Types Of Data
Theory
Data splits into two main families: categorical (labels, like eye colour) and numerical (measurements you can average, like height). Each splits in two: categorical → nominal (no order) or ordinal (ranked); numerical → discrete (counted) or continuous (measured). Watch out for things like postcodes that look numerical but are really categorical labels.
Statistics begins with one question: what type of data are you working with? The type tells you which graphs to use, which summary statistics make sense, and which conclusions are valid.
Data falls into two main families:
- Categorical data describes a quality, category, or label. Examples: eye colour, blood type, country of origin.
- Numerical data is a measurement or count expressed as a number that can be meaningfully averaged. Examples: height, weight, number of pets.
Categorical data has two sub-types:
- Nominal — categories with no natural order. Examples: blood type, state of residence, favourite colour.
- Ordinal — categories with a meaningful order. Examples: clothing size (S, M, L), star ratings, agreement scales.
Numerical data also has two sub-types:
- Discrete — values that can be counted, usually whole numbers. Examples: number of siblings, cars sold per day.
- Continuous — values measured on a scale that can take any value in an interval. Examples: height, weight, temperature, time.
The first diagram is the data-type tree — the classification map you should picture for every dataset. The second shows the most common trick cases: data that looks numerical but is really categorical.
There are no calculation formulas — just the four-way classification and a quick test.
The four types of data
| Family | Type | Description | Examples |
|---|---|---|---|
| Categorical | Nominal | No natural order | blood type, eye colour, state |
| Ordinal | Meaningful ranking | S/M/L, star rating, agreement scale | |
| Numerical | Discrete | Counted, whole numbers | siblings, cars sold, goals scored |
| Continuous | Measured on a scale | height, weight, time, temperature |
How to classify any dataset
- Are the values labels or measurements? Labels (words, codes, IDs) → categorical. Measurements (numbers you'd average) → numerical.
- If categorical, is there a natural order? No order → nominal. Ordered ranking → ordinal.
- If numerical, are the values counted or measured? Counted (whole numbers) → discrete. Measured on a scale (can be any value, often with decimals) → continuous.
Spotting the trick cases
- If the values are digits but represent identifiers (postcode, jersey number, phone number, ID), it's categorical, not numerical.
- Apply the averaging test: ask whether the average makes sense. If not, the data is a label disguised as a number.
- Continuous data stays continuous even when written to the nearest unit. "5 seconds" is still continuous if time was being measured.
Eye colour is a label, not a measurement. There is no natural order between brown, blue, and hazel.
Answer: categorical, nominal.
The values are whole-number counts (
Answer: numerical, discrete.
Postcodes look numerical, but they are labels for areas. Applying the averaging test: the "average postcode" of a group of customers is meaningless.
Answer: categorical, nominal.
Weight is a measurement on a continuous scale; it can take any value in an interval, not just whole numbers. The average weight makes perfect sense.
Answer: numerical, continuous.
Common pitfalls
Frequently asked questions
What is the difference between categorical and numerical data?
Categorical data describes a quality, category, or label — like eye colour or blood type. Numerical data is a measurement or count that can be meaningfully averaged — like height or number of pets.
What is the difference between nominal and ordinal data?
Both are categorical. Nominal categories have no natural order — like blood type or favourite colour. Ordinal categories have a meaningful order or ranking — like clothing size (S, M, L) or a 1 to 5 star rating.
What is the difference between discrete and continuous data?
Both are numerical. Discrete data takes values that can be counted — usually whole numbers like the number of siblings. Continuous data is measured on a scale and can take any value in an interval — like height, weight, or time.
Is a postcode numerical data?
No. Postcodes look numerical, but they are labels for areas. The average postcode is meaningless, which is the clue: postcodes are categorical, nominal data. The same applies to jersey numbers, phone numbers, and student ID numbers.
If I record time to the nearest second, is it discrete or continuous?
Time is continuous data. Whether the data is discrete or continuous depends on what values are POSSIBLE, not how you record them. Time, weight, and height are continuous even when written to the nearest unit.
Why does the type of data matter?
The data type tells you which graphs are appropriate, which summary statistics make sense, and which conclusions are valid. A bar chart works for categorical data; a histogram works for numerical. Calculating a mean makes sense for numerical data but not for nominal categories.
Practice Questions
20 questions available.
Practice Questions