01 / 22

Are Variables
Related?

Explore how statisticians detect and measure relationships between two variables using scatterplots, correlation, and regression.

AP Statistics

What Does "Related" Mean?

In statistics, two variables are related (or associated) when knowing the value of one variable gives you information about the likely value of the other. If study hours go up and test scores tend to go up too, those variables are related.

Explanatory vs. Response Variables

Explanatory Variable
The variable we suspect explains or predicts changes in the other. Plotted on the \(x\)-axis.
Response Variable
The variable we think is affected or predicted. Plotted on the \(y\)-axis.

The Scatterplot

A scatterplot displays the relationship between two quantitative variables. Each point represents one observation with its \(x\) and \(y\) values.

Hours Studied Exam Score 0 2 4 6 8 50 60 70 80 90 100

Describing a Scatterplot

When you describe a scatterplot, always mention four features:

Direction
Positive, negative, or no direction
Form
Linear, curved, or clustered
Strength
Strong, moderate, or weak
Outliers
Any unusual points that deviate from the pattern

The Correlation Coefficient \(r\)

The correlation coefficient \(r\) measures the strength and direction of a linear relationship between two quantitative variables.

1
\(r\) is always between \(-1\) and \(+1\).
2
Positive \(r\): as \(x\) increases, \(y\) tends to increase.
3
Negative \(r\): as \(x\) increases, \(y\) tends to decrease.
4
\(r = 0\) means no linear relationship (but a curved one may exist).

Visualizing Different \(r\) Values

\(r = +0.95\) \(r = +0.50\) \(r \approx 0\) \(r = -0.85\) \(r = -1.0\) \(r = +1.0\)

Correlation vs. Causation

Correlation
Two variables move together: a statistical association exists. Measured by \(r\).
Causation
Changes in one variable directly produce changes in the other. Requires a controlled experiment.

Important Properties of \(r\)

1
\(r\) has no units. It is a pure number, unaffected by measurement units.
2
Switching \(x\) and \(y\) does not change \(r\). Correlation is symmetric.
3
\(r\) only measures linear association. A strong curve can have \(r \approx 0\).
4
Outliers can dramatically change \(r\). Always check the scatterplot first.
5
Adding or multiplying a constant to all values of \(x\) or \(y\) does not change \(r\).

The Least-Squares Regression Line

The least-squares regression line (LSRL) is the line that minimizes the sum of squared vertical distances (residuals) from each point to the line.

Coefficient of Determination \(r^2\)

\(r^2\) tells you the proportion of the variation in \(y\) that is explained by the linear relationship with \(x\).

\(r^2 = 0.81\)
81% of the variation in \(y\) is explained by \(x\). Strong model.
\(r^2 = 0.25\)
25% of the variation in \(y\) is explained by \(x\). Weak model.

How to Analyze a Relationship

1
Identify variables. Which is explanatory (\(x\))? Which is response (\(y\))?
2
Make a scatterplot. Plot the data and look at D-F-S-O.
3
Compute \(r\). Quantify the linear strength and direction.
4
Fit the LSRL \(\hat{y} = a + bx\). Interpret slope and intercept in context.
5
Check residuals. Look for patterns: if curved, the linear model may be inappropriate.
6
Report \(r^2\). State how much variation is explained by the model.

Example 1: Study Hours and Scores

A teacher records study hours and test scores for 8 students. The data yields \(\bar{x} = 4\), \(\bar{y} = 75\), \(s_x = 2\), \(s_y = 10\), and \(r = 0.90\).

Example 2: Predicting with the LSRL

Using \(\hat{y} = 57 + 4.5x\), predict the score for a student who studies 5 hours.

Hours (x) Score (y) 0 1 2 3 4 5 6 7 8 50 60 70 80 90 100 79.5

Example 3: Residuals

A residual is the difference between an observed value and the predicted value from the LSRL.

If a student studies 5 hours and actually scores 85, but the model predicts 79.5:

Common Mistakes

✕ Confusing \(r\) and \(r^2\)
\(r = 0.70\) does not mean 70% explained. It is \(r^2 = 0.49\), so only 49% of variation is explained.
✕ Claiming Causation
A strong \(r\) from an observational study does not prove one variable causes the other. Only experiments can show causation.
✕ Extrapolating
Using the LSRL to predict far beyond the range of the data is unreliable. The relationship may not hold outside the observed range.

Summary Table

Concept Key Detail
Scatterplot Displays relationship between two quantitative variables Describe: D-F-S-O
\(r\) Measures strength and direction of linear association \(-1 \leq r \leq 1\)
LSRL \(\hat{y} = a + bx\) Passes through \((\bar{x}, \bar{y})\)
\(r^2\) Proportion of variation in \(y\) explained by \(x\) Always between 0 and 1
Residual \(y - \hat{y}\) Check for random scatter
Causation Requires controlled experiment Correlation ≠ causation

Question 1 of 5

A scatterplot of hours of exercise per week vs. resting heart rate shows a downward trend with points closely clustered around a line. Which best describes this relationship?

Question 2 of 5

If the correlation between height and weight for a sample of adults is \(r = 0.72\), what is the value of \(r^2\) and what does it mean?

Question 3 of 5

The LSRL for predicting final exam score from midterm score is \(\hat{y} = 10 + 0.85x\). A student scored 80 on the midterm and 82 on the final. What is the residual?

Question 4 of 5

A researcher finds that cities with more fire stations have more crimes. She concludes that fire stations cause crime. What is the flaw?

Question 5 of 5

Which of the following is true about the correlation coefficient \(r\)?