# Build banner tables and apply weights

Canonical page: https://www.questionpunk.com/support/banner-tables-and-weighting

Define the groups used in tables and create a weighting scheme with a visible diagnostic preview.

Last reviewed: 2026-10-12

## Before you start

Tables must be enabled for your workspace. Open Results → Tools and check for Tables. Viewing needs study access; editing banners or weighting schemes needs edit access. For population weighting, have defensible population targets and record how the sample was recruited.

## Steps

### 1. Define the banner

Open **Results → Tools → Tables**, then **Edit banner**. Add column groups from single-choice or numeric questions. Keep one column per answer, recode or merge categories, or define numeric ranges. Save the banner.

### 2. Check the tables

Compare totals and groups, and read the base sizes and calculation notes. Confidence letters refer to differences between columns under the displayed test settings. A column with too small an effective base may be shown without being tested.

### 3. Create a weighting scheme

Open **Weighting → New scheme**, name it, and choose **Rake to population targets** or **Use a weight variable already in the data**.

### 4. Enter targets or choose weights

For raking, add variables and enter population percentages that total 100% for each variable. Define numeric bands carefully: both range endpoints are included. For supplied weights, select the numeric weight variable. Inspect missing values and unusual codes before continuing.

### 5. Review diagnostics and activate

Check target fit, convergence, effective sample size, design effect, weight distribution and warnings. Adjust targets or optional caps and floors if justified. Create or save the scheme, then choose **Make active**. Return to Tables to verify the active scheme is used.

## Details

The active weighting scheme applies in Tables. Other Results or analysis views are not automatically weighted; analysis views that offer a weight variable need their own selection.

Weighting can adjust imbalance on chosen variables but cannot turn an opt-in sample into a random sample. Use the displayed effective sample and sampling assumptions when interpreting results. Open-ended numeric bands can accidentally include codes such as 999; check observed ranges and missing-value coding.

## Check the result

Tables use your saved banner and active weighting scheme, with diagnostic information available for interpretation.

## If something goes wrong

If Tables is absent, the feature may not be enabled. If edits are unavailable, check study permissions. Targets that do not sum to 100% prevent a valid raking setup. If the preview does not converge or shows extreme weights, review coding, target coverage and sparse groups before activating. Saving a scheme alone does not make it active.

## Related articles

- [Quantitative analysis](https://www.questionpunk.com/support/quantitative-analysis.md)
- [Import responses into an existing survey](https://www.questionpunk.com/support/import-responses.md)