Carrying Out a Test for the Difference Between Two Population Proportions

AP Statistics· Unit 6, Topic 10: Significance Test for Difference in Two Proportions· 25 min read

1. Step 1: State Hypotheses and Define Parameters★★☆☆☆⏱ 7 min

The first step of any AP Statistics hypothesis test requires you to explicitly define your population parameters before writing null and alternative hypotheses to avoid losing easy rubric points.

📘 Definition

Null Hypothesis for Two-Proportion Test

$H_{0} : p_{1} - p_{2} = 0$

The claim that there is no difference between the two population proportions of successes, or $p_{1} = p_{2}$

Example:

$H_{0} : p_{m} - p_{f} = 0$ , where $p_{m}$ is the proportion of male high school students who have tried vaping, and $p_{f}$ is the proportion of female high school students who have tried vaping.

📐 Worked Example

A researcher wants to test if the proportion of teens who own a smartphone is different between rural and urban areas. State the correct hypotheses and define parameters.

1
First, define parameters clearly in context:
2
$p_1 = \text{True proportion of urban teens who own a smartphone}$
3
$p_2 = \text{True proportion of rural teens who own a smartphone}$
4
Write null hypothesis of no difference:
5
$H_0: p_1 = p_2 \quad \text{or} \quad H_0: p_1 - p_2 = 0$
6
Write two-sided alternative hypothesis for 'different':
7
$H_a: p_1 \neq p_2 \quad \text{or} \quad H_a: p_1 - p_2 \neq 0$

Exam tip:

You will lose the full hypotheses point on the AP rubric if you use sample statistics $\overset{p}{^}_{1}, \overset{p}{^}_{2}$ instead of population parameters $p_{1}, p_{2}$ .

2. Step 2: Verify Conditions for Inference★★★☆☆⏱ 6 min

All three conditions must be checked and explicitly justified to earn full points on the AP exam, even if they are implied in the problem prompt.

Random: Data comes from two independent random samples, or groups are assigned randomly in a controlled experiment
Independence: Observations within each group are independent, and each sample is no larger than 10% of its respective population
Normality: The sampling distribution of $\overset{p}{^}_{1} - \overset{p}{^}_{2}$ is approximately normal, verified by pooled successes and pooled failures all ≥ 10

📐 Worked Example

For a sample of 200 urban teens with 172 smartphone owners, and 180 rural teens with 135 smartphone owners, verify the normality condition for the test.

1
First calculate pooled total successes: 172 + 135 = 307
2
Calculate pooled total failures: (200 - 172) + (180 - 135) = 28 + 45 = 73
3
Check that all values are greater than 10: 307 ≥ 10, 73 ≥ 10. The normality condition is satisfied.

✓ Quick check

Which of the following scenarios violates the normality condition for a two-proportion z-test?

Sample 1 has 8 successes out of 50, sample 2 has 7 successes out of 50
Sample 1 has 3 successes out of 40, sample 2 has 4 successes out of 40
Reveal answer
true —
Pooled successes = 7 < 10, normality condition fails.
Sample 1 has 12 successes out of 30, sample 2 has 11 successes out of 30

3. Step 3: Calculate Test Statistic and P-Value★★★☆☆⏱ 7 min

Once conditions are verified, you calculate the pooled standard error, z test statistic, and corresponding p-value based on your pre-defined alternative hypothesis.

z = \frac{\hat{p}_1 - \hat{p}_2}{\sqrt{\hat{p}_{pooled}(1-\hat{p}_{pooled}) \left( \frac{1}{n_1} + \frac{1}{n_2} \right)}}

📘 Definition

Pooled Sample Proportion

$\overset{p}{^}_{p oo l e d} = \frac{x _{1} + x _{2}}{n _{1} + n _{2}}$

Combined proportion of successes across both samples, used to estimate the common population proportion assumed under the null hypothesis

📐 Worked Example

Using the urban/rural teen smartphone data: n1=200, x1=172, n2=180, x2=135, calculate the z test statistic.

1
Calculate individual sample proportions:
2
$\hat{p}_1 = 172/200 = 0.86, \quad \hat{p}_2 = 135/180 = 0.75$
3
Calculate pooled proportion:
4
$\hat{p}_{pooled} = (172 + 135)/(200 + 180) = 307/380 \approx 0.8079$
5
Compute standard error:
6
$SE = \sqrt{0.8079 * 0.1921 * (1/200 + 1/180)} \approx 0.0393$
7
Compute z statistic:
8
$z = (0.86 - 0.75)/0.0393 \approx 2.799$

4. Step 4: Draw a Conclusion in Context★★☆☆☆⏱ 5 min

Your conclusion must explicitly link the p-value to the significance level, reference the hypotheses, and state your result in the full context of the problem to earn all rubric points.

Compare your calculated p-value to the pre-stated significance level α (almost always 0.05 for AP exams)
State whether you reject or fail to reject the null hypothesis, referencing the p-value comparison
Write a final contextual conclusion that directly addresses the original research question

📐 Worked Example

Given p-value ≈ 0.005 for the two-sided smartphone ownership test, α=0.05, write the full AP-aligned conclusion.

1
Compare p-value to α: 0.005 < 0.05
2
Decision: Reject the null hypothesis
3
Contextual conclusion: There is convincing statistical evidence that the true proportion of teens who own smartphones is different between urban and rural populations.

5. Common Pitfalls

Wrong move:

Using unpooled standard error for a two-proportion significance test

Why:

The null hypothesis assumes $p_{1} = p_{2}$ , so pooling is required to get the correct standard error, and AP graders will deduct full points for this mistake

Correct move:

Always use the pooled proportion to calculate standard error for two-proportion z-tests, never use individual sample proportions

Wrong move:

Defining parameters using the collected sample groups instead of the full population

Why:

Hypotheses make claims about population values, not the data you collected, so this is a fundamental conceptual error

Correct move:

Explicitly state that $p_{1}$ and $p_{2}$ refer to the true, unknown proportions for the entire population of interest, not just your sample

Wrong move:

Stating 'we accept the null hypothesis' when p-value > α

Why:

You do not have evidence to prove the null is true, only that you do not have enough evidence to reject it

Correct move:

Always use the phrase 'fail to reject the null hypothesis' for non-significant results

Wrong move:

Checking the normality condition using individual sample counts instead of pooled counts

Why:

This can lead you to incorrectly think the normality condition fails when it actually passes, or vice versa

Correct move:

Sum successes across both groups, sum failures across both groups, confirm both totals are ≥ 10

Wrong move:

Writing a conclusion that does not reference the context of the original problem

Why:

AP Stats rubrics require conclusions to be tied to the specific scenario, generic conclusions get zero points for that step

Correct move:

Name the populations and the specific success outcome in your final conclusion, do not use generic terms like 'the two proportions'

6. Quick Reference Cheatsheet

Test Step	Required Action for Full AP Points
Hypotheses	Define $p_{1}, p_{2}$ in full context, write $H_{0} : p_{1} - p_{2} = 0$ and correct one/two-sided $H_{a}$
Conditions	Explicitly check Random, Independence (10% condition), Normality (pooled successes/failures ≥10)
Calculations	Compute pooled p, pooled SE, z statistic, one or two-tailed p-value
Conclusion	Compare p-value to α, state reject/fail to reject $H_{0}$ , write full contextual statement

7. Frequently Asked

Do I always need to use the pooled proportion for a two-proportion z-test?

Yes, for AP Statistics significance tests for two proportions, pooling is required when the null hypothesis states $p_{1} = p_{2}$ , as we assume the two populations have identical success proportions. You never use pooling for confidence intervals for the difference in two proportions.

What happens if the normality condition fails for my two-proportion test?

If pooled successes or pooled failures are less than 10, the sampling distribution of $\overset{p}{^}_{1} - \overset{p}{^}_{2}$ will not be approximately normal, and your calculated p-value will be unreliable. In this case, you should use a randomization test or Fisher's exact test instead.

When this came up on past exams

AI-estimated based on syllabus patterns — cross-check with official past papers for accuracy. Use only as revision-focus signals.

2025 · FRQ 2
Two-proportion test for vaccine efficacy
2024 · MCQ Set 4
Verify conditions for two-proportion z-test
2023 · FRQ 1
Calculate p-value for two-tailed test

Going deeper

official guideAP Statistics CED Inference for ProportionsOfficial College Board syllabus reference for this topic

What's Next

Mastery of two-proportion z-tests is one of the highest-weighted skills on the AP Statistics exam, appearing in nearly every recent FRQ section focused on inference. This skill directly builds on your prior work with one-proportion tests and sampling distributions for difference in proportions, and it is the required prerequisite for learning chi-square tests for homogeneity later in Unit 6. You will see this scenario tested both as a full 4-step FRQ and as multiple choice questions that ask you to identify correct conditions, calculate p-values, or spot flawed conclusions. Make sure to practice writing full, context-rich conclusions to avoid losing easy rubric points on exam day.