Choice Analysis.

Choice is the simplest question type: respondents pick one answer (single choice) or several (multiple choice). Its variations — Scale, Numeric Scale, NPS, Mood, Rating, Hidden Variables and Score — all analyse the same way, so this one guide covers them all. (New to Analyse? Start with the Analysis Overview.)

Open a choice view and you get a chart and a Glow Table. The controls above the table are its Table Slots — one slot per dimension, each set to where that dimension sits — and the defaults are:

• Choices = Rows, Filters = Columns
• Values = Cell % — each group’s share of the people actually asked
• Chart = Grouped Bars

Here’s "Q37. Thinking about your checkout experience at the store, how satisfied were you with it?" — the example every number below comes from:

Compare groups side by side

The All / Male / Female / Metro / Regional columns above are cross filters: click the + at the end of the table’s filter chips, tick the groups you want, and choose Add to > Current View (or selected views, or every view). Each chip becomes a column in the table and a series in the chart — a visual crosstab. Reading it: 52% of males were somewhat satisfied vs 48% of females. (The ⚙ on the Filters slot manages the same list in one place — reorder, remove, add or create.)

To narrow the whole view to one audience instead — say, one retailer’s customers — use + Prefilter above the chart and watch the n in the top corner drop. Prefilters and cross filters combine happily; the full model (including removing chips with × / Shift+×) is in Prefilters & Cross Filters, and building your own filters is in Create & Import Filters.

How the chart reads the table

Whatever sits in your table’s Rows becomes the chart’s axis; whatever sits in Columns becomes the bars (or lines). Choice views default to Grouped Bars — the Rotate button flips horizontal/vertical, and the chart-type button offers all six types (see Chart Types).

Transpose — flip choices and filters

Want your filters down the side and choices across the top? Set the Choices slot to Columns — the Filters slot automatically flips to Rows (and vice versa; the two dimensions swap axes in one move). On the default Cell % the numbers don’t change: Male / Somewhat satisfied is 52% in either orientation.

See trends over time

If your survey runs in waves, trend any choice question in three moves:

1. Set the Filters slot to Rows and add your wave filters as cross filters — e.g. W10 (Oct), W11 (Nov), W12 (Dec).
2. Keep Values on Cell % or use Row % — with waves in the rows, both show each wave’s own distribution (on a single-choice question they read the same).
3. Switch the chart type to Line.

Each wave becomes a point on each choice’s line — a time series per choice:

Score & average your choices

A whole distribution is sometimes more detail than the story needs. If your choices are ordered — a 1–5 or 1–7 Likert scale, a 0–10 NPS scale, a frequency ladder — you can score each choice and collapse the question to a single average per group:

• Satisfaction: Very dissatisfied = 1 … Very satisfied = 5 → average satisfaction out of 5
• NPS: Detractors (0–6) = −100, Passives (7–8) = 0, Promoters (9–10) = 100 → Net Promoter Score
• Frequency: Fortnightly = 24, Monthly = 12, Bi-monthly = 6, Quarterly = 4, Half-yearly = 2, Don’t see them = 0 → visits per year

To set it up:

1. Click the ⚙ on the Choices slot.
2. Type a score next to each choice (e.g. 1–5), and exclude anything that shouldn’t count — "N/A", "Don’t know".
3. Tick Combine Weighted and name the combined row ("Average", "NPS", "Visits p/year" …), then click Done.

Then set the Values slot to Average. Q37’s average lands at 4.1 out of 5 — "somewhat satisfied" overall — and with gender cross filters you can compare groups on one line: 4.07 for males vs 4.20 for females. (Sum uses the same scores when you want a total instead of an average.)

Sig-test your differences

Open the dashboard Settings (⚙) and tick Subset Significance Test: on Column % views, cells go green or red when they’re significantly higher or lower than the Total column (on Row %, the Total row), at a 95% confidence level by default. Tick Margin of Error to see how reliable each base is. Details and worked calculations: Significance Testing and Margin of Error.

Tell the story across views

The classic pattern: duplicate your choice view a few times and cut each copy differently — one by household income, one by generation, one by retailer, one by wave (that’s exactly the Q37A–Q37E sequence in the demo dashboard). Duplicate and reorder in the View Manager; rename a view by clicking its title and typing; add Commentary (the speech-bubble icon on the chart) so stakeholders read it right. To merge, hide, reorder or relabel the choices themselves, see Configure Options.

And when it all looks right: click Save — switching dashboards discards unsaved changes.

FAQs

My multi-select percentages add up to more than 100%.

They should — on a multiple-choice question each choice is its own share of the people asked, and one person can tick several. Only single-choice questions split 100% between choices.

Why did my numbers change when I transposed?

You’re on Column % or Row % — their denominator flips with the orientation. Switch to Cell % and transpose all you like.

My older dashboards show Column % instead of Cell %.

Cell % is the default for new tables on percentage questions; existing views keep whatever they were saved with. Change any view via its Values slot.

My average says 4 but the chart says otherwise.

The table rounds to your dashboard’s Decimal Places (default 0dp); the chart plots exact values. Raise Decimal Places in Settings to see 4.07 vs 4.20 in the table too.

Where did "Add Global" go?

Retired. The Add to menu now offers Current View, Selected Views or Global (add to every view in one go) — and chips added that way stay ordinary local chips. See Prefilters & Cross Filters.