A banner is the set of variables you put in the columns — typically gender, age group, or region. Every other variable in your file is then tabulated against it, one table per variable.
Each banner column gets a letter. When a figure is significantly higher than another column in the same banner variable, the letters of the columns it beats appear next to it. So 27.1% B means this column is significantly higher than column B.
Columns are only ever compared within the same banner variable. Comparing “Men” to “25–34” would be meaningless — the two groups overlap, and the test assumes they don’t.
Proportions use a two-tailed z-test with pooled variance; means use Welch’s t-test. Columns with a base under 30 are not tested; under 100 they are flagged as fragile.
As is standard in banner books, no multiple-comparison correction is applied — across many tables and columns, a few significant letters will appear by chance alone. Treat isolated letters as leads, not proof.
Nothing is uploaded. The file is read and all calculations run inside your browser — close the tab and the data is gone.
1
CSV, XLSX or XLS — one row per respondent
Drop your data file here
or click to browse — CSV, XLSX, XLS
2
Types are detected automatically — correct anything that looks wrong
3
Pick 1 to 5 variables for the columns