In AP Statistics, the difference in two population proportions (p₁ - p₂) compares the true proportion of a characteristic in two distinct populations; a two-proportion z-test checks whether that difference is plausibly zero, using a pooled proportion p̂c to build the test statistic (Topic 6.11).
The difference in two population proportions, written p₁ - p₂, is the gap between the true proportion of some outcome in one population and the true proportion in another. Think of it as the answer to a comparison question. Is the cure rate higher with the new drug than the old one? Do seniors support the policy at a different rate than freshmen? You never see p₁ - p₂ directly. You estimate it with the difference in sample proportions, p̂₁ - p̂₂, and then use inference to decide what the sample gap says about the population gap.
In Topic 6.11, the inference tool is the two-proportion z-test. The null hypothesis says the two populations are identical, H₀: p₁ = p₂ (equivalently p₁ - p₂ = 0). Because the null assumes both populations share one common proportion, you pool the samples into a combined proportion p̂c = (n₁p̂₁ + n₂p̂₂)/(n₁ + n₂) and use it in the standard error. The test statistic is z = (p̂₁ - p̂₂ - 0) / √[p̂c(1 - p̂c)(1/n₁ + 1/n₂)]. That z tells you how many standard errors your observed difference sits from zero, and the p-value tells you how surprising that is if the populations really were equal.
This term sits at the center of Topic 6.11 in Unit 6 (Inference for Categorical Data: Proportions), the capstone of the proportions unit. It supports three learning objectives directly. AP Stats 6.11.A asks you to calculate the test statistic, including the pooled proportion p̂c. AP Stats 6.11.B asks you to interpret the p-value, and the CED is picky here. Your interpretation has to acknowledge that the p-value is computed assuming the null is true, meaning assuming the two population proportions are actually equal. AP Stats 6.11.C asks you to make the formal decision by comparing the p-value to α and then translate that decision back into an answer about the real-world research question. One practical note from the CED: the test statistic formula isn't printed on the exam formula sheet, but you can rebuild it from the general (statistic - parameter)/standard error template plus the standard error formulas that are provided.
Keep studying AP Statistics Unit 6
Sampling Distribution of Differences in Sample Proportions (Unit 5)
The whole test only works because Unit 5 told you how p̂₁ - p̂₂ behaves across repeated samples. It's approximately normal with mean p₁ - p₂. The two-proportion z-test is just that sampling distribution put to work under the assumption p₁ - p₂ = 0.
Confidence Interval for a Difference of Two Proportions (Unit 6)
The interval and the test answer related questions in different ways. The interval estimates how big p₁ - p₂ is, while the test asks whether it's plausibly zero. Heads up on the math, though. The confidence interval does not pool, because it makes no assumption that the proportions are equal.
One-Sample z-Test for a Proportion (Unit 6)
Same logic, one extra population. Both tests use z = (statistic - parameter)/standard error and both end with a p-value compared to α. The two-sample version just swaps in p̂₁ - p̂₂ as the statistic and the pooled standard error underneath.
Chi-Square Test for Homogeneity (Unit 8)
When you have more than two populations to compare, the two-proportion z-test runs out of room. The chi-square test for homogeneity is the upgrade that compares proportions across many groups at once. With exactly two groups, the two methods agree.
Multiple-choice questions hit this term from a few angles. Some test the conditions, asking which situation would invalidate the pooled z formula (think non-random samples or failed large counts checks). Others test the mechanics, like identifying the correct pooled proportion p̂c or recognizing that you need critical values or a p-value to finish the test. On the free-response side, a two-proportion z-test is a classic inference FRQ format. You'd be expected to do the full four-step routine. State hypotheses in terms of p₁ and p₂, name the procedure and check conditions, compute z and the p-value, then compare the p-value to α and write a conclusion in context. Two scoring traps to avoid. First, your p-value interpretation must reference the assumption that p₁ = p₂. Second, never say you "accept" the null. You either reject it or fail to reject it.
The population difference p₁ - p₂ is the unknown parameter, a fixed number you'll never observe. The sample difference p̂₁ - p̂₂ is the statistic you calculate from your data, and it changes from sample to sample. Hypotheses are always written about the parameter (H₀: p₁ - p₂ = 0), never about the statistic. Writing H₀: p̂₁ = p̂₂ on an FRQ loses credit, because there's nothing uncertain about numbers you already computed.
The difference in two population proportions, p₁ - p₂, is the parameter you test; the difference in sample proportions, p̂₁ - p̂₂, is the statistic you use to test it.
The null hypothesis is H₀: p₁ = p₂ (or p₁ - p₂ = 0), which is why you pool the two samples into a combined proportion p̂c for the standard error.
The test statistic is z = (p̂₁ - p̂₂ - 0) / √[p̂c(1 - p̂c)(1/n₁ + 1/n₂)], and you can rebuild it from the formula sheet even though it isn't printed there.
A correct p-value interpretation must state that it was computed assuming the null is true, meaning the two population proportions are equal.
If the p-value is less than or equal to α, reject H₀ and conclude there is convincing evidence of a difference; if the p-value is greater than α, fail to reject H₀ (never 'accept' it).
The significance test pools the samples, but the confidence interval for p₁ - p₂ does not, because the interval doesn't assume the proportions are equal.
It's the parameter p₁ - p₂, the gap between the true proportions of some outcome in two distinct populations. In Topic 6.11 you test whether this difference is zero using a two-proportion z-test built on the pooled proportion p̂c.
Because the null hypothesis assumes p₁ = p₂, meaning both populations share one common proportion. The pooled proportion p̂c = (n₁p̂₁ + n₂p̂₂)/(n₁ + n₂) is your best estimate of that shared value, so it goes into the standard error.
Not exactly, but you don't need to memorize it. The formula sheet gives you the general test statistic template and the standard error formulas, and the CED explicitly says you can construct the two-proportion z statistic from those pieces.
No, it doesn't prove anything. A small p-value (≤ α) gives convincing statistical evidence to reject H₀: p₁ = p₂, but significance tests never prove a hypothesis. Your conclusion should say there is convincing evidence of a difference, in the context of the problem.
The test asks a yes-or-no question (is zero a plausible difference?) while the interval estimates a range of plausible values for p₁ - p₂. Computationally, the test uses the pooled standard error because it assumes p₁ = p₂, but the interval uses separate p̂₁ and p̂₂ since it makes no such assumption.