Correlation: From Intuition to Precision

Imagine you're tracking two things at once — say, hours spent studying and exam scores. You notice a pattern: when study hours go up, scores tend to go up too. That's correlation in its simplest form: a measure of how two variables move together.

But here's the key intuition: correlation is not about causation. Just because two things move together doesn't mean one causes the other. Ice cream sales and drowning incidents both rise in summer — they're correlated, but ice cream doesn't cause drowning. Both are driven by a third factor (heat).

The Core Idea

Correlation quantifies the strength and direction of a linear relationship between two variables. It answers three questions:

1. Direction: Do they move in the same direction (positive) or opposite directions (negative)?
2. Strength: How tightly do they follow that pattern — is it a perfect line, or a loose cloud of points?
3. Form: Is the relationship linear (a straight line) or something else? Correlation only measures linear relationships.

The Precise Statement: Pearson's Correlation Coefficient

The most common measure is Pearson's correlation coefficient, denoted by $r$. It's a single number between $-1$ and $+1$.

Where:

• $x_i, y_i$ are individual data points
• $\bar{x}, \bar{y}$ are the means of $x$ and $y$
• $n$ is the number of pairs

This is the same formula written more compactly: covariance divided by the product of standard deviations.

What the Numbers Mean

The Intuition Behind the Formula

The numerator $\sum (x_i - \bar{x})(y_i - \bar{y})$ is the covariance. For each point, it asks: is this point above or below the mean in both variables?

• If a point is above the mean in $x$ and above the mean in $y$, the product $(x_i - \bar{x})(y_i - \bar{y})$ is positive.
• If a point is above in $x$ but below in $y$, the product is negative.

Summing all these products tells you whether the overall pattern is positive or negative.

The denominator $\sqrt{\sum (x_i - \bar{x})^2} \sqrt{\sum (y_i - \bar{y})^2}$ is just a scaling factor — it ensures $r$ always falls between $-1$ and $+1$, regardless of the units of measurement.

A Simple Example

Suppose you have five students:

Here $\bar{x} = 3$, $\bar{y} = 60$. Every point lies exactly on the line $y = 10x + 30$. Computing $r$ gives exactly $+1$ — perfect positive correlation.

Now change the last point to (5, 50). The pattern is still upward but not perfect. $r$ drops to about $+0.6$ — moderate positive correlation.

What Correlation Does NOT Tell You

1. Causation: $r = 0.9$ does not mean $x$ causes $y$. There could be a third variable, or the relationship could be coincidental.
2. Non-linear relationships: A perfect circle has $r = 0$ even though $x$ and $y$ are perfectly related.
3. Outliers: A single extreme point can dramatically change $r$. Always plot your data first.
4. Slope: $r$ doesn't tell you how steep the line is — only how tightly points cluster around it.

The Bottom Line

Correlation is a precise, standardized measure of linear association. It tells you whether two variables move together, how consistently they do so, and in which direction. But it's a tool, not an oracle — use it with scatter plots, context, and a healthy skepticism about causation.