Σ
SigmaPrep
AP Statistics worksheetsBlog
← All articles
AP StatisticsAP Exams

AP Statistics Unit 4 Review: The Most-Tested Inference for Means Questions

Σ
Sigma Prep
Math Instructor · 10+ Years Experience
November 2026·13 min read
Key takeaways
  • Unit weight: a little under a fifth of the multiple choice, 76 of the 406 official questions on today’s course (18.7%), but only about a tenth of the free response: about 102 of the 1,040 official points we counted (9.8%).
  • Top types: the sampling distribution of xˉ\bar{x}xˉ, hypotheses and conclusions for a one-sample or paired ttt-test, and choosing between a one-sample, paired and two-sample procedure.
  • Practice: Unit 4 is 15.0% of the practice questions we tagged but 18.7% of the official ones. Stating a test’s conclusion and interpreting the confidence level get a tenth to a third of their share, while computing ttt intervals and one-sample ttt statistics is over-drilled.

Unit 4 is a little under a fifth of the AP Statistics multiple choice, and the means half of the exam’s inference. College Board’s Course and Exam Description (the CED) gives it 10–20% of the multiple choice. On the six official practice exams from 2012–2019 it was 40 of 215 questions (18.6%), 4 to 8 per exam. In the 24 sample questions of the 2026 CED, the only official questions in the new format, it is 4.

The unit gathers the means half of the old inference units: the sampling distribution of xˉ\bar{x}xˉ, one-sample and paired ttt intervals and tests, and the same for two means. The sampling distribution of xˉ1−xˉ2\bar{x}_1 - \bar{x}_2xˉ1​−xˉ2​ is a new topic (4.6), and inference for the slope of a regression line is no longer in the course.

The exam changes in May 2027. Section I becomes 42 multiple-choice questions with four choices instead of five, and some come in sets: two or three questions on one shared stimulus (6 of the 24 CED sample questions are in 2 sets). A sample with its mean and standard deviation, then an interval and a claim, is a natural set. Section II becomes 4 free-response questions of 10 points each, scored point by point. At the 2012–2019 rate, about 8 of the 42 multiple-choice questions would be Unit 4; the CED’s 10–20% range puts it at about 4 to 8.

On the free response, Unit 4 earned about 102 of the 1,040 points on the 124 official questions we counted, and 23 of those 124 questions had a Unit 4 part. In the four 2026 sample questions it is 6 of the 40 points, all in sample Question 4.

Unit 4 on the free response: Questions 3 and 4

From 2027 the free-response questions come in a fixed order: Question 1 assesses Practices 1 and 2 (the investigative question and how data are collected), Question 2 Practices 3 and 4 (analyzing data and drawing conclusions), Question 3 inference, and Question 4 a mix of Practices 2, 3 and 4. Question 3 is a full significance test or confidence interval, from the conditions to the conclusion.

Of the 21 official free-response questions that fit the new Question 3, 7 had a Unit 4 part; of the 22 that fit Question 4, 8 did. Inference for means was the setting of 7 operational free-response questions from 2012 to 2026, 3 of them since 2019, most recently in 2026. The operational ones that fit Question 3 were 2 matched-pairs ttt tests (2014 and 2023), 2 two-sample ttt tests (2018 and 2026) and 1 one-sample ttt interval (2013). The 2026 sample Question 4 builds to a two-sample ttt interval for a difference in means.

The 10 most-tested Unit 4 question types

Ranked by the number of official multiple-choice questions on today’s course (13 College Board sources, 1997–2026), with how many of the 13 sources had one, how many were on the six 2012–2019 practice exams, and how many of the 84 operational free-response questions since 2012 had a part on it.

1. The sampling distribution of xˉ\bar{x}xˉ: shape, mean and standard deviation

The mean of xˉ\bar{x}xˉ is μ\muμ, its standard deviation is σn\frac{\sigma}{\sqrt{n}}n​σ​, and it is approximately normal when the population is normal or the sample is large.

Topic 4.17 official questions on 6 of 13 sources, one of them a 2026 CED sample5 on the 2012–2019 practice exams4 FRQs since 2012, 1 of them since 2019
Try it
The hold time of calls to an electric company's help line has a distribution that is skewed to the right, with mean 8 minutes and standard deviation 5 minutes. A random sample of 50 calls will be selected, and the sample mean xˉ\bar{x}xˉ will be computed. Which of the following best describes the sampling distribution of xˉ\bar{x}xˉ?
  1. Approximately normal, with mean 8 minutes and standard deviation 0.100 minutes
  2. Approximately normal, with mean 8 minutes and standard deviation 0.707 minutes
  3. Skewed to the right, with mean 8 minutes and standard deviation 0.707 minutes
  4. Approximately normal, with mean 8 minutes and standard deviation 5 minutes
Show solutionHide solution
Solution

(B) Approximately normal, with mean 8 minutes and standard deviation 0.707 minutes. The mean of xˉ\bar{x}xˉ is μ=8\mu = 8μ=8 minutes and its standard deviation is 550=0.707\frac{5}{\sqrt{50}} = 0.70750​5​=0.707 minutes. The population is skewed to the right, but n=50n = 50n=50 is large, so by the central limit theorem the sampling distribution of xˉ\bar{x}xˉ is approximately normal.

Common mistake
dividing σ\sigmaσ by nnn instead of n\sqrt{n}n​: 550=0.100\frac{5}{50} = 0.100505​=0.100 (A). And (C) keeps the population’s skew: with 50 calls the central limit theorem applies. For a small sample from a skewed population it would not, and the shape would stay skewed.

Practice: Sampling Distributions worksheet · AP Statistics Unit 4 worksheet

2. Hypotheses for a one-sample or paired ttt-test

State H0H_0H0​ and HaH_aHa​ about the parameter, μ\muμ or μd\mu_dμd​, with the null value from the question and the direction from what is suspected.

Topic 4.46 official questions on 6 of 13 sources3 on the 2012–2019 practice exams3 FRQs since 2012, 2 of them since 2019
Try it
15 adult volunteers each took a computer reaction-time test twice, once in silence and once with loud music playing, in a random order. Their reaction times, in milliseconds, are shown below. Let μd\mu_dμd​ be the true mean difference in reaction time (music minus silence) for adults like these volunteers. Which of the following are appropriate hypotheses for a test of whether there is a difference?
{"cols":["Silence","Music"],"pairs":[[236,205],[271,304],[295,254],[249,273],[276,295],[258,257],[248,206],[330,364],[323,318],[324,339],[275,280],[252,201],[297,311],[274,267],[258,273]],"diff":true}VolunteerSilenceMusicDifference (Music − Silence)1236205-312271304333295254-414249273245276295196258257-17248206-428330364349323318-5103243391511275280512252201-51132973111414274267-71525827315
  1. H0:μd=0H_0: \mu_d = 0H0​:μd​=0 and Ha:μd≠0H_a: \mu_d \neq 0Ha​:μd​=0
  2. H0:μd=−1.27H_0: \mu_d = -1.27H0​:μd​=−1.27 and Ha:μd≠−1.27H_a: \mu_d \neq -1.27Ha​:μd​=−1.27
  3. H0:xˉd=0H_0: \bar{x}_d = 0H0​:xˉd​=0 and Ha:xˉd≠0H_a: \bar{x}_d \neq 0Ha​:xˉd​=0
  4. H0:μd≠0H_0: \mu_d \neq 0H0​:μd​=0 and Ha:μd=0H_a: \mu_d = 0Ha​:μd​=0
Show solutionHide solution
Solution

(A) H0:μd=0H_0: \mu_d = 0H0​:μd​=0 and Ha:μd≠0H_a: \mu_d \neq 0Ha​:μd​=0. Each volunteer took the test twice, so the parameter is μd\mu_dμd​, the true mean difference in reaction time (music minus silence). “No difference” is H0:μd=0H_0: \mu_d = 0H0​:μd​=0, and a test of whether there is a difference in either direction is two-sided: Ha:μd≠0H_a: \mu_d \neq 0Ha​:μd​=0.

Common mistake
putting the sample into H0H_0H0​. (B) tests μd=−1.27\mu_d = -1.27μd​=−1.27, the mean of these 15 differences; the null value comes from the question (“no difference”), not from the data. (C) writes the hypotheses about xˉd\bar{x}_dxˉd​, a statistic that is already known.

Practice: Hypothesis Tests worksheet · AP Statistics Unit 4 worksheet

3. The conclusion of a one-sample or paired ttt-test

Compare the P-value with α\alphaα, decide, and state the conclusion about the population mean, in context and in terms of HaH_aHa​.

Topic 4.56 official questions on 6 of 13 sources, one of them a 2026 CED sample3 on the 2012–2019 practice exams2 FRQs since 2012, 1 of them since 2019
Try it
A regional airport promises that the first checked bag reaches the carousel 15 minutes after a flight arrives, on average. The airport's operations manager suspects that the mean wait for the first checked bag of all arriving flights at a regional airport is greater than 15 minutes, and selected a random sample of 30 flights and recorded the wait for the first checked bag of each. The sample mean was 15.18 minutes, and the sample standard deviation was 2.49 minutes. The airport's operations manager will test H0:μ=15H_0: \mu = 15H0​:μ=15 against Ha:μ>15H_a: \mu > 15Ha​:μ>15, where μ\muμ is the mean wait for the first checked bag of all arriving flights at a regional airport. All conditions for inference were met, and the P-value of the test was 0.3475. Which of the following is an appropriate conclusion at the α=0.05\alpha = 0.05α=0.05 level?
  1. The P-value of 0.3475 is greater than α=0.05\alpha = 0.05α=0.05, so the null hypothesis is not rejected. There is convincing statistical evidence that the mean wait for the first checked bag of all arriving flights at a regional airport is 15 minutes.
  2. The P-value of 0.3475 is greater than α=0.05\alpha = 0.05α=0.05, so the null hypothesis is rejected. There is convincing statistical evidence that the mean wait for the first checked bag of all arriving flights at a regional airport is greater than 15 minutes.
  3. The P-value of 0.3475 is greater than α=0.05\alpha = 0.05α=0.05, so the null hypothesis is not rejected. There is not convincing statistical evidence that the mean wait for the first checked bag of all arriving flights at a regional airport is greater than 15 minutes.
  4. The P-value of 0.3475 is greater than α=0.05\alpha = 0.05α=0.05, so the null hypothesis is not rejected. There is convincing statistical evidence that the mean wait for the first checked bag of all arriving flights at a regional airport is greater than 15 minutes.
Show solutionHide solution
Solution

(C) The P-value of 0.3475 is greater than α=0.05\alpha = 0.05α=0.05, so the null hypothesis is not rejected. There is not convincing statistical evidence that the mean wait for the first checked bag of all arriving flights at a regional airport is greater than 15 minutes. The P-value, 0.3475, is greater than α=0.05\alpha = 0.05α=0.05, so H0H_0H0​ is not rejected. The conclusion is about the population mean and in terms of HaH_aHa​: there is not convincing statistical evidence that the mean wait for the first checked bag of all arriving flights is greater than 15 minutes.

Common mistake
accepting H0H_0H0​ (A). Failing to reject does not show that the mean wait is 15 minutes; a sample mean of 15.18 is consistent with many values near 15. (D) makes the right decision and then states the evidence backwards.

Practice: Hypothesis Tests worksheet · AP Statistics Unit 4 worksheet

4. Choosing a one-sample or matched-pairs procedure

Two measurements on the same units are paired: analyze the differences with a one-sample ttt procedure. A question about whether something changed calls for a test; one about how much calls for an interval.

Topic 4.46 official questions on 5 of 13 sources5 on the 2012–2019 practice exams1 FRQ since 2012 (2024)
Try it
A district selected a random sample of 20 seventh graders, who each took a 50-point vocabulary quiz before and after using a word-learning app for a month. The district wants to know whether the vocabulary quiz score tends to be higher after. Which of the following is the most appropriate procedure?
  1. A two-sample ttt-test for a difference between two population means
  2. A matched-pairs ttt-test for a mean difference
  3. A matched-pairs ttt-interval for a mean difference
  4. A two-sample zzz-test for a difference between two population proportions
Show solutionHide solution
Solution

(B) A matched-pairs ttt-test for a mean difference. Each student took the quiz twice, before and after, so the two sets of scores are paired. The district asks whether scores tend to be higher, which is a test, so the procedure is a matched-pairs ttt-test on the 20 differences (after minus before).

Common mistake
a two-sample ttt-test (A). The before and after scores come from the same 20 students, so they are not independent samples; the pairing removes the student-to-student variation. (C) uses the right data but estimates the size of the change instead of testing for one.

Practice: Hypothesis Tests worksheet · AP Statistics Unit 4 worksheet

5. Interpreting the confidence level

The confidence level is the long-run share of intervals, from repeated random samples, that capture the population mean.

Topic 4.35 official questions on 5 of 13 sources, one of them a 2026 CED sample3 on the 2012–2019 practice examsno FRQ since 2012
Try it
To estimate the mean noise level of all lunch periods in the cafeterias of a large district, the district's facilities director selected a random sample of 23 lunch periods and recorded the noise level of each. The sample mean was 77.54 decibels, and the sample standard deviation was 4.55 decibels. A 95% confidence interval for the mean noise level of all lunch periods in the cafeterias of a large district is (75.57,79.51)(75.57, 79.51)(75.57,79.51). Which of the following is a correct interpretation of the 95% confidence level?
  1. If many random samples of 23 lunch periods were selected and an interval were built the same way from each, about 95% of the intervals would capture the mean noise level of all lunch periods in the cafeterias of a large district.
  2. About 95% of all lunch periods in the cafeterias of a large district have a noise level between 75.57 and 79.51 decibels.
  3. About 95% of all random samples of 23 lunch periods will have a sample mean between 75.57 and 79.51 decibels.
  4. If many random samples of 23 lunch periods were selected and an interval were built the same way from each, about 95% of the intervals would contain 77.54 decibels.
Show solutionHide solution
Solution

(A) If many random samples of 23 lunch periods were selected and an interval were built the same way from each, about 95% of the intervals would capture the mean noise level of all lunch periods in the cafeterias of a large district. The 95% describes the method, not this one interval: if many random samples of 23 lunch periods were taken and an interval built the same way from each, about 95% of those intervals would capture the true mean noise level.

Common mistake
reading the level as a statement about individual lunch periods (B) or about sample means (C). Single lunch periods vary far more than the interval, and the intervals aim at μ\muμ. (D) aims them at this sample’s mean, 77.54 decibels, instead of the population mean.

Practice: Confidence Intervals worksheet · AP Statistics Unit 4 worksheet

6. Carry out a one-sample or paired ttt-test

The test statistic t=xˉ−μ0s/nt = \frac{\bar{x} - \mu_0}{s/\sqrt{n}}t=s/n​xˉ−μ0​​ with n−1n - 1n−1 degrees of freedom; for paired data, the same on the differences.

Topic 4.54 official questions on 4 of 13 sources3 on the 2012–2019 practice exams2 FRQs since 2012, 1 of them since 2019
Try it
In a physical education class, 14 student volunteers each stood on one foot as long as they could, once with eyes open and once with eyes closed, in a random order. The times, in seconds, are shown below. Let μd\mu_dμd​ be the true mean difference in time (eyes closed minus eyes open) for students like these volunteers. The hypotheses are H0:μd=0H_0: \mu_d = 0H0​:μd​=0 and Ha:μd<0H_a: \mu_d < 0Ha​:μd​<0. Assume the conditions for inference are met. What is the value of the test statistic?
{"cols":["Eyes open","Eyes closed"],"pairs":[[38.5,42.2],[41.3,43],[30.9,13.8],[45.2,47.8],[37,34.4],[38,30.1],[45,39.2],[47,30.4],[39.1,31.9],[33.9,24.9],[39.5,39.3],[46.1,29],[39.2,24.6],[35,30.7]],"diff":true}StudentEyes openEyes closedDifference (Eyes closed − Eyes open)138.542.23.7241.343.01.7330.913.8-17.1445.247.82.6537.034.4-2.6638.030.1-7.9745.039.2-5.8847.030.4-16.6939.131.9-7.21033.924.9-9.01139.539.3-0.21246.129.0-17.11339.224.6-14.61435.030.7-4.3
  1. −12.73-12.73−12.73
  2. −3.40-3.40−3.40
  3. −2.50-2.50−2.50
  4. −0.91-0.91−0.91
Show solutionHide solution
Solution

(B) −3.40-3.40−3.40. Work with the 14 differences (eyes closed minus eyes open): xˉd=−6.743\bar{x}_d = -6.743xˉd​=−6.743 and sd=7.414s_d = 7.414sd​=7.414 seconds. The standard error is 7.41414=1.981\frac{7.414}{\sqrt{14}} = 1.98114​7.414​=1.981, so t=−6.743−01.981=−3.40t = \frac{-6.743 - 0}{1.981} = -3.40t=1.981−6.743−0​=−3.40 with df=13df = 13df=13.

Common mistake
treating the two columns as independent samples. The two-sample ttt for these data is −2.50-2.50−2.50 (C), but the same students stood with eyes open and closed, so the test uses the 14 differences. (A) divides sds_dsd​ by nnn instead of n\sqrt{n}n​.

Practice: Hypothesis Tests worksheet · AP Statistics Unit 4 worksheet

7. Carry out a two-sample ttt-test

The unpooled standard error s12n1+s22n2\sqrt{\frac{s_1^2}{n_1} + \frac{s_2^2}{n_2}}n1​s12​​+n2​s22​​​, the ttt statistic and its P-value, with the degrees of freedom from technology.

Topic 4.104 official questions on 3 of 13 sources1 on the 2012–2019 practice exams2 FRQs since 2012, 1 of them since 2019
Try it
A baker wants to know whether a new yeast makes dough rise faster than the yeast she uses now. 45 batches of dough were randomly assigned to the new yeast or the current yeast. The computer output below summarizes the rising time, in minutes. The baker will test H0:μ1−μ2=0H_0: \mu_1 - \mu_2 = 0H0​:μ1​−μ2​=0 against Ha:μ1−μ2<0H_a: \mu_1 - \mu_2 < 0Ha​:μ1​−μ2​<0, where μ1\mu_1μ1​ is the mean rising time of dough made with the new yeast and μ2\mu_2μ2​ is the mean rising time of dough made with the current yeast. What is the value of the two-sample ttt statistic?
{"kind":"twot","labels":["New","Current"],"alt":"less","C":0.95,"full":false}Two-Sample T-Test and CI: New, CurrentSampleNMeanStDevSE MeanNew2554.46.61.32Current2060.45.01.12
  1. −4.86-4.86−4.86
  2. −3.47-3.47−3.47
  3. −2.46-2.46−2.46
  4. 3.473.473.47
Show solutionHide solution
Solution

(B) −3.47-3.47−3.47. The output gives each group’s standard error, so SE=1.322+1.122=1.73SE = \sqrt{1.32^2 + 1.12^2} = 1.73SE=1.322+1.122​=1.73 and t=54.4−60.41.73=−3.47t = \frac{54.4 - 60.4}{1.73} = -3.47t=1.7354.4−60.4​=−3.47, with df=42.90df = 42.90df=42.90 from technology.

Common mistake
adding the standard errors instead of their squares: 54.4−60.41.32+1.12=−2.46\frac{54.4 - 60.4}{1.32 + 1.12} = -2.461.32+1.1254.4−60.4​=−2.46 (C). Variances add; standard deviations do not. (D) subtracts in the other order: follow the order in the hypotheses, new minus current.

Practice: Hypothesis Tests worksheet · AP Statistics Unit 4 worksheet

8. Using a confidence interval for a mean to judge a claim

A claimed value inside the interval is plausible; only a value outside it is ruled out.

Topic 4.34 official questions on 4 of 13 sources1 on the 2012–2019 practice exams1 FRQ since 2012 (2025)
Try it
To estimate the mean noise level of all lunch periods in the cafeterias of a large district, the district's facilities director selected a random sample of 50 lunch periods and recorded the noise level of each. The sample mean was 71.83 decibels, and the sample standard deviation was 3.61 decibels. A 90% confidence interval for the mean noise level of all lunch periods in the cafeterias of a large district is (70.97,72.69)(70.97, 72.69)(70.97,72.69). The district's facilities director has heard a claim that the mean noise level of all lunch periods in the cafeterias of a large district is more than 71.7 decibels. Which of the following is the best conclusion about the claim?
  1. Because 71.7 is inside the interval and so is a plausible value of the mean, the interval does not give convincing evidence that the mean noise level of all lunch periods in the cafeterias of a large district is more than 71.7 decibels.
  2. Because 71.7 is inside the interval, the interval shows that the mean noise level of all lunch periods in the cafeterias of a large district is 71.7 decibels.
  3. Because the sample mean, 71.83, is greater than 71.7, the sample gives convincing evidence that the mean noise level of all lunch periods in the cafeterias of a large district is more than 71.7 decibels.
  4. Because most of the interval is above 71.7, the interval gives convincing evidence that the mean noise level of all lunch periods in the cafeterias of a large district is more than 71.7 decibels.
Show solutionHide solution
Solution

(A) Because 71.7 is inside the interval and so is a plausible value of the mean, the interval does not give convincing evidence that the mean noise level of all lunch periods in the cafeterias of a large district is more than 71.7 decibels. 71.7 is inside the interval from 70.97 to 72.69, so it is a plausible value of the mean noise level. The interval does not give convincing evidence that the mean is more than 71.7 decibels, even though most of the interval lies above 71.7.

Common mistake
judging the claim from the sample mean. 71.83 is above 71.7 (C), but another sample would give another mean; the interval already allows for that. (B) goes too far the other way: 71.7 is plausible, not proven, and so is every other value in the interval.

Practice: Confidence Intervals worksheet · AP Statistics Unit 4 worksheet

9. Probability about a sample mean

Find a probability for xˉ\bar{x}xˉ from the normal model with mean μ\muμ and standard deviation σn\frac{\sigma}{\sqrt{n}}n​σ​.

Topic 4.14 official questions on 4 of 13 sources1 on the 2012–2019 practice exams1 FRQ since 2012 (2023)
Try it
The daily milk amount of cows in a large dairy herd has a distribution that is approximately normal, with mean 30 liters and standard deviation 6 liters. A random sample of 25 cows will be selected. What is the approximate probability that the mean daily milk amount of the 25 cows is greater than 30.61 liters?
  1. 0.19440.19440.1944
  2. 0.30560.30560.3056
  3. 0.45950.45950.4595
  4. 0.69440.69440.6944
Show solutionHide solution
Solution

(B) 0.30560.30560.3056. The population is normal, so xˉ\bar{x}xˉ is normal with mean 30 and standard deviation 625=1.2\frac{6}{\sqrt{25}} = 1.225​6​=1.2 liters. Then z=30.61−301.2=0.51z = \frac{30.61 - 30}{1.2} = 0.51z=1.230.61−30​=0.51, and P(xˉ>30.61)=0.3056P(\bar{x} > 30.61) = 0.3056P(xˉ>30.61)=0.3056.

Common mistake
using σ=6\sigma = 6σ=6 for the mean of 25 cows: normalcdf(30.61,∞,30,6)=0.4595\text{normalcdf}(30.61, \infty, 30, 6) = 0.4595normalcdf(30.61,∞,30,6)=0.4595 (C). One cow’s milk varies with standard deviation 6 liters; the mean of 25 cows varies with standard deviation 1.2. (D) is the other tail.

Practice: Sampling Distributions worksheet · AP Statistics Unit 4 worksheet

10. Interpreting a confidence interval for a mean

Name the confidence level, the interval and the population mean, in context.

Topic 4.34 official questions on 4 of 13 sources2 on the 2012–2019 practice examsno FRQ since 2012
Try it
To estimate the mean wait for the first checked bag of all arriving flights at a regional airport, the airport's operations manager selected a random sample of 19 flights and recorded the wait for the first checked bag of each. The sample mean was 17.89 minutes, and the sample standard deviation was 2.65 minutes. A 95% confidence interval for the mean wait for the first checked bag of all arriving flights at a regional airport is (16.61,19.17)(16.61, 19.17)(16.61,19.17). Which of the following is a correct interpretation of the interval?
  1. We are 95% confident that the interval from 16.61 to 19.17 minutes captures the mean wait for the first checked bag of all arriving flights at a regional airport.
  2. About 95% of all random samples of 19 flights will have a sample mean between 16.61 and 19.17 minutes.
  3. There is a 95% probability that the mean wait for the first checked bag of all arriving flights at a regional airport is between 16.61 and 19.17 minutes.
  4. About 95% of all arriving flights at a regional airport have a wait for the first checked bag between 16.61 and 19.17 minutes.
Show solutionHide solution
Solution

(A) We are 95% confident that the interval from 16.61 to 19.17 minutes captures the mean wait for the first checked bag of all arriving flights at a regional airport. The interval estimates μ\muμ, the mean wait for the first checked bag of all arriving flights at the airport: we are 95% confident that the interval from 16.61 to 19.17 minutes captures it.

Common mistake
describing individual flights (D). The sample’s standard deviation is 2.65 minutes, so single waits spread far beyond 16.61 to 19.17; the interval is about the mean. (C) puts a probability on the population mean, which is fixed: this interval either captures it or not.

Practice: Confidence Intervals worksheet · AP Statistics Unit 4 worksheet

On the free response: what multiple choice rarely asks

  • The conclusion of a two-sample ttt-test. 2 official multiple-choice questions, but 2 operational free-response questions, both since 2019 (most recently in 2026). State it about the population (or true treatment) means, and say whether the design allows a cause-and-effect conclusion.
  • Conditions for a one-sample ttt procedure. 1 official multiple-choice question, 2 operational free-response questions, most recently in 2024. With n<30n < 30n<30, look at a graph of the data (or the differences) for strong skewness or outliers, and say what you see.
  • Hypotheses for a two-sample ttt-test. 1 official multiple-choice question; the 2026 exam asked for them on the free response. Define μ1\mu_1μ1​ and μ2\mu_2μ2​ in context and keep the order of subtraction.
  • Conditions for two-sample ttt procedures. 1 official multiple-choice question, but 4 official free-response questions with a part on them (1 operational, 3 on the practice exams). Each sample needs n≥30n \ge 30n≥30 or data with no strong skewness or outliers.

Rising and fading in Unit 4

We compare the 42 operational free-response questions of 2012–2018 with the 42 of 2019–2026 (both 7 exams; 2020 was never released).

  • Unit 4 as a whole is steady on the free response: 8.5% of the free-response points in 2012–2018, 9.4% in 2019–2026 (7 questions with a Unit 4 part in each period).
  • Rising: two-sample decisions. The conclusion of a two-sample ttt-test went from no question to 2, and choosing a two-sample procedure and its hypotheses each from none to 1. Hypotheses for one mean went from 1 to 2.
  • Steady: carrying out the tests. The one-sample (or paired) ttt-test and the two-sample ttt-test were each in 1 question in each period.
  • Fading: the sampling distribution of xˉ\bar{x}xˉ on its own. It went from 3 questions to 1, and the one-sample ttt interval and two-sample conditions each from 1 to none.

New and never asked yet: the sampling distribution of xˉ1−xˉ2\bar{x}_1 - \bar{x}_2xˉ1​−xˉ2​

Topic 4.6 is new in the 2026 CED, and no official question has asked about it yet. It works like the difference of two proportions in Unit 3: the mean of xˉ1−xˉ2\bar{x}_1 - \bar{x}_2xˉ1​−xˉ2​ is μ1−μ2\mu_1 - \mu_2μ1​−μ2​, and for independent samples the variances add, so its standard deviation is σ12n1+σ22n2\sqrt{\frac{\sigma_1^2}{n_1} + \frac{\sigma_2^2}{n_2}}n1​σ12​​+n2​σ22​​​. Our free-response set has a part on it.

What students over- and under-practice in Unit 4

We tagged 1,130 multiple-choice questions from teacher tests, review packets, AP Classroom exports and prep books (reposted official questions removed) and compared each type’s share of that practice with its share of the 406 official questions. Unit 4 is 15.0% of the practice questions against 18.7% of the official ones.

Unit 4 question types: share of 1,130 practice multiple-choice questions against share of the 406 official ones
Question typePractice vs official shareRatio
Construct a two-sample ttt interval0.80% vs 0.25%3.2× over
Construct a one-sample ttt interval1.33% vs 0.74%1.8× over
Carry out a one-sample or paired ttt-test1.50% vs 0.99%1.5× over
Sampling distribution of xˉ\bar{x}xˉ2.12% vs 1.72%1.2× over
Sample size for a margin of error0.09% vs 0.99%0.1× under
Conclusion of a one-sample or paired ttt-test0.18% vs 1.48%0.1× under
Interpret the confidence level0.35% vs 1.23%0.3× under
Hypotheses for a one-sample or paired ttt-test0.62% vs 1.48%0.4× under
Choose a one-sample or paired procedure0.62% vs 1.48%0.4× under

Official counts per type are small (1 to 7 questions), so read the ratios as directions.

  • Computing intervals and test statistics is over-drilled. The two-sample ttt interval has only 1 official multiple-choice question but gets three times that share of practice.
  • The words around a test are under-practiced. The conclusion, the hypotheses and the choice of procedure each have 6 official questions, among the most in the unit, but get a tenth to under half of their share.
  • Interpreting the confidence level gets a third of its share although 5 of the 13 sources asked about it, one of them the 2026 CED.

How we counted

Every number here comes from public official AP Statistics material: 406 multiple-choice questions on today’s course from 13 College Board sources (the 1997, 2002 and 2007 released exams, the 2008, 2012–2015, 2018 and 2019 practice exams, and the sample questions in the 2005, 2020 and 2026 course descriptions), and 124 free-response questions (the 2012–2026 exams except 2020, the practice exams and the four 2026 samples; 1,040 points). Free-response points before 2027 are split from each question’s holistic score in proportion to its scoring sections. Each question is tagged to its CED topic and question type. Every example problem is our own.

Practice Unit 4

Our free AP Statistics Unit 4 worksheet has 20 original multiple-choice questions with four choices, including a set of questions on one shared stimulus (a sample’s mean and standard deviation, then a one-sample ttt interval and a claim), weighted by how often the official exams ask each type. It comes with an answer key, worked solutions and a printable PDF.

For free response, the AP Statistics Unit 4 FRQ set has 3 questions in the style of Questions 2, 3 and 4 of the 2027 exam, including a full two-sample ttt-test like Question 3, each scored out of 10 points, with a scoring guide and a PDF. Both make good practice before the Unit 4 Progress Check in AP Classroom.

The formulas this unit uses, each with a worked example and the calculator command, are on our free AP Statistics formula sheet.

Unit 4 follows the same steps as Unit 3’s inference for proportions, with ttt in place of zzz: see the AP Statistics Unit 3 review and the Unit 3 worksheet.

← Previous unitUnit 3: Inference for Categorical Data: ProportionsNext unit →Unit 5: Regression Analysis

Common Questions

The CED gives Unit 4 10–20% of the multiple choice. On the six official practice exams from 2012–2019 it was 40 of 215 multiple-choice questions (18.6%), 4 to 8 per exam. On the free response it earned about 102 of the 1,040 points on the 124 official questions we counted (9.8%), and 23 of those questions had a Unit 4 part.

Mostly Questions 3 and 4. Question 3 is a full inference question; of the 21 official free-response questions that fit it, 7 had a Unit 4 part, matched-pairs and two-sample t tests most often. Of the 22 that fit Question 4, 8 had a Unit 4 part, and the 2026 sample Question 4 builds to a two-sample t interval.

TInterval and T-Test for one mean (run them on the list of differences for paired data), 2-SampTInt and 2-SampTTest for two means (not pooled), and invT and tcdf for critical values and P-values. On the free response, name the procedure, state the hypotheses and check the conditions in writing; a calculator result alone does not earn those points.

Start with the most-tested question types above and the common mistake for each, then take a timed 20-question practice test with an answer key (about 2 minutes a question, as on the exam) and the unit’s free-response questions with a scoring guide. Read the worked solution of every question you missed, then redo those types.

Practice this unit: free AP Statistics Unit 4 worksheet

Exam-style multiple choice in the 2027 format, with an answer key and a printable PDF. No sign-up needed.

Get the Unit 4 worksheet →

Keep reading

AP Statistics

AP Statistics Unit 1 Review: The Most-Tested Data and Study Design Questions

AP Statistics Unit 1 review for the 2027 exam: the data and study design questions the official exams ask most, with examples and a practice test.

AP Statistics

AP Statistics Unit 2 Review: The Most-Tested Probability and Distributions Questions

AP Statistics Unit 2 review for the 2027 exam: the probability and distribution questions the official exams ask most, with examples and a practice test.

AP Statistics

AP Statistics Unit 3 Review: The Most-Tested Inference for Proportions Questions

AP Statistics Unit 3 review for the 2027 exam: the inference for proportions questions the official exams ask most, with examples and a practice test.

Σ
SigmaPrep

Sigma Prep (sigmaprep.io) makes free AP Statistics practice: exam-style multiple choice and free response for all five units, with answer keys and scoring guides, plus a formula sheet.

AP Statistics

Practice tests by unitFormula sheetUnit reviewsStatistics worksheets

Sigma Prep

Math worksheetsAboutContact

Legal

TermsPrivacy
© 2026 Sigma Prep. All rights reserved.AP® is a trademark registered by the College Board, which is not affiliated with, and does not endorse, this site.