Analyse

1. Use your own data

Drop a CSV here, or click to choose a file

...or paste CSV text

...or type in a 4×4 matrix

Counts only; response times need trial-level data.

Or load a published example

A brief selection of real confusion matrices from the GRT literature, so you can see and compare GRIN against a classical maximum-likelihood fit on published data.

Format

One row per trial: stimulus, response (as A1B1-style codes, your own labels like Happy/Male, or 0–3), optional participant. A rt column is accepted but ignored: the response-time model is withdrawn pending validation of its replacement, so inference here is counts-only.

2. Choose who to analyse

The classic GRT figure

Show comparison ellipses (dashed)

Compare GRIN's estimate against

Show marginal distributions

Style this export Colour palette

Background Figure title

One colour for all 4 stimuli

The four stimuli are told apart by position either way, colour is a bonus cue, not the only one, so this is safe to turn off if it's more distracting than helpful.

Include legend in plot

Show axis names

Show level labels (A1/A2/B1/B2)

Edit figure labels

Dimension A

Levels (A1, A2)

Dimension B

Levels (B1, B2)

Renames the axes, the confusion matrix headers, and the export figure.

GRIN posterior

Structural conclusions

MLE, full model

MLE model selection (BIC)