In AP Statistics, the significance level (alpha, α) is the predetermined probability of rejecting a true null hypothesis. You compare your p-value to α in every significance test. If the p-value ≤ α, reject H₀; if the p-value > α, fail to reject H₀. Common values are 0.05, 0.01, and 0.10.
The significance level, written as alpha (α), is the cutoff you set before running a hypothesis test. It answers one question. How small does my p-value have to be before I'm willing to reject the null hypothesis? Per the CED (AP Stats 6.6.A), α is "the predetermined probability of rejecting the null hypothesis given that it is true." In plain terms, it's the amount of false-positive risk you're willing to accept.
That second part is the piece most people miss. Alpha isn't just a cutoff line, it IS the probability of a Type I error (rejecting a null hypothesis that was actually true). Setting α = 0.05 means you're accepting a 5% chance of crying wolf when nothing is going on. That's why the choice of α depends on consequences. If a false positive would be costly (say, approving an ineffective drug), you pick a smaller α like 0.01. The decision rule never changes across tests: if the p-value ≤ α, reject H₀; if the p-value > α, fail to reject H₀.
The significance level is the single most reused idea in the entire inference half of AP Stats. It shows up in Unit 6 (proportion tests, Topics 6.6, 6.7, 6.10, 6.11), Unit 7 (mean tests, Topics 7.4, 7.5, 7.8, 7.9), Unit 8 (chi-square tests, Topics 8.2, 8.3), and Unit 9 (regression slope tests, Topic 9.5). Every "justify a claim" learning objective, including AP Stats 6.6.A, 7.5.C, 7.9.C, 8.3.D, and 9.5.C, requires the same move. You make a formal decision by explicitly comparing the p-value to α. Alpha also anchors the error analysis in Topic 6.7. AP Stats 6.7.B states that α is the probability of a Type I error, and AP Stats 6.7.C tells you that increasing α decreases the probability of a Type II error. Learn the p-value vs. α comparison once and you've learned the conclusion step for every test on the exam.
Keep studying AP Statistics Unit 7
Type I Error (Unit 6)
These are two names for the same number. The significance level α equals the probability of a Type I error, which is rejecting a null hypothesis that's actually true. That's why Topic 6.7 says the consequences of a false positive should drive your choice of α. Scary consequences mean a smaller alpha.
P-Value (Units 6-9)
Alpha is the bar; the p-value is the jump. The p-value is calculated from your data assuming H₀ is true, while α is fixed before you collect anything. The conclusion of every test is just a comparison of the two, and that comparison is identical whether you're testing a proportion, a mean, a chi-square statistic, or a regression slope.
Power and Type II Error (Unit 6)
Alpha and Type II error pull against each other. Per AP Stats 6.7.C, increasing α makes it easier to reject H₀, which lowers the chance of a Type II error and raises power. Lowering α does the opposite. There's no free lunch, so choosing α is choosing which mistake you'd rather risk.
Test for the Slope of a Regression Model (Unit 9)
Topic 9.5 is proof that α travels everywhere. Even in the last inference topic of the course, AP Stats 9.5.C asks for the exact same decision rule. Compare the p-value of the t-test for slope to α, then reject or fail to reject H₀: β = β₀. New test statistic, same conclusion logic.
Significance levels appear in nearly every released inference FRQ, including 2017 Q5 (chi-square), 2018 Q6, 2021 Q4 (two-proportion test for repeat purchases), and 2023 Q4 (two-sample comparison from an experiment). The prompt usually hands you α (often 0.05) and expects you to use it in your conclusion. Full credit requires three linked pieces: the explicit comparison ("since p-value = 0.022 > α = 0.01"), the formal decision ("we fail to reject H₀"), and a conclusion about the alternative hypothesis in context ("there is not convincing evidence that..."). Multiple-choice questions love edge cases. A favorite setup gives you a p-value near a common α and asks which researcher correctly rejects, testing whether you know that p = 0.022 means reject at α = 0.05 but fail to reject at α = 0.01. Two traps to avoid in writing: never say you "accept" the null hypothesis, and never claim the test "proves" anything.
The significance level α is chosen before the test and never changes based on your data. The p-value is computed from your data, assuming the null hypothesis is true. Alpha is the standard you set; the p-value is the evidence you found. You compare evidence to standard. A second mix-up worth killing: α is not "the probability the null is true." It's the probability of rejecting H₀ when H₀ is true, which is a conditional probability about your decision, not about reality.
The significance level α is the predetermined probability of rejecting the null hypothesis given that it is true, which makes it exactly equal to the probability of a Type I error.
The decision rule is universal across all AP Stats tests: if the p-value ≤ α, reject H₀; if the p-value > α, fail to reject H₀.
Alpha is set before collecting data, while the p-value is computed from the data, so changing α after seeing your p-value is cheating.
Increasing α decreases the probability of a Type II error and increases power, so choosing α is a trade-off between false positives and false negatives.
When the consequences of a Type I error are serious, choose a smaller α (like 0.01 instead of 0.05).
Failing to reject H₀ means there is insufficient evidence for the alternative hypothesis, not that the null hypothesis has been proven true.
It's the threshold (alpha, α) you compare your p-value to when deciding whether to reject the null hypothesis. The CED defines it as the predetermined probability of rejecting H₀ given that H₀ is true, and the most common value is 0.05.
No. Alpha is fixed before the test and represents your tolerance for a false positive, while the p-value is calculated from your sample data assuming H₀ is true. You make a decision by comparing them, so they have to be different things.
No, and this misreading costs points. α = 0.05 means that IF the null hypothesis is true, there's a 5% probability your test will incorrectly reject it. It's a statement about the test's error rate, not about whether H₀ is true.
You reject the null hypothesis. The CED decision rule is reject H₀ when the p-value ≤ α, so equality counts as rejection. In practice, a p-value landing exactly on α is rare, but MCQs can test the boundary.
Alpha IS the probability of a Type I error (rejecting a true null). Raising α makes Type I errors more likely but Type II errors less likely, and per Topic 6.7 you should weigh which error is more consequential before choosing α.