Scribe logoScribeNotes & practice
Sign inStart free
  1. Notes
  2. ›AP Statistics
  3. ›5 Regression Analysis
  4. ›5.4 Residuals

AP Statistics · Unit 5 · Regression Analysis

5.4 Residuals

Calculate and interpret a residual, and use a residual plot to judge whether a linear model fits the data.

  • 4 min read
  • 5 parts
  • 10 questions
  • Draft
Start a Quick 5 on this topic→Jump to the 10 questions
1/5A residual is the gap between observed and predicted values
  1. 1A residual is the gap between observed and predicted values
  2. 2A positive residual means the point lies above the line
  3. 3A residual plot reveals whether a line fits the data
  4. 4A curved pattern in residuals signals a poor linear model
  5. 5Four steps turn residuals into a conclusion about fit
  6. Check your recall
  7. Practice: 10 questions
1

Part 1 of 5: A residual is the gap between observed and predicted values

A regression line predicts, but real points rarely sit exactly on it. Take a student who studied 5 hours and scored 81 points. The model y^=55+4.5x\hat{y}=55+4.5xy^​=55+4.5x predicts 77.5 points.

The difference between the observed and predicted scores is a residual. In words, residual = observed value − predicted value. In symbols, it is y−y^y-\hat{y}y−y^​. Here the residual is 81−77.5=+3.581-77.5=+3.581−77.5=+3.5 points. The student did better than the model expected.

Common mistake

Order matters. Subtract the predicted value from the observed value. Reversing the order flips the sign and reverses the meaning.

Key term

residual

A residual is the observed response value minus the predicted value, y−y^y-\hat{y}y−y^​, measured in the units of the response variable.

Part 1 done

2

Part 2 of 5: A positive residual means the point lies above the line

Now we can read what a residual says about a point. On a scatterplot, a residual is the vertical distance from the point to the line. A point above the line has a positive residual.

A point below the line has a negative residual. A point exactly on the line has a residual of zero. The size of the residual shows how far the model missed.

To interpret a residual, name the variables and give units. Then say whether the observed value was above or below the prediction. Mention the size of the miss in the units of yyy.

Worked example

Calculate and interpret a residual

A student studied 7 hours and scored 82 points. Use y^=55+4.5x\hat{y}=55+4.5xy^​=55+4.5x to calculate and interpret the residual.
  1. 1
    Find the predicted score: y^=55+4.5(7)=86.5\hat{y}=55+4.5(7)=86.5y^​=55+4.5(7)=86.5.
  2. 2
    Subtract: residual =y−y^=82−86.5=−4.5=y-\hat{y}=82-86.5=-4.5=y−y^​=82−86.5=−4.5.
  3. 3
    Read the sign: negative means the observed score is below the prediction.

Answer: The residual is -4.5 points. The student scored 4.5 points below the model’s prediction.

Part 2 done

Quick check

One question on the part you just read

Easy

Identify the expression that gives the residual for a data point.
Show answer and explanation

Answer: A · Worked explanation

A residual is the observed value minus the predicted value, y−y^y-\hat{y}y−y^​. The expression y^−y\hat{y}-yy^​−y has the right size but the wrong sign, which reverses the meaning. Adding or dividing the two values does not measure the miss at all.
3

Part 3 of 5: A residual plot reveals whether a line fits the data

One residual describes one point, but we need to judge the whole model. A residual plot shows the residuals on the vertical axis. The horizontal axis shows the explanatory variable.

When a linear model is appropriate, the residuals scatter randomly around zero. There is no clear pattern, and the spread stays roughly even from left to right.

Figure 1. Random scatter around zero supports a linear model. A U shape shows that a curve remains in the data.

In the left plot, the residuals look random, so the line captures the trend. In the right plot, a clear curve remains. The line has missed something about the relationship. A residual plot makes such misses easy to see.

Key term

residual plot

A residual plot is a scatterplot of residuals against the explanatory variable or the predicted values, used to judge whether a linear model is appropriate.

Part 3 done

4

Part 4 of 5: A curved pattern in residuals signals a poor linear model

Now compare that random scatter with plots that show structure. A U shape means the relationship is curved, not linear. The line sits below the points at the ends and above them in the middle.

A fan shape is different. There, residuals spread out as xxx grows, so predictions get less precise for larger xxx. In both cases, the plain line is not telling the whole story.

Common residual plot patterns

PatternWhat it showsLinear model appropriate?
Random scatter around zeroThe line captures the trendYes
U shape or other curveThe relationship is not linearNo
Fan shapeThe spread changes with xxxUse with caution
Pattern
Random scatter around zero
What it shows
The line captures the trend
Linear model appropriate?
Yes
Pattern
U shape or other curve
What it shows
The relationship is not linear
Linear model appropriate?
No
Pattern
Fan shape
What it shows
The spread changes with xxx
Linear model appropriate?
Use with caution

Exam tip

Name the pattern and link it to the model. For example: the residual plot shows a curve, so a linear model is not appropriate.

Part 4 done

Quick check

One question on the part you just read

Easy

A data point lies above the regression line. Identify the sign of its residual.
Show answer and explanation

Answer: A · Worked explanation

Above the line, the observed value exceeds the predicted value, so y−y^y-\hat{y}y−y^​ is positive. A negative residual means the point is below the line, and zero means it lies on the line. The sign can always be found from the position of the point.
5

Part 5 of 5: Four steps turn residuals into a conclusion about fit

Putting these ideas together gives a routine you can reuse.

  • Calculate each residual as y−y^y-\hat{y}y−y^​.
  • Plot the residuals against xxx, with a line at zero.
  • Look for a curve, a fan shape, or any other pattern.
  • Conclude whether a linear model is appropriate, and say why.

Size matters too. Judge a residual against the units of yyy. A residual of 3.5 points is small on a 100-point quiz but large on a 10-point quiz.

A single point with a much larger residual than the rest deserves a second look. It may be an unusual observation rather than a sign that the whole model is wrong.

Note read

Check your recall[12 marks]

6 written questions. Answer each one in your head or on paper, then open the mark scheme.

  1. 1.Calculate[2 marks]

    A model for the price of a used laptop, in dollars, from its age in years is y^=800−150x\hat{y}=800-150xy^​=800−150x. A 2-year-old laptop sold for 540 dollars. Calculate the residual.

    Show mark schemeHide mark scheme

    Model answer

    y^=800−150(2)=500\hat{y}=800-150(2)=500y^​=800−150(2)=500 dollars. Residual =y−y^=540−500=+40=y-\hat{y}=540-500=+40=y−y^​=540−500=+40 dollars.

    • 1 mark for finding the predicted price of 500 dollars
    • 1 mark for subtracting predicted from observed to get +40 dollars
  2. 2.Describe[2 marks]

    Describe the meaning of the residual for the 2-year-old laptop, in context.

    Show mark schemeHide mark scheme

    Model answer

    The laptop sold for 40 dollars more than the model predicted for a 2-year-old laptop.

    • 1 mark for stating that the observed price was above the predicted price
    • 1 mark for giving the size, 40 dollars, in context
  3. 3.Identify[1 mark]

    Identify the quantity plotted on the vertical axis of a residual plot.

    Show mark schemeHide mark scheme

    Model answer

    The residuals.

    • 1 mark for identifying the residuals
  4. 4.Describe[2 marks]

    Describe what a random scatter of points around zero in a residual plot tells you about a linear model.

    Show mark schemeHide mark scheme

    Model answer

    There is no clear pattern in the residuals, so the line captures the trend. A linear model is appropriate.

    • 1 mark for stating that there is no clear pattern in the residuals
    • 1 mark for concluding that a linear model is appropriate
  5. 5.Justify[3 marks]

    A residual plot for a linear model shows a clear U shape. Justify whether the linear model is appropriate.

    Show mark schemeHide mark scheme

    Model answer

    The linear model is not appropriate. The U shape is a systematic curve in the residuals. It shows that the relationship is curved, so the line misses the trend.

    • 1 mark for identifying the pattern as a curve or U shape
    • 1 mark for concluding that a linear model is not appropriate
    • 1 mark for linking the pattern to a relationship that is not linear
  6. 6.Explain[2 marks]

    Explain why a residual of -3.5 points can be small in one context but large in another.

    Show mark schemeHide mark scheme

    Model answer

    The size of a residual must be judged against the units and scale of the response. On a 100-point quiz, 3.5 points is small, but on a 10-point quiz it is large.

    • 1 mark for stating that residual size is judged against the scale of yyy
    • 1 mark for giving a contrast such as a 100-point quiz and a 10-point quiz

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.

3.

Easy·Identify[1 mark]

Identify the pattern in a residual plot that supports using a linear model.
Show answer and explanation

Answer: A · Worked explanation

Random scatter around zero with no pattern means the line has captured the trend. A U shape shows a curved relationship, and a fan shape shows uneven spread. A steady rise in the residuals would also be a pattern, so structure would remain unexplained.
4.

Medium·Calculate[2 marks]

A model is y^=20+2.5x\hat{y}=20+2.5xy^​=20+2.5x. A point has x=6x=6x=6 and y=31y=31y=31. Calculate its residual.
Show answer and explanation

Answer: A · Worked explanation

The predicted value is 20+2.5(6)=3520+2.5(6)=3520+2.5(6)=35. The residual is y−y^=31−35=−4y-\hat{y}=31-35=-4y−y^​=31−35=−4. The value +4+4+4 comes from subtracting in the wrong order. The value 35 is the predicted value, not the residual. The negative sign tells us the point lies below the line, which agrees with 31<3531<3531<35.
5.

Medium·Identify[2 marks]

A model predicts the selling price of a house, in thousands of dollars, from its size. One house has a residual of -12. Identify the correct interpretation.
Show answer and explanation

Answer: A · Worked explanation

A residual is observed minus predicted, so -12 means the observed price was 12 thousand dollars below the prediction. A positive residual would mean the house sold for more than predicted. The residual is in the units of the response, thousands of dollars, not a percentage.
6.

Medium·Determine[2 marks]

A residual plot for a linear model shows residuals of +5,+2,−1,−3,−1,+2,+5+5, +2, -1, -3, -1, +2, +5+5,+2,−1,−3,−1,+2,+5 for x=1,2,…,7x=1,2,\ldots,7x=1,2,…,7. Determine the best conclusion.
Show answer and explanation

Answer: A · Worked explanation

The residuals fall from positive to negative and then rise again, which is a U shape. This systematic curve shows the relationship is not linear. Having both signs is not enough, because random scatter also needs no pattern. The equal end values are part of the curve, not a separate reason.
7.

Medium·Calculate[2 marks]

A data point has an observed value of 18.6 and a residual of +2.4+2.4+2.4. Calculate its predicted value.
Show answer and explanation

Answer: A · Worked explanation

Since the residual is y−y^y-\hat{y}y−y^​, we get y^=18.6−2.4=16.2\hat{y}=18.6-2.4=16.2y^​=18.6−2.4=16.2. The value 21.0 adds the residual instead of subtracting it. The value 7.8 divides 18.6 by 2.4, and 44.6 multiplies them. Neither follows from the definition of a residual. Check by substituting: 18.6−16.2=2.418.6-16.2=2.418.6−16.2=2.4.
8.

Hard·Determine[2 marks]

A residual plot for a linear model shows residuals close to zero for small xxx values and spread far above and below zero for large xxx values. Determine the best conclusion.
Show answer and explanation

Answer: A · Worked explanation

A widening spread, called a fan shape, means the residuals are larger for large xxx, so predictions there are less precise. A curved relationship would show a U or an arch, not a widening spread. Equal precision would show an even band around zero. The sign of the slope cannot be read from this pattern.
9.

Hard·Identify[2 marks]

A model is y^=3+2x\hat{y}=3+2xy^​=3+2x. Point P is (4,9)(4, 9)(4,9) and point Q is (6,17)(6, 17)(6,17). Identify the correct statement about their residuals.
Show answer and explanation

Answer: A · Worked explanation

For P, y^=3+2(4)=11\hat{y}=3+2(4)=11y^​=3+2(4)=11, so the residual is 9−11=−29-11=-29−11=−2. For Q, y^=3+2(6)=15\hat{y}=3+2(6)=15y^​=3+2(6)=15, so the residual is 17−15=+217-15=+217−15=+2. Both miss by the same amount in opposite directions. Comparing the raw yyy values ignores that each point has a different prediction.
10.

Hard·Explain[2 marks]

A student says, “All the residuals are small, so the linear model must be appropriate.” The residual plot, however, shows a clear curve. Identify the best response.
Show answer and explanation

Answer: A · Worked explanation

Judging fit means looking for a pattern, not only for size. A curve in the residuals shows the line systematically misses the trend, even when each miss is small. The residual plot of a good model shows no pattern at all, not a rising line, and a curve does not require large residuals.

Keep going

Next topic5.5 Least-squares regressionPrevious topic5.3 Linear regression models

More in Unit 5 · Regression Analysis

  1. 5.1ScatterplotsNotes · 10 questions
  2. 5.2CorrelationNotes · 10 questions
  3. 5.3Linear regression modelsNotes · 10 questions
  4. 5.4ResidualsNotes · 10 questions
  5. 5.5Least-squares regressionNotes · 10 questions

All AP Statistics topics →

Draft: written by Scribe and checked automatically, not yet by a subject reviewer.

Practice questions are original, written by Scribe and checked automatically for correctness.

Topic codes follow the official numbering. Not endorsed by College Board or any exam board.

On this page

0/5 read

  1. 1A residual is the gap between observed and predicted values
  2. 2A positive residual means the point lies above the line
  3. 3A residual plot reveals whether a line fits the data
  4. 4A curved pattern in residuals signals a poor linear model
  5. 5Four steps turn residuals into a conclusion about fit
  6. Check your recall
  7. Practice: 10 questions
All coursesScribeHow it worksPricingPrivacyTerms

Notes and questions are written by Scribe in our own words. Topic codes follow the official course numbering so you can find your place. Not produced or endorsed by the IB, College Board, AQA, OCR, Pearson Edexcel or any exam board.

© 2026 Scribe