In AP Statistics, the slope of a regression model is the coefficient b in ŷ = a + bx that gives the predicted change in the response variable for each one-unit increase in the explanatory variable; in Unit 9 you use b to estimate and test the true population slope β with t-procedures.
The slope of a regression model tells you how fast the response variable changes as the explanatory variable goes up. If ŷ = 12 + 2.3x, then for every one-unit increase in x, the predicted value of y goes up by 2.3. That number, b, is the sample slope, and it's calculated from your data.
Here's the Unit 9 twist. Your sample slope b is just an estimate of the true population slope β, the slope of the line that fits the entire population (μy = α + βx). A different random sample would give you a slightly different b. So b is a statistic with a sampling distribution, just like x̄ or p̂ from earlier units. That's what makes inference possible. You can build a confidence interval for β using b ± t*(SEb), or run a t-test with hypotheses about β, using a t-distribution with n − 2 degrees of freedom. The slope stops being just a description of your scatterplot and becomes a parameter you can make claims about.
The slope of a regression model is the entire point of Unit 9 (Inference for Quantitative Data: Slopes). The CED dedicates four topics to it. You identify and calculate confidence intervals for β (AP Stats 9.2.A and 9.2.D), interpret them and use them to justify claims (AP Stats 9.3.A and 9.3.B), set up a t-test for a slope with hypotheses like H₀: β = 0 (AP Stats 9.4.A and 9.4.B), and carry out that test by calculating t = (b − β)/SEb and interpreting the p-value (AP Stats 9.5.A and 9.5.B). It also ties back to Unit 2, where you first learned to interpret slope in context. Unit 9 is the capstone of the course's inference arc, so expect it to show up on both MCQs and the inference FRQ.
Keep studying AP® Statistics Unit 9
Confidence Interval (Units 6-9)
A confidence interval for β follows the same recipe you learned for proportions and means, estimate ± (critical value)(standard error). Here it's b ± t*(SEb). If 0 is inside the interval, you don't have convincing evidence of a linear relationship, which is exactly what a Fiveable practice question with the interval (-0.12, 0.89) is testing.
Hypothesis Test (Units 6-9)
The t-test for a slope is the last hypothesis test in the course, and it follows the same logic as every other one. Assume H₀: β = 0 is true, compute a t-statistic, find a p-value, and compare it to α. The only new pieces are the standard error formula and the degrees of freedom.
Degrees of Freedom (Units 7-9)
For inference on a slope, df = n − 2, not n − 1. The intuition is that you estimated two things from the data (the slope and the intercept), so you lose two degrees of freedom. Mixing up n − 1 and n − 2 is one of the easiest MCQ traps in Unit 9.
Constant Variance and Independence Condition (Unit 9)
Before any slope inference, you check conditions per the CED. The relationship is linear, the standard deviation of y doesn't change with x (check a residual plot for both), data come from a random sample or randomized experiment, n ≤ 10% of N when sampling without replacement, and responses for each x are approximately normal.
Multiple-choice questions love handing you computer output (b, t, p-value, SEb) and asking you to read it correctly. Practice questions in this style give you something like β̂ = 2.3, t = 1.82, p-value = 0.078 at α = 0.05 and ask for the right conclusion, which is fail to reject H₀ since 0.078 > 0.05. Others test whether you can interpret a p-value of 0.03 properly, meaning the probability of getting a sample slope at least this extreme assuming the true slope equals the null value. You'll also see confidence-interval conclusion questions, where checking whether 0 falls inside the interval decides the claim. On the FRQ side, a full inference problem for a slope means stating hypotheses about β (not b), naming the t-test for a slope, checking all the conditions, computing t with n − 2 degrees of freedom, and writing a conclusion in context that explicitly compares the p-value to α. Vague conclusions or hypotheses written about the sample slope b lose points every year.
b is the slope of the line of best fit calculated from your sample. β is the true slope of the population regression line, which you never actually see. All Unit 9 inference uses b as a point estimate for β. Your hypotheses must always be written about β, because there's no uncertainty about b. You computed it. Writing H₀: b = 0 instead of H₀: β = 0 is a classic FRQ point-loser.
The slope b is the predicted change in the response variable for each one-unit increase in the explanatory variable, and it estimates the true population slope β.
A confidence interval for the slope is b ± t*(SEb), and if zero falls inside the interval, you lack convincing evidence of a linear relationship between x and y.
The test for a slope is a t-test with t = (b − β₀)/SEb and n − 2 degrees of freedom, almost always with H₀: β = 0.
The p-value is calculated by assuming the null hypothesis is true, so a small p-value means your sample slope would be unlikely if the true slope really equaled the null value.
Before any slope inference, check the conditions: linearity, constant standard deviation of y across x (both via residual plots), independence from random sampling plus the 10% condition, and approximately normal responses for each x.
Holding everything else constant, a larger sample size makes the confidence interval for the slope narrower.
It's the coefficient b in ŷ = a + bx, interpreted as the predicted change in y for each one-unit increase in x. In Unit 9, you treat b as an estimate of the true population slope β and run t-procedures on it.
Because simple linear regression estimates two parameters from the data, the slope and the intercept, so you lose two degrees of freedom. A t-test for a slope with 25 data points uses df = 23.
No. It means you lack convincing evidence of a linear relationship at that confidence level, not that the true slope is zero. For example, an interval like (-0.12, 0.89) is consistent with β = 0 but also with β = 0.5, so you fail to reject H₀ rather than accepting it.
b is the sample slope you compute from your data, and β is the unknown true slope of the population regression line. Hypotheses are always written about β (like H₀: β = 0), never about b, because b is a known number with no uncertainty.
Assuming the true population slope equals the value in the null hypothesis (usually 0), the p-value is the probability of getting a sample slope at least as extreme as yours by chance. So a p-value of 0.032 means results this extreme would happen about 3.2% of the time if there were truly no linear relationship.
Connect this key term to the AP exam workflow: review the course, practice questions, and check related study tools.
Review units, study guides, and course resources.
Check this vocabulary in multiple-choice context.
Apply key concepts in written AP responses.
Estimate the exam score you are working toward.
Review the highest-yield facts before practice.
Put the full course together before test day.