Results
Compare answers and respondent segments
Choose between a cross-tab and a segment comparison, check the denominator, and interpret the reported tests.
Compare two questions
In Analysis, use Compare two questions to put one question in rows and another in columns. For example, compare a participant’s role with their preferred product. Check the counts as well as the percentages: a large percentage in a small cell can represent very few people.
Supported categorical comparisons report a chi-square test and Cramér’s V. Cell markers highlight differences in the table. These are unweighted calculations; turning on survey weighting does not turn them into tests for a weighted sample.
Compare one group with the others
Use Compare a segment to select one answer to a single-choice question. The comparison group consists of other respondents who answered that question, not people who skipped it. This helps answer questions such as whether first-time customers report a different experience from returning customers.
Choice results use chi-square comparisons, while supported numeric and rating results use Welch’s t-test. Active dataset filters and cleaning rules determine which responses are available. Check the sample sizes and missing answers before drawing a conclusion.
Report what the comparison supports
Report the question wording, group definitions, filters, counts, and size of the observed difference. A statistical flag does not establish cause or make a convenience sample representative.
Looking through many comparisons creates more opportunities to find a chance difference. Treat an unexpected pattern as a reason to investigate, and keep a record of which comparisons you planned before collecting data. For configured scales, regressions, or condition tests, use Research analysis and review its complete-item counts separately.