Banner table
Weighted crosstabs with nested banners and significance letters.
| Brand | Total | 18–34 | 35–54 |
|---|---|---|---|
| Northwind | 42% | 51% A | 33% B |
| Harbor | 28% | 22% B | 36% A |
| Field & Co | 18% | 16% | 21% |
When to use it
Use a banner table when the deliverable is the book a client already reads: a stub down the side, breaks across the top, a total column, and letters where columns differ. A single unweighted 2×2 without letters is chi-square. A map of the same association is correspondence analysis.
Running it in Tensr
Options
Prop
Type
Reading the output
A banner is the book a client already reads: rows of brands or statements, columns of a break, a total column, and letters where columns differ. The grid stores the weight, the base, and the row identities next to every cell.
Compare each column with every other column at 95%. A letter means that cell is higher than the named column. Put region under age and the total column stays the weighted total of the rows, not a sum of the displayed breaks.
Letters are withheld when a column base is below the cutoff, even if the percentage looks decisive. Unweighted counts stay available beside the weighted percent so a client can see the base. The grid is keyboard reachable, and each letter has a text alternative in the cell, not colour alone.
Coming from SPSS
SPSS is Analyze → Tables → Custom Tables. Tensr’s item uses the same path. SPSS letters are a custom-tables option; Tensr’s column letters start on, with a low-base cutoff of 30.
Related
Chi-square is the unweighted 2×2 without letters. Correspondence analysis maps the same stub-by-banner association.