Correlation vs cause: why “these two move together” isn’t the same as “this causes that”

Correlation is a starting point

When two things move together, there are at least three possible explanations: the first causes the second, the second causes the first, or something else causes both. A stressful day could lead to both late coffee and poor sleep, with the coffee itself doing little. That “something else” is called a confounder.

Why it’s hard with everyday data

Real lives are messy. Habits cluster: a night out brings a later dinner, a drink and a short night all at once. You can’t easily run a clean experiment on yourself, and a handful of days can show a pattern purely by luck, especially if you look at many possible patterns.

What gets you closer to cause

Even then, analysing everyday data can’t prove cause the way a controlled experiment can. It can show what seems to affect you, and you can then test it deliberately, for example by changing one thing for a week.

• Enough days, with and without the habit, for the pattern to be seen repeatedly.
• Accounting for the other things going on, such as the day of the week, how you slept the night before, and how hard you trained.
• Correcting for the fact that testing many patterns will turn up some that look real by chance.
• Being honest about confidence: “early” is not the same as “confirmed”.

How The Gap approaches it

The Gap runs statistical tests on your own days that adjust for the other factors above and for chance, and labels each finding early or confirmed. It only calls something confirmed after it has held up, and it tells you how far along each pattern is.

A finding is something to think about, not an instruction. This is general information, not medical advice.