QuestionPunkQuestionPunk
Start free

Quantitative analysis

Run descriptive statistics, correlations, regression, and segmentation on your survey data, and save analyses you can rerun later.

Explore structured data with summaries, group comparisons, paired measurements, rating scales, reliability, correlations, regression, segmentation, and principal components. Filter and save analyses to rerun or export later. Group summaries can also use weights.

Last reviewed
analysisquantitativestatisticsreport10-20 minutesIntermediateResearchersAnalystsProduct teams

Before you start

Make sure your survey already has responses, or prepare a CSV dataset to import. Check that the variables you want to analyze are typed correctly in the dataset.

Find the analysis workspaces

A navigation overview using actual English-interface demo screenshots, generated English narration, and English/Japanese captions. This is a narrated screenshot sequence, not a continuous recording. Examples are fictional.

Download video
Read the video transcript

In Results, open Tools and choose Themes to code open-ended text responses and review your codebook. Open Tools and choose Stats to run descriptive statistics, crosstabs, and hypothesis tests on structured variables.

Steps

  1. Open Quantitative

    Click Quantitative in the sidebar to see the studies available to you. You can also open it from a survey by using the top section switcher, going to Results, and opening the quantitative workspace for that survey. You can also import an external CSV dataset into its own quantitative workspace.

    English interface · Results → Tools → Stats showing descriptive statistics across 24 demo responses.

  2. Check your data and variables

    In Data & variables, review how each column is typed: Numeric, Single category, Multiple choices, or Open-ended text. Exclude any rows you do not want included in the analysis.

  3. Choose a method

    In Analysis, choose a method: Summarize & compare groups, Likert & rating items, Scale reliability (Cronbach’s alpha), Pearson correlation, Spearman rank correlation, Multiple linear regression, Segmentation (k-means), or Principal components (PCA). Likert needs at least one rating item. Reliability needs at least two items. Other methods, except summaries, need at least two numeric variables. Compare the same people compares two numeric ratings from each respondent with a paired t-test.

  4. Filter and weight the analysis

    Apply filters to narrow the data. For Summarize & compare groups, you can choose a numeric weight column when you have a justified weighting scheme. Check its values and how it was constructed; selecting a column alone does not make a sample representative.

  5. Save and rerun analyses

    Save the analysis with a name so you can return to the same settings later. If the underlying data changes, saved analyses are marked stale so you know to rerun them before exporting.

Details and limitations

Quantitative is a standalone workspace you can open from the sidebar, and you can also open it for a survey from Results. In survey results, Report, Summary, and Responses are the main sections, and Ask opens the analysis assistant alongside them.

Summarize & compare groups covers descriptive statistics, group comparisons, and cross-tabs. Likert & rating items summarizes rating questions. Scale reliability (Cronbach’s alpha) checks how consistently multiple items move together. Pearson correlation and Spearman rank correlation measure how closely two numeric measures move together. Multiple linear regression estimates associations with an outcome while accounting for other factors. Segmentation (k-means) finds groups of respondents with similar answer patterns and lets you export segments. Principal components (PCA) reduces correlated measures to a smaller set of patterns and exports component scores. Compare the same people is for paired measurements: both numeric values must belong to the same respondent.

Open Results → Tools → Stats for descriptive statistics, crosstabs, hypothesis tests, regression, factor analysis, and cluster analysis. On Conjoint and MaxDiff questions in Results → Summary, click Run HB analysis under Hierarchical Bayes (HB) to estimate respondent-level utilities, break results out by segment with Cut by, run the interactive Conjoint Market simulator (Share of preference or First choice, overall or by Segment, across up to 5 product profiles), and download Excel with simulator (Conjoint), Excel (MaxDiff), or CSV (individual utilities). For Price Ladder (Gabor-Granger) questions, the question report computes demand and Revenue index across price points and recommends the revenue-maximizing price.

Saved analyses are reproducible. Each one keeps its setup, including variables, groups, weights, filters, and excluded rows. If the dataset changes, the saved analysis is marked stale so you can rerun it before you export or present it.

You can analyze responses from an existing survey or import an external CSV dataset that does not have a parent survey.

Interpret the result with its limits. Check the included sample, missing values, excluded rows, variable coding, and method assumptions before reporting. A correlation or regression shows an association; it does not establish cause and effect. Keep synthetic test data separate from real participant findings.

Check the result

You should have a quantitative analysis run with the method, filters, and any weights you chose. If you saved it, you should also see the named analysis available to open again later.

If something goes wrong

If a method is unavailable or does not give the setup you expect, review the variable types in Data & variables and confirm you have enough rating items or numeric variables for that method.