AP Statistics · Unit 5 · Regression Analysis
5.3 Linear regression models
Write the equation of a linear regression model, interpret its slope and intercept in context, and use the model to make predictions.
1/5A linear regression model predicts a response from an explanatory value
- A linear regression model predicts a response from an explanatory value
- The equation of the line has an intercept and a slope
- Slope and intercept must be interpreted in context
- Substituting an x value gives a prediction
- Predictions outside the data range are unreliable
- Check your recall
- Practice: 10 questions
Part 1 of 5: A linear regression model predicts a response from an explanatory value
Suppose a teacher wants to predict quiz scores, out of 100, from weekly study time. A scatterplot of the data trends upward in a roughly straight pattern. A linear regression model summarizes that pattern with a straight line used for prediction.
The value we want to predict is the response variable, written . The value we use to make the prediction is the explanatory variable, written . Here, quiz score is the response and study time is explanatory.
Key terms · tap one for its definition
linear regression model
A linear regression model is an equation that predicts the response variable from the explanatory variable using a straight line.
response variable
The response variable is the variable whose value the model predicts, plotted on the vertical axis.
explanatory variable
The explanatory variable is the variable used to predict the response, plotted on the horizontal axis.
Part 2 of 5: The equation of the line has an intercept and a slope
Now the model needs an equation. It has the form . The symbol , read as y-hat, is a predicted value of the response, not an observed one.
The constant is the y-intercept. The coefficient is the slope. For the quiz example, suppose the model is .
Here is hours studied per week and is the predicted quiz score. The model was built from students who studied between 1 and 8 hours.
Key terms · tap one for its definition
y-intercept
The y-intercept is the value in , the predicted response when .
slope
The slope is the value in , the predicted change in the response for each one-unit increase in .

Quick check
One question on the part you just read
Easy
Show answer and explanation
Answer: B · Worked explanation
Part 3 of 5: Slope and intercept must be interpreted in context
The numbers 55 and 4.5 mean little until they are tied to the variables. A full interpretation names both variables, uses units, and says predicted.
- Slope: for each additional hour studied per week, the predicted quiz score increases by 4.5 points.
- Intercept: when hours studied is 0, the predicted quiz score is 55 points.
In this model, 0 hours lies outside the data, which run from 1 to 8 hours. The intercept is then more a starting point for the line than a real prediction. Mention this when it applies.
Part 4 of 5: Substituting an x value gives a prediction
A fitted model earns its keep through prediction. To predict, substitute the given value into the equation and calculate . The result is a predicted value, so real students will score above or below it.
Worked example
Predict a quiz score
- Write the equation: .
- Substitute : .
- Multiply: .
- Add: .
Answer: The predicted quiz score is 82 points.
On the graph, each prediction is a point on the line. The observed scores scatter around it. Report a prediction with units and context. For example, say that a student who studies 6 hours is predicted to score 82 points. A bare number is incomplete, and leaving out the word predicted suggests the value is certain.

Quick check
One question on the part you just read
Easy
Show answer and explanation
Answer: A · Worked explanation
Part 5 of 5: Predictions outside the data range are unreliable
A line extends forever, but the pattern in the data may not. Predicting for an value inside the range of the data is interpolation. Predicting beyond that range is extrapolation.
The quiz model came from students who studied 1 to 8 hours. For 20 hours, it gives points. That is impossible on a quiz scored out of 100.
Extrapolation fails because the relationship may change outside the observed range. The data give no evidence about what happens there.
Key terms · tap one for its definition
interpolation
Interpolation is predicting the response for an value that lies within the range of the observed data.
extrapolation
Extrapolation is predicting the response for an value outside the range of the observed data, which is unreliable.
Check your recall[13 marks]
6 written questions. Answer each one in your head or on paper, then open the mark scheme.
- 1.Identify[1 mark]
A model for the resale price of a car, in dollars, from its age in years is . Identify the explanatory variable.
Show mark schemeHide mark scheme
Model answer
The explanatory variable is the age of the car, in years.
- 1 mark for identifying the age of the car (in years) as the explanatory variable
- 2.Describe[2 marks]
Describe the meaning of the slope in the car model , in context.
Show mark schemeHide mark scheme
Model answer
For each additional year of age, the predicted resale price decreases by 1,500 dollars.
- 1 mark for stating that the predicted price changes by 1,500 dollars (a decrease) for each additional year of age
- 1 mark for using the word predicted and giving units and context
- 3.Describe[2 marks]
Describe the meaning of the y-intercept, 21,000, in the car model, in context.
Show mark schemeHide mark scheme
Model answer
A car of age 0 years is predicted to have a resale price of 21,000 dollars.
- 1 mark for stating that the intercept is the predicted price when age is 0 years
- 1 mark for giving the value 21,000 dollars in context
- 4.Calculate[2 marks]
Calculate the predicted resale price of a 4-year-old car using .
Show mark schemeHide mark scheme
Model answer
dollars.
- 1 mark for substituting into the equation
- 1 mark for obtaining 15,000 dollars with units
- 5.Justify[4 marks]
The car data came from cars aged 1 to 9 years. Justify why the model should not be used to predict the price of a 20-year-old car.
Show mark schemeHide mark scheme
Model answer
The model gives dollars, which is impossible for a price. An age of 20 years lies outside the data range, so the prediction is extrapolation. The relationship may change outside the observed range, so the line gives no evidence there.
- 1 mark for calculating a predicted value of dollars
- 1 mark for noting that a negative price is impossible
- 1 mark for stating that 20 years is outside the range of the data (extrapolation)
- 1 mark for explaining that the pattern may not continue beyond the observed range
- 6.Explain[2 marks]
Explain why the actual price of a particular 4-year-old car usually differs from the predicted price.
Show mark schemeHide mark scheme
Model answer
The model gives a predicted value that follows the overall trend. Individual cars scatter above and below the line, so an observed price rarely equals the prediction.
- 1 mark for stating that the model gives a predicted value for the trend
- 1 mark for explaining that individual observations scatter around the line

Practice this topic
10 quick-fire multiple-choice questions, easy to hard. Tap an answer to see if you're right and the worked explanation. (The written questions with mark schemes are under “Check your recall” above.)
0 right0 of 10 answered
Includes the 2 quick checks in the note above.
Easy[1 mark]
Show answer and explanation
Answer: A · Worked explanation
MediumCalculate[2 marks]
Show answer and explanation
Answer: A · Worked explanation
MediumIdentify[2 marks]
Show answer and explanation
Answer: A · Worked explanation
MediumIdentify[2 marks]
Show answer and explanation
Answer: A · Worked explanation
MediumCalculate[2 marks]
Show answer and explanation
Answer: A · Worked explanation
HardJustify[2 marks]
Show answer and explanation
Answer: A · Worked explanation
HardDetermine[2 marks]
Show answer and explanation
Answer: A · Worked explanation
HardExplain[2 marks]
Show answer and explanation
Answer: A · Worked explanation