Cross-tab analysis (cross-tabulation, banner tables) is the most-used analytical method in survey research. Take a question's responses, break them out by demographic or behavioral segments, and check whether the differences between segments are statistically significant.

If you've worked in market research for more than a year, you've stared at hundreds of cross-tabs. Done well, they answer the most common stakeholder question: "is this difference real?"

What a cross-tab looks like

Question: "How likely are you to recommend our product?" (NPS, top-2-box %)

Banner: Age × Gender × Region

	Total	18–34	35–54	55+	Male	Female	West	East
Sample size	1,000	320	380	300	480	520	450	550
Top-2-box	42%	48%	41%	36%	39%	45%	47%	38%

A "stub" is the row variable (the question). A "banner" is the column variable (the segments). A full report might have 20 questions in stubs and 8 banner segments — that's 160 cross-tabs, often delivered as a single Excel workbook.

Significance testing

The thing that separates a market research cross-tab from a pivot table is significance testing. The 48% (18–34) vs 36% (55+) difference looks meaningful, but is it?

Standard tests:

Column proportions test (z-test): for percentage differences between two columns
Chi-square: for an overall test of independence
Means test: when the stub is a numeric (e.g., average score)

Tabular Pro flags significant cells with letters: column A's 48% is significantly higher than column C (55+)'s 36% — annotated as "C" in column A.

When to use cross-tabs

Subgroup comparison of any survey question
Banner reports — "give me the top 20 questions by age, gender, region, income"
Stakeholder requests — "is this gap statistically significant?"
Tracker reporting — same banner every wave, deltas highlighted

When not to use cross-tabs

Three-way interactions. Cross-tabs handle one breakout at a time. For "age × gender × region simultaneously," a regression is cleaner.
Continuous outcomes. Cross-tabs work on percentages and means; for relationships between continuous variables, use correlation/regression.
Open-ends. Cross-tabs can break out coded open-ends, but the open-ends themselves need coding first.

Weighted cross-tabs

When samples don't match population demographics, you weight. Tabular Pro's survey weighting flows through to cross-tabs automatically — weighted percentages, weighted significance testing, weighted bases.

How Tabular Pro runs cross-tabs

Native cross-tab analysis built into the platform
Column proportions, chi-square, and means tests
Letter-annotated significance flags
Banner builder with custom column groups (nets, top-box)
Weighted analysis built in
Export to Excel, SPSS, or directly into a Tabular Pro dashboard