# Behaviour insights

How each behaviour you log goes with your next day’s scores.

Canonical page: https://pulse.portlabs.in/metrics/behaviour-insights/

## What goes in

- **Journal check-ins:** Each behaviour you answered yes or no, over the last 90 days.

- **Next-day scores:** The next morning’s Recovery and HRV, and that night’s Sleep Performance.

## How it is weighted

For each behaviour, Pulse compares the days you answered yes with the days you answered no: the difference in average next-day score. HRV shows as the change in standard deviations from your baseline. To see how sure that difference is, Pulse resamples your days 1,000 times and keeps the middle 90% of the results.

## What the bands mean

- **Positive or negative:** That whole 90% range sits above or below zero: a clear effect.

- **No clear effect:** The range crosses zero.

- **Needs more data:** Fewer than 5 yes days or 5 no days.

## A worked example

Say someone logs alcohol for a month. Over 29 days they answer yes on 8 and no on 21. Pulse pairs each answer with the next morning's Recovery, because the question is what the day after looks like.

On the 21 no days, next-morning Recovery averages higher than after the 8 yes days. The difference of the two averages, yes minus no, comes out at minus 11.6 points: on the day after alcohol, Recovery was about 12 points lower.

That difference alone could be luck, so Pulse resamples. It redraws 8 yes days and 21 no days with replacement, takes the difference again, repeats 1,000 times, and keeps the middle 90% of the results. For this data the range runs from minus 14.7 to minus 8.4.

The whole range sits below zero, so the label is Negative, a clear effect. Had the range run from, say, minus 4 to plus 6, the label would be No clear effect. Had there been only 4 yes days, Pulse would say Needs more data, because it needs at least 5 yes and 5 no days. These figures come from running Pulse's behaviour insights function on 29 check-ins and 30 daily Recovery scores. It shows an association, not proof of cause.

## Limits

This is an association, not proof of cause. It does not adjust for other behaviours or training, so a tag you tend to log on hard days can look harmful. About 1 in 10 behaviours with no real effect will still show a clear one by chance. Days without an answer are left out.

## Sources

- [Efron B, Tibshirani RJ. An Introduction to the Bootstrap. Chapman & Hall, 1993](https://doi.org/10.1201/9780429246593)

## Questions

### Does alcohol lower my recovery and HRV?

Pulse can show you what it did for you. Log alcohol yes or no each day, and after at least 5 days of each, it compares next-morning Recovery and HRV on the two groups. It cannot say alcohol caused the change, because late nights and poor sleep tend to come with drinking. A clear Negative label over 90 days is a strong hint, not proof.

### How many days of journalling do I need?

At least 5 yes days and 5 no days for the same behaviour within the last 90 days. Below that, Pulse says Needs more data. A behaviour you do twice a month will take months to qualify. Fewer behaviours logged consistently work better than many logged now and then, and days you skip are left out for that behaviour.

### Why does a harmless habit show a clear effect?

Pulse checks several behaviours against three scores, and its 90% range means about 1 in 10 behaviours with no real effect will still show a clear one by chance. It also does not adjust for other habits or training, so a tag you log on hard days can look harmful. Treat one result as a prompt to test, not a verdict.

### How is this different from WHOOP Journal's Behavior Insights?

The idea is the same: compare yes days with no days on next-day scores. WHOOP's help page sets a 5 yes and 5 no threshold in 90 days and does not publish its statistics. Pulse uses the same threshold, shows the size of the difference and a bootstrap range, and explains that it is an association. See [WHOOP Journal explained](https://pulse.portlabs.in/blog/whoop-journal-behaviours/).

## Related

- [Recovery](https://pulse.portlabs.in/metrics/recovery/index.md)
- [Heart rate variability (HRV)](https://pulse.portlabs.in/metrics/hrv/index.md)
- [Sleep Performance](https://pulse.portlabs.in/metrics/sleep-performance/index.md)
