01 / 21

Justifying Claims:
Slope of Regression Models

Learn how to use a confidence interval for the slope \(\beta_1\) to justify whether a linear relationship truly exists between two variables.

AP Statistics

What Is the Regression Slope?

In a linear regression model \(\hat{y} = b_0 + b_1 x\), the slope \(b_1\) tells us how much the predicted response \(\hat{y}\) changes for each one-unit increase in the explanatory variable \(x\).

\(b_1\) (Sample Slope)
The slope calculated from your sample data: a point estimate of the true relationship.
\(\beta_1\) (Population Slope)
The true slope of the population regression line: the parameter we want to make claims about.

Why Does \(\beta_1 = 0\) Matter?

Linear Relationship
\(\beta_1 \neq 0\)
No Relationship
\(\beta_1 = 0\)

If \(\beta_1 = 0\), the explanatory variable \(x\) has no linear effect on the response. The regression line is flat. If \(\beta_1 \neq 0\), knowing \(x\) helps us predict \(y\).

The Confidence Interval for \(\beta_1\)

A confidence interval gives a range of plausible values for the true population slope \(\beta_1\). The formula is:

\(b_1\)
Center of the interval: your best single estimate of \(\beta_1\).
\(t^*\)
Depends on your confidence level and degrees of freedom \(n - 2\).
\(SE_{b_1}\)
Measures how much \(b_1\) varies from sample to sample.

How the CI Justifies a Claim

0 Does not contain 0 Evidence! Contains 0 No evidence
0 Not in CI
We have convincing evidence that \(\beta_1 \neq 0\): a linear relationship likely exists.
0 Is in CI
We lack convincing evidence that \(\beta_1 \neq 0\): the slope could plausibly be zero.

The Decision Rule

1
Construct a confidence interval for \(\beta_1\) at your chosen confidence level (typically 95%).
2
Check whether the value 0 falls inside the interval.
3
Conclude: If 0 is not in the interval, reject \(H_0\). If 0 is in the interval, fail to reject \(H_0\).

Confidence Interval vs. Hypothesis Test

Confidence Interval Hypothesis Test
Provides Range of plausible values for \(\beta_1\) A yes/no decision about \(\beta_1 = 0\)
Key output \((b_1 - t^* \cdot SE,\; b_1 + t^* \cdot SE)\) \(p\text{-value}\)
Reject \(H_0\) when 0 is outside the interval \(p\text{-value} < \alpha\)
Extra information Direction and magnitude of effect Strength of evidence only

Conditions for Valid Inference

Before constructing a CI for \(\beta_1\), you must verify the LINE conditions:

Linearity
The scatterplot (or residual plot) shows a linear pattern with no obvious curves.
Independence
Observations are independent: random sample or \(n < 10\%\) of the population.
Normality
Residuals are approximately Normal, or the sample size is large enough (\(n \geq 30\)).
Equal Variance
The scatter around the line is roughly the same for all values of \(x\) (homoscedasticity).

How to Justify a Claim: Step by Step

1
State the hypotheses. \(H_0\!: \beta_1 = 0\) vs. \(H_a\!: \beta_1 \neq 0\) (or one-sided).
2
Check LINE conditions. Verify linearity, independence, normality, and equal variance using residual plots.
3
Compute the CI. Use \(b_1 \pm t^* \cdot SE_{b_1}\) with \(df = n - 2\) from computer output or a table.
4
Interpret the interval. "We are C% confident the true slope \(\beta_1\) is between (lower, upper)."
5
Make the decision. Does the interval contain 0? State your conclusion in context.

Example 1: Study Hours and Exam Scores

A teacher collects data from 25 students on hours studied (\(x\)) and exam scores (\(y\)). Computer output gives:

\(b_1 = 4.2\)
Each additional hour of study is associated with a 4.2-point increase in score.
\(SE_{b_1} = 1.1\)
The standard error of the slope estimate.
\(df = 23\)
Degrees of freedom: \(n - 2 = 25 - 2\).

Example 1: Conclusion

-2 0 2 4 6 (1.924, 6.476)

Example 2: Shoe Size and GPA

A researcher investigates whether shoe size (\(x\)) predicts college GPA (\(y\)) using data from 40 students. Computer output gives:

\(b_1 = 0.03\)
A very small estimated slope.
\(SE_{b_1} = 0.05\)
Standard error is large relative to the slope.
\(df = 38\)
Degrees of freedom: \(n - 2 = 40 - 2\).

Example 2: Conclusion

-0.2 -0.1 0 0.1 0.2 (-0.071, 0.131)

Example 3: Reading Computer Output

A regression analysis of temperature (\(x\), in \(^{\circ}\text{F}\)) and ice cream sales (\(y\), in dollars) for \(n = 30\) days produces:

Predictor Coef SE Coef T P
Constant -21.4 15.8 -1.35 0.187
Temperature 3.84 0.62 6.19 < 0.001

Common Mistakes to Avoid

Confusing \(b_1\) with \(\beta_1\)
The sample slope \(b_1\) is an estimate. The CI is about the population parameter \(\beta_1\), not the sample statistic.
Claiming Causation
A significant slope shows association, not causation, unless the data come from a randomized experiment.
Ignoring Conditions
If LINE conditions are violated, the CI may not be valid. Always check residual plots before interpreting results.

Summary: Decision Framework

Scenario 0 Not in CI 0 Is in CI
Decision Reject \(H_0\) Fail to reject \(H_0\)
Conclusion Convincing evidence of a linear relationship No convincing evidence of a linear relationship
About the slope \(\beta_1\) is likely nonzero; CI gives plausible range \(\beta_1\) could be zero; slope may be negligible
Corresponds to \(p\text{-value} < \alpha\) \(p\text{-value} \ge \alpha\)

Question 1 of 5

A 95% confidence interval for the slope of a regression model is (0.45, 1.23). Which conclusion is justified?

Question 2 of 5

Computer output shows \(b_1 = 2.5\), \(SE_{b_1} = 1.8\), and \(n = 22\). Using \(t^* = 2.086\) for a 95% CI, what is the interval?

Question 3 of 5

A 99% confidence interval for \(\beta_1\) is (-0.12, 0.88). Which statement is correct?

Question 4 of 5

Which of the following is not a required condition (LINE) for constructing a valid confidence interval for \(\beta_1\)?

Question 5 of 5

A 95% CI for the slope is (2.1, 5.3). A student writes: "There is a 95% chance that the true slope is between 2.1 and 5.3." What is wrong with this interpretation?