An insight can point to records worth exploring. It cannot, on its own, explain why a change happened. Strata's correlation views compare recorded measures and show supporting detail. Treat the result as a question grounded in a particular dataset, not as a general rule about your body.
Read the claim and its scope
Open an insight and identify the measures it compares, the time period, and any lag. A same-day association and a next-day association are different comparisons. Keep that distinction when explaining what the card says.
Review the confidence label and supporting details. A stronger statistical association in the records does not establish a causal mechanism or guarantee that it will recur.
Ask which days could be compared
An analysis depends on overlapping observations. Seven sleep records and three wellbeing check-ins do not automatically produce seven complete pairs. Missing data can change which days are represented.
Logging habits may also vary with the day itself. If you record meals more carefully on some days, a nutrient pattern can partly reflect the available logs. Inspect coverage rather than treating every point as equally complete.
- Name the two measures.
- Check the period and lag.
- Inspect the paired observations and missing data.
- Read the uncertainty before forming a conclusion.
Keep association separate from cause
Two measures can move together because of another factor, because the recording routine changed, or by chance. A personal observational record does not randomly assign the conditions needed to isolate a cause.
If several candidate patterns are explored, some may appear interesting without being stable. Return to the individual records and see whether the observation remains useful as more data becomes available.
Choose a proportionate next step
A useful next step may be improving logging consistency or writing down a question for a professional. Avoid changing medication, treatment, or a major dietary plan solely because an insight card suggests an association.
Ask an assistant to summarize the evidence and uncertainties separately. The answer should preserve the original comparison rather than turn it into a confident health explanation.
An illustrative example
An association with a boundary
An illustrative card compares fictional sleep and wellbeing records on matching dates. It invites Alex to inspect the paired records. It does not claim that sleep caused the wellbeing ratings or display a fabricated statistical significance result.
Keep this in mind
Use an insight to investigate a record. Keep its dataset, coverage, and uncertainty attached.