Skip to content
SAT · Problem Solving and Data Analysis

Correlation

Correlation appears in 5 reviewed SAT questions and sits in Problem Solving and Data Analysis. Below are worked examples showing exactly how the exam tests it, with the answer marked and every distractor explained.

How Correlation is tested

Example 1Medium
A scatter plot shows a strong negative correlation between hours of TV watched per day and GPA. Which of the following best describes this relationship?
  1. AGPA has no relationship with TV watching.
  2. BStudents who watch more TV tend to have higher GPAs.
  3. CWatching TV causes a student's GPA to decrease.
  4. DStudents who watch more TV tend to have lower GPAs.Correct

Why D is right

A negative correlation means that as one variable increases, the other tends to decrease. More TV → lower GPA (on average). The data or units point to this result when the values are interpreted correctly.
Why the other 3 choices miss
A
Incorrect: A strong correlation (even negative) indicates a relationship, not the absence of one.
B
Incorrect: This describes a positive correlation, not a negative one.
C
Incorrect: Correlation does not imply causation; the scatter plot shows association, not cause-and-effect.
Example 2Harder
A study finds that the number of ice cream cones sold per day is positively correlated with the daily high temperature. The correlation coefficient is r = 0.92. Which of the following conclusions is justified?
  1. AHigher temperatures cause people to buy more ice cream.
  2. BTemperature explains exactly 92% of variation in ice cream sales.
  3. CIce cream sales and temperature are likely linearly related with a strong positive association.Correct
  4. DOn cold days, ice cream sales are zero.

Why C is right

r = 0.92 indicates a strong positive linear association. Correlation does not prove causation, and r^2 = 0.846 (not 0.92) explains the proportion of variance. The data or units point to this result when the values are interpreted correctly.
Why the other 3 choices miss
A
Incorrect: Correlation does not prove causation; other factors might drive both temperature and ice cream sales.
B
Incorrect: r^2 = 0.92^2 ≈ 0.846 (about 84.6%) explains the variation, not r itself (92%).
D
Incorrect: Correlation predicts a trend, not an exact minimum value; some ice cream is sold on cold days.
Example 3Harder
The number of hours a student studies per week and their GPA are tracked for 6 weeks. The data shows a moderate positive correlation (r = 0.6). Which action would most likely change the correlation coefficient toward 1.0?
  1. AAdding more students who all study the same amount
  2. BDecreasing the number of weeks tracked
  3. CRemoving an outlier student who studies a lot but has a low GPACorrect
  4. DConverting GPA from a 4.0 scale to a 100-point scale for every student

Why C is right

An outlier (high study hours, low GPA) creates a data point that goes against the positive trend, reducing r. Removing it would strengthen the positive correlation. The data or units point to this result when the values are interpreted correctly.
Why the other 3 choices miss
A
Adding students who all study the same amount introduces a vertical cluster with no variation in x, which disrupts the correlation.
B
Fewer data points increase uncertainty in r; the correlation strength depends on the underlying relationship, not sample size alone.
D
Converting every GPA to a 100-point scale is a linear transformation and does not affect the correlation coefficient r.

Studied alongside

Other concepts from Problem Solving and Data Analysis.