Course contentsShow
← Mathematics
Lesson 2949 of 3,12158. Joint and Limit BehaviorFree lesson

Properties of Covariance

Explore key properties: linearity, symmetry, relationship to variance, and how constants affect covariance.

Properties of Covariance

You'll learn: How covariance behaves under key mathematical operations, making calculations easier and insights clearer.

Core Idea

Covariance measures how two random variables vary together, but it's not just a black-box calculation. It follows predictable rules that mirror properties you've seen with expectation and variance. Understanding these properties lets you simplify complex expressions and reason about relationships between variables without computing everything from scratch.

Key Properties

Linearity in Each Argument

Covariance is linear in both variables:

  • Cov(aX + b, Y) = a·Cov(X, Y)
  • Cov(X, aY + b) = a·Cov(X, Y)

Constants shift the variables but don't change how they co-vary. Scaling one variable scales the covariance proportionally.

Symmetry

Cov(X, Y) = Cov(Y, X)

The order doesn't matter—if X tends to be high when Y is high, then Y tends to be high when X is high.

Relationship to Variance

Cov(X, X) = Var(X)

A variable's covariance with itself is just its variance. This connects our two key measures of spread and association.

Constants Have Zero Covariance

Cov(X, c) = 0 for any constant c

A constant never varies, so it can't co-vary with anything. This explains why adding constants doesn't affect covariance.

Bilinearity (Extension)

For sums: Cov(X₁ + X₂, Y) = Cov(X₁, Y) + Cov(X₂, Y)

Covariance distributes over addition, just like expectation does.

Analogy

Think of covariance like measuring how two dancers move together. Scaling one dancer's movements (multiplying by a) scales their coordination proportionally. Adding a constant just shifts their position without changing how synchronized they are.

Key Takeaway: Covariance is symmetric, linear in each argument, equals variance when applied to a single variable, and unaffected by constant shifts—properties that simplify analysis of joint distributions.