Relative risk: find the two risks behind the comparison

The word “relative” asks us to find the reference group. A comparison can sound dramatic while leaving the underlying event counts unclear. We use invented nonmedical arithmetic below; it is not a finding about ADHD or advice about a health decision.

Two comparisons with the same ratio

In one fictional parcel record, an event occurs for 2 of 100 parcels in group A and 1 of 100 in group B. The proportions are 2% and 1%. Dividing the first by the second gives a ratio of 2.

In another invented record, the event occurs for 20 of 100 parcels in A and 10 of 100 in B. The ratio is also 2, but the underlying proportions are 20% and 10%. The first pair differs by one percentage point; the second differs by ten.

The same ratio therefore leaves unanswered how common the event was in either invented group. Neither calculation tells us why the groups differ. We have not described how the parcels entered the groups or what else differed between them.

Ask for the reference, not just the headline

Event being compared:
Group A and its reported risk or proportion:
Reference group B and its reported risk or proportion:
Time frame for each:
Which number is divided by which:
Underlying event counts, if available:
Uncertainty or limitations reported:

Reversing the comparison changes the ratio. In the first example, 1% divided by 2% is 0.5. Always say which group is the reference; do not turn “half” and “twice” into conflicting claims by dropping that detail.

Do not turn a ratio into a personal prediction

When reading a medical headline, locate the original source rather than borrowing the parcel example's numbers. Check that the groups concern the same event and a comparable time frame. A label such as “increased risk” does not by itself describe a particular person's future.

Our page explains the comparison, not how to select a treatment or calculate individual risk. The study design, uncertainty, and relevance to a population belong to the actual report and appropriate professional interpretation.

Different questions in the same family

A parent might ask, “How often did it happen?” A teen might ask, “Compared with whom?” A grandparent may want to know whether a headline concerns children or adults. These questions can sit together without someone having to declare the headline either reassuring or frightening before the details are known.

Source

[1] National Cancer Institute. Relative risk. NCI Dictionary of Cancer Terms. Publication date not stated on the inspected entry; accessed September 5, 2026. The opening definition and alternate name support [1]. The cancer-specific discussion is not adopted as Ready advice. All parcel numbers are invented for this explanation.

Related guides

• Absolute risk

  : recover the underlying event chance.

• Sample size

  : ask what the figures were based on.

• Confounding

  : consider what a comparison may leave unresolved.

• Causation

  : distinguish a ratio from evidence about cause.