Comparison Charts
The Comparison Charts section lets you compare how sub topics are associated with two user-defined categories — for example, comparing which topics are most linked to positive versus negative sentiment, or to two different customer segments. It uses the same differentiating terms approach as the Diff Terms analysis type and shows the results as a Dumbbell chart by default.
Configuring the comparison
Section titled “Configuring the comparison”The controls at the top of the section define what is being compared. After making changes, click Apply to update the chart.
| Control | What it does |
|---|---|
| Variable | The supporting variable to compare by (for example, Overall Sentiment or a customer type variable) |
| Category 1 | The first value from that variable (for example, Positive) |
| Category 1 Label | The display label for Category 1 shown in the chart. Required. |
| Category 2 | The second value to compare against (for example, Negative) |
| Category 2 Label | The display label for Category 2 shown in the chart. Required. |
Reading the chart
Section titled “Reading the chart”Sub topics are listed down the left of the chart, and the horizontal axis shows Proportion % — the proportion of responses in which each sub topic appears.
- Sub topics above the dashed line are most associated with Category 1 — they appear more often in responses from that group.
- Sub topics below the dashed line are most associated with Category 2.
- Each sub topic has two data points connected by a line: one dot for its proportion in Category 1, and one for its proportion in Category 2. The further apart the dots, the more strongly differentiated that sub topic is between the two categories.
- Your category labels appear on the right-hand side of the chart under Most Associated With:, as coloured bars.
Chart controls
Section titled “Chart controls”Below the chart, the Chart Controls section lets you adjust how results are displayed.
Significance Filter
Section titled “Significance Filter”Toggle the Significance Filter on to limit the chart to sub topics where the difference between the two categories is statistically significant — filtering out differences that could be due to chance.
When the Significance Filter is on, two additional controls appear:
| Control | What it does |
|---|---|
| Significance Filter Mode | How the filter is applied — either highlighting or labelling significant results |
| Significance Filter Threshold | The p-value threshold used to determine significance (for example, 0.05) |
Display Amount
Section titled “Display Amount”Sets how many sub topics are shown per category in the chart. The minimum is 6 and the maximum is 20. Increasing this value shows more sub topics but may make the chart harder to read.
Frequently asked questions
Section titled “Frequently asked questions”What variable should I use for the comparison?
Any supporting variable in your project can be used. The most common choice is a sentiment or satisfaction variable — comparing, for example, respondents with high satisfaction scores against those with low scores. You can also compare demographic segments, time periods, or any other categorical variable from your data.
Why does the chart look the same as Diff Terms in the Workbench?
Comparison Charts uses the same underlying differentiating terms model as the Diff Terms analysis type, but is pre-configured within the Text Auto Report for quick access without opening the Workbench. The chart type and controls are the same.
Can I change the chart type from Dumbbell to something else?
The Comparison Charts section in the Text Auto Report uses the Dumbbell chart by default. To switch to a Lollipop chart, select Show Lollipop chart from the section menu (⋮). To use other Diff Terms chart types (Bubble, Diagonal, Horizontal), run a Diff Terms analysis in the Workbench.