How to read survey crosstabs

Start with a table you can check

This table comes from the October 15, 2026 oat-latte naming study, an actual model-run study with 250 responses. It shows how often Morning Cloud was ranked first. Download all responses or the full crosstabs.

Ask what each percentage divides by

Among male responses, 41 of 128 rank Morning Cloud first (32.0%). Among female responses, 56 of 122 do so (45.9%). Each percentage uses its own group as the denominator. These two percentages do not need to add to 100%.

Recompute the overall result from counts

Across both groups, 41 + 56 = 97 responses rank Morning Cloud first. Dividing 97 by 250 gives 38.8%. Because the group bases differ, calculate the overall result from counts rather than taking a simple average of the two percentages. Results are unweighted.

Describe the difference before explaining it

The absolute difference between the two group percentages is 13.9 percentage points. This describes the output; it does not explain why the groups differ or establish a statistically significant effect.

Check exclusions and multiple-choice rules

All 250 responses are included in this ranking comparison. For any other question, inspect its answered base before calculating. Distinguish missing, invalid and not-applicable answers from substantive choices.

For select-all-that-apply questions, percentages may sum above 100%. For exclusive single-choice questions, valid category counts should add to the answered base.

Use audience differences as a question, not automatic proof

Choose relevant comparisons before inspecting every possible subgroup. Keep the exact question, group bases and differences together. These are model-generated responses, and a larger generated count does not establish population accuracy.

Use a group difference to guide follow-up investigation. It does not establish actual buying behavior, a causal explanation or a significance result.