TukeyHSD
TukeyHSD
The TukeyHSD node performs Tukey’s Honest Significant Difference post-hoc test from a fitted linear model using TukeyHSD(aov(model_lm)). It reports pairwise comparisons between factor levels.
What it does
- Retrieves a stored linear model from the upstream Linear Model node
- Runs
TukeyHSD(aov(model_lm))on the backend - Returns a comparison table with effect sizes, confidence intervals, and adjusted p-values
- Optionally generates and displays a Tukey differences plot below the table
How to use it
- Fit a linear model - add a Linear Model node, set a formula with at least one categorical predictor, and tick “Output to storr”
- Connect the model - drag an edge from the Linear Model node to the TukeyHSD node
- Click Run - Tukey multiple comparisons are computed and displayed as a table
Configuration
| Setting | Required | Description |
|---|---|---|
| Model connection | Yes | A Linear Model node with “Output to storr” enabled |
| Plot differences | No | If enabled, also generates a Tukey differences plot (plot(TukeyHSD(aov(model_lm)))) |
| Comment | No | Annotation for generated R code |
Output
Displays a table with these columns:
comparison- factor levels being compareddifference- estimated mean differencelower_ci- lower confidence interval boundupper_ci- upper confidence interval boundp_value- adjusted p-value
When Plot differences is enabled, the Output tab also shows a Tukey differences plot below the table with Zoom and Download controls.
Generated R code
TukeyHSD(aov(model_lm))With optional plot:
TukeyHSD(aov(model_lm))
plot(TukeyHSD(aov(model_lm)))Tips
- Use at least one factor (categorical) predictor in your model formula to get pairwise comparisons.
- If no factor levels are present, the node will return an informative error.
- Enable Plot differences when you want a quick visual check of pairwise confidence intervals in addition to the numeric table.