Risk Difference

The risk difference (RD) subtracts the risk in the unexposed group from the risk in the exposed group (or, for an intervention, control risk minus treated risk). Unlike the relative risk, it is an absolute measure expressed on the original probability scale. An RD of 0 means no effect; a positive RD means excess risk from the exposure; a negative RD (an absolute risk reduction) means the exposure prevents cases.

RD matters in population science because it captures public-health impact: it answers "how many cases per 100 people does this exposure add or prevent?" Its reciprocal is the number needed to treat, and multiplying RD by a population size estimates the absolute number of cases attributable to the exposure.

The common pitfall is reporting only relative measures, which can make a tiny absolute effect sound dramatic. A treatment that cuts risk from 0.0002 to 0.0001 has an impressive RR of 0.5 but an RD of just 0.0001 — one case prevented per 10,000 treated. Always pair relative and absolute measures.

For example, if a community program lowers a year's incidence from 8% to 5%, the risk difference is 3 percentage points, meaning 3 fewer cases per 100 participants.