Slope vs correlation coefficient in regression

Slope vs correlation: the short answer

Both come from the same scatterplot, but they answer different questions. The slope $b$ of the regression line tells you how much the predicted response goes up or down for each one-unit rise in the explanatory variable, in the real units of the data. The correlation coefficient $r$ tells you how tightly the points cluster around a straight line, on a fixed scale from -1 to 1.

A large slope does not mean a strong correlation, and a strong correlation does not mean a large slope. The two numbers are linked, but they measure separate things, and mixing them up is one of the most common regression errors.

What the slope tells you

In the regression equation $\hat{y} = a + bx$, the slope $b$ is the predicted change in the response variable $y$ for a one-unit increase in the explanatory variable $x$. The AP course states this directly: the slope is the predicted increase or decrease in the response for a one-unit increase in the explanatory variable, and you interpret it in context.

Because the slope is a rate of change, it carries units. If $x$ is hours studied and $y$ is exam score, a slope of 4 means about 4 more points per extra hour. Change the units of either variable and the number for the slope changes, even though the underlying relationship is the same. On the AP exam you read the slope from calculator output; the course expects it to be found using technology.

What the correlation coefficient tells you

The correlation coefficient $r$ summarizes the strength and direction of the linear association between two quantitative variables. It is unit-free and always lands between -1 and 1, inclusive. A value near -1 or 1 means the points sit close to a straight line; a value of 0 means no linear association.

The sign of $r$ gives the direction: negative for a downward trend, positive for an upward trend. Because $r$ is unit-free, rescaling the data (say, converting inches to centimeters) leaves it unchanged. That is exactly what you want from a strength score: it should not depend on the units you happened to measure in.

Slope vs correlation side by side

How the two are linked

Slope and correlation share a sign but not a size. They are connected by the standard deviations of the two variables through $b = r \cdot \frac{s_y}{s_x}$, where $s_y$ is the standard deviation of the response and $s_x$ is the standard deviation of the explanatory variable. Since both standard deviations are positive, the fraction $\frac{s_y}{s_x}$ is positive, so $b$ and $r$ always carry the same sign.

That relationship also shows why the size of the slope and the size of $r$ can drift apart: the slope scales with the spread ratio $\frac{s_y}{s_x}$, while $r$ does not. Note that this exact formula is not printed on the Fall 2026 AP formula sheet, which lists only $\hat{y} = a + bx$; on the exam you get $b$ and $r$ from technology. The link is still worth knowing, because it explains why a steep line can have a weak correlation and a gentle line can have a strong one.

When to report which

Report the slope when the question is about size of effect: how much the response moves per unit of the explanatory variable. It is the number you use to make a prediction or to describe a rate in context, such as dollars per square foot or points per hour.

Report the correlation when the question is about strength: how well a line describes the cloud of points. Use $r$ (or its square, $r^2$) to judge how much of the pattern is linear before you trust a prediction. A full description of a linear relationship usually names both: the direction and strength from $r$, and the rate from $b$.

The classic mix-up and how to avoid it

The classic error is treating a big slope as proof of a strong relationship, or reading $r$ as if it were a rate of change. A slope of 50 sounds large, but if the points scatter widely the correlation can still be near 0, and the line predicts poorly. A correlation of 0.95 sounds strong, but it tells you nothing about how many units $y$ rises per unit $x$.

Three habits keep them apart. First, check units: a slope has them, $r$ never does. Second, check the range: $r$ must sit between -1 and 1, so any value outside that is not a correlation. Third, describe strength with $r$ and rate with $b$, and never let one stand in for the other.

Where this fits in AP Statistics

Both ideas live in Unit 5. Correlation is topic 5.2, and the slope of the least-squares regression line is topic 5.5, where you also meet the coefficient of determination $r^2$ covered in r vs r-squared. The Fall 2026 course expects you to interpret both in context and to find them with technology rather than by a printed formula.

A common AP trap pairs a large slope with a weak fit, or asks you to interpret $r$ as though it were a slope. Practice reading calculator output so you can pull the right number for the right question, and remember that neither one proves a cause, as covered in correlation vs causation.