Written by Dackohn; last reviewed November 2026.
Variables are categorical (nominal or ordinal) or numeric (interval or ratio). Nominal data are unordered categories, ordinal data are ordered categories, interval data have equal steps but no true zero, and ratio data have a true zero. The type determines which averages and tests make sense.
The four levels of measurement
| Level | What it means | Examples | Typical average |
|---|---|---|---|
| Nominal | Categories with no order | Blood type, country, eye colour | Mode |
| Ordinal | Ordered categories with uneven gaps | Satisfaction rating, education level, race position | Median |
| Interval | Equal steps, but no true zero | Temperature in °C, calendar year | Mean |
| Ratio | Equal steps and a true zero | Height, weight, income, reaction time | Mean, or median if skewed |
How to tell them apart
- Can you put the values in a meaningful order? If not, the variable is nominal.
- Are the gaps between values equal? If not, it is ordinal. The step from “good” to “very good” need not equal the step from “poor” to “fair”.
- Does zero mean “none of it”? If yes, it is ratio, and “twice as much” makes sense. 80 kg is twice 40 kg, but 20 °C is not twice as hot as 10 °C, because 0 °C is not an absence of temperature.
Discrete and continuous
Numeric variables can also be discrete, meaning they are counted and can only take certain values (children in a family, website visits), or continuous, meaning they are measured and can take any value in a range (height, time). Discrete counts with many possible values are often analysed as if continuous.
Which test fits which type
| Outcome variable | Comparing groups | Measuring association |
|---|---|---|
| Nominal | Chi-square, Fisher’s exact | Cramér’s V |
| Ordinal | Mann–Whitney, Kruskal–Wallis | Spearman |
| Interval or ratio | t-test, ANOVA | Pearson, regression |
The tricky cases
- Likert items. A single 1–5 rating is ordinal. A total across many items behaves more like interval data, which is why summed scales are often analysed with t-tests.
- Numbers as labels. Zip codes and jersey numbers are nominal despite being digits. Their average means nothing.
- Binned data. Age recorded in bands (18–24, 25–34) is ordinal, even though age itself is ratio. Collect the exact value if you can.
Try it yourself
Which Test Should I Use? →Answer a few questions about your variables and get a recommendation. Descriptive Statistics Calculator →Summarise numeric data with the mean, median, SD and more.Frequently asked questions
What are the four levels of measurement?
Nominal, ordinal, interval and ratio. They were proposed by the psychologist S. S. Stevens in 1946.
Is age nominal, ordinal, interval or ratio?
Age in years is ratio, because zero means no age and 40 is twice 20. Age recorded in bands such as 18-24 is ordinal.
Is a Likert scale ordinal or interval?
A single item is ordinal. A sum of many items is often treated as interval, which is usually reasonable with enough items.
Related reading: Choosing a statistical test · Mean, median and mode · Parametric vs nonparametric tests
Sources & further reading: NIST/SEMATECH e-Handbook of Statistical Methods · OpenStax Introductory Statistics. How we compute & validate: methodology. Written & reviewed by Dackohn; last reviewed August 2026.