Properties of Variance: Scaling and Translation
You'll learn how multiplying or adding constants to a random variable affects its variance—and discover a surprising fact about shifting data.
The Core Idea
Variance measures the spread of a random variable around its mean. When you transform a random variable by scaling (multiplying) or translating (adding a constant), the variance changes in predictable ways:
- Translation (adding a constant): If you add any constant b to every outcome, the variance stays exactly the same: Var(X + b) = Var(X)
- Scaling (multiplying by a constant): If you multiply every outcome by a constant a, the variance gets multiplied by a²: Var(aX) = a²Var(X)
- Combined: When you do both at once: Var(aX + b) = a²Var(X)
Why This Makes Sense
Translation doesn't affect spread. Think of test scores: if everyone gets 5 bonus points, the average shifts up by 5, but the spread between scores stays identical. Everyone moved together, so the variability is unchanged.
Scaling amplifies spread quadratically. If you convert measurements from meters to centimeters (multiplying by 100), distances between values grow by factor 100. Since variance involves squared deviations, it gets multiplied by 100² = 10,000. Doubling all values (a = 2) quadruples the variance.
Example
If Var(X) = 9, then:
- Var(X + 10) = 9 (shifting doesn't change spread)
- Var(3X) = 3² × 9 = 81 (tripling triples distances, squares variance)
- Var(3X - 7) = 9 × 9 = 81 (the -7 doesn't matter)
Key Takeaway: Adding constants to a random variable leaves variance unchanged, while multiplying by a constant a scales variance by a²—capturing how spread grows with the square of the scaling factor.