An outcome with ordered categories: strongly disagree through strongly agree, low, medium, high. Why treating it as a number is a compromise, the assumption that decides whether the model holds, and how to report it in APA 7.
An ordinal outcome has categories with a meaningful order but no guaranteed equal spacing. Strongly disagree, disagree, neutral, agree, strongly agree are ordered, but the distance from neutral to agree is not necessarily the distance from agree to strongly agree.
Treating such a variable as numeric and running linear regression assumes those gaps are equal. It is a widespread compromise and often survives contact with reality, particularly on summed multi-item scales. On a single item with five categories, the assumption is harder to defend and the predictions can fall between categories that do not exist.
Ordinal regression, in its proportional-odds form, models the probability of falling at or below each category boundary. It uses the ordering without pretending the spacing is known.
This is the assumption that makes the model compact, and the one that most often fails.
Proportional odds means a predictor's effect is the same at every threshold. Whatever autonomy does to the odds of moving from disagree to neutral, it does the same to the odds of moving from agree to strongly agree. That is why the model reports one coefficient per predictor rather than one per boundary.
It is a strong claim, and it is frequently untrue. A predictor might strongly separate the bottom of the scale while doing nothing at the top. When it fails, a single coefficient averages effects that genuinely differ and hides the interesting part of the finding.
Test it rather than assuming it. Where it fails, the honest alternatives are a model that allows effects to vary by threshold, or a series of binary logistic models at each cut point reported together. Reporting a proportional-odds model whose assumption fails, without saying so, is the main way this analysis misleads.
Lay the data out one row per case, with the ordered outcome in one column and predictors in others.
Upload the outcome as text with its category labels, not as numeric codes. If Likert responses arrive as 1 through 5, Quanta will read them as numbers, and the ordinal analysis you want will not be offered. Bringing the labels in as text keeps the variable typed correctly.
Confirm the category order. An ordinal model is defined by its ordering, and a scrambled order produces a fitted model that is quietly meaningless.
Configure the outcome and predictors in the Setup panel on the right and the result appears in the center panel.
Check the category counts first. Sparse categories at the ends of the scale destabilize the thresholds, and collapsing them may be the more honest model.
Report coefficients as odds ratios with confidence intervals, state the direction of the outcome ordering, and state that proportional odds was checked.
In a proportional-odds model, higher autonomy was associated with greater odds of endorsing a higher agreement category, OR = 1.48, 95% CI [1.19, 1.84], p < .001. The proportional odds assumption was tested and retained.
Say which direction the odds refer to. An ordinal model can be parameterized either way, and without the statement a reader cannot tell whether the effect is positive or negative.
Report the assumption test explicitly, whatever its result. Where it failed, say what you did instead.
Give the category distribution. Thresholds estimated from a handful of cases in an end category are the least stable part of the model.
Every result carries a ReliCheck Intelligence card with one action: explain this result in plain language. It runs entirely on your Mac with Apple Intelligence. No upload, no API key, nothing to configure. The model only explains the result Quanta has already computed and validated. It never computes a statistic, and it never applies a decision on your behalf.
From the action bar you can add the result to a report, copy the table, or copy the APA text. Exported reports carry the APA table, its note, and a caution specific to that analysis beneath it.
Quanta's logistic, Poisson, and ordinal regressions are validated against R's glm and MASS::polr, with coefficients matched to 1e-3. The engine is pure Swift with no R or Python at runtime. Full record on the validation page.
Regression is part of every Quanta subscription, $9.99 a month or $79.99 a year after the 30-day trial. One price for everyone. See pricing.
The analysis, the report builder, and the on-device explanations all work with the network off. Your dataset is never uploaded.
Logistic regression for a binary outcome · Multiple linear regression for a numeric outcome · Spearman correlation for two ordinal variables without controls · the complete list on the analyses page.