IRT — Polytomous

Coming soon

Psychometrics (legacy hub)

This test is implemented and is currently going through StatMinds’ production verification: every statistic is independently checked against a trusted reference (scipy / R), locked with regression tests, and the screen is exercised across assumption-met/violated and significant/non-significant scenarios before it opens up.

See what’s live now

Item Response Theory calibration for ordered polytomous items (Likert scales, partial-credit scoring).

Three models: GRM (Graded Response Model — Samejima 1969, the standard for Likert), PCM (Partial Credit Model — Masters 1982, Rasch family with item-specific category thresholds), RSM (Rating Scale Model — Andrich 1978, common rating-scale structure across all items). Reports per-item discrimination (a) + category thresholds (b_k) + category characteristic curves + test information function + person θ. Engine auto-picks model by IC + theoretical fit.

Worked example

How do Likert items function across the trait continuum?

A graded-response IRT model was fit to 15 five-point items, estimating discrimination and ordered thresholds per item.

Result

Discriminations were strong (1.2–2.4) and category thresholds were correctly ordered for all items, indicating well-functioning categories.

How you'd report it (APA)

A graded-response model showed strong discriminations (1.2–2.4) with correctly ordered thresholds across items.

When to use it

  • Likert questionnaire calibration (GRM default)
    20-item depression scale (5-point Likert: not at all → extremely) on n=800.
  • Partial Credit / Rating Scale Model (Rasch family)
    10-item motor-functioning rating scale (0-3 partial credit per item) on n=400 patients.

When NOT to — use instead

Assumptions (and what to do if they fail)

Items use an ordered (Likert-style) response scalemedium

Check: See the assumption diagnostics in the workspace.

If violated: The workspace flags this and suggests a robust or nonparametric alternative.

Single underlying latent trait (unidimensionality)medium

Check: See the assumption diagnostics in the workspace.

If violated: The workspace flags this and suggests a robust or nonparametric alternative.

Local independence (item responses independent given θ)medium

Check: See the assumption diagnostics in the workspace.

If violated: The workspace flags this and suggests a robust or nonparametric alternative.

Listwise exclusion of incomplete response setsmedium

Check: See the assumption diagnostics in the workspace.

If violated: The workspace flags this and suggests a robust or nonparametric alternative.

Ready to run a IRT — Polytomous on your own data?

Guided setup, automatic assumption checks, effect sizes, figures and an APA write-up.

Run this test →