Skip to main content

Discuss your survey

Tell us what you need to understand from your student feedback. Name and email are required; the other fields are optional.

Include your survey, reporting question or deadline. Please leave student comments and identifying student details out of this enquiry.

We’ll use these details to respond to your enquiry. You can also email info@studentvoice.ai.

Sentiment analysis for UK universities: a practical guide

Updated Nov 06, 2026

Sentiment analysis helps a university distinguish what students value from what they want changed. Its most useful unit is a topic within a comment: praise for teaching and criticism of feedback can appear in the same response.

Consider this synthetic comment: “The seminars made the ideas clear, but I received feedback too late to use it on the next assignment.” A useful analysis retains both observations. Teaching is positive; assessment feedback is negative. A single overall label would lose part of the experience.

Follow a result back to the comment

The interactive worked example lets you select a survey and topic, inspect the synthetic comments and see how the summary changes. Its labels were assigned manually to explain the process. It is an illustration of counting and interpretation, rather than a live model prediction.

For each result, ask three connected questions: what experience is being described, which topic does it concern, and how many distinct comments contain it? Keep the source text available to the authorised reviewer so that a short label never becomes the only account of what was said.

Count comments consistently

Suppose a synthetic set contains ten comments. Four mention assessment feedback. One of those four mentions late feedback in two sentences. The topic still appears in four comments, so its mention rate is 40% of that set. Counting the repeated sentence again would inflate the result.

One comment can also mention teaching and feedback. It contributes to each relevant topic, so the topic shares can add to more than 100%. Within one topic, a comment can contain both positive and negative observations. State whether your report counts labels, comments or model probabilities; those are different measures.

Distinguish a percentage from an index

A percentage tells the reader the proportion of a defined set. A sentiment index combines positive and negative information according to a specified calculation.

In the public NSS research methodology, our index uses positive and negative model probabilities. Matching sentences are averaged within each comment and topic before comment-level values are averaged. The index is not calculated by subtracting the simple label counts in the worked example. Its unit is index points, not the percentage of students satisfied.

A report should name the measure and link its definition beside the chart. The glossary explains mention rate, classification coverage and percentage-point change in the same terms.

Turn the pattern into a useful review

Start with the topic and then read the examples. “Feedback is late”, “feedback is difficult to understand” and “feedback arrives after the next assignment” suggest different checks. The first concerns delivery, the second explanation, and the third assessment sequencing.

For an assessment review, an analyst could group those experiences separately, ask the programme team to check the relevant process and agree a response with students. Retain positive examples too: a valued feedback discussion may offer a practice worth sharing.

Record the action owner and the next review point. Compare the same survey question and population when examining a later cycle, and use the assessment and feedback reading pathway to inform the discussion.

Specify the workflow you need

Before commissioning analysis, agree topic definitions, reporting populations, review responsibilities and the outputs colleagues will receive. Use the governance checklist for the working record and the buyer’s guide for evaluation questions. Student Voice Analytics brings these elements into an agreed university survey analysis service.

Buyer guidance

Choose an approach to NSS comment analysis

Read the NSS buyer’s guide for analysis options, reporting requirements, and questions to take into procurement.

  • Covers NSS, PTES, PRES, UKES, module evaluations.
  • Explains benchmarks, taxonomy, and reproducibility.
  • Includes procurement checklist prompts.

Related Entries