A two-sample z-interval for a difference between two population proportions gives you a range of plausible values for how much two groups differ on a yes/no variable. You build it as $(\hat{p}_1 - \hat{p}_2) \pm z^*$ times the standard error, check independence and normal conditions for both samples, and then read whether 0 is inside the interval to decide if the groups likely differ.

Why This Matters for the AP Statistics Exam

This topic is part of inference for proportions, which carries real weight on the AP Statistics exam. Knowing how to build and read a two-sample proportion interval helps you on multiple-choice questions that test conditions, formulas, and interpretation, and it sets you up for free-response work where you choose a procedure, check conditions, show the calculation, and explain the interval in context.

It also connects directly to later topics. Once you can construct this interval, you can justify claims about a difference in proportions and recognize how the same setup relates to the two-proportion significance test.

more resources to help you study

practice multiple choice FRQ practice & scoring cheatsheets score calculator key terms

Key Takeaways

The correct procedure for comparing two proportions on one categorical variable is a two-sample z-interval for a difference between population proportions.
The interval is (p̂1 - p̂2) ± z* √(p̂1(1-p̂1)/n1 + p̂2(1-p̂2)/n2). The point estimate is p̂1 - p̂2 and the rest is the margin of error.
For a confidence interval, do not pool the proportions. Use each sample's own p̂ in the standard error.
Check independence (two independent random samples or a randomized experiment, plus the 10% condition for each sample) and the normal condition for both samples.
If the whole interval is above or below 0, you have evidence of a real difference. If 0 is inside the interval, a difference of 0 is plausible.
You do not need to memorize the formula. You can build it from the standard error pieces on the formula sheet.

Building a Two-Sample Z-Interval

A two-sample z-interval lets you compare the proportion of a categorical outcome between two independent groups. Think of comparing the proportion of free throws made by two players, or the proportion of voters who support a measure in two cities.

The interval follows the same structure as every confidence interval: a point estimate plus or minus a margin of error.

Point Estimate

The point estimate is the difference between the two sample proportions, p̂1 - p̂2. You just subtract one sample proportion from the other.

Margin of Error

The margin of error is the buffer you add and subtract so the interval can capture the true difference in population proportions. It is built from two pieces:

the critical value z* for your confidence level
the standard error of the difference

For a confidence interval, the standard error uses each sample's own proportion (no pooling):

SE = √(p̂1(1-p̂1)/n1 + p̂2(1-p̂2)/n2)

Full Formula

Putting it together:

(p̂1 - p̂2) ± z* √(p̂1(1-p̂1)/n1 + p̂2(1-p̂2)/n2)

You can construct this from the standard error formulas on the AP Statistics formula sheet, so you do not have to memorize it cold.

Reading the Interval and Zero

The interval gives a range of plausible values for the true difference p1 - p2. Because the null idea of "no difference" is a difference of 0:

If 0 is not inside the interval, a difference of 0 is not plausible, so you have evidence the two proportions differ.
If 0 is inside the interval, then no difference is plausible, and you cannot say the proportions differ.

You will work with interpretations more in the next topic, so focus here on the setup and calculation.

Conditions to Check

Inference always requires checking conditions before you trust the interval. You are using two samples to estimate a difference between two populations, so check each condition for both samples.

Random

Both samples must come from independent random samples, or the groups must come from a randomized experiment. Without randomness, sampling bias can make your interval misleading and you cannot generalize to the populations.

Independence (10% Condition)

When sampling without replacement, each sample should be no more than 10% of its population: n1 ≤ 0.10N1 and n2 ≤ 0.10N2. In a randomized experiment, random assignment of treatments is what makes the groups independent.

Normal (Large Counts)

Use the Large Counts idea for both samples. Check that the successes and failures in each sample are large enough (commonly at least 10): n1p̂1, n1(1-p̂1), n2p̂2, and n2(1-p̂2). This is what lets you treat the sampling distribution of p̂1 - p̂2 as approximately normal.

How to Use This on the AP Statistics Exam

Free Response

When a question asks you to estimate a difference in proportions:

Name the procedure: two-sample z-interval for a difference between population proportions.
Check all three conditions for both samples, using numbers from the problem when possible.
Show the point estimate, standard error, and z* so your work is clear and easy to follow.
State the interval and tie it back to the context of the two groups.

Calculator

To compute quickly, use the "2-PropZInt" option in your calculator's Stats/Tests menu. Enter the successes and sample size for each group and the confidence level. Even when you use the calculator, write out the procedure name, conditions, and interpretation so your reasoning is visible.

Common Trap

For a confidence interval, do not pool the proportions. Pooling shows up in the two-proportion significance test, not in the interval. For the interval, plug each sample's own p̂ into the standard error.

Worked Example

Suppose you compare two players' field goal proportions in their first NBA season. Player 1 attempted 1623 shots and made 836. Player 2 attempted 1493 shots and made 622. Build a 95% confidence interval for the difference in their proportions of shots made.

Conditions

Random: The problem treats each first season as the sample. If a problem does not state random selection, you note that you are assuming it so you can proceed.
Independence: It is reasonable that each player took far more than 10 times their first-season attempts over a career, so n1 ≤ 0.10N1 and n2 ≤ 0.10N2 are satisfied, making the samples independent.
Normal: Each player had well over 10 makes and over 10 misses, so n1p̂1, n1(1-p̂1), n2p̂2, and n2(1-p̂2) are all large enough to use a normal approximation for the difference.

Calculation

The sample proportions are p̂1 = 836/1623 and p̂2 = 622/1493. Put these into:

(p̂1 - p̂2) ± z* √(p̂1(1-p̂1)/n1 + p̂2(1-p̂2)/n2)

with z* for 95% confidence. Using "2-PropZInt" gives the interval. Then check whether 0 falls inside it: if it does not, you have evidence the two proportions differ. Interpretation in full context comes in the next topic.

Common Misconceptions

Pooling for the interval: The pooled proportion is for the two-proportion significance test. For a confidence interval, use each sample's own p̂ in the standard error.
Forgetting to check both samples: The normal condition has four parts (successes and failures for each sample). Checking only one sample is incomplete.
Saying the interval gives the probability the true difference is inside: Once computed, the interval either contains the true difference or it does not. Confidence refers to the long-run capture rate of the method, not a single interval.
Treating "0 in the interval" as proof of no difference: It only means a difference of 0 is plausible. It does not prove the proportions are equal.
Mixing up the point estimate: The center of the interval is p̂1 - p̂2, the difference of sample proportions, not a single proportion.
Ignoring direction: The order you subtract in (group 1 minus group 2) sets the sign, so keep track of which group is which when you read positive or negative values.

Vocabulary

The following words are mentioned explicitly in the College Board Course and Exam Description for this topic.

Term	Definition
10% condition	The requirement that sample size n is at most 10% of the population size N to ensure independence when sampling without replacement.
approximately normal	A distribution that closely follows the shape of a normal distribution, allowing for the use of normal probability methods.
categorical variable	A variable that takes on values that are category names or group labels rather than numerical values.
confidence interval	A range of values, calculated from sample data, that is likely to contain the true population parameter with a specified level of confidence.
difference in proportions	The difference between two population proportions, calculated as p₁ - p₂, used to compare the prevalence of a characteristic across two populations.
difference of two population proportions	The comparison between two population proportions, expressed as p₁ - p₂, to determine if they differ significantly.
independence	The condition that observations in a sample are not influenced by each other, typically ensured through random sampling or randomized experiments.
population proportion	The true proportion or percentage of a characteristic in an entire population, typically denoted as p.
randomized experiment	A study design where subjects are randomly assigned to treatment groups to establish cause-and-effect relationships.
sample proportion	The proportion of individuals in a sample that have a particular characteristic, denoted as p-hat (p̂).
sampling distribution	The probability distribution of a sample statistic (such as a sample proportion) obtained from repeated sampling of a population.
sampling without replacement	A sampling method in which an item selected from a population cannot be selected again in subsequent draws.
simple random sample	A sample selected from a population such that every possible sample of the same size has an equal chance of being chosen.
standard error	The standard deviation of a sampling distribution, which measures the variability of a sample statistic across repeated samples.
success-failure condition	A requirement that the expected number of successes and failures in each sample (np̂ and n(1-p̂)) meet a minimum threshold, typically 5 or 10, to ensure the sampling distribution is approximately normal.
test statistic	A calculated value used to determine whether to reject the null hypothesis in a hypothesis test, computed from sample data.
two-sample z-interval	A confidence interval procedure that uses the standard normal distribution to estimate the difference between two population proportions based on sample data.

Frequently Asked Questions

What is a two-proportion z-interval?

A two-proportion z-interval estimates the difference between two population proportions. It uses the sample difference, p-hat1 minus p-hat2, plus or minus a z critical value times the standard error.

When do I use a confidence interval for the difference of two proportions?

Use it when you have two independent groups and one categorical yes/no outcome, and you want a range of plausible values for p1 minus p2.

What conditions do I check for a two-proportion z-interval?

Check that the samples or treatments are independent, each sample is random or comes from a randomized experiment, each sample is less than 10 percent of its population when sampling without replacement, and both groups meet the normal condition.

What is the standard error for a two-proportion z-interval?

For a confidence interval, the standard error is the square root of p-hat1(1 - p-hat1)/n1 plus p-hat2(1 - p-hat2)/n2. Do not pool the sample proportions for an interval.

What does it mean if zero is inside the interval?

If 0 is inside the interval, a difference of zero is plausible, so the interval does not give strong evidence that the two population proportions differ.

Do I need to memorize the two-proportion z-interval formula for AP Statistics?

You should know how the interval works and how to build it from the formula sheet. The AP formula sheet gives the general confidence interval structure and standard error pieces, so focus on procedure, conditions, and interpretation.