An odds ratio (OR) compares the odds of an outcome in one group with the odds in another. It is the effect size for 2 × 2 tables and the main result of a logistic regression. An OR of 1 means no difference; above 1, the outcome is more likely in the first group; below 1, less likely. The confidence interval matters as much as the value: if it includes 1, the difference is not significant.
What you need before you write a single word
- The counts and percentages with the outcome in each group.
- The odds ratio and which group is the reference.
- Its 95% confidence interval.
- The test it came with (χ², Fisher's exact, or a logistic regression coefficient) and its p.
The APA 7 format template
[Outcome] was more common among [group 1] (XX%) than [group 2] (XX%), χ²(1, N = XXX) = X.XX, p = .XXX. The odds of [outcome] were X.XX times higher for [group 1] than for [group 2], OR = X.XX, 95% CI [X.XX, X.XX].
All 53 templates, ready to paste with the italics set: the Rescue Vault, $27 →
Formatting rules reviewers actually check:
- Keep the leading zero (an OR can exceed 1): OR = 0.45, not .45.
- Give the 95% CI in square brackets.
- State the reference group: an OR of 3.14 for smokers is 0.32 for non-smokers.
- Italicize OR when used as a statistic.
A worked example
In a sample of 200, 30 of 100 smokers (30%) and 12 of 100 non-smokers (12%) had the condition. The test gives χ²(1, N = 200) = 9.76, p = .002, and OR = 3.14, 95% CI [1.50, 6.58]. Written up in APA 7:
The condition was more common among smokers (30%) than non-smokers (12%), χ²(1, N = 200) = 9.76, p = .002. Smokers had 3.14 times the odds of having the condition, OR = 3.14, 95% CI [1.50, 6.58].
Mistakes reviewers catch
- "3.14 times more likely". That describes relative risk. An odds ratio is about odds: write "3.14 times the odds". Here the risk ratio is 30% / 12% = 2.5.
- No confidence interval. With small samples the CI is often very wide; the reader needs to see it.
- Unclear reference group. Say which group is compared with which.
- Dropping the leading zero. Odds ratios below 1 still keep it: 0.45.
Let KyroStat do the write-up for you
KyroStat reports odds ratios with their confidence intervals for every 2 × 2 table and logistic model, and words them correctly as odds.
Frequently asked questions
How do I report odds ratios from a logistic regression? Give each predictor's B, SE, Wald, p and OR with its 95% CI. See how to report a logistic regression.
What does an odds ratio below 1 mean? Lower odds in the first group. An OR of 0.50 means half the odds; some authors flip the comparison (1/0.50 = 2.00) to make it easier to read.
Which test goes with an odds ratio from a small table? Fisher's exact test when expected counts are below 5; otherwise the chi-square test.