
How to interpret r² in AP Statistics




r² (r-squared), also called the coefficient of determination, is the proportion of the variation in the response variable (y) that is explained by its linear relationship with the explanatory variable (x). If r² = 0.64, then 64% of the variation in y is explained by the linear relationship with x, and the other 36% comes from everything else.
On the AP Statistics exam, calculating r² is rarely the hard part, because technology gives it to you. The points go to the sentence that interprets it in context. This guide covers where r² comes from, how to write the interpretation, and what r² cannot tell you.
Where r² fits in AP Statistics
In the AP Statistics Course and Exam Description, r² belongs to Unit 5, Regression Analysis. The unit is about 10–20% of the multiple-choice section and about nine class periods, and it covers graphs of two quantitative variables, correlation, linear regression models, residuals and least-squares regression.
The course description defines r² in one sentence: in simple linear regression, the square of the correlation coefficient is called the coefficient of determination, and it is the proportion of variation in the response variable that is explained by the linear relationship with the explanatory variable. The unit also tells teachers to give students practice translating calculator output into appropriate statistical language, which is exactly the skill r² tests.
If you want the surrounding vocabulary first, start with the AP Statistics vocabulary list, the definition of the correlation coefficient and the definition of least-squares regression.
What r² means
Suppose you fit a least-squares line to predict a student’s exam score (y) from the hours they studied (x), and the output says r² = 0.64. Two readings say the same thing:
- Variation explained: 64% of the variation in exam scores is explained by the linear relationship with hours studied.
- Prediction error: the squared prediction errors from the line are 64% smaller than the squared errors you would make by predicting the mean score for everyone.
The first reading is the one the exam wants. r² always falls between 0 and 1 (software often prints it as a percent). An r² of 0 means the line explains none of the variation, and an r² of 1 means every point sits exactly on the line.
How r² is calculated
In simple linear regression, r² is the square of the correlation coefficient, r. The AP course expects you to read r and r² from technology rather than compute them by hand.
It still helps to know where the number comes from. Each point’s distance from the mean of y is its total variation, and the regression line splits that into a part it explains and a part it leaves over (the residuals):
- Total variation (SST): the sum of the squared differences between each y and the mean of y.
- Unexplained variation (SSE): the sum of the squared residuals.
- r² = 1 − SSE / SST. If SST = 400 and SSE = 144, then r² = 1 − 144/400 = 0.64, so r = 0.80 or r = −0.80.
That last point matters: squaring throws away the sign. A positive and a negative association can have the same r². The direction comes from r, which always has the same sign as the slope of the line. If r² = 0.81 and the slope is negative, r = −0.90.
How to write the interpretation
Use this template and fill in the context every time:
- "About [r² as a percent] of the variation in [response variable] is explained by the linear relationship with [explanatory variable]."
Both variable names have to be in the sentence, in the right order. A generic sentence such as "r² is the proportion of variation explained" earns nothing, because it never mentions the data.
These are the interpretations that lose points:
- "64% of the students’ scores are explained." r² is about variation, not a share of the individuals or the points.
- "The model is 64% accurate." r² is not an accuracy rate or the percentage of predictions that are right.
- "Studying causes 64% of the change in scores." Regression shows association. Only a well-designed experiment supports a causal claim.
- Swapping the variables. The variation explained is in the response (y), by the explanatory variable (x), never the other way around.
Practice question
The AP Statistics Course and Exam Description includes this sample multiple-choice question:
Emmalyn collects data from her track teammates: their 60-meter-dash time in seconds (x) and long-jump distance in inches (y). Her model is ŷ = 313.875 − 30.805x, with r = −0.591 and r² = 0.349. Which is a correct interpretation of the coefficient of determination?
- (A) Approximately 35% of the variation in dash time can be explained by long-jump distance.
- (B) Approximately 59% of the variation in dash time can be explained by long-jump distance.
- (C) Approximately 35% of the variation in long-jump distance can be explained by the linear relationship with dash time.
- (D) Approximately 59% of the variation in long-jump distance can be explained by the linear relationship with dash time.
The answer is (C), and the wrong choices are the mistakes above. Choices (A) and (B) swap the variables, so they explain dash time (x) instead of long-jump distance (y). Choices (B) and (D) report 59%, which is the size of r, not r². The correct answer names the response variable, uses 35% (0.349 rounded), and says "the linear relationship with" the explanatory variable.
What r² does not tell you
A high r² is not a verdict that the model is good. It cannot tell you:
- Whether a line is the right model. Several very different scatterplots can share the same r², so r² alone never proves the pattern is linear. Check the residual plot too: apparent randomness supports a linear model, and curvature suggests a different one. This is Topic 5.4 in the course description.
- Whether the association is causal. Two variables can move together without one causing the other. The course description’s own example asks whether shark attacks and ice cream sales rising together means ice cream causes shark attacks.
- The direction of the relationship. Use the sign of r or of the slope.
- Whether outliers are driving the result. A single influential point can raise or lower r² a great deal, so look at the scatterplot.
- Whether predictions outside your data are safe. Predicting for x-values beyond the range you observed is extrapolation, and r² says nothing about it.
Also keep a low r² in context. In a field where many factors drive the outcome, such as human behavior, an r² of 0.25 can still point to a real and useful relationship. It just means individual predictions will be noisy.
r² and adjusted r²
AP Statistics only covers simple linear regression, with one explanatory variable, so plain r² is all you need. If you take statistics further, you will meet multiple regression, where a model has several explanatory variables. There, r² never goes down when you add a variable, even a useless one, so analysts report adjusted r², which penalizes extra variables. The idea of "variation explained" stays the same.
Study tips
- Memorize the template sentence and practice filling it in for new contexts until the variable order is automatic.
- When you see r² in a problem, ask which variable is the response, and put it first.
- If a question gives you r² and asks about r, find the sign from the slope or the direction of the scatterplot.
- Practice with released questions from the AP Statistics past exam questions page and the sample questions in the course description.
- For a second explanation with a worked example, see Penn State’s page on the coefficient of determination.
- Keep the AP Stats formula sheet handy while you practice, and use the AP Statistics score calculator to see how your practice results translate to an exam score.
Key takeaways
- r² is the proportion of variation in y explained by the linear relationship with x, and in simple regression it equals r squared.
- Interpret it in context: "About __% of the variation in [response] is explained by the linear relationship with [explanatory]."
- r² has no sign. Get the direction from r or the slope.
- It is not an accuracy rate, it does not show causation, and it does not prove a line is the right model, so check the residual plot as well.
- It is a Unit 5 skill, and Unit 5 is about 10–20% of the multiple-choice section.