Explainer · Plain-language

What Is a Variable in Research? Types, Examples & Measurement | CASRAI

A variable in research is any characteristic, attribute, or quantity that can take on different values across people, time points, or experimental conditions. Variables are the building blocks of empirical research: they are what researchers measure, manipulate, and examine in relation to one another.

CASRAI plain-language explainers — clear answers to recurring research-administration questions

The step most authors miss

Doing CRediT right? Don’t stop at the statement.

A CRediT statement credits you inside one paper. The recognition CRediT was built for happens when those roles are tied to you, persistently. Sign in with your ORCID — free — and claim your CRediT contributions on casrai.org, the home of the standard. They become a verified, portable part of your identity, not a line that disappears into one PDF.

Free: claim your contributions, then export a journal-ready CRediT statement, schema.org structured data, JATS XML, CSV or BibTeX — and preview your public profile. A membership publishes that profile publicly and verifies the journals you serve.

Variables classified by role

The role a variable plays determines how it is treated analytically. An independent variable (IV) is the presumed cause — the factor a researcher manipulates (in an experiment) or selects to compare (in an observational study). A dependent variable (DV) is the measured outcome expected to change in response to the IV. A confounding variable is an outside factor correlated with both the IV and DV that can produce a spurious relationship between them, threatening internal validity. A control variable is a potential confound held constant deliberately. A mediating variable explains the mechanism by which the IV influences the DV (the "how"). A moderating variable changes the strength or direction of the IV–DV relationship (the "when" or "for whom").

Levels of measurement: Stevens (1946)

Stanley Smith Stevens (1946) proposed four hierarchical levels of measurement that determine the mathematical operations and statistical tests applicable to a variable. Nominal variables classify cases into named, unordered categories with no numerical meaning — blood type, nationality, diagnosis. Ordinal variables rank cases in order but with unequal intervals — Likert-scale responses, educational attainment levels. Interval variables have equal intervals between values but no true zero — temperature in Celsius, IQ scores. Ratio variables have equal intervals and a meaningful zero point — weight, height, reaction time, income. The level of measurement determines appropriate statistics: means are valid for interval and ratio variables but not for nominal or ordinal ones; only ratio variables support meaningful ratio statements ("twice as heavy").

Operationalisation

Operationalisation is the process of turning an abstract concept into a concrete, measurable variable. The construct "student motivation" has no ruler; to measure it, a researcher must choose an observable indicator — for example, scores on a validated motivation questionnaire, number of voluntary study hours, or grade point average. The operational definition specifies exactly how the variable will be measured in the study, enabling replication and making implicit assumptions explicit. Different operationalisations of the same construct can produce different results, which is one reason replication across studies using different operationalisations (conceptual replication) builds stronger scientific knowledge than replication using identical procedures.

Continuous and discrete variables

Within numerical (quantitative) variables, a further distinction applies. Continuous variables can take any numerical value within a range — height could in principle be 1.734 m or 1.7341 m. They are measured rather than counted. Discrete variables take only specific, separate values — typically whole numbers. Number of publications, number of children, and number of hospital admissions are discrete: a researcher cannot have 2.7 publications. This distinction matters for statistical modelling: methods such as linear regression assume a continuous outcome; count data require Poisson or negative binomial regression; binary outcomes require logistic regression. Mismatching the statistical model to the variable type is a common analytical error.

Key facts

At a glance

Definition: Any characteristic that can take different values across cases or conditions
Roles: Independent, dependent, control, confounding, mediating, moderating
Stevens (1946): Nominal, ordinal, interval, ratio — four levels of measurement
Operationalisation: Converting an abstract concept into a measurable variable
Continuous: Can take any value in a range (measured)
Discrete: Can only take specific separate values (counted)

Common misconceptions

What people often get wrong

Often heard: A variable is always a number.

Actually: No — a variable is any attribute that can take different values, including categorical ones such as nationality, blood type, or diagnosis. Nominal and ordinal variables are non-numerical but are still valid variables in research.

Often heard: The independent variable is always the more important one.

Actually: No — independent and dependent are roles that depend on the study's design and question, not the variable's importance. The same characteristic (e.g., income) can be an IV in one study and a DV in another.

Often heard: Ordinal and interval data can be analysed with the same statistical methods.

Actually: No — the level of measurement constrains valid operations. Calculating a mean on ordinal data (e.g., averaging Likert ratings) is statistically controversial because the intervals between response categories cannot be assumed equal.

Going deeper