Scribe logoScribeNotes & practice
Sign inStart free
  1. Notes
  2. ›AP Statistics
  3. ›5 Regression Analysis
  4. ›5.2 Correlation

AP Statistics · Unit 5 · Regression Analysis

5.2 Correlation

Interpret the correlation coefficient in context, recognize its properties, and explain what correlation cannot tell you.

  • 4 min read
  • 5 parts
  • 10 questions
  • Draft
Start a Quick 5 on this topic→Jump to the 10 questions
1/5The correlation coefficient measures linear association
  1. 1The correlation coefficient measures linear association
  2. 2The sign and size of r show direction and strength
  3. 3Correlation has properties that follow from how it is built
  4. 4A value of r near zero can hide a curved pattern
  5. 5Correlation alone cannot establish that one variable causes another
  6. Check your recall
  7. Practice: 10 questions
1

Part 1 of 5: The correlation coefficient measures linear association

A scatterplot shows a pattern, but one number can summarize part of it. The correlation coefficient, rrr, measures the direction and strength of a linear association between two quantitative variables. Both variables must be measured on the same individuals.

It is calculated from standardized values: r=1n−1∑(xi−xˉsx)(yi−yˉsy)r=\frac{1}{n-1}\sum\left(\frac{x_i-\bar{x}}{s_x}\right)\left(\frac{y_i-\bar{y}}{s_y}\right)r=n−11​∑(sx​xi​−xˉ​)(sy​yi​−yˉ​​). Here sxs_xsx​ and sys_ysy​ are the sample standard deviations, and nnn is the number of pairs.

In practice, technology computes rrr. Its value always satisfies −1≤r≤1-1\le r\le 1−1≤r≤1. Check the scatterplot first, because rrr only makes sense for a roughly linear pattern.

Key term

correlation coefficient

The correlation coefficient is the statistic rrr, between -1 and 1, that measures the direction and strength of a linear association.

Part 1 done

2

Part 2 of 5: The sign and size of r show direction and strength

Reading rrr takes two steps: look at its sign, then its distance from zero. A positive rrr means yyy tends to increase as xxx increases. A negative rrr means yyy tends to decrease. A value near 0 shows little linear pattern, while a value near 1 or -1 shows points hugging a line.

Reading the correlation coefficient

Value of rInterpretation
r=1r=1r=1 or r=−1r=-1r=−1Points fall exactly on a line
Near 1 or -1Strong linear association
About 0.5 or -0.5Moderate linear association
Near 0Weak or no linear association
Value of r
r=1r=1r=1 or r=−1r=-1r=−1
Interpretation
Points fall exactly on a line
Value of r
Near 1 or -1
Interpretation
Strong linear association
Value of r
About 0.5 or -0.5
Interpretation
Moderate linear association
Value of r
Near 0
Interpretation
Weak or no linear association

Exam tip

There are no strict cutoffs for strong, moderate, and weak. Describe rrr in words and always name both variables in context.

The size of rrr cannot reveal the form of the pattern. Two plots can share the same rrr but look very different. Always examine the scatterplot as well as the value of rrr.

Worked example

Interpreting r in context

A study of households finds r=−0.82r=-0.82r=−0.82 between daily high temperature and gas used for heating. Interpret rrr.
  1. 1
    Sign: it is negative, so gas use tends to decrease as temperature increases.
  2. 2
    Size: 0.82 is close to 1, so the association is strong.
  3. 3
    Form: rrr describes linear association only.

Answer: There is a strong, negative, linear association between daily high temperature and gas used for heating.

Part 2 done

Quick check

One question on the part you just read

Easy

Identify the set of possible values of the correlation coefficient rrr.
Show answer and explanation

Answer: C · Worked explanation

The correlation coefficient always satisfies −1≤r≤1-1\le r\le 1−1≤r≤1, with the sign giving the direction. The range of 0 to 1 would allow only positive associations. Values below -1 or above 1 never occur, and 0 to 100 is a percentage scale that rrr does not use.
3

Part 3 of 5: Correlation has properties that follow from how it is built

Because rrr uses standardized values, it behaves in predictable ways. These properties are worth memorizing.

  • rrr has no units.
  • rrr is unchanged when xxx and yyy are switched.
  • rrr is unchanged when either variable is rescaled by adding a constant or multiplying by a positive constant.
  • rrr changes sign, but not size, when one variable is multiplied by a negative constant.
  • rrr is not a resistant statistic: one outlier can change it a lot.

A resistant statistic changes little when extreme values appear. Because rrr is not resistant, check the scatterplot for outliers before trusting its value.

Key term

resistant statistic

A resistant statistic is one whose value changes little when extreme values or outliers are added or removed.

Part 3 done

4

Part 4 of 5: A value of r near zero can hide a curved pattern

The properties above assume the pattern is a line. Correlation measures only linear association. A strong curved pattern can still give rrr near 0, so look at the scatterplot before using rrr.

Figure 1. These points follow a clear U shape, yet the correlation coefficient is exactly 0.

In the plot, the points follow a clear U shape, yet r=0r=0r=0. The left side and the right side cancel each other. A high rrr also does not prove a pattern is a line. A steadily rising curve can give rrr near 1.

Common mistake

r=0r=0r=0 does not mean there is no association. It means there is no linear association.

Part 4 done

Quick check

One question on the part you just read

Easy

Identify the best interpretation of r=−0.9r=-0.9r=−0.9 between two quantitative variables.
Show answer and explanation

Answer: A · Worked explanation

The sign is negative, so yyy tends to decrease as xxx increases. The size 0.9 is close to 1, so the linear association is strong. Weak would need a value near 0, and positive would need a positive sign. The value does not say there is no association of any kind.
5

Part 5 of 5: Correlation alone cannot establish that one variable causes another

A strong correlation tempts people to claim cause and effect. Suppose ice cream sales and sunburn cases are strongly correlated across summer days. Ice cream does not cause sunburn. Hot, sunny weather increases both.

Here the weather is a lurking variable. It can make two variables appear linked without either one affecting the other.

Only a well-designed experiment with random assignment can support a cause-and-effect conclusion. Data from an observational study show association, not causation.

Exam tip

When asked about causation, name a plausible lurking variable and point out that the data are observational.

Key term

lurking variable

A lurking variable is a variable not included among those studied that is related to both and can create or hide an apparent association.

Note read

Check your recall[13 marks]

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

  1. 1.Interpret[3 marks]

    Interpret r=−0.78r=-0.78r=−0.78 between weekly training hours and 5 km run time, in context.

    Show mark schemeHide mark scheme

    Model answer

    There is a moderately strong to strong, negative, linear association between weekly training hours and run time. Run time tends to decrease as training hours increase.

    • 1 mark for identifying the direction as negative
    • 1 mark for identifying the strength as moderately strong to strong and the form as linear
    • 1 mark for referring to both variables in context
  2. 2.Identify[2 marks]

    Identify two properties of the correlation coefficient rrr.

    Show mark schemeHide mark scheme

    Model answer

    Any two of: rrr has no units; rrr is unchanged when xxx and yyy are switched; rrr is unchanged by adding a constant or multiplying by a positive constant; rrr is not resistant to outliers.

    • 1 mark for identifying one valid property of rrr
    • 1 mark for identifying a second valid property of rrr
  3. 3.Explain[3 marks]

    Explain why r=0r=0r=0 does not mean there is no association between two variables.

    Show mark schemeHide mark scheme

    Model answer

    The correlation coefficient measures only linear association. Points can follow a strong curved pattern, such as a U shape, and still give r=0r=0r=0. So r=0r=0r=0 means no linear association, not no association.

    • 1 mark for stating that rrr measures only linear association
    • 1 mark for giving a curved pattern as a case with a strong association but r near 0
    • 1 mark for concluding that r=0r=0r=0 means no linear association
  4. 4.Justify[3 marks]

    A study finds a strong positive correlation between the number of firefighters at a fire and the damage the fire causes. Justify why this does not show that firefighters cause damage.

    Show mark schemeHide mark scheme

    Model answer

    Correlation shows association, not causation, and the data are observational. A lurking variable, the size of the fire, can lead to both more firefighters and more damage.

    • 1 mark for stating that correlation does not show causation
    • 1 mark for identifying a plausible lurking variable such as fire size
    • 1 mark for explaining that the lurking variable can affect both measured variables
  5. 5.Predict[2 marks]

    Predict the new value of rrr when r=0.80r=0.80r=0.80 and every height is converted from inches to centimeters. Give a reason.

    Show mark schemeHide mark scheme

    Model answer

    The value of rrr stays 0.80. Rescaling a variable by multiplying by a positive constant does not change rrr.

    • 1 mark for predicting that rrr stays 0.80
    • 1 mark for giving the reason that multiplying by a positive constant does not change rrr

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 which correlation coefficient shows the strongest linear association: 0.45, -0.82, 0.10, or -0.30.
Show answer and explanation

Answer: D · Worked explanation

Strength depends on the distance of rrr from 0, not on the sign. The value -0.82 is the farthest from 0, so it is the strongest. The value 0.45 is the largest positive value, but it is closer to 0, so it shows a weaker association.
4.

Medium·Determine[1 mark]

Two variables have r=0.70r=0.70r=0.70 when yyy is measured in inches. Every yyy value is converted to centimeters by multiplying by 2.54. Determine the new correlation coefficient.
Show answer and explanation

Answer: B · Worked explanation

Multiplying a variable by a positive constant leaves rrr unchanged, so rrr stays 0.70. The value 1.78 comes from multiplying rrr by 2.54, but rrr has no units to rescale. The value 0.28 divides rrr by 2.54, and -0.70 would need a negative multiplier.
5.

Medium·Determine[1 mark]

Two variables have r=0.64r=0.64r=0.64. Every yyy value is multiplied by -1. Determine the new correlation coefficient.
Show answer and explanation

Answer: B · Worked explanation

Multiplying one variable by a negative constant reverses the direction, so the sign of rrr changes while its size stays the same. The new value is -0.64. Keeping 0.64 ignores the reversal. The values 0.36 and -0.36 come from computing 1−0.641-0.641−0.64, which has no meaning here.
6.

Medium·Predict[1 mark]

A scatterplot has a strong positive linear pattern with r=0.95r=0.95r=0.95. One new point is added far below the line at the right end. Predict the effect on rrr.
Show answer and explanation

Answer: D · Worked explanation

The correlation coefficient is not a resistant statistic, so one extreme point can change it noticeably. A point far from the line weakens the linear pattern and moves rrr toward 0. It does not stay the same, because rrr is not resistant. It also does not become exactly 0 from one point.
7.

Medium·Identify[1 mark]

A study finds r=0.78r=0.78r=0.78 between hours studied and exam score. Identify the best interpretation.
Show answer and explanation

Answer: A · Worked explanation

The positive sign shows that score tends to rise with hours studied, and 0.78 is fairly close to 1, so the linear association is strong. The value of rrr is not a percentage of students or of a score. It also cannot establish that studying causes the rise, since correlation alone does not show causation.
8.

Hard·Identify[2 marks]

Identify the situation in which the correlation coefficient rrr is not an appropriate summary.
Show answer and explanation

Answer: C · Worked explanation

The correlation coefficient measures only linear association. A U-shaped pattern is strongly related but not linear, so rrr could be near 0 and would mislead. In the other three situations the pattern is roughly linear, even when the association is weak, so rrr still describes it.
9.

Hard[2 marks]

A study finds r=0.88r=0.88r=0.88 between the number of firefighters at a fire and the cost of the damage. Which explanation is best?
Show answer and explanation

Answer: D · Worked explanation

Larger fires bring more firefighters and also cause more damage, so fire size is a lurking variable that explains the association. Concluding that firefighters cause damage mistakes correlation for causation. A large rrr is not evidence of error, and rrr is not a percentage of fires.
10.

Hard·Determine[2 marks]

Five points are (−2,4)(-2,4)(−2,4), (−1,1)(-1,1)(−1,1), (0,0)(0,0)(0,0), (1,1)(1,1)(1,1), and (2,4)(2,4)(2,4). Determine the value of rrr.
Show answer and explanation

Answer: A · Worked explanation

The points form a symmetric U shape. With xˉ=0\bar{x}=0xˉ=0 and yˉ=2\bar{y}=2yˉ​=2, the products (x−xˉ)(y−yˉ)(x-\bar{x})(y-\bar{y})(x−xˉ)(y−yˉ​) are -4, 1, 0, -1, and 4, which sum to 0. So r=0r=0r=0. The value 1.00 would need the points on a rising line, and the pattern here is curved, not linear.

Keep going

Next topic5.3 Linear regression modelsPrevious topic5.1 Scatterplots

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. 1The correlation coefficient measures linear association
  2. 2The sign and size of r show direction and strength
  3. 3Correlation has properties that follow from how it is built
  4. 4A value of r near zero can hide a curved pattern
  5. 5Correlation alone cannot establish that one variable causes another
  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