# Correlation

A statistical association between two variables that does not alone establish causation.

Canonical page: https://givegigs.com/wiki/correlation
Last updated: 2026-11-09T19:00:06.411Z
Parent topic: [research](https://givegigs.com/wiki/research)

## Contents
- [Overview](#wiki-section-overview-1)
- [Read an association carefully](#wiki-section-interpret-2)
- [From association to cause](#wiki-section-causal-questions-3)

<a id="wiki-section-overview-1"></a>
## Overview

Correlation describes a pattern of statistical association between variables: values tend to vary together in some way. The pattern may be positive, negative, nonlinear, or absent under the chosen measure. Correlation alone does not establish that one variable causes another.

Explore related topics:
- [statistics](https://givegigs.com/wiki/statistics)
- [causation](https://givegigs.com/wiki/causation)
- [sampling](https://givegigs.com/wiki/sampling)

<a id="wiki-section-interpret-2"></a>
## Read an association carefully

Inspect a graph and the underlying observations, not just a coefficient. Check the scale, outliers, subgroups, missing data, and whether the association holds over the range being discussed. A strong relationship in one dataset may change when the population or measurement changes.

<a id="wiki-section-causal-questions-3"></a>
## From association to cause

A third factor may influence both variables, the direction may run the other way, or selection and measurement may create a misleading pattern. Causal conclusions need a design and reasoning that address these alternatives, such as a well-conducted experiment or a justified observational method. Do not infer an intervention's effect from correlation alone.

## Sources

- [NIST Engineering Statistics Handbook: Scatter Plots](https://www.itl.nist.gov/div898/handbook/eda/section3/eda33q.htm)

---
This article is collaboratively maintained. Check the canonical page and cited sources for the latest context.