Examples
Worked examples
- Is an instance
In a study testing whether a tutoring method (standard vs. new) improves exam scores (0-100), tutoring method is a nominal-level independent variable and exam score is a ratio-level dependent variable -- the two classification systems are applied together, not as alternatives to each other.
- Is an instance
In a survey where self-reported pain (rated 1 to 5) is used to predict medication adherence (adherent vs. non-adherent), pain rating is an ordinal explanatory variable and adherence is a nominal response variable.
Counter-examples
Looks similar, but isn't
- Not an instance
Treating "independent variable" and "nominal variable" as synonyms is a common drafting error: an independent variable can be measured at any level. Dosage in milligrams is a ratio-level independent variable, not a nominal one, so measurement level has to be identified as a separate step from design role rather than assumed from it.
Editorial commentary
“Types of variables” is not one classification but shorthand for two independent frameworks researchers use to describe any variable in a study. The first, measurement level (or scale of measurement), classifies a variable by the mathematical properties of the values it can take: nominal, ordinal, interval, or ratio. The second, design role, classifies a variable by the function it plays in a specific study’s causal or descriptive structure: independent/dependent, explanatory/response, and confounding. Every variable in a study is classified along both dimensions at once, and confusing the two is one of the most common errors in a methods section draft.
Two Classification Systems, Not One
Measurement level describes how a variable was recorded — what a value of it can mathematically do (be ranked, be averaged, have a meaningful zero). Design role describes what a variable is doing in the analysis — whether it is the presumed cause, the outcome, a predictor in a correlational design, or a nuisance factor the analysis needs to account for. A single variable, such as age in years, always has exactly one measurement level (ratio) but can take different design roles depending on the study: it might be the independent variable in one analysis, a confounder in another, and neither in a third. Reviewers and methods checklists ask about both separately because they govern different decisions — measurement level determines which statistical test is valid, while design role determines what the test is being asked to show.
Measurement-Level Types: Nominal, Ordinal, Interval, Ratio
This four-level scheme, formalized by psychologist S.S. Stevens in 1946, orders variables by how much mathematical information their values carry:
| Level | What it captures | Example | Valid operations |
|---|---|---|---|
| Nominal | Unordered categories | Blood type, treatment arm, country of enrollment | Counts, mode, chi-square |
| Ordinal | Ordered categories, unequal or unknown spacing | Cancer stage, Likert-scale agreement, pain severity (mild/moderate/severe) | Median, rank-based tests |
| Interval | Equal spacing, no true zero | Calendar date, temperature in Celsius | Mean, standard deviation (not ratios) |
| Ratio | Equal spacing, true zero | Age, weight, dosage in milligrams, duration in days | Mean, ratios, all arithmetic operations |
Nominal and ordinal variables are often grouped as categorical (or qualitative); interval and ratio are grouped as quantitative (or continuous). This grouping is what most statistical-test decision trees actually key off — a chi-square test needs categorical variables, a Pearson correlation needs quantitative ones — which is why getting the measurement level right is a precondition for choosing a valid test, not a formality.
Design-Role Types: Independent/Dependent, Explanatory/Response, Confounding
Design role answers a different question: given a study’s hypothesis, what is each variable being asked to do?
- Independent and dependent variables — the vocabulary used for experimental designs, where the researcher manipulates or assigns the independent variable and measures the dependent variable as the outcome.
- Explanatory and response variables — the vocabulary preferred for observational and correlational designs, where nothing is manipulated but one variable is still modeled as predicting or explaining another.
- Confounding variables — a third factor, not the focus of the hypothesis, that is associated with both the independent/explanatory variable and the dependent/response variable, and can distort the apparent relationship between them if not identified, measured, or controlled for.
These roles are assigned relative to a specific research question and analysis, not fixed properties of the variable itself. Smoking status might be the independent variable in a study of smoking cessation programs, but a confounder in a study of coffee consumption and heart disease.
How the Two Systems Combine
Because measurement level and design role are independent dimensions, any combination is possible, and a manuscript’s methods section typically needs to specify both for every variable:
- A nominal independent variable — e.g., treatment arm (drug vs. placebo) predicting a ratio-level dependent variable (blood pressure in mmHg).
- An ordinal explanatory variable — e.g., self-reported pain rating (1-5) predicting a nominal response variable (medication adherence: yes/no).
- A ratio-level confounder — e.g., baseline age (years) controlled for statistically because it is associated with both an exposure and an outcome.
A frequent drafting error is treating “independent variable” and “nominal variable” as synonyms. They are not: an independent variable can be measured at any level. Dosage in milligrams is a ratio-level independent variable, not a nominal one. Identifying measurement level and design role are two separate steps, and skipping straight from “what’s my hypothesis” to “what test do I run” without doing both steps is a common source of an invalid statistical test choice caught late in peer review.
Why the Distinction Matters in a Methods Section
Reporting guidelines and statistical-test selection both depend on getting this right. A reviewer checking a manuscript’s methods section is typically verifying three things: that every variable’s measurement level is stated explicitly (not left to be inferred from the results table), that its design role is unambiguous given the stated hypothesis, and that the chosen statistical test is valid for that combination. Getting measurement level right also matters for how a variable should be operationalized in the first place — the decision to record pain as an ordinal category versus a continuous visual-analog-scale score is a measurement-level choice made at the design stage, before any statistical test is picked.
Frequently Asked Questions
What are the four types of variables in statistics?
The four measurement-level types are nominal, ordinal, interval, and ratio. This is a separate classification from design-role types like independent, dependent, explanatory, response, and confounding — a study reports both for each variable, not one or the other.
Is “independent variable” a type of measurement level?
No. Independent/dependent describes a variable’s role in a study’s design, not the mathematical properties of its values. An independent variable can be nominal, ordinal, interval, or ratio depending on how it was measured.
What is the difference between explanatory/response and independent/dependent variables?
They describe the same underlying relationship — one variable is modeled as predicting or accounting for another — but independent/dependent is the vocabulary used in experimental designs where a variable is manipulated, while explanatory/response is used in observational or correlational designs where nothing is manipulated.
How is a confounding variable different from an independent variable?
An independent variable is the presumed cause a study is specifically investigating. A confounding variable is a third factor, outside the study’s main hypothesis, that is associated with both the independent and dependent variables and can distort the apparent relationship between them if not accounted for.
Machine-readable encodings
Use in your systems
<role vocab="credit"
vocab-identifier="https://casrai.org/dictionary/"
vocab-term="Types of Variables"
vocab-term-identifier="https://casrai.org/dictionary/term/types-of-variables" />{
"@context": "https://schema.org",
"@type": "DefinedTerm",
"@id": "https://casrai.org/dictionary/term/types-of-variables",
"name": "Types of Variables",
"identifier": "https://casrai.org/dictionary/term/types-of-variables",
"description": "\"Types of variables\" is not one classification but shorthand for two independent frameworks used to describe any variable in a study. The first, measurement level (or scale of measurement), classifies a variable by the mathematical properties of the values it can take: nominal (unordered categories), ordinal (ordered categories without equal spacing), interval (equal spacing, no true zero), or ratio (equal spacing with a true zero). The second, design role, classifies a variable by the function it plays in a specific study's causal or descriptive structure: independent/dependent (manipulated or predictor vs. measured outcome in an experimental design), explanatory/response (the same relationship in an observational or correlational design), and confounding (a third variable associated with both that can distort the apparent relationship). Every variable in a study is classified along both dimensions at once -- measurement level is a property of how the variable was recorded, while design role is a property of what a specific analysis is asking the variable to do.",
"inDefinedTermSet": "https://casrai.org/dictionary/domain/research-outputs#set",
"url": "https://casrai.org/dictionary/term/types-of-variables",
"sameAs": [],
"license": "https://creativecommons.org/licenses/by/4.0/",
"publisher": {
"@id": "https://casrai.org/#organization"
},
"dateModified": "2026-07-23T06:55:38",
"inLanguage": "en"
}
