Ordinal Logistic Regression calculator

Fit ordinal logistic regression models (cumulative link models) online. Assess proportional odds assumptions and coefficients.

What it does

Models how predictors relate to an ordered categorical outcome (e.g. mild / moderate / severe), respecting the ordering.

When to use it

Your outcome has 3+ ordered categories and you want a single summary of the shift toward higher categories per predictor.

Worked example

You want to know what drives a satisfaction rating of poor / fair / good / excellent. The outcome has a natural order but the gaps between the levels are not equal, so it is neither a number nor an unordered category. Predictors might be waiting time, whether the issue was resolved first time, and which site handled it.

Alternatives

• a partial proportional-odds model or separate logistics — the proportional-odds assumption fails for some predictors
• Logistic — the outcome really has only two categories
• Linear — the outcome is continuous (or has many levels and is roughly symmetric — state that shortcut explicitly)

How to read the output

Coefficient (Estimate) — log-odds scale

A common log-odds effect across the category thresholds; exp(b) is the odds ratio of being in a higher category per 1-unit increase. The app shows the raw log-odds coefficient. One OR summarises all thresholds — it is only meaningful if proportional odds holds.

Proportional-odds assumption

The model assumes each predictor's effect is the same across every category split. This is NOT tested here. If effects plausibly differ by threshold, the single-OR summary can mislead — inspect per-threshold coefficients.

z-statistic / 95% CI

Wald z and a CI per coefficient, as for other GLMs. A CI spanning 0 (log-odds) / 1 (OR) is undetermined.

How this calculator is validated · Which statistical test should I use?