Scribe logoScribeNotes & practice
Sign inStart free
  1. Notes
  2. ›AP Statistics
  3. ›3 Inference for Categorical Data: Proportions
  4. ›3.10 Interval for a difference of proportions

AP Statistics · Unit 3 · Inference for Categorical Data: Proportions

3.10 Interval for a difference of proportions

Calculate and interpret a confidence interval for the difference between two population proportions after checking the conditions.

  • 4 min read
  • 5 parts
  • 10 questions
  • Draft
Start a Quick 5 on this topic→Jump to the 10 questions
1/5A confidence interval estimates the gap between two proportions
  1. 1A confidence interval estimates the gap between two proportions
  2. 2The standard error replaces unknown proportions with sample proportions
  3. 3Conditions must be checked before computing the interval
  4. 4The calculation takes four steps from counts to interval
  5. 5An interval is interpreted in context with its confidence level
  6. Check your recall
  7. Practice: 10 questions
1

Part 1 of 5: A confidence interval estimates the gap between two proportions

One sample difference rarely equals the true gap exactly. Another pair of samples would give a different value. A confidence interval offers a range of reasonable values for p1−p2p_1-p_2p1​−p2​.

The two-proportion z-interval follows the pattern of every interval: statistic plus or minus margin of error. The statistic is p^1−p^2\hat{p}_1-\hat{p}_2p^​1​−p^​2​. The margin of error is the distance from the center to either endpoint.

That distance is a critical value times the standard error. Both pieces come next. An interval is useful because it shows both the estimate and its uncertainty, which a single number cannot do.

Key term

two-proportion z-interval

A two-proportion z-interval is a confidence interval for p1−p2p_1-p_2p1​−p2​ built from two independent samples and a critical value from the standard normal distribution.

Part 1 done

2

Part 2 of 5: The standard error replaces unknown proportions with sample proportions

The earlier standard deviation formula needs p1p_1p1​ and p2p_2p2​, which are unknown. We substitute the sample proportions instead. The result is the standard error, an estimate of the statistic’s standard deviation.

Written out in full, the interval is (p^1−p^2)±z∗p^1(1−p^1)n1+p^2(1−p^2)n2(\hat{p}_1-\hat{p}_2)\pm z^*\sqrt{\frac{\hat{p}_1(1-\hat{p}_1)}{n_1}+\frac{\hat{p}_2(1-\hat{p}_2)}{n_2}}(p^​1​−p^​2​)±z∗n1​p^​1​(1−p^​1​)​+n2​p^​2​(1−p^​2​)​​.

The critical value z∗z^*z∗ comes from the standard normal distribution and depends on the confidence level. A higher confidence level needs a larger critical value. Each sample contributes one term under the root, and larger samples make their terms smaller.

Common critical values

Confidence levelCritical value z∗z^*z∗
90%1.645
95%1.96
99%2.576
Confidence level
90%
Critical value z∗z^*z∗
1.645
Confidence level
95%
Critical value z∗z^*z∗
1.96
Confidence level
99%
Critical value z∗z^*z∗
2.576

Key terms · tap one for its definition

standard error

The standard error is the standard deviation of a statistic estimated from sample data, with sample proportions replacing the unknown population proportions.

critical value

The critical value z∗z^*z∗ is the standard normal value that captures the middle CCC% of the distribution for a confidence level of CCC%.

Part 2 done

Quick check

One question on the part you just read

Easy

Identify the general form of a two-proportion z-interval.
Show answer and explanation

Answer: C · Worked explanation

An interval is the statistic plus or minus a critical value times the standard error, and the statistic here is the difference p^1−p^2\hat{p}_1-\hat{p}_2p^​1​−p^​2​. Adding the sample proportions estimates nothing useful. Multiplying the difference by z∗z^*z∗ instead of the standard error scales the wrong piece.
3

Part 3 of 5: Conditions must be checked before computing the interval

A formula returns a number even when the model does not fit. So check three conditions first.

  • Random: each sample is random, or treatments were randomly assigned.
  • Independent: each sample is less than 10% of its population, and the two samples are independent of each other.
  • Large counts: each sample has at least 10 successes and at least 10 failures.

Check the counts with the observed data, because the true proportions are unknown. That means n1p^1n_1\hat{p}_1n1​p^​1​, n1(1−p^1)n_1(1-\hat{p}_1)n1​(1−p^​1​), n2p^2n_2\hat{p}_2n2​p^​2​, and n2(1−p^2)n_2(1-\hat{p}_2)n2​(1−p^​2​) must all reach 10.

Exam tip

Write out the four counts and compare each with 10. A single count below 10 breaks the condition.

Part 3 done

4

Part 4 of 5: The calculation takes four steps from counts to interval

Now put the formula to work on data. Follow the same steps each time. Check the conditions, find the statistic, find the standard error, then find the margin of error.

Worked example

A 95% interval for a difference of click proportions

In a random sample of 200 customers shown design A of a web page, 120 clicked a button. In an independent random sample of 200 customers shown design B, 90 clicked. Each group is less than 10% of its population. Construct a 95% confidence interval for p1−p2p_1-p_2p1​−p2​, where group 1 saw design A.
  1. 1
    Counts: 120, 80, 90, and 110 are all at least 10, so the conditions are met.
  2. 2
    Statistic: p^1=120200=0.60\hat{p}_1=\frac{120}{200}=0.60p^​1​=200120​=0.60 and p^2=90200=0.45\hat{p}_2=\frac{90}{200}=0.45p^​2​=20090​=0.45, so p^1−p^2=0.15\hat{p}_1-\hat{p}_2=0.15p^​1​−p^​2​=0.15.
  3. 3
    Standard error: 0.60(0.40)200+0.45(0.55)200=0.0024375≈0.0494\sqrt{\frac{0.60(0.40)}{200}+\frac{0.45(0.55)}{200}}=\sqrt{0.0024375}\approx0.04942000.60(0.40)​+2000.45(0.55)​​=0.0024375​≈0.0494.
  4. 4
    Margin of error: 1.96(0.0494)≈0.09681.96(0.0494)\approx0.09681.96(0.0494)≈0.0968. Interval: 0.15±0.09680.15\pm0.09680.15±0.0968.

Answer: The 95% interval is approximately (0.053, 0.247)(0.053,\,0.247)(0.053,0.247).

The order of subtraction sets the signs. Reversing the order would give (−0.247, −0.053)(-0.247,\,-0.053)(−0.247,−0.053), which carries the same information. Keep extra digits during the work and round only at the end. State the final interval with its endpoints in increasing order.

Exam tip

Name the order of subtraction in your answer, such as design A minus design B.

Part 4 done

Quick check

One question on the part you just read

Easy

Identify the critical value z∗z^*z∗ for a 95% confidence interval.
Show answer and explanation

Answer: D · Worked explanation

A 95% interval captures the middle 95% of the standard normal distribution, which needs z∗=1.96z^*=1.96z∗=1.96. The value 1.645 belongs to a 90% interval, and 2.576 belongs to a 99% interval. Higher confidence requires a larger critical value, because capturing more of the distribution means reaching farther from the center.
5

Part 5 of 5: An interval is interpreted in context with its confidence level

The final step is a sentence in context. A good interpretation names the confidence level, the endpoints, the two populations, and the order of subtraction.

For the example, write the following. We are 95% confident that the interval from 0.053 to 0.247 captures the true difference. The difference is between the proportions of all customers who would click, design A minus design B.

Every value in this interval is positive. So the plausible values say that design A has the larger click proportion.

Common mistake

The confidence level describes the method, not this one interval. About 95% of intervals built this way capture the true difference.

Note read

Check your recall[13 marks]

6 written questions. Answer each one in your head or on paper, then open the mark scheme.

  1. 1.Identify[1 mark]

    Identify the statistic that sits at the center of a confidence interval for p1−p2p_1-p_2p1​−p2​.

    Show mark schemeHide mark scheme

    Model answer

    The difference in sample proportions, p^1−p^2\hat{p}_1-\hat{p}_2p^​1​−p^​2​.

    • 1 mark for identifying p^1−p^2\hat{p}_1-\hat{p}_2p^​1​−p^​2​ as the center of the interval
  2. 2.Calculate[3 marks]

    In a random sample of 150 adults from city A, 90 favor a recycling program. In an independent random sample of 150 adults from city B, 60 favor it. Each sample is less than 10% of its city. Calculate a 95% confidence interval for p1−p2p_1-p_2p1​−p2​, where city A is group 1.

    Show mark schemeHide mark scheme

    Model answer

    The statistic is 0.60−0.40=0.200.60-0.40=0.200.60−0.40=0.20. The standard error is 0.60(0.40)150+0.40(0.60)150≈0.0566\sqrt{\frac{0.60(0.40)}{150}+\frac{0.40(0.60)}{150}}\approx0.05661500.60(0.40)​+1500.40(0.60)​​≈0.0566. The margin of error is 1.96(0.0566)≈0.1111.96(0.0566)\approx0.1111.96(0.0566)≈0.111, so the interval is about (0.089, 0.311)(0.089,\,0.311)(0.089,0.311).

    • 1 mark for calculating the standard error as about 0.0566
    • 1 mark for calculating the margin of error as about 0.111
    • 1 mark for giving the interval as about (0.089, 0.311)(0.089,\,0.311)(0.089,0.311)
  3. 3.Determine[2 marks]

    One random sample has 8 successes in 60 trials. An independent random sample has 35 successes in 70 trials. Determine whether the large counts condition for an interval is met.

    Show mark schemeHide mark scheme

    Model answer

    The counts are 8 and 52 for the first sample, and 35 and 35 for the second. Since 8 is below 10, the condition is not met.

    • 1 mark for computing the successes and failures in each sample correctly
    • 1 mark for concluding the condition is not met because 8 is below 10
  4. 4.Interpret[3 marks]

    Interpret the interval (0.089, 0.311)(0.089,\,0.311)(0.089,0.311) for p1−p2p_1-p_2p1​−p2​ in the context of the recycling program, where city A is group 1.

    Show mark schemeHide mark scheme

    Model answer

    We are 95% confident that the interval from 0.089 to 0.311 captures the true difference in the proportions of all adults who favor the program, city A minus city B. Every plausible value is positive, so city A has the larger proportion.

    • 1 mark for stating the confidence level and the interval endpoints
    • 1 mark for stating that the interval captures the true difference in proportions of all adults favoring the program
    • 1 mark for naming the order, city A minus city B, and noting that city A appears larger
  5. 5.Explain[2 marks]

    Explain why the standard error uses p^1\hat{p}_1p^​1​ and p^2\hat{p}_2p^​2​ instead of p1p_1p1​ and p2p_2p2​.

    Show mark schemeHide mark scheme

    Model answer

    The population proportions p1p_1p1​ and p2p_2p2​ are unknown. The sample proportions estimate them, so they replace them to estimate the standard deviation.

    • 1 mark for stating that p1p_1p1​ and p2p_2p2​ are unknown
    • 1 mark for explaining that sample proportions are substituted to estimate the standard deviation
  6. 6.Predict[2 marks]

    Predict how the width of an interval changes when the confidence level is raised from 95% to 99% with the same data.

    Show mark schemeHide mark scheme

    Model answer

    The interval becomes wider, because the critical value rises from 1.96 to 2.576, which increases the margin of error.

    • 1 mark for predicting that the interval becomes wider
    • 1 mark for linking the change to a larger critical value and margin of error

Practice this topic

10 quick-fire multiple-choice questions, easy to hard. Tap an answer to see if you're right and the worked explanation. (The written questions with mark schemes are under “Check your recall” above.)

0 right0 of 10 answered

Includes the 2 quick checks in the note above.

3.

Easy·Identify[1 mark]

Identify what a confidence interval for p1−p2p_1-p_2p1​−p2​ is estimating.
Show answer and explanation

Answer: B · Worked explanation

The interval gives plausible values for the true difference p1−p2p_1-p_2p1​−p2​ between population proportions. The sample difference p^1−p^2\hat{p}_1-\hat{p}_2p^​1​−p^​2​ is already known, since it sits at the center. It is the statistic used to build the interval, not the target. The unknown true difference is what we want to estimate, so the interval is built to capture it.
4.

Medium·Calculate[2 marks]

Random independent samples give p^1=0.45\hat{p}_1=0.45p^​1​=0.45 with n1=100n_1=100n1​=100 and p^2=0.30\hat{p}_2=0.30p^​2​=0.30 with n2=100n_2=100n2​=100. Calculate the margin of error for a 90% interval, using z∗=1.645z^*=1.645z∗=1.645.
Show answer and explanation

Answer: A · Worked explanation

The standard error is 0.45(0.55)100+0.30(0.70)100=0.004575≈0.0676\sqrt{\frac{0.45(0.55)}{100}+\frac{0.30(0.70)}{100}}=\sqrt{0.004575}\approx0.06761000.45(0.55)​+1000.30(0.70)​​=0.004575​≈0.0676. Then 1.645(0.0676)≈0.1111.645(0.0676)\approx0.1111.645(0.0676)≈0.111. The value 0.133 uses z∗=1.96z^*=1.96z∗=1.96 instead, and 0.068 is the standard error alone. The value 0.150 is the difference between the sample proportions, which is the center of the interval and not its margin of error.
5.

Medium·Identify[2 marks]

One random sample has 12 successes in 100 trials. An independent random sample has 7 successes in 100 trials. Identify the correct statement about the large counts condition.
Show answer and explanation

Answer: C · Worked explanation

Each sample needs at least 10 successes and at least 10 failures. The second sample has only 7 successes, so the condition fails. The first sample passes with 12 successes. Adding counts across groups, or pointing to large sample sizes, does not meet the condition.
6.

Medium·Identify[2 marks]

A 95% interval for p1−p2p_1-p_2p1​−p2​ is (0.02, 0.14)(0.02,\,0.14)(0.02,0.14), where p1p_1p1​ and p2p_2p2​ are the proportions of all men and all women who own a smartwatch. Identify the correct interpretation.
Show answer and explanation

Answer: A · Worked explanation

The interval is about the true difference in population proportions, with the order named. The sample difference is already known exactly, so it is not uncertain. Claims about individuals or about each sample proportion misstate what is being estimated, and a probability statement about the sample difference ignores that it is known.
7.

Medium·Identify[2 marks]

Identify the change that makes a 95% interval for p1−p2p_1-p_2p1​−p2​ narrower.
Show answer and explanation

Answer: D · Worked explanation

Larger samples reduce the standard error, which shrinks the margin of error. A 99% level uses a larger critical value, so it widens the interval. Smaller samples raise the standard error. Reversing the order only flips the signs and leaves the width unchanged.
8.

Hard·Determine[2 marks]

A 95% interval for p1−p2p_1-p_2p1​−p2​ is (0.05, 0.25)(0.05,\,0.25)(0.05,0.25). Determine the 95% interval for p2−p1p_2-p_1p2​−p1​ computed from the same data.
Show answer and explanation

Answer: B · Worked explanation

Reversing the order of subtraction flips each sign and swaps the endpoints, giving (−0.25, −0.05)(-0.25,\,-0.05)(−0.25,−0.05). The same interval would be wrong because the difference now points the other way. The pair (0.75, 0.95)(0.75,\,0.95)(0.75,0.95) treats the endpoints like proportions, subtracting them from 1, which is not valid for a difference.
9.

Hard[2 marks]

A 95% interval for p1−p2p_1-p_2p1​−p2​ is (0.053, 0.247)(0.053,\,0.247)(0.053,0.247). Which statement correctly describes what 95% means?
Show answer and explanation

Answer: C · Worked explanation

The confidence level describes how often the method works: across many repeated pairs of samples, about 95% of the intervals capture the true difference. Individual people and sample differences are not what the interval describes. The true difference is a fixed number, not a random value.
10.

Hard·Determine[2 marks]

A random sample of 90 drivers from region 1 has 54 who wear seat belts. An independent random sample of 120 drivers from region 2 has 36 who wear them. Each sample is less than 10% of its population. Determine the 95% confidence interval for p1−p2p_1-p_2p1​−p2​.
Show answer and explanation

Answer: D · Worked explanation

The statistic is 0.60−0.30=0.300.60-0.30=0.300.60−0.30=0.30. The standard error is 0.60(0.40)90+0.30(0.70)120=0.0044167≈0.0665\sqrt{\frac{0.60(0.40)}{90}+\frac{0.30(0.70)}{120}}=\sqrt{0.0044167}\approx0.0665900.60(0.40)​+1200.30(0.70)​​=0.0044167​≈0.0665, so the margin of error is 1.96(0.0665)≈0.1301.96(0.0665)\approx0.1301.96(0.0665)≈0.130. The interval is (0.170, 0.430)(0.170,\,0.430)(0.170,0.430). The interval (0.191, 0.409)(0.191,\,0.409)(0.191,0.409) uses z∗=1.645z^*=1.645z∗=1.645, which is a 90% interval and is too narrow for the stated confidence level. Forgetting the square root in the standard error would give a far narrower interval still.

Keep going

Next topic3.11 Claims from a difference intervalPrevious topic3.9 Differences in sample proportions

More in Unit 3 · Inference for Categorical Data: Proportions

  1. 3.6p-valuesNotes · 10 questions
  2. 3.7Carrying out a test for a proportionNotes · 10 questions
  3. 3.8Errors in testsNotes · 10 questions
  4. 3.9Differences in sample proportionsNotes · 10 questions
  5. 3.10Interval for a difference of proportionsNotes · 10 questions
  6. 3.11Claims from a difference intervalNotes · 10 questions
  7. 3.12Setting up a two-proportion testNotes · 10 questions
  8. 3.13Carrying out a two-proportion testNotes · 10 questions
  9. 3.14Setting up a chi-square test for two-way tablesNotes · 10 questions

All AP Statistics topics →

Draft: written by Scribe and checked automatically, not yet by a subject reviewer.

Practice questions are original, written by Scribe and checked automatically for correctness.

Topic codes follow the official numbering. Not endorsed by College Board or any exam board.

On this page

0/5 read

  1. 1A confidence interval estimates the gap between two proportions
  2. 2The standard error replaces unknown proportions with sample proportions
  3. 3Conditions must be checked before computing the interval
  4. 4The calculation takes four steps from counts to interval
  5. 5An interval is interpreted in context with its confidence level
  6. Check your recall
  7. Practice: 10 questions
All coursesScribeHow it worksPricingPrivacyTerms

Notes and questions are written by Scribe in our own words. Topic codes follow the official course numbering so you can find your place. Not produced or endorsed by the IB, College Board, AQA, OCR, Pearson Edexcel or any exam board.

© 2026 Scribe