📦 Why is there survy?

Survey data has a type of structure that no Python tool handles correctly: multiple-choice questions — the type of question "select all that apply" where respondents can select multiple options at once. The raw data often stores them as multiple columns (hobby_1, hobby_2, hobby_3) or as a delimited string ("Sport;Book"), but pandas treats them as unrelated columns or plain text. For each project, you need to rewrite the familiar code to separate, group, count, and export the data — and you always encounter small, hard-to-detect errors: counting the number of selections instead of the number of respondents, losing the connection between columns, or making mistakes when the formatting changes.

SPSS solved this problem decades ago with built-in multiple response sets. R has some scattered solutions in expss, MRCV, and surveydata. But Python — the language where AI tools actually generate code — still has nothing.

survy makes MULTISELECT an official variable type. Just load the data, and survy automatically identifies the format, combines the columns into meaningful variables, and maintains awareness of the data type throughout the frequency calculation, crosstab, filtering, and exporting results processes. Correct code is also simple code — which means AI assistants can reliably generate accurate survy code.

✨ Features

🔹 Multiselect is a first-layer variable — automatically recognizes both compact and wide formats

🔹 Read & write multiple formats: CSV, Excel, JSON, SPSS

🔹 Integrated tools for validation, monitoring, and data analysis

🔹 Cross-tabulation with statistical testing

🔹 AI-friendly — includes agent skill to help LLM coding assistants generate accurate survy code