Perfect Linear Relationships
What you'll learn: You'll discover exactly when correlation reaches its extreme values of +1 or −1 and what that reveals about the mathematical relationship between two variables.
The Extreme Case of Correlation
You've learned that the correlation coefficient ρ (or sample correlation r) measures linear association and always falls between −1 and +1. But when does it actually equal these boundary values?
The answer: Correlation equals exactly +1 or −1 if and only if the two variables have a perfect linear relationship—meaning one variable can be written as an exact linear function of the other.
Mathematical Characterization
For random variables X and Y:
- ρ = +1 occurs when Y = aX + b for some constants a > 0 and b (positive slope)
- ρ = −1 occurs when Y = aX + b for some constants a < 0 and b (negative slope)
In these cases, there's exact linear dependence: every point lies precisely on a straight line with no scatter whatsoever. The variables are perfectly predictable from one another.
The Intuition
Think of correlation as measuring "how close to a line" your data falls. Most relationships have some scatter, giving correlations like 0.7 or −0.4. But when ρ = ±1, there's zero scatter—the data forms a perfect line. It's like saying "X and Y are really just the same information, stretched and shifted."
Note that if a = 0 (making Y constant), correlation is undefined because there's no variation to correlate.
Key Takeaway: Correlation equals ±1 precisely when one variable is an exact linear transformation of the other, representing complete linear dependence with zero deviation from a straight line.