect size should convince me?
For 1-5 ratings across two honest weeks, a gap of 0.5 or more is worth acting on; 0.2-0.5 deserves a repeat run; under 0.2 is noise. Bigger claims need bigger gaps - or more weeks.
Can I test supplements this way?
Over-the-counter ones like magnesium, yes - it’s one of the most popular experiments, and “no effect” saves you money. Prescribed medication, never: that’s a conversation with your doctor, who can run a real supervised n-of-1 if warranted.
SOURCES
[1] Guyatt G et al. Determining optimal therapy - randomized trials in individual patients. NEJM, 1986.
[2] Kravitz RL, Duan N (eds). Design and Implementation of N-of-1 Trials: A User’s Guide. AHRQ, 2014.
[3] Barnett AG et al. Regression to the mean: what it is and how to deal with it. Int J Epidemiology, 2005.
Claritips shows patterns, not medical advice.