Skip to content

Correlation

Origin domain
Mathematics
Also from
Economics & Finance, Public Administration & Policy, Physics, Computer Science & Software Engineering
Aliases
Co Variation, Statistical Association, Dependence

Core Idea

Correlation is the structural pattern in which two or more variables systematically co-vary — values of one tend to track values of another above what independence would predict — without any implied mechanism, direction, or production relation between them. The defining commitment is statistical association as a self-standing fact: knowing one variable updates expectations about the other, yet the association is silent about which (if either) drives which, leaving open common-cause, reverse-cause, mediated, or coincidental explanations.

How would you explain it like I'm…

Goes Together

When ice cream sales go up, sunburns go up too. They go together! But ice cream does not cause sunburns. The sun causes both. Things can move together without one making the other happen.

Things That Move Together

Correlation means two things tend to change together. When one goes up, the other often goes up too (or down). Tall parents usually have tall kids. Cold weather and hot chocolate sales both rise in winter. But just because two things move together doesn't mean one causes the other. Something else might be making them both happen, or it could even be a coincidence.

Statistical Association

Correlation is a measurable pattern where two variables tend to change together more than chance would predict. If you know one value, you can make a better guess about the other. Scientists measure this with a number between minus one and plus one. A high positive number means they rise together; a high negative number means one rises as the other falls. Crucially, correlation says nothing about what causes what. Maybe A causes B, maybe B causes A, maybe a hidden third factor causes both, or maybe it's coincidence.

 

Correlation is the structural pattern in which two or more variables systematically co-vary, such that knowing one variable's value updates your probability distribution over the other beyond what statistical independence would predict. Francis Galton first quantified this in 1888 measuring the co-variation of human stature across kin, and Karl Pearson formalized the product-moment coefficient in 1896, a normalized measure of linear co-movement bounded between minus one and plus one. The defining commitment is that correlation is a self-standing fact about joint variation, silent about mechanism: it leaves open common-cause explanations, reverse causation, mediated chains, or sheer coincidence. The same structural shape recurs across finance (co-moving asset returns), epidemiology (exposure-outcome associations), physics (entangled-particle statistics), machine learning (predictive features), and ecology (species co-occurrence). The minimal commitment is always the same: together, but not necessarily because of one another.

Broad Use

  • Statistics / mathematics: the correlation coefficient measuring linear co-movement of two random variables.
  • Finance: correlated asset returns, central to portfolio diversification and systemic risk.
  • Epidemiology / public health: observed association between an exposure and an outcome that may or may not be causal.
  • Physics (non-obvious): quantum correlations between entangled particles whose measurements covary without a classical signal between them.
  • Machine learning: feature correlations that aid prediction yet mislead when mistaken for causal levers.
  • Ecology: species co-occurrence patterns that may reflect interaction or shared habitat preference.

Clarity

Naming correlation lets practitioners assert a real, exploitable relationship while withholding the stronger claim of causation — the single most important hygiene rule in empirical reasoning. It distinguishes "moves together" from "makes happen" and exposes the gap that confounders, selection, and coincidence can fill.

Manages Complexity

It compresses a cloud of joint observations into a directionless summary of dependence, enough to predict and to flag where deeper mechanism-finding is warranted, without committing to the much harder causal model. This lets analysts prioritize: prediction needs only correlation; intervention needs causation.

Abstract Reasoning

Recognizing correlation as distinct supports the inferences "association does not license intervention," "a third variable may explain both," and "a strong predictor need not be a usable lever." It motivates the whole apparatus of confounding, randomization, and causal identification built to upgrade a correlation to a causal claim.

Knowledge Transfer

The "correlation is not causation" caution transfers across every empirical field: the epidemiologist's confounder, the economist's omitted variable, and the ML practitioner's spurious feature are one structure. The diversification insight from finance — combine weakly correlated components to reduce variance — transfers to ensemble learning and to portfolio-style risk pooling in engineering reliability.

Example

Ice-cream sales correlate with drowning deaths; neither causes the other — summer heat drives both. The same directionless co-variation describes correlated mortgage defaults that amplified 2008 systemic risk and the perfectly correlated measurement outcomes of entangled photons, where the association is real and predictive yet carries no transmissible cause between the sites.

Relationships to Other Abstractions

Local relationship map for CorrelationParents appear above the current abstraction, mutual partners to the right, and children below. Node labels state whether each abstraction is prime or domain-specific; colors identify relation types.CorrelationPRIMEDomain-specific abstraction: Wrong Direction (Reverse Causation Fallacy) — presupposesWrong Direction…DOMAINPrime abstraction: Correlated-Source Attribution Failure — presupposesCorrelated-Sour…PRIMEPrime abstraction: Cross-Dimensional Leakage — presupposes, typicalCross-Dimension…PRIMEDomain-specific abstraction: Feldstein-Horioka Puzzle — is a decomposition ofFeldstein-Horio…DOMAINDomain-specific abstraction: Gibson's Paradox — is a decomposition ofGibson's ParadoxDOMAINDomain-specific abstraction: Ecological Correlation — is a kind ofEcologicalCorrelationDOMAINDomain-specific abstraction: Gloger's Rule — is a kind ofGloger's RuleDOMAIN

Current abstraction Correlation Prime

Foundational — no parent edges in the catalog.

Children (7) — more specific cases that build on this

  • Ecological Correlation Domain-specific is a kind of Correlation

    Ecological correlation is a correlation specialized to variables measured on partition-aggregated groups rather than individuals.

  • Gloger's Rule Domain-specific is a kind of Correlation

    Gloger's rule is correlation specialized to positive covariation between melanin pigmentation and environmental humidity within endotherm taxa.

  • Wrong Direction (Reverse Causation Fallacy) Domain-specific presupposes Correlation

    Wrong Direction presupposes correlation because the fallacy begins with a symmetric observed association that carries no information about which variable is upstream.

  • Correlated-Source Attribution Failure Prime presupposes Correlation

    Correlated-Source Attribution Failure presupposes Correlation, whose structure must already obtain for the child mechanism to be meaningful or operational.

  • Cross-Dimensional Leakage Prime presupposes, typical Correlation

    A specific generative story for WHY an observed cross-output correlation is inflated above the true cross-source signal: a shared channel loads onto multiple outputs (cov(y_i,y_j) = cov(t_i,t_j) and lambda_ilambda_jvar©).

Not to Be Confused With

Correlation is not causality, which adds a productive, asymmetric, mechanism-bearing connection; correlation is exactly the association stripped of that productive link, which is why the two are famously conflated and must be separated. It is not coupling, where a specified mechanism makes a change in one produce a change in the other; correlation may exist with no mechanism at all. It is more specific than relation: correlation is the statistical co-variation species of association, not any pattern of standing-together.