The difference of two population proportions (p₁ - p₂) is the true gap between the proportion of successes in two distinct populations. In AP Stats Topic 6.8, you estimate this unknown parameter with a two-sample z-interval built from sample data (p̂₁ - p̂₂).
The difference of two population proportions, written p₁ - p₂, is the parameter you're after whenever a question compares two groups on one categorical variable. Think "what fraction of population 1 has this trait, minus what fraction of population 2 has it." You almost never know p₁ or p₂ directly, so you estimate the difference using the sample statistic p̂₁ - p̂₂ and build a two-sample z-interval for a difference between population proportions around it.
The interval is (p̂₁ - p̂₂) ± z* √(p̂₁(1-p̂₁)/n₁ + p̂₂(1-p̂₂)/n₂). Before you compute anything, the CED requires two checks. First, independence, meaning the data come from two independent random samples or a randomized experiment, and if you're sampling without replacement, each sample is at most 10% of its population. Second, normality of the sampling distribution of p̂₁ - p̂₂, which you verify with the Large Counts condition applied four times (n₁p̂₁, n₁(1-p̂₁), n₂p̂₂, and n₂(1-p̂₂) all large enough). One useful reframe is that the resulting interval is a set of plausible values for the true gap between the groups, and whether 0 sits inside that interval tells you whether "no difference at all" is still plausible.
This parameter is the heart of Topic 6.8 in Unit 6 (Inference for Categorical Data: Proportions). It directly supports four learning objectives. AP Stats 6.8.A asks you to name the right procedure (a two-sample z-interval for a difference between population proportions). AP Stats 6.8.B asks you to verify independence and the Large Counts condition for both samples. AP Stats 6.8.C asks you to calculate the interval, and AP Stats 6.8.D asks you to turn it into an interval estimate with real units, like "the difference in the proportion of teens versus adults who own a smartwatch." This is also where AP Stats levels up from describing one group to comparing two, which is what most real studies (and most experiment-based FRQs) actually do.
Keep studying AP Statistics Unit 6
Sample Proportion and the Sampling Distribution of p̂₁ - p̂₂ (Units 5-6)
p₁ - p₂ is the fixed, unknown parameter; p̂₁ - p̂₂ is the statistic that varies from sample to sample. Everything in Topic 6.8 works because the sampling distribution of p̂₁ - p̂₂ is approximately normal when the conditions hold, which is pure Unit 5 logic applied to two groups at once.
Confidence Interval Interpretation (Unit 6)
A two-sample interval gets interpreted just like a one-sample interval, except the parameter is a difference. The extra payoff is the zero check. If the entire interval is above 0, you have convincing evidence p₁ is larger; if 0 is inside, a true difference of zero is still plausible.
Independence Condition and Large Counts Condition (Unit 6)
With two samples, every condition doubles. You need two independent random samples (or a randomized experiment), the 10% condition for both samples, and Large Counts checked on all four counts. Skipping any of these is one of the most common ways to lose points on inference FRQs.
Difference of Two Population Means (Unit 7)
Unit 7 runs the exact same playbook for quantitative data. Swap proportions for means and z for t, and the structure (parameter is a difference, check conditions, build interval, check whether 0 is plausible) carries over completely. Master it here and Unit 7 feels familiar.
On multiple choice, expect stems that make you pick the correct procedure (two-sample z-interval, not one-sample, not a test for means), identify which condition fails, or interpret an interval like (0.02, 0.11) in context. On the FRQ side, two-proportion inference is a classic full-inference question. You need to define the parameter p₁ - p₂ in context, name the procedure, verify independence and Large Counts for both samples, compute the interval, and interpret it with units and direction. A clarifying statement in the CED says the interval formula isn't printed on the formula sheet as-is, but you can rebuild it from the general statistic ± (critical value)(standard error) structure and the standard error formulas that are provided. The biggest scoring trap is interpreting the interval as if it were about a single proportion instead of a difference.
p₁ - p₂ is the parameter, the true difference between two populations, fixed but unknown. p̂₁ - p̂₂ is the statistic, calculated from your data, and it changes from sample to sample. The confidence interval is centered at p̂₁ - p̂₂ but estimates p₁ - p₂. On FRQs, defining the parameter with hats (or interpreting the interval as being about the sample) costs you the parameter-definition point, because there's nothing uncertain about your own sample data.
The difference of two population proportions, p₁ - p₂, is an unknown parameter that you estimate with the sample statistic p̂₁ - p̂₂.
The correct procedure for comparing two population proportions on one categorical variable is a two-sample z-interval for a difference between population proportions.
Conditions come in pairs: two independent random samples or a randomized experiment, the 10% condition for each sample, and the Large Counts condition checked on all four counts.
The interval formula is (p̂₁ - p̂₂) ± z* √(p̂₁(1-p̂₁)/n₁ + p̂₂(1-p̂₂)/n₂), and you can rebuild it from the standard error formulas on the AP formula sheet instead of memorizing it.
If 0 falls inside the interval, a true difference of zero is plausible, so you don't have convincing evidence that the population proportions differ.
Always interpret the interval as a difference with direction, for example "we are 95% confident the proportion for group 1 exceeds the proportion for group 2 by between 2 and 11 percentage points."
It's the parameter p₁ - p₂, the true gap between the proportion of successes in two distinct populations. In Topic 6.8 you estimate it with a two-sample z-interval built around the sample difference p̂₁ - p̂₂.
No. It only means a difference of zero is one of the plausible values, so you lack convincing evidence of a difference. "No evidence of a difference" is not the same as "evidence of no difference," and that wording distinction earns or loses FRQ points.
p₁ - p₂ is the population parameter, fixed and unknown. p̂₁ - p̂₂ is the sample statistic you compute from data, and it's the center of your confidence interval. Define the parameter without hats on the exam, since the inference is about the populations, not your samples.
No. The CED notes the interval formula isn't printed verbatim on the formula sheet, but you can construct it from the general statistic ± (critical value)(standard error) structure and the standard error formulas that are provided.
No. Pooling p̂₁ and p̂₂ into a combined proportion belongs to the two-sample significance test, which assumes p₁ = p₂ under the null. The confidence interval makes no such assumption, so its standard error uses p̂₁ and p̂₂ separately.