How to analyse open-ended survey responses

Why open-ended responses are hard to analyse

Two respondents can describe the same experience in completely different words. Eyeballing responses doesn't scale, and word clouds or keyword counts throw away meaning — “the lectures felt rushed” and “we never had time to ask questions” are the same theme with no shared keyword. What you need is coding: labels attached to meanings, not words.

A repeatable process

1. 1. Clean the data. Remove blanks and non-answers (“n/a”, “none”) before coding.
2. 2. Code the responses. Attach short labels to the ideas in each answer, keeping a quote per code.
3. 3. Cluster into themes. Group related codes and name the recurring patterns — see the six phases for the full method.
4. 4. Quantify. Count how many responses express each theme to show prevalence.
5. 5. Evidence it. Report each theme with representative quotes so it is checkable.

Using the analysis tool for survey data

The thematic analysis tool works on survey data too: split your responses into natural groups — one questionnaire item per box, or one respondent group per box — and treat each group as a “study”. The engine codes each group separately, then finds the themes they share, so you can see whether, say, first-years and finalists raise the same issues. Each code carries a verbatim response quote, ready for your results chapter.