Definition · Plain-language
Correlational research
Correlational research measures whether and how two or more variables are related, without manipulating any of them, quantifying the strength and direction of their association.
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.
Measuring relationships, not manipulating
Correlational research sets out to discover whether variables are associated and, if so, how closely. Unlike an experiment, it involves no manipulation: the researcher measures the variables as they naturally occur and examines whether they tend to vary together. For example, a study might record students’ sleep and their exam scores to see whether more sleep accompanies higher marks. The method is well suited to questions where manipulation is impossible or unethical, to studying variables in natural settings, and to making predictions — knowing that two variables are related allows one to be estimated from the other.
Direction and strength of correlation
A correlation has both a direction and a strength, usually summarised by a correlation coefficient ranging from −1 to +1. A positive correlation means the variables rise and fall together; a negative correlation means one rises as the other falls; and a coefficient near zero means little or no linear relationship. The magnitude indicates strength: values near +1 or −1 signal a strong, tight association, while values near zero signal a weak one. Correlations describe linear patterns, so a coefficient near zero does not rule out a strong non-linear relationship that the statistic simply fails to capture.
Correlation is not causation
The cardinal rule of correlational research is that correlation does not imply causation. Even a strong, reliable correlation cannot establish that one variable causes the other, for three reasons: a third, confounding variable may drive both (the third-variable problem); the causal direction may be the reverse of what is assumed (the directionality problem); or the association may be coincidental. Establishing causation requires the manipulation and control of an experiment. Correlational findings are nonetheless valuable for prediction and for generating hypotheses that experiments can then test.
Key facts
At a glance
- Definition: measures the relationship between variables without manipulating them
- Summarised by: a correlation coefficient from −1 to +1
- Positive: variables increase and decrease together
- Negative: one variable increases as the other decreases
- Cardinal rule: correlation does not imply causation
- Why no cause: confounding, reverse causation or coincidence may explain the link
Common misconceptions
What people often get wrong
Often heard: A strong correlation proves that one variable causes the other.
Actually: It does not. A confounding variable, reversed causal direction, or coincidence can all produce a strong correlation, so causation requires the manipulation and control of an experiment.
Often heard: A correlation near zero means the variables are completely unrelated.
Actually: Not necessarily. The coefficient measures linear association, so a near-zero value can still hide a strong non-linear (for example, U-shaped) relationship that the statistic does not detect.
Often heard: Correlational research is just a weaker substitute for an experiment.
Actually: It serves different purposes — prediction, studying non-manipulable variables, and generating hypotheses — and is essential where manipulation is impossible or unethical, not merely a fallback experiment.
