Paired or independent — the question that changes the test
Are your two sets of numbers from the same people, or different people?
Independent — two separate groups of people. Group A got the supervised programme, group B got the home programme. Knowing one person's score tells you nothing about anyone in the other group.
Paired (related) — the same people measured twice, or deliberately matched pairs. Before and after. Left knee and right knee. A case matched to a control of the same age and sex.
Why it matters
Paired data has less noise, because each person acts as their own control — their age, their build, their baseline severity all cancel out. A paired test uses that and detects smaller differences. Analysing paired data as if it were independent throws the pairing away and usually turns a real finding into a non-significant one.
An example
Forty-two patients each have VAS pain measured at baseline and at twelve weeks. That is paired — 42 people, 84 measurements, one pair each. A paired t-test applies.
Forty-two patients randomised, 21 to each of two programmes, measured once at twelve weeks. That is independent — an independent t-test applies.
Common mistakes
- Reporting n = 84 when you have 42 people measured twice.
- Using an independent t-test on before-and-after data.
- Forgetting that a matched case-control study is paired by design.
Read next
- How the right statistical test is chosen — Four questions decide it, and none of them require you to know statistics first.
- Which study design you are actually running — Observational or interventional; and if observational, which direction time runs in.
Chavery Research Companion applies this to your own study: it asks the questions in plain language, checks the assumptions against your data, recommends the test, and writes the sentence that reports it. Start free — planning and the master chart cost nothing.