Some text on this page is in English for now.
What numbers can't tell you
Health numbers describe groups. You are one person, and you get the final say.
Averages describe groups
Most health numbers come from studying large groups of people. They tell you what tends to happen across many lives, and they can't tell you exactly what will happen in yours. In one study, a new daily habit took 66 days on average to feel automatic, but some people got there in 18 days and others were projected to need 254 Lally and colleagues, 2010. The average was real, and almost nobody was average.
A link is not a cause
Some studies can only show that two things show up together. People who spent 120 minutes a week in nature reported better health, but healthier people may simply get outside more White and colleagues, 2019. Every source on this site is labeled with what kind of study it is, so you can weigh it yourself.
Big-sounding numbers
"Twice as likely" can mean 2 in 1,000 becoming 4 in 1,000 (my example). People read counts like "3 in 100" more accurately than percentages, and relative changes tend to sound bigger than they are Akl and colleagues, 2011. When I give a number, I try to give the count, and to say compared with what.
Tracking can crowd out noticing
Writing down progress helps many people reach their goals Harkin and colleagues, 2016. But a number only sees what it measures. Hours in bed don't show how rested you feel, and a step count doesn't show a good walk with a friend. If tracking starts to feel like a test you keep failing, stop for a while. How you feel is information too.
You know your own life
Your body, your work, your family and your budget aren't in any study. Use the numbers here as a starting point and a way to compare, then decide for yourself. "I'm keeping this as it is" is a fine answer.
Sources for this page
Lally and colleagues (2010). How are habits formed: modelling habit formation in the real world.
Single study, 82 people over 12 weeks: one group of people; results may not repeat elsewhere.
What we use: A new daily habit took 66 days on average to feel automatic, from 18 days to a projected 254; missing one day made little difference.
Checked Oct 11, 2026 · https://doi.org/10.1002/ejsp.674 · DOI 10.1002/ejsp.674
White and colleagues (2019). Spending at least 120 minutes a week in nature is associated with good health and wellbeing.
Survey, 19,806 people in England: shows a link, not a cause.
What we use: People who spent 120 minutes or more a week in nature were more likely to report good health and wellbeing; it did not matter how the time was split up. A link, not proof of cause.
Checked Oct 11, 2026 · https://doi.org/10.1038/s41598-019-44097-3 · DOI 10.1038/s41598-019-44097-3 · PubMed 31197192
Akl and colleagues (Cochrane) (2011). Using alternative statistical formats for presenting risks and risk reductions.
Systematic review, 35 studies: pulls together every study it could find on one question.
What we use: People understood natural frequencies ('3 in 100') better than percentages or probabilities; relative changes looked bigger and more persuasive than absolute ones.
Checked Oct 11, 2026 · https://doi.org/10.1002/14651858.CD006776.pub2 · DOI 10.1002/14651858.CD006776.pub2 · PubMed 21412897
Harkin and colleagues (2016). Does monitoring goal progress promote goal attainment? A meta-analysis of the experimental evidence.
Pooled analysis of many studies, 138 studies, 19,951 people: an average across studies; only as good as the studies inside.
What we use: Checking progress helped people reach goals, more so when the progress was written down.
Checked Oct 11, 2026 · https://doi.org/10.1037/bul0000025 · DOI 10.1037/bul0000025 · PubMed 26479070