Run significance testing and read the shading
When you are comparing segments in a banner or crosstab, AddMaple can highlight cells that are higher or lower than expected—so you can spot reliable differences without exporting to another stats package.
This guide covers the workflow: build the table, turn shading on, read the results, and decide what belongs in your deliverable. For how the calculations work (Holm adjustment, Cohen's h, weighting), see Significance Testing.
When to use significance shading
Use shading when you need to:
- Compare answer distributions across segments (for example region, wave, or buyer type)
- Scan a banner table for lifts or drop-offs worth explaining
- Decide which subgroup differences to include in a deck or Insight Hub
Shading works on pivot tables with at least two segment columns and more than one category per axis. For broader driver work across the dataset, start with Understand what drives outcomes.
If the stats summary says "no relationship" but individual cells still show shading, see Why does the stats summary say there's no relationship when details show differences?.
Step 1: Build the pivot table
- Pivot the columns you want to compare (for example Satisfaction by Region)
- Confirm both axes have more than one category
- If you use respondent weights, apply them in the pivot builder so counts and shading reflect weighted bases
Step 2: Turn on significance shading
- Open the chart menu on the left of the page
- Toggle Significance Testing on
- AddMaple switches the table to % of column view so segments share a comparable base
- A legend appears under the table explaining the colors
Step 3: Read the colors
| Visual | Meaning |
|---|---|
| No color (white/grey) | Holm-adjusted p > 0.10 — no clear signal |
| Faint circle | Marginal signal (0.05 < p ≤ 0.10) — worth a second look |
| Background shading | p ≤ 0.05 — statistically significant after correction |
| Warmer shading | Observed share higher than expected for that cell |
| Cooler shading | Observed share lower than expected |
| Deeper color | Stronger effect size (Cohen's h) |
Hover a colored cell to see the z-score, Holm-adjusted p-value, effect size, and expected baseline percentage.
Colors stay the same when you switch between % of column, % of row, and % of all—significance is computed from the underlying counts, not the display format.
Step 4: Focus on segments with enough base
Combine shading with filters so you are not over-interpreting thin cells:
- Filter to the audience or wave you are reporting
- Check base sizes before calling out a shaded cell in a deliverable
- Open supporting rows or text comments when a difference needs a "why" story
See Exploring your data for filter workflows.
Step 5: Move findings into your deliverable
Once you trust a difference:
- Save the chart to My Collection or Add to Dashboard
- Add a short callout explaining the segment and direction of the lift
- For ranked discovery across the whole dataset, use Related columns alongside table shading