In statistics, machine learning, and quantitative economics, Pearson's Correlation Coefficient (\(r\)) is the definitive metric measuring the strength and linear direction of the association between two continuous variables. From determining whether marketing spend drives sales revenue to testing whether exercise duration lowers resting blood pressure, correlation allows analysts to quantify relationships on a normalized dimensionless scale.
Mastering the \(-1.0 \le r \le +1.0\) scale, the Coefficient of Determination (\(R^2\)), and the correlation vs. causation fallacy ensures rigorous scientific interpretation.
Pearson's r: The -1.0 to +1.0 Bivariate Scale #
| Correlation Value (\(r\)) | Direction & Strength | Linear Behavior |
|---|---|---|
| \(r = +1.00\) | Perfect Positive Correlation | As Variable X increases, Variable Y increases in exact linear lockstep. |
| \(+0.70 \le r \le +0.99\) | Strong Positive | Firm upward linear trend (e.g. Height vs. Shoe Size). |
| \(+0.30 \le r \le +0.69\) | Moderate Positive | General upward tendency with substantial scatter. |
| \(r = 0.00\) | Zero Linear Correlation | No linear association (e.g. Shoe Size vs. Intelligence). |
| \(-0.70 \ge r \ge -0.99\) | Strong Negative Correlation | As Variable X increases, Variable Y decreases (e.g. Car Speed vs. Travel Time). |
| \(r = -1.00\) | Perfect Negative Correlation | Exact inverse linear lockstep. |
Visualizing Positive, Negative, and Zero Correlation Scatter Plots #
Scatter plot patterns for Strong Positive (\(r = +0.95\)), Zero (\(r = 0.0\)), and Strong Negative (\(r = -0.95\)):
Scatter Plot Patterns: Positive, Zero, and Negative Correlation
Visualizing r = +0.95 (Green), r = 0.00 (Slate), and r = -0.95 (Red).
| Pattern | Correlation Coefficient (\(r\)) | Practical Meaning |
|---|---|---|
| Upward Linear Sloping | \(+0.80\) to \(+1.00\) | Variables increase together in strong proportion. |
| Random Cloud / No Slope | \(-0.20\) to \(+0.20\) | No linear relationship between variables. |
| Downward Linear Sloping | \(-0.80\) to \(-1.00\) | Inverse relationship: as X increases, Y decreases. |
The Mathematical Pearson r & Covariance Formulas #
r = Σ[ (x_i − x̄)(y_i − ȳ) ] / √[ Σ(x_i − x̄)^2 × Σ(y_i − ȳ)^2 ] = Cov(X, Y) / [ s_x × s_y ] Pearson's \(r\) is simply sample covariance normalized by the product of the individual standard deviations of X and Y.
The Coefficient of Determination (R²): Explaining Variance #
Squaring Pearson's \(r\) yields the Coefficient of Determination (\(R^2\)), which represents the proportion of total variance in Variable Y directly explained by Variable X:
- If \(r = 0.80\): \(R^2 = (0.80)^2 = \mathbf{0.64 \text{ (64%)}}\). Exactly 64% of the variation in Y is explained by X; the remaining 36% is unexplained residual variance.
- If \(r = 0.50\): \(R^2 = (0.50)^2 = \mathbf{0.25 \text{ (25%)}}\).
Worked Step-by-Step Calculation (5 Data Pairs) #
Evaluate paired data for Ad Spend \(X\) and Sales \(Y\): \((1, 2), (2, 3), (3, 5), (4, 4), (5, 6)\):
- \(ar{x} = 3.0\), \(ar{y} = 4.0\).
- \(\text{SS}_x = \sum(x - 3)^2 = 4 + 1 + 0 + 1 + 4 = \mathbf{10.0}\).
- \(\text{SS}_y = \sum(y - 4)^2 = 4 + 1 + 1 + 0 + 4 = \mathbf{10.0}\).
- \(\sum[(x - 3)(y - 4)] = (-2)(-2) + (-1)(-1) + (0)(1) + (1)(0) + (2)(2) = 4 + 1 + 0 + 0 + 4 = \mathbf{9.0}\).
- Calculate \(r\):
\[r = \frac{9.0}{\sqrt{10.0 \times 10.0}} = \frac{9.0}{10.0} = \mathbf{+0.90 \text{ (Strong Positive)}}\] - Calculate \(R^2\): \(R^2 = (0.90)^2 = \mathbf{0.81 \text{ (81% Variance Explained)}}\).
Check statistical significance testing in our p-value guide.
The Correlation vs. Causation Fallacy (Spurious Correlations) #
"Correlation proves relationship, but correlation does NOT prove causation."
A classic example: Ice cream sales and drowning deaths correlate strongly (\(r = +0.85\)). Eating ice cream does not cause drowning; both variables are independently driven by a confounding third variable: hot summer temperatures. Establishing true causal mechanisms requires controlled randomized experiments (A/B tests).
Key Takeaways #
- Pearson's \(r\) ranges strictly from \(-1.0\) to \(+1.0\).
- \(R^2 = r^2\) quantifies the percentage of variance explained.
- Correlation measures only LINEAR relationships (fails on non-linear curves like parabolas).
- Calculate Pearson's r, R², and linear regressions: use our free Correlation Calculator.
Frequently Asked Questions #
What is the difference between Pearson and Spearman correlation?
Pearson's r measures linear relationships between continuous parametric variables. Spearman's rank correlation (\( ho\)) measures monotonic relationships between ranked or non-normal ordinal variables, making it immune to extreme outlier distortions.
What is Anscombe's Quartet?
Anscombe's Quartet consists of four datasets that have the exact same mean, variance, regression line, and correlation coefficient (\(r = 0.816\)), yet look completely different when graphed (linear, curved parabola, vertical with single outlier), proving why you must always visualize scatter plots before trusting correlation math.
How does standard deviation affect Pearson's r?
Pearson's \(r\) divides covariance by the product of both variables' standard deviations (\(s_x \times s_y\)), which standardizes the measurement and ensures \(r\) is independent of units (inches, dollars, kilograms). See our standard deviation guide.
Primary Sources & Citations #
- Pearson, K. (1895). "Note on regression and inheritance in the case of two parents." Proceedings of the Royal Society of London, 58, 240–242.
- Rodgers, J. L., & Nicewander, W. A. (1988). "Thirteen ways to look at the correlation coefficient." The American Statistician, 42(1), 59–66.
- Vigen, T. (2015). Spurious Correlations. Hachette Books.
Our team of financial analysts, engineers, and researchers builds precision calculation tools and evidence-based guides. Every article is peer-reviewed for mathematical accuracy and tested against primary source data. Learn about our editorial standards.
Looking for more? Browse all free resources including guides, comparisons, and glossary terms.