Significance Testing
The Significance card runs up to three independent statistical tests over the chart and decorates the bars whose difference from a reference is statistically significant. It is the same card as the Line Chart — same modes, same statistics, shared with the SDM Smart Table — with display modes suited to bars.
Prerequisites
| Data Role | Type | Role in the test |
|---|---|---|
| Base (sample size) | Measure | n of each bar. Bound once, it applies to every series; bound through the legend, each series carries its own base. Required by every mode except an external reference that carries its own result. |
| Standard Deviation | Measure | Per series and per bar. When present, switches the test from proportions to means. Optional. |
| Significance reference | Measure | One to three measures read by the External reference mode. Not drawn. Optional. |
The Significance card appears in the Format pane as soon as a Base or a Significance reference measure is bound.
Card layout
| Group | Contains |
|---|---|
| Tests | Significance Level, Variance Method, Value scale, and the three mode dropdowns Test 1 / Test 2 / Test 3 |
| Test 1, Test 2, Test 3 | Reference settings and display options of each test |
The level, the variance method, the value scale, the regular expressions and the match status work exactly as in the Line Chart — see that page for the details and the statistical conventions.
Comparison modes
| Mode | Each bar is compared with |
|---|---|
| None | — test disabled |
| vs. Preceding category | the previous bar of the same series |
| Regular expression on X | the bar of the same series in the category matching the pattern |
| Regular expression on Y | the bar of the series matching the pattern, in the same category |
| External reference | a measure of the Significance reference role read for that bar — last year, a benchmark, another scope. See External reference |
With a multi-level axis, categories are named by all their levels, so a pattern on X can target a
parent level (2024) or a full path.
Display options
| Property | Description | Default |
|---|---|---|
| Display mode | Icon, Text, Border color, Background color, Bar outline | Icon |
| Positive / Negative color | Used when the bar is significantly higher / lower than its reference | Green / red |
| Custom icon positive / negative (base64) | Replace the default arrows — Icon mode | (empty) |
| Positive / Negative text | Written for the two directions — Text mode | + / - |
Display modes
| Mode | With a data label | Without a data label |
|---|---|---|
| Icon / Text | Right after the label (before it for a label placed left of a negative bar; above or below a rotated label) | Past the end of the bar — above a column, below a negative one, right of a bar, left of a negative one. In a stacked chart, at the centre of the segment |
| Border color | A ring around the label | A ring around the bar |
| Background color | Behind the label text, replacing the label's own background | A translucent tint over the bar |
| Bar outline | The bar border is drawn in the colour of the direction, inside the bar | same |
When several tests decorate the same bar with icons or texts, they queue up one after the other. Several border or outline tests on the same bar nest inside each other.
Good to know
- 100 % stacked — the tests run on the raw values and bases, not on the displayed percentages.
- Rows hidden by an Add. Table critical threshold are not decorated, since they are not drawn, but they still act as references: vs. Preceding category compares with the preceding category even when it is hidden, and a pattern on X can target a hidden category.
- Blank values — the column chart draws a blank value as 0, but a blank value is never tested, neither as a tested bar nor as a reference.