Is a Likert scale ordinal or interval data?

Technically, a Likert scale (1=strongly disagree to 5=strongly agree) is ordinal — the intervals between categories are not guaranteed to be equal. However, in practice, many researchers treat Likert responses as interval data when the scale has 5+ points, respondents tend to use it in an approximately equal-interval way, and they are combining multiple Likert items into a composite score. This is a pragmatic compromise — the parametric statistics (mean, SD, t-test) applied are slightly incorrect but often robust. For a single Likert item or when precision matters, use median and non-parametric tests.

Can temperature be ratio data?

Temperature in Celsius and Fahrenheit is interval data — the zero is arbitrary. Temperature in Kelvin is ratio data — 0 K represents the theoretical absence of thermal energy (absolute zero). This means: 200 K is truly twice as warm as 100 K (in terms of thermal energy). But 20°C is NOT twice as warm as 10°C, because the zero point is arbitrary. For most statistical analysis of temperatures, the distinction between interval and ratio rarely matters since arithmetic on differences (not ratios) is usually performed.

What happens if I use the wrong statistical method for my data type?

Using parametric methods (mean, t-test, ANOVA) on nominal or ordinal data can produce misleading results. For example, computing the mean of a Likert satisfaction scale and comparing groups with a t-test — if scale intervals are unequal, differences in means don't mean what you think. Using a Pearson correlation on ordinal data inflates or deflates the correlation estimate. The errors are usually smaller with larger samples and approximately equal intervals, which is why the practice is common — but it's technically incorrect.

Is age discrete or continuous?

Age is fundamentally continuous — it changes moment by moment. But it is commonly measured discretely (in whole years). When recorded as "35 years old," it is treated as discrete. When recorded as "35.4 years" from a birth date, it is continuous. The statistical treatment depends on how it was measured. For most analyses, age is treated as a continuous ratio variable, allowing means, standard deviations, Pearson correlation, and regression.

What are qualitative vs quantitative data?

Qualitative data (also called categorical) describes qualities or categories — it cannot be meaningfully added or averaged. Nominal and ordinal data are qualitative. Quantitative data represents numerical quantities — arithmetic operations are meaningful. Interval and ratio data are quantitative. Note: ordinal data sits on the boundary — it has order (a quantitative property) but unequal intervals (limiting arithmetic operations). This boundary is why Likert scale treatment is debated.

Types of Data in Statistics — Nominal, Ordinal, Interval & Ratio Explained

Name: Types of Data in Statistics — Nominal, Ordinal, Interval & Ratio
Availability: OnlineOnly
Author: CalcMulti Editorial Team

By CalcMulti Editorial Team·Updated: February 2026·8 min read

In statistics, not all data is the same. The type of data you have determines which statistical methods are appropriate — using the wrong method for your data type leads to incorrect analyses and misleading conclusions.

There are four levels of measurement — nominal, ordinal, interval, and ratio — arranged in order of information content. There is also the fundamental distinction between categorical data (which describes groups) and numerical data (which represents quantities). This guide explains each type clearly with examples and tells you which statistical tools apply to each.

The Four Levels of Measurement

Stanley Stevens (1946) proposed four levels of measurement, each with different mathematical properties. Higher levels contain all the properties of lower levels, plus additional ones.

Level	Categories?	Order?	Equal Intervals?	True Zero?	Examples
Nominal	✓	✗	✗	✗	Blood type (A/B/AB/O), eye colour, country, yes/no
Ordinal	✓	✓	✗	✗	Education level, Likert scale (1–5), race position
Interval	✓	✓	✓	✗	Temperature (°C/°F), calendar year, IQ score
Ratio	✓	✓	✓	✓	Height, weight, age, income, time, kelvin temperature

Nominal Data — Categories Without Order

Nominal data groups observations into named categories with no natural order or ranking. The only mathematical operation valid for nominal data is counting — you can tell whether two values are the same or different, but not which is "greater."

Examples: blood type (A, B, AB, O), gender, eye colour, nationality, type of smartphone, yes/no survey responses, zip codes (even though they look like numbers).

Nominal data with only two categories (e.g., yes/no, male/female, defective/not defective) is called binary or dichotomous data.

Appropriate statistics: mode (most frequent value), frequency counts, percentages, chi-square tests, logistic regression (for binary outcomes).

Ordinal Data — Categories With Meaningful Order

Ordinal data has categories with a meaningful order, but the gaps between categories are not necessarily equal. You know which is higher, but not by how much.

Examples: education level (no degree < bachelor's < master's < PhD), race positions (1st, 2nd, 3rd), Likert scale responses (strongly disagree → strongly agree), movie ratings (1–5 stars), pain scale (0–10), socioeconomic class (low/middle/high).

Key point: the difference between 1st and 2nd place is not necessarily the same as between 2nd and 3rd. Similarly, the difference in satisfaction between "disagree" and "neutral" may not equal the difference between "neutral" and "agree."

Appropriate statistics: median, percentiles, mode, Spearman correlation, Mann-Whitney U test, Kruskal-Wallis test. Avoid: mean (differences not equal), Pearson r, t-test (unless sample is large and ordinal scale is approximately interval).

Interval Data — Equal Spacing, No True Zero

Interval data has a meaningful order AND equal intervals between values, but lacks a true zero point. The zero is arbitrary — it does not mean "absence of the quantity."

Examples: temperature in Celsius or Fahrenheit (0°C does not mean "no temperature" — it is the freezing point of water), IQ scores (an IQ of 0 is not "no intelligence"), calendar years (year 0 is arbitrary), Celsius-scale exam scores if the minimum possible is not 0.

Because zero is arbitrary, ratios are not meaningful: 30°C is NOT "twice as hot" as 15°C (in Celsius). But differences are meaningful: the difference between 20°C and 25°C is the same as between 30°C and 35°C.

Appropriate statistics: mean, standard deviation, Pearson correlation, t-tests, ANOVA (all arithmetic operations on differences are valid). Avoid: ratio statements ("twice as much"), geometric mean, coefficient of variation.

Ratio Data — Full Measurement Scale

Ratio data has all the properties of interval data PLUS a true zero point — zero means "complete absence" of the quantity. This makes ratio comparisons (twice as much, half as long) meaningful.

Examples: height (0 cm = no height), weight, age, income ($0 = no income), time (0 seconds = no time elapsed), distance, counts (0 items = none), temperature in Kelvin.

With ratio data, all arithmetic operations are valid: you can meaningfully say someone who earns $80,000 earns twice as much as someone who earns $40,000.

Appropriate statistics: all statistics are valid — mean, SD, geometric mean, coefficient of variation, Pearson r, regression, parametric tests. The geometric mean is especially meaningful for ratio data (growth rates, multiplicative relationships).

Discrete vs Continuous Data

A separate but overlapping classification is discrete vs continuous.

Discrete data: can only take specific values, typically integers with no valid intermediate values. Examples: number of children (you cannot have 2.3 children), number of defects, number of votes, number of items sold. Typically counted.

Continuous data: can take any value within a range, including fractions and decimals. Examples: height (1.734 m is valid), weight, temperature, time, blood pressure. Typically measured.

Most ratio and interval data is continuous; most nominal and ordinal data is discrete — but exceptions exist. Counts are discrete ratio data (you can have 0 items — a true zero).

Type	Discrete Examples	Continuous Examples
Nominal	Blood type, gender (discrete by nature)	— (not applicable)
Ordinal	Likert scale, education level, star rating	Occasional in personality measures
Interval	IQ scores (often treated as discrete)	Temperature, calendar dates
Ratio	Number of children, defect count	Height, weight, income, speed

Which Statistical Methods for Which Data Type

Using a method designed for ratio data on ordinal data is a common mistake that can produce misleading results. Here is a quick reference guide.

Analysis Goal	Nominal	Ordinal	Interval/Ratio
Central tendency	Mode	Median	Mean
Spread	Frequency table	IQR, percentiles	SD, variance
Association	Chi-square, Cramér's V	Spearman ρ	Pearson r, regression
Compare 2 groups	Chi-square test	Mann-Whitney U	Independent t-test
Compare 3+ groups	Chi-square test	Kruskal-Wallis	One-way ANOVA
Prediction	Logistic regression	Ordinal regression	Linear regression

Related Calculators

Descriptive Statistics Measures Comparison

Choosing the right measure for your data

When to Use Mean Median Mode

Match central tendency to data type

Parametric vs Nonparametric

Choosing the right statistical test

Chi-Square Calculator

Test for categorical data

Descriptive vs Inferential Statistics

Two branches of statistical analysis

Statistics Hub

All statistics calculators & guides

← Back to Statistics Hub

Frequently Asked Questions

Educational use only. Content is based on publicly documented mathematical formulas and reviewed for accuracy by the CalcMulti Editorial Team. Last updated: February 2026.