In AP Statistics, a p-value is the probability of getting a test statistic as extreme or more extreme than the one observed, computed by assuming the null hypothesis is true. If the p-value is less than or equal to α, you reject H₀; if it's greater than α, you fail to reject H₀.
A p-value answers one specific question. If the null hypothesis were actually true, how likely is it that random chance alone would produce data at least as extreme as what you got? A small p-value means your data would be weird in a world where H₀ is true, so you have evidence against H₀. A large p-value means your data are pretty ordinary under H₀, so you don't have enough evidence to reject it.
The phrase "assuming the null hypothesis is true" is not optional decoration. The CED repeats it in every single test, from one-proportion z-tests (AP Stats 6.5.B) to two-sample t-tests (AP Stats 7.9.B) to chi-square tests (AP Stats 8.3.C) to slope tests (AP Stats 9.5.B). The p-value is computed FROM the null distribution, the sampling distribution your test statistic would follow if H₀ were true. Where you measure "extreme" depends on the alternative hypothesis. With Hₐ: >, it's the area at or above your test statistic. With Hₐ: <, it's the area at or below. With Hₐ: ≠, you take both tails. What a p-value is NOT is the probability that the null hypothesis is true. That misreading is the single most common p-value error, and AP graders watch for it.
The p-value is the connective tissue of the entire second half of AP Stats. The same logic repeats across four units. You meet it in Unit 6 with proportions (Topics 6.4-6.6, 6.10-6.11), reuse it in Unit 7 with means and matched pairs (Topics 7.5, 7.8-7.9), again in Unit 8 with chi-square tests (Topics 8.2-8.3, 8.6), and once more in Unit 9 with regression slopes (Topic 9.5). Each unit has a learning objective specifically about interpreting the p-value (6.5.B, 7.5.B, 7.9.B, 8.3.C, 8.6.C, 9.5.B) and one about justifying a claim by comparing it to α (6.6.A, 7.5.C, 7.9.C, 8.3.D, 8.6.D, 9.5.C). Master the interpretation once and you've effectively prepped six learning objectives at the same time. Only the parameter changes (p, μ₁-μ₂, the distribution of a categorical variable, β); the reasoning never does. For the deep dive on the logic itself, start with the Topic 6.5 study guide on interpreting p-values.
Keep studying AP Statistics Unit 8
Significance Level (α) (Unit 6)
The p-value is what your data produced; α is the cutoff you set before collecting data. The formal decision rule from 6.6.A compares them directly. If p ≤ α, reject H₀. If p > α, fail to reject. The p-value is evidence; α is the bar that evidence has to clear.
Null Hypothesis (Unit 6)
You literally cannot compute a p-value without H₀, because the p-value comes from the null distribution. Every CED interpretation objective requires you to say the p-value was "computed assuming the null hypothesis is true." Drop that phrase on an FRQ and you lose credit.
Type I and Type II Errors (Unit 6, Topic 6.7)
Since you reject H₀ whenever p ≤ α, the significance level α is exactly the probability of a Type I error (rejecting a true null). A tiny p-value doesn't make your conclusion certain. There's always a chance the null was true and you just got unlucky data.
Chi-Square Statistic (Unit 8)
Chi-square tests show that the p-value idea isn't tied to z or t. The χ² statistic measures distance between observed and expected counts, and because chi-square distributions are skewed right, the p-value is always the area in the upper tail (8.6.B). Same logic, different curve.
Confidence Interval (Units 6-9)
Confidence intervals and two-sided significance tests are two views of the same inference. If a 95% interval for a parameter misses the null value, a two-sided test at α = 0.05 would reject H₀. Topic 7.10 expects you to move between these procedures fluently.
On the multiple-choice section, p-value questions usually hand you computer output or a stated p-value and ask which interpretation or conclusion is correct. The wrong answers are predictable. They claim the p-value is the probability H₀ is true, they "accept" the null, or they reject when p > α. For example, a stem giving a slope test with p = 0.07 at α = 0.05 wants you to fail to reject H₀ and say there is insufficient evidence of a linear relationship, not that you've proven the slope is zero.
On FRQs, the p-value shows up inside full significance tests, which appear nearly every year. The 2017 exam asked whether data on age at schizophrenia diagnosis gave convincing evidence of a difference between men and women, and the 2021 exam tested whether a $10 coupon increased repeat purchases at an online pet supply company. In both, you state hypotheses, check conditions, compute the test statistic and p-value, then write a conclusion that (1) compares p to α explicitly, (2) states the decision about H₀, and (3) answers in context about the alternative. "Because the p-value of 0.012 is less than α = 0.05, we reject H₀ and have convincing evidence that..." is the sentence pattern graders reward.
Both are probabilities used in the same decision, which is why they blur together. The significance level α is chosen in advance and never depends on your data; it's the probability of a Type I error you're willing to tolerate. The p-value is calculated from your data after the fact and measures how extreme your results are under H₀. Think of α as the bar and the p-value as the jump. You compare the two, but you never confuse which one came from the data.
A p-value is the probability of getting a test statistic as extreme or more extreme than the observed one, calculated by assuming the null hypothesis is true.
The decision rule is the same in every AP Stats test: if the p-value ≤ α, reject H₀; if the p-value > α, fail to reject H₀ (never "accept" H₀).
The direction of the alternative hypothesis determines where "extreme" lives: upper tail for >, lower tail for <, and both tails for ≠.
A p-value is NOT the probability that the null hypothesis is true, and a large p-value does not prove H₀; it only means there isn't sufficient evidence against it.
The same p-value logic repeats across Units 6-9 for proportions, means, chi-square tests, and regression slopes; only the parameter and the null distribution change.
A full-credit FRQ conclusion explicitly compares the p-value to α, states the decision about H₀, and answers the research question in context.
It's the probability, computed assuming the null hypothesis is true, of getting a test statistic as extreme or more extreme than the one your sample produced. Small p-values (typically ≤ 0.05) give convincing evidence against H₀.
No, and this is the most-penalized p-value error on the AP exam. The p-value is a probability about the data given that H₀ is true, not a probability about H₀ itself. P(data this extreme | H₀ true) is not P(H₀ true).
α is a cutoff you choose before collecting data (often 0.05), and it equals the probability of a Type I error. The p-value is computed from your actual data. You make a decision by comparing them: reject H₀ when p ≤ α.
No, you fail to reject it. A p-value like 0.12 means there's insufficient evidence for the alternative, not proof that H₀ is true. Writing "accept H₀" on an FRQ costs you points.
Use the template: "Assuming H₀ is true (the true proportion/mean/slope equals the null value), there is a [p-value] probability of getting a sample result as extreme or more extreme than the one observed." Then compare p to α, state your decision, and answer in context, just like the 2017 and 2021 FRQs required.
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.