AP Statistics Unit 9: Inference for Quantitative Data: Slopes

The Big Picture

Welcome to the final new inference procedure of the course! So far, you've learned to make inferences about proportions and means. Now, we'll apply those same core ideas—confidence intervals and significance tests—to the relationship between two quantitative variables.

Think back to when you first studied scatterplots. You learned to describe the form, direction, and strength of a relationship and calculate the line of best fit (LSRL) for your sample data. But a crucial question remained: If we see a linear trend in our sample, how do we know it's not just a coincidence? Is it possible that in the greater population, there's actually no relationship at all, and we just happened to get a sample that looked linear by random chance?

This unit is all about answering that question. We will use the slope of our sample regression line, b, as a point estimate for the true slope of the population regression line, β (beta). By performing inference on the slope, we can determine if there is statistically significant evidence of a linear relationship between two variables for an entire population.

Key Questions

• How can we determine if the linear association we observe in a sample scatterplot is strong enough to indicate a real linear association in the entire population?

• What is a plausible range of values for the true slope of the linear relationship between two quantitative variables in a population?

• How do we perform a formal hypothesis test to provide convincing evidence that a change in an explanatory variable is associated with a linear change in a response variable?

Your Learning Path

1. The Foundation of Inference for Slopes

Topic 9.1: Introducing Statistics: Do Those Points Align?

This is where we build the conceptual bridge from descriptive statistics to inference. You'll learn that just like sample means and proportions, the slope of a sample regression line (b) is a statistic that varies from sample to sample. We'll explore the idea of a sampling distribution for slopes, which describes the pattern of sample slopes we'd expect to see if we could take many random samples from a population. This concept is the bedrock for everything that follows.

2. Estimating the True Slope with Confidence

Topics 9.2 - 9.3: Constructing and Interpreting Confidence Intervals for a Slope

Here, you'll move from a single point estimate (b) to a more informative interval estimate. You will learn the formula and conditions for a t-interval for the slope. The key is not just calculating the interval, but also interpreting it in context and using it to decide if a linear relationship is plausible. For instance, if the interval of plausible values for the true slope contains zero, we don't have convincing evidence of a linear relationship.

3. Testing for a Linear Relationship

Topics 9.4 - 9.5: Setting Up and Carrying Out a Significance Test for a Slope

This is the formal hypothesis testing procedure. You'll learn to write the null and alternative hypotheses for the slope (H₀: β = 0 vs. Hₐ: β ≠ 0, > 0, or < 0). You'll master checking the specific conditions for this test, calculating the t-test statistic, finding the p-value, and drawing a conclusion in the context of the problem. This is the primary tool statisticians use to formally declare if evidence supports a linear association.

4. Putting It All Together

Topic 9.6: Skills Focus: Selecting an Appropriate Inference Procedure

You've now learned a wide array of inference procedures for proportions, means, and slopes. This topic is a crucial capstone where you'll practice analyzing a scenario and choosing the correct procedure from all the options you've learned this year. This is a critical skill for the AP Exam, requiring you to distinguish between problems about means, proportions, differences, or relationships.

How to Succeed in This Unit

• Master the Conditions (L.I.N.E.R.): The conditions for inference on a slope are specific. Use the acronym L.I.N.E.R. (Linear, Independent, Normal, Equal Variance, Random) to remember them. Pay special attention to checking the "Linear," "Normal," and "Equal Variance" conditions using a residual plot. Be prepared to sketch a residual plot and explain what you are looking for (no leftover pattern, roughly equal scatter, no outliers).

• Know Your Notation and Hypotheses: Be precise. The sample slope is b, but the population slope is β (beta). Your hypotheses must be about the population parameter β. For nearly every test you'll encounter, the null hypothesis will be H₀: β = 0, which corresponds to the claim of "no linear relationship" between the variables.

• Become an Expert at Reading Computer Output: You will rarely be asked to calculate the t-statistic for slope from scratch. Instead, you'll be given computer regression output. You must be able to quickly locate the slope, the standard error of the slope (SE Coef), the t-statistic, and the p-value from this table. Practice identifying these key values in different formats.

• Connect the Interval and the Test: Understand the link between a confidence interval and a two-sided hypothesis test. If a 95% confidence interval for the slope does not contain 0, then you would reject the null hypothesis H₀: β = 0 at the α = 0.05 significance level. This dual understanding demonstrates a deep mastery of inference.