Quantitative UX runs on measurement. Get the four scale types right, then master the three that matter most: Likert, Semantic Differential and Guttman.
Quantitative methodology studies phenomena with numerical, statistical data to test hypotheses and generalize. Measurement uses four scale types: nominal, ordinal, interval and ratio. In UX, three scales matter most: the Likert scale (agreement with statements, used in the SUS), the Semantic Differential (feelings between opposite adjectives) and the Guttman scale (cumulative, hierarchical agreement). Choosing the right scale determines which statistics you can run.
Key takeaways
- Quantitative research seeks measurable, generalizable, objective results.
- The four scale types are nominal, ordinal, interval and ratio.
- Likert measures agreement; it powers the System Usability Scale (SUS).
- Semantic Differential captures feeling between two opposite adjectives.
- Guttman is cumulative and hierarchical: agree with item 4 and you agree with 1 to 3.
The four measurement scales
| Scale type | What it measures | Example |
|---|---|---|
| Nominal | Categories with no order or hierarchy. | Gender, ethnicity |
| Ordinal | Order and rank, but not the distance between ranks. | Education level |
| Interval | Ranked with equal distances, but no true zero. | Celsius, Fahrenheit |
| Ratio | Like interval, but with a true zero. | Age, income, weight |
The 3 scales to master in UX
| Scale | What it captures and how | Best for |
|---|---|---|
| Likert | Level of agreement with a statement, usually 5, 7 or 9 points. Sum, average, and use median and mode. Avoid double-barreled or open-ended items. Powers the SUS. | General agreement |
| Semantic Differential | A concept rated on a continuum between two opposite adjectives (Intuitive vs Confusing). Pairs must be true antonyms. | Feelings & perception |
| Guttman | Cumulative, hierarchical statements; agreeing with a higher item implies agreeing with lower ones. Needs correct sequencing to hold. | Cumulative / stages |
How to analyze Likert data
- Collect responses, assigning each point a value (Strongly Disagree = 1 to Strongly Agree = 5).
- Sum each participant's scores across the statements.
- Average across participants to get a mean sentiment.
- Add median and mode to see distribution and outliers.
- Visualize with bar charts (per statement) and histograms (overall distribution).
Watch the wording. “How satisfied are you with the website’s speed and visual design?” is double-barreled: it mixes two things a user may feel differently about, and terms like speed and design are ambiguous without context. Split it into separate, specific statements.
Nominal (categories), ordinal (rank), interval (equal distances, no true zero) and ratio (true zero).
A scale that measures agreement with a statement, usually on 5, 7 or 9 points from Strongly Disagree to Strongly Agree. It powers the System Usability Scale.
A scale where users rate a concept on a continuum between two opposite adjectives, like Intuitive versus Confusing, to capture feeling and perception.
A cumulative, hierarchical scale where agreeing with a higher-ranked item implies agreement with all lower ones. It must be sequenced correctly to work.
Avoid double-barreled items that combine two topics, keep statements specific and unambiguous, and never use open-ended questions on a Likert scale.
Want research measured with the right scale, not guesswork? Talk to Dorve.
Start a project