Levels of Measurement: Nominal, Ordinal, Interval, and Ratio

Welcome back. Last lesson focused on the first classification of a variable: categorical versus quantitative, and then discrete versus continuous for quantitative variables. This lesson adds a second classification system: the variable’s level of measurement.

Levels of measurement tell us what information the recorded values actually contain. Can the values only name groups? Can they be ranked? Do equal numerical differences mean the same thing everywhere on the scale? Does zero mean “none”? Those answers determine whether a variable is nominal, ordinal, interval, or ratio.

By the end, you should be able to classify a variable using a reliable set of questions rather than trying to memorize isolated examples.

---

The four levels build on one another

Think of the four levels as adding one useful property at a time.

Level	What the values tell us	Key question
Nominal	Category or name only	Is there no meaningful order?
Ordinal	Categories with an order	Can values be ranked, but gaps are unclear?
Interval	Ordered numerical values with equal differences	Are differences meaningful, but zero arbitrary?
Ratio	Ordered numerical values with equal differences and a true zero	Does zero mean none of the quantity?

The key idea is that the scale becomes more informative as you move from nominal to ratio:

1. All four levels can distinguish one recorded value from another.
2. Ordinal, interval, and ratio variables can be placed in a meaningful order.
3. Interval and ratio variables have meaningful, equal-sized differences.
4. Only ratio variables have a true zero, so statements involving “twice as much” are meaningful.

The last two distinctions—equal intervals and a true zero—cause most confusion. We will work slowly through them.

---

Nominal: names with no natural order

A nominal variable places observations into categories, but those categories have no meaningful ranking. “Nominal” resembles name: it names the group an observation belongs to.

Examples:

• eye colour: brown, blue, green
• program of study: biology, business, history
• blood type: A, B, AB, O
• favourite music genre
• device type: phone, laptop, tablet
• whether a person has a driver’s licence: yes or no

There is no fact-based sense in which one eye colour is higher or lower than another. You can count categories and find proportions, but you cannot meaningfully average them.

Numbers do not automatically make data quantitative

A nominal variable can be represented by digits:

• Postal code: $905$, $416$, and $647$ identify areas.
• Player number: $7$, $18$, and $42$ identify players.
• Student ID: a number identifies a student record.
• A survey may code “yes” as $1$ and “no” as $0$.

In each case, the number is a label, not an amount. It would not make sense to say that postal code $800$ is twice postal code $400$, or to compute an average jersey number as if it described a typical quantity.

Classification: if categories have no meaningful order, the variable is nominal.

---

Ordinal: ordered categories, but unknown gaps

An ordinal variable has categories that can be put in a meaningful order. The word ordinal is related to order.

Common examples include:

• satisfaction: very dissatisfied, dissatisfied, neutral, satisfied, very satisfied
• class rank: first, second, third
• letter grades: F, D, C, B, A
• shirt size: small, medium, large
• pain rating: mild, moderate, severe
• agreement on a survey: strongly disagree through strongly agree

With ordinal data, you can safely say one observation is higher or lower than another. For example, “satisfied” represents more satisfaction than “neutral.”

But you cannot assume the distance between categories is equal.

Consider this satisfaction scale:

Code	Response
1	Strongly disagree
2	Disagree
3	Neutral
4	Agree
5	Strongly agree

The codes are useful for recording the ordered categories, but a change from $1$ to $2$ is not proven to represent the same size change in opinion as a change from $4$ to $5$. The numbers communicate rank, not a precisely measured amount.

Similarly, a runner who finishes second was slower than the runner who finishes first. But without their actual times, you do not know how much slower. The gap between first and second might be $0.01$ seconds, while the gap between second and third might be 15 seconds.

A common trap: rating scales

A $1$-to-$5$ rating scale is usually ordinal in an introductory statistics course, even though it uses numbers. The values tell you the direction of the opinion, but they do not guarantee equal spacing in feeling or attitude.

Classification: if values have a natural order but differences are not known to be equal, the variable is ordinal.

---

Interval: equal differences, but zero is not “none”

An interval variable is numerical and has:

• a meaningful order;
• equal intervals, meaning equal differences represent equal changes in the quantity;
• no true zero.

The standard examples are temperature in degrees Celsius and degrees Fahrenheit.

On the Celsius scale:

• The difference between $10^{\circ }\text{C}$ and $20^{\circ }\text{C}$ is $10^{\circ }\text{C}$.
• The difference between $70^{\circ }\text{C}$ and $80^{\circ }\text{C}$ is also $10^{\circ }\text{C}$.

Those differences are comparable. So it is meaningful to calculate a temperature change or average temperatures measured on the same scale.

However, $0^{\circ }\text{C}$ does not mean “no temperature” or “no thermal energy.” It is a selected reference point on the Celsius scale. Therefore, it is not meaningful to say:

$20^{\circ }\text{C}$ is twice as hot as $10^{\circ }\text{C}$.

The division $20÷10=2$ is mathematically correct, but the interpretation is not. The zero point is not an absolute absence of the measured quantity.

The same applies to Fahrenheit. A temperature of $60^{\circ }\text{F}$ is $30^{\circ }\text{F}$ warmer than $30^{\circ }\text{F}$, but not “twice as warm.”

Other introductory examples sometimes treated as interval include:

• calendar years, such as 1990 and 2020;
• IQ scores.

For calendar years, the difference of 10 years has a consistent meaning, but year 0 is not the absence of time. It is a calendar reference point.

Classification: if values are numerical, differences are meaningful and equally spaced, but zero does not mean “none,” the variable is interval.

---

Ratio: equal differences and a true zero

A ratio variable has all the properties of interval data plus a true zero. A true zero means the quantity is absent.

Examples:

• height
• weight
• age measured from birth
• distance travelled
• elapsed time
• amount of money held
• number of siblings
• number of courses taken
• number of messages received
• temperature measured in kelvin

For these variables, both differences and ratios make sense.

Suppose one student studies 2 hours and another studies 6 hours:

• The difference is $6-2=4$ hours.
• The second student studied three times as long as the first.

This works because 0 hours means no study time.

Similarly, a person who has 4 siblings has twice as many siblings as a person who has 2. Having 0 siblings means the complete absence of siblings, so zero is genuine.

Celsius versus kelvin

The contrast between Celsius and kelvin is especially helpful:

• Celsius temperature is interval because $0^{\circ }\text{C}$ is not the absence of thermal energy.
• Kelvin temperature is ratio because 0 K is absolute zero, the physical lower limit of that scale.

So a temperature of 200 K has twice the absolute temperature of 100 K. The scale used to record the same general idea—temperature—matters.

Classification: if zero means none of the measured quantity and ratio statements such as “twice as much” make sense, the variable is ratio.

---

A classification routine that works under test pressure

When a question gives you a variable, do not begin by staring at the numbers. Identify what was actually measured or recorded, then use these questions in order.

1. Are the values only labels for categories?

• No meaningful order: nominal.
• Examples: major, country of citizenship, hair colour, type of vehicle.

If the answer is yes, stop. It is nominal.

2. If there is an order, are the gaps between values known to be equal?

• Order only; gaps unknown: ordinal.
• Examples: rankings, satisfaction categories, letter grades, pain categories.

If values are only ranks or ordered responses, stop. It is ordinal.

3. If equal differences are meaningful, does zero mean none of the quantity?

• No true zero: interval.
• Examples: Celsius temperature, Fahrenheit temperature, calendar year.
• True zero: ratio.
• Examples: height, time, counts, distance, mass.

Here is the compact version:

Ask this question	If yes	If no
Is there a meaningful order?	Keep checking	Nominal
Are equal differences meaningful?	Keep checking	Ordinal
Does zero mean none of the quantity?	Ratio	Interval

Notice that this routine starts with meaning, not with the format of the data. Numerical codes can be nominal or ordinal. A variable written using words can be ordinal. What matters is what the recorded values communicate.

---

Worked classifications

1. “Primary social-media platform used”

Possible responses: TikTok, Instagram, YouTube, X.

These are categories with no inherent ranking. It is nominal.

2. “Customer service rating: poor, fair, good, excellent”

The categories clearly have an order. But we cannot show that the step from poor to fair equals the step from good to excellent. It is ordinal.

3. “Outdoor temperature in degrees Fahrenheit”

A difference of $10^{\circ }\text{F}$ means the same temperature change anywhere on the scale. But $0^{\circ }\text{F}$ does not mean no temperature. It is interval.

4. “Number of times a student visited the library this month”

This is a count. Zero visits means no visits, and 8 visits is twice 4 visits. It is ratio. It is also quantitative and discrete, using the classification from the previous lesson.

5. “Height in centimetres”

Height is a measured quantity. Zero centimetres would mean no height, and 180 cm is twice 90 cm in length. It is ratio. It is also quantitative and continuous.

6. “Age group: under 18, 18–24, 25–34, 35–44”

The groups have an order, but they are categories rather than exact ages. The class widths may not even be equal. It is ordinal.

Compare that with exact age in years, which is generally ratio: 0 years means no elapsed time since birth, and ratios are meaningful.

7. “Final exam score out of 100 points”

In many introductory statistics settings, this is treated as ratio when it is a count of points earned: 0 points means no points earned, and 80 points is twice 40 points. Always use your instructor’s convention if one is specified, but the important reasoning is to connect the scale to what the score represents.

---

Keep the two classification systems separate

The previous lesson’s categories and today’s levels answer different questions.

Question	Possible answers
What general kind of variable is it?	Categorical or quantitative
If quantitative, how does it arise?	Discrete or continuous
What information does its measurement scale contain?	Nominal, ordinal, interval, or ratio

Some common pairings are:

Variable	General type	Measurement level
Eye colour	Categorical	Nominal
Satisfaction level	Categorical	Ordinal
Celsius temperature	Quantitative, continuous	Interval
Number of pets	Quantitative, discrete	Ratio
Weight	Quantitative, continuous	Ratio

Nominal and ordinal variables are normally categorical. Interval and ratio variables are quantitative.

The important warning is that a coding choice does not upgrade the measurement level. Recording “strongly disagree” through “strongly agree” as $1$ through $5$ does not make the ratings interval data. They remain ordinal unless there is a justified reason to treat every step as equal.

---

Takeaways

The level of measurement tells you what statements and calculations are justified for a variable:

• Nominal: categories with no order, such as eye colour or program of study.
• Ordinal: ordered categories with unequal or unknown gaps, such as rankings and satisfaction ratings.
• Interval: numerical values with equal differences but no true zero, such as Celsius or Fahrenheit temperature.
• Ratio: numerical values with equal differences and a true zero, such as height, time, distance, and counts.

When stuck, use the three-question routine:

1. Is there a meaningful order?
2. Are equal differences meaningful?
3. Does zero mean none of the quantity?

Next, you will shift from the variables themselves to the study design: distinguishing experiments from cross-sectional, retrospective, and prospective observational studies.