The Four Scales of Measurement (Stevens' Taxonomy)
Nominal, Ordinal, Interval, and Ratio: the foundational classification that determines which summary metrics and tests are mathematically valid.
In this lesson you will learn
- 1Classify any clinical or survey variable into Nominal, Ordinal, Interval, or Ratio scales.
- 2Match each scale of measurement to its valid central tendency and dispersion pair (n%, Median [IQR], or Mean (SD)).
- 3Identify why numeric coding of categories (such as 1 = Male, 2 = Female) does not make a variable quantitative.
On this page▼
Stanley Smith Stevens' 1946 Measurement Framework
This section defines the four measurement scales—Nominal, Ordinal, Interval, and Ratio—and shows which mathematical operations each scale supports.
Misclassifying an ordinal disease stage or coded category as continuous ratio data leads to selecting invalid parametric tests that reviewers flag immediately.
In 1946, psychologist Stanley Smith Stevens published a foundational paper in Science establishing four hierarchical levels of measurement. Each level adds one mathematical property to the level below it:
- Nominal (Identity only,
= / ≠): Unordered qualitative categories such as ABO blood group, marital status, or genotype. Summarised strictly using counts and percentagesn (%)and the Mode. - Ordinal (Magnitude order,
< / >): Ranked categories such as cancer TNM stage (I–IV), Glasgow Coma Scale items, or 5-point Likert items. Summarised usingMedian [IQR]or frequencies. - Interval (Equal step distance,
+ / −): Continuous scales with equal unit distances but a convention-based zero, such as temperature in Celsius (°C). Differences are meaningful, but ratios are not. - Ratio (True absolute zero,
× / ÷): Continuous clinical measurements where zero signifies complete absence of the quantity—such as serum albumin (g/dL), weight (kg), or survival duration (months).
Interactive Variable Classifier
Test your ability to classify scales of measurement and assign valid statistical tests
Clinical Example
An oncology database exports three columns: Primary Site (1 = Colon, 2 = Lung, 3 = Breast), TNM Stage (1 = I, 2 = II, 3 = III, 4 = IV), and Tumour Diameter (cm).
Interpretation: Even though all three columns contain digits in the raw export, only Tumour Diameter is a true ratio variable where 4.8 cm is twice 2.4 cm. Site must be summarised as n (%), and TNM Stage as n (%) or Median [IQR].
Common Pitfall
Allowing statistical software to treat numerically coded nominal categories (1, 2, 3) as continuous numbers and reporting a meaningless 'mean gender' or 'mean hospital ward'.
Key takeaways
- 1The scale of measurement dictates permissible mathematical operations: computing an arithmetic mean on nominal codes (such as blood group) produces meaningless numbers.
- 2Nominal scales label unordered categories; Ordinal scales rank categories in a natural order where intervals between ranks are not necessarily equal.
- 3Interval scales have equal numeric steps with an arbitrary zero (such as Celsius temperature), whereas Ratio scales possess an absolute zero that supports multiplication ratios.
Check your understanding
One question per screen with immediate methodological rationale after each answer
A researcher records tumour stage as Stage I, Stage II, Stage III, and Stage IV. Which scale of measurement is this?
Finished reading this lesson?
Mark as completed to track your progress through the handbook