Direct comparison
Odds Ratio vs. Risk Ratio
They match at low baseline risk but split apart as it rises — a worked table shows an odds ratio of 6.0 next to a true risk ratio of 2.0.
Ask about Odds Ratio vs. Risk Ratio
Answers are drawn from this comparison and the rest of the CASRAI corpus, with a link to every source.
Answers are AI-generated from CASRAI’s own published pages and can be wrong, so check the linked sources before relying on one; your question is logged without personal data — never sold, never used to train a third-party model — to show us what CASRAI is missing, so please do not type personal or confidential details. How we use this
How do Odds Ratio, Risk Ratio compare side by side?
The table below compares Odds Ratio, Risk Ratio across 10 procurement-relevant dimensions, from what it measures through when to prefer it.
Side-by-side comparison
| Dimension | Odds Ratio | Risk Ratio |
|---|---|---|
| What it measures | Odds of the outcome in the exposed group vs. odds in the unexposed group | Probability (risk) of the outcome in the exposed group vs. probability in the unexposed group |
| Formula | [p₁ ÷ (1 − p₁)] ÷ [p₀ ÷ (1 − p₀)] | p₁ ÷ p₀ |
| Study designs it can be validly calculated from | Case-control, cohort, cross-sectional, and any logistic regression output | Cohort studies and randomized trials only — requires a true incidence denominator, which a standard case-control sample does not have |
| At low baseline risk (rare outcome) | Closely approximates the risk ratio — the classical "rare disease assumption" | Reference value the odds ratio is approximating |
| At high baseline risk (common outcome) | Diverges sharply from the risk ratio, always farther from 1 in the same direction | Stays the actual proportional change in risk, by definition |
| Worked example — baseline risk 40%, true RR = 2.0 | 6.0 (would read, mistakenly, as a 6-fold effect) | 2.0 (the actual doubling of risk) |
| Direction of the divergence when RR ≠ 1 | Always overstates the magnitude of effect relative to RR (exaggerates away from 1) | Never needs the correction — it is the quantity being exaggerated |
| Non-collapsibility | The crude (unadjusted) and covariate-adjusted odds ratio can differ even with zero confounding, purely as a property of the odds scale | Approximately collapsible — adjusting for a variable that is not a real confounder moves the risk ratio only slightly |
| How it typically gets misreported | Narrated in "X% more/less likely" language, which is risk-ratio phrasing, without checking whether the rare-disease condition actually held | The quantity that "X% more/less likely" language is actually describing |
| When to prefer it | Case-control data, or any adjusted estimate coming out of a logistic regression model | Whenever incidence is directly observable (cohort study or RCT) and the outcome is not rare |
Common questions
Common questions about Odds Ratio vs Risk Ratio
Why is my odds ratio so much bigger than my risk ratio?
+
Because the outcome in your data is common rather than rare. Odds and risk are mathematically identical only in the limit as risk approaches zero; as baseline risk climbs, 1 − p₁ shrinks faster than 1 − p₀ in the odds formula, which pulls the odds ratio away from 1 faster than the risk ratio moves. The exact relationship is OR = RR × [(1 − p₀) ÷ (1 − p₁)] — when RR is above 1, that bracketed term is always above 1 too, so OR is always the larger number.
When is it safe to read a case-control odds ratio as if it were a risk ratio?
+
When the outcome under study is rare in the source population — conventionally, under roughly 10% cumulative incidence, though there is no hard cutoff and the approximation degrades gradually rather than failing at a specific threshold. This is the "rare disease assumption" that makes case-control odds ratios a usable stand-in for the risk ratio a cohort study of the same population would have produced. It does not apply to case-control studies of common outcomes.
Can a case-control study report a risk ratio directly?
+
Not from a standard case-control sample. Case-control designs fix the ratio of cases to controls by study design rather than by true population prevalence, which means the sample cannot be used to calculate incidence, and risk ratio requires incidence in both groups. Odds ratio is the statistic that design can validly produce; risk ratio requires a cohort or trial design with a real denominator.
Is an odds ratio of 2.0 the same effect size as a risk ratio of 2.0?
+
Only if the outcome is rare enough that the rare-disease assumption holds. At a 40% baseline risk, an odds ratio of 2.0 corresponds to a risk ratio closer to about 1.4, not 2.0 — the two scales converge only as baseline risk falls toward zero, so treating "OR = 2.0" and "RR = 2.0" as interchangeable statements is only ever an approximation, and a worse one the more common the outcome is.
Going deeper
