Student feedback analysis glossary for UK HE
Updated Nov 06, 2026
Use the same definitions in the analysis, the report and the meeting where decisions are made. These terms explain how a comment becomes a topic count, a sentiment measure and an item for investigation.
The interactive worked example puts the counting terms into practice. For the precise definitions used in our public aggregate research, see the NSS methodology.
Comments, sentences and topics
Comment
One submitted text response in the analysis dataset. A student may submit more than one response, so a comment count should not automatically be described as a student count.
Sentence-level analysis
Examining the separate statements within a comment. This preserves different topics and sentiments in a response such as praise for a seminar followed by criticism of assessment feedback.
Categorisation (topic classification)
Assigning text to defined topics. A comment can concern several topics. The undergraduate category framework shows one organised set of themes and categories.
Taxonomy
A maintained structure of topics and definitions. Keep its version with an analysis so that a category label means the same thing to each reader.
Coding (manual coding)
A person assigns labels or develops interpretations through reading. A review log records how ambiguous cases and disagreements were resolved. Manual review can inform and check an automated workflow as well as support a qualitative study.
Counts and denominators
Denominator
The population used as the base of a calculation. “Four of ten comments” and “four of twenty students” answer different questions. Name the base beside the measure.
Mention rate
The proportion of the defined comment population that mentions a topic. In our public NSS briefs, it is the percentage of classified comments in the same year and population that mention that topic. A comment counts once per topic, even when several sentences mention it.
All-comment coverage
An analysis workflow includes every usable comment rather than selecting a sample for the main classification. Report the rules for unusable, excluded and unclassified text separately.
Classification coverage
The share of source comments that received a classification. Coverage describes how much of the source material was classified. Accuracy describes whether classifications are correct; assess it through a separate review.
Percentage-point change
The arithmetic difference between two percentages. A synthetic increase from 30% to 35% is five percentage points. It is a relative increase of about 16.7%, because five divided by thirty is approximately 0.167. Name the unit you are reporting.
Sentiment and comparison
Sentiment
The positive, negative or neutral evaluation expressed about a topic. One response can include more than one sentiment. Read it alongside the relevant text and topic rather than assigning a single mood to the student.
Sentiment index
A combined measure with a defined formula. In our public NSS analysis it is 100 multiplied by positive probability minus negative probability, with matching sentences averaged within each comment and topic before averaging comments. See the sentiment guide for the distinction between this index and simple label counts.
Benchmarking (sector benchmark)
Comparing a result with a specified reference population. Record the survey, period, subject grouping and definition of the measure so readers understand the comparison.
Cohort mix (composition)
The distribution of responses across groups such as subject or study mode. A change in that distribution changes the composition of the result being compared.
CAH / CAH3
The Common Aggregation Hierarchy groups higher education subjects. CAH3 is its detailed subject level, used in our public subject briefs.
Review and publication
Redaction
Removing or obscuring specified information from text. The remaining context still needs review before sharing an example.
Reproducibility (repeatability)
Being able to repeat an analysis using the same inputs and documented method, or explain differences through a version record.
Governance pack
The working records that explain data handling, the method, review decisions and approval of outputs. Use the governance checklist to assign owners.
Closing the feedback loop
Explaining the response to feedback and checking what happens next. A useful record connects the concern, evidence, decision, owner and review date. Explore the closing the loop pathway for examples and reading.
Put the terms to work
Use the software decision guide to choose a workflow and Student Voice Analytics to discuss an institutional analysis. Keep these definitions available alongside the report so colleagues can spend their meeting on the decision.
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.