Positive vs negative correlation
By Jude Wallis · Published
A positive correlation means that as one variable increases, the other tends to increase. A negative correlation means that as one increases, the other tends to decrease. The sign of gives the direction of a linear relationship, and the size of , ignoring the sign, gives its strength.
AP Statistics: Unit 5 (topics 5.1 Graphical Representations Between Two Quantitative Variables, 5.2 Correlation). Describing the direction of an association as positive or negative is part of topic 5.1, and the correlation coefficient is topic 5.2, both in Unit 5, Regression Analysis, of the Fall 2026 AP Statistics course. The correlation is normally found with technology and interpreted in context.
Positive vs negative correlation: the short answer
The correlation coefficient is a number from -1 to 1 that describes the linear relationship between two quantitative variables. It carries two separate pieces of information.
The sign is the direction. A positive means higher values of one variable tend to go with higher values of the other, and the scatterplot rises from left to right. A negative means higher values of one tend to go with lower values of the other, and the scatterplot falls from left to right.
The size, the distance of from 0, is the strength. Values near 1 or -1 mean the points lie close to a line. Values near 0 mean a weak linear relationship.
Negative does not mean weak or bad. A correlation of describes a stronger linear relationship than .
What a positive correlation means
With a positive correlation, individuals who are above average on tend to be above average on , and those below average on tend to be below average on . Examples: height and shoe size, hours studied and exam score, outdoor temperature and ice cream sales.
The word "tend" is doing work. A positive correlation is a statement about the overall pattern. Unless exactly, the points do not all lie on one line, and individuals can go against the pattern, such as a tall person with small feet.
A least-squares line fitted to positively correlated data has a positive slope. At the extreme, means every point lies exactly on a line that rises.
What a negative correlation means
With a negative correlation, individuals who are above average on tend to be below average on . Examples: a car's age and its resale value, hours of screen time before bed and hours of sleep, elevation and average temperature.
A negative correlation is as real and as useful for prediction as a positive one of the same size. Knowing a car's age helps predict its value because the relationship is strong. The direction just tells you which way the prediction moves.
A least-squares line fitted to negatively correlated data has a negative slope. At the extreme, means every point lies exactly on a line that falls.
The value that signals no linear relationship is , which is a different thing from a negative value. Even then the variables may be related by a curve, as does a correlation of zero mean no relationship shows.
The differences side by side
| Feature | Positive correlation | Negative correlation |
|---|---|---|
| Sign of | ||
| As increases | tends to increase | tends to decrease |
| Scatterplot, left to right | Rises | Falls |
| Slope of the least-squares line | Positive | Negative |
| Perfect case | ||
| Strength measured by | Distance of from 0 | Distance of from 0 |
| Value of | Same for and | Same for and |
The last two rows are the same in both columns on purpose. Direction and strength are separate, so and describe relationships that are equally strong, and both give .
What the sign does and does not depend on
The sign of comes from the products of standardized values. For each point, multiply its -score for by its -score for . A point above average on both variables, or below average on both, gives a positive product. A point above average on one and below on the other gives a negative product. The correlation is the sum of these products divided by , so its sign shows which kind of point dominates.
Two things do not change the sign or the size of :
- Swapping which variable is called and which is called .
- Changing units, such as inches to centimeters or degrees Celsius to degrees Fahrenheit.
One thing does flip the sign: reversing the direction of a scale. Test score and number of questions missed measure the same performance in opposite directions. If hours studied has with number correct on a fixed-length test, it has with number missed. Before interpreting a sign, check which direction each variable runs.
Computing step by step is covered in how to find r by hand.
The classic mix-up and how to avoid it
The most common error is ranking correlations as numbers on a number line and concluding that is weaker than because it is smaller. For strength, drop the sign and compare 0.8 with 0.3. The negative correlation is the stronger one.
The second error is reading a negative correlation as "no correlation" or as a sign that something is wrong. The value for no linear relationship is 0.
The third error is treating the sign as proof of cause. A positive correlation between ice cream sales and sunburns does not show that ice cream causes sunburn. Both rise with hot weather. Correlation describes association, and the sign says only which way the association runs. See correlation vs causation.
When you describe a relationship, give direction and strength as separate words and put them in context: there is a strong, negative, linear relationship between a car's age and its resale value.
Same strength, opposite signs
Data set A has points (1, 2), (2, 4), (3, 5), (4, 4), (5, 5). Data set B has points (1, 5), (2, 4), (3, 5), (4, 4), (5, 2). Find the correlation for each and compare.
Find the means. Both sets have . Both have , because they contain the same five -values in a different order.
Find the deviations, the same for both sets: -2, -1, 0, 1, 2. Their squares sum to .
Find the deviations. For A: -2, 0, 1, 0, 1. For B: 1, 0, 1, 0, -2. In both sets the squares sum to .
For A, multiply paired deviations and add: .
For B, do the same: .
Compute each correlation as the sum of products divided by the square root of the product of the two sums of squares. For A: . For B: .
Compare. The sizes are equal, so the linear relationships are equally strong. The signs are opposite: in A, tends to rise with , and in B it tends to fall. Both give .
Set A has and set B has . The two relationships have the same strength and opposite directions, and both have .
From the sign of r to the sign of the slope
For a group of teenagers, nightly screen time has standard deviation 1.5 hours, nightly sleep has standard deviation 0.9 hours, and the correlation between screen time and sleep is . Describe the relationship, find the slope of the least-squares line for predicting sleep from screen time, and find .
Read the sign: is negative, so teenagers with more screen time tend to get less sleep.
Read the size: 0.62 is well away from 0 and well short of 1, which is usually described as a moderate linear relationship.
Compute the slope with : hours of sleep per hour of screen time.
Check the signs agree. Standard deviations are positive, so the slope always has the same sign as . Both are negative here.
Interpret the slope: for each additional hour of screen time, predicted sleep decreases by about 0.372 hours, roughly 22 minutes.
Compute . About 38.4% of the variation in sleep is explained by the linear relationship with screen time. The same would result from .
There is a moderate, negative, linear relationship: more screen time goes with less sleep. The slope is hours of sleep per hour of screen time, negative like , and . These are observational data, so the correlation does not show that screen time causes the lost sleep.
Frequently asked questions
Is a negative correlation weaker than a positive correlation?
No. Strength is measured by how far is from 0, ignoring the sign. A correlation of is stronger than a correlation of . The sign tells you only whether the variables tend to move in the same direction or in opposite directions.
What is the difference between a negative correlation and no correlation?
A negative correlation means a linear relationship in which one variable tends to decrease as the other increases. No correlation means is 0 or close to it, so there is no linear relationship in either direction. A negative far from 0 signals a clear relationship.
Does the slope of the regression line always have the same sign as the correlation?
Yes. The least-squares slope equals times the ratio of the standard deviation of to the standard deviation of , and both standard deviations are positive. A positive correlation gives a positive slope and a negative correlation gives a negative slope.