Worksheets · AP Statistics

AP Statistics Unit 3 Practice Test: Inference for Proportions

Unit 3: Inference for Categorical Data: Proportions

20 exam-style multiple-choice questions on inference for proportions, free to use in class. Good practice before the Unit 3 Progress Check in AP Classroom. Free response for this unit: Unit 3 FRQ set.

Sigma Prep · sigmaprep.io/worksheets/ap-statistics

Name

AP Statistics Unit 3: Inference for Proportions

Date

Directions: Work each problem in the space below it, then choose the best of the four answer choices. A graphing calculator with statistical capabilities may be used on every question, as on the exam, and the exam’s formula sheet and tables apply. Some questions come in a set that refers to one shared stimulus.

Multiple choice

Questions 1–20
  1. A state agency compared how many rural and urban households in the state have a backup generator. The state has about 400,000 rural households and about 1,100,000 urban households. Independent random samples of 120 rural households and 120 urban households were selected. Of the 120 rural households, 28 have a backup generator; of the 120 urban households, 12 have a backup generator. The state agency tested H0:p1−p2=0H_0: p_1 - p_2 = 0 against Ha:p1−p2>0H_a: p_1 - p_2 > 0, where p1p_1 is the proportion of all rural households in the state that have a backup generator and p2p_2 is the proportion of all urban households in the state that have a backup generator. All conditions were met, and the P-value was 0.0028. Which of the following is an appropriate conclusion at the α=0.05\alpha = 0.05 level?
    1. (A)The P-value of 0.0028 is less than α=0.05\alpha = 0.05, so the null hypothesis is not rejected. There is not convincing statistical evidence that the proportion of all rural households in the state that have a backup generator is greater than the proportion of all urban households in the state that have a backup generator.
    2. (B)The P-value of 0.0028 is less than α=0.05\alpha = 0.05, so the null hypothesis is rejected. There is not convincing statistical evidence that the proportion of all rural households in the state that have a backup generator is greater than the proportion of all urban households in the state that have a backup generator.
    3. (C)The P-value of 0.0028 is less than α=0.05\alpha = 0.05, so the null hypothesis is rejected. There is convincing statistical evidence that proves that the proportion of all rural households in the state that have a backup generator is greater than the proportion of all urban households in the state that have a backup generator.
    4. (D)The P-value of 0.0028 is less than α=0.05\alpha = 0.05, so the null hypothesis is rejected. There is convincing statistical evidence that the proportion of all rural households in the state that have a backup generator is greater than the proportion of all urban households in the state that have a backup generator.
  2. A national report states that 35% of adults have a library card. The city library director will select a random sample of 150 adults in the city and test H0:p=0.35H_0: p = 0.35 against Ha:p>0.35H_a: p > 0.35 at the α=0.10\alpha = 0.10 level, where pp is the proportion of all adults in the city who have a library card. For which of the following true values of pp would the probability of a Type II error be smallest?
    1. (A)p=0.40p = 0.40
    2. (B)p=0.35p = 0.35
    3. (C)p=0.30p = 0.30
    4. (D)p=0.45p = 0.45
  3. A city parks department compared weekly park use by younger and older adults in the city. The city has about 210,000 adults under 40 and about 260,000 adults 40 and over. Independent random samples of 300 adults under 40 and 300 adults 40 and over were selected. Of the 300 adults under 40, 77 use a city park weekly; of the 300 adults 40 and over, 91 use a city park weekly. Which of the following correctly justifies treating the responses as independent for a two-sample zz procedure?
    1. (A)The responses are independent because both sample sizes are at least 30.
    2. (B)The two samples were selected independently and at random, and 300 is less than 10% of the 210,000 adults under 40, and 300 is less than 10% of the 260,000 adults 40 and over, so the responses can be treated as independent.
    3. (C)The responses are independent because each sample has at least 10 successes and 10 failures.
    4. (D)The responses are independent because the adults under 40 and the adults 40 and over were randomly assigned to their groups.
  4. The city transit agency has a target that no more than 15% of trips arrive more than five minutes late. The city auditor will select a random sample of 250 trips on the city's bus routes and test H0:p=0.15H_0: p = 0.15 against Ha:p>0.15H_a: p > 0.15 at the α=0.10\alpha = 0.10 level, where pp is the proportion of all trips on the city's bus routes that arrive more than five minutes late. Consider the following changes to the plan, each made with everything else the same.
    I. Using a sample of 125 instead of 250 would increase the power of the test.
    II. Using a significance level of α=0.05\alpha = 0.05 instead of 0.100.10 would increase the power of the test.
    III. The true value of pp being 0.23 rather than 0.18 would increase the power of the test.
    Which of the changes would increase the power of the test?
    1. (A)III only
    2. (B)I and III only
    3. (C)II and III only
    4. (D)I, II, and III
  5. Suppose that 78% of all seeds in a shipment of lettuce seeds sprout within ten days (the shipment has about 2 million seeds). Random samples of 60 seeds will be selected, and p^\hat{p} is the proportion in a sample that sprout within ten days. What is the standard deviation of the sampling distribution of p^\hat{p}?
    1. (A)0.00290.0029
    2. (B)0.00690.0069
    3. (C)0.01300.0130
    4. (D)0.05350.0535
  6. Questions 6–7 refer to the following.

    To estimate the proportion of all smartphone owners in a city who use a daily screen-time limit, a researcher selected a random sample of 220 smartphone owners. In the sample, 36 of the smartphone owners use a daily screen-time limit. (The city has about 900,000 smartphone owners.)
  7. Which of the following is a 99% confidence interval for the proportion of all smartphone owners in the city who use a daily screen-time limit?
    1. (A)(0.099,0.228)(0.099, 0.228)
    2. (B)(0.106,0.222)(0.106, 0.222)
    3. (C)(0.123,0.205)(0.123, 0.205)
    4. (D)(0.139,0.189)(0.139, 0.189)
  8. The researcher plans a 99% confidence interval for the proportion of all smartphone owners in the city who use a daily screen-time limit. Consider these statements about the conditions for the interval.
    I. Large counts: np^=36n\hat{p} = 36 and n(1−p^)=184n(1 - \hat{p}) = 184 are both at least 10, so the sampling distribution of p^\hat{p} is approximately normal.
    II. For the 10% condition, 220 is less than 10% of the population, so sampling without replacement still gives (nearly) independent responses.
    III. For the 10% condition, 220 is less than 10% of the population, so the sampling distribution of p^\hat{p} is approximately normal.
    Which of the statements are correct?
    1. (A)I only
    2. (B)II only
    3. (C)I and II only
    4. (D)I, II, and III
  9. Questions 8–9 refer to the following.

    A regional planning office surveyed a random sample of 150 workers in the region and recorded each worker's home location and usual way of getting to work. The region has about 900,000 workers. The table below shows the results.
    {"rowVar":"Home location","colVar":"Way to work","rows":["City","Suburb"],"cols":["Drive","Transit","Walk or bike"],"counts":[[46,30,24],[31,12,7]]}Home locationDriveTransitWalk or bikeTotalCity463024100Suburb3112750Total774231150
  10. If there is no association between the variables, what is the expected count of workers in the City and Transit cell?
    1. (A)14.0014.00
    2. (B)21.0021.00
    3. (C)25.0025.00
    4. (D)28.0028.00
  11. How many degrees of freedom does the chi-square test for these data have?
    1. (A)df=3df = 3, because (2−1)+(3−1)=3(2 - 1) + (3 - 1) = 3.
    2. (B)df=2df = 2, because the table has 2 rows and 3 columns: (2−1)(3−1)=2(2 - 1)(3 - 1) = 2.
    3. (C)df=5df = 5, because the table has 6 cells, minus 1: 6−1=56 - 1 = 5.
    4. (D)df=6df = 6, because the table has 6 cells: 2(3)=62(3) = 6.
  12. A large farm's standard allows 5% of its packed eggs to be cracked. The quality inspector at the farm believes that the proportion of all eggs packed at the farm that are cracked is greater than 5%. The inspector selected a random sample of 400 eggs packed at the farm; 24 of them were cracked. The inspector tested H0:p=0.05H_0: p = 0.05 against Ha:p>0.05H_a: p > 0.05, where pp is the proportion of all eggs packed at the farm that are cracked. Assume the conditions for inference are met. Which of the following are the test statistic and the P-value of the test?
    1. (A)z=0.92z = 0.92 and P-value =0.1794= 0.1794
    2. (B)t=0.92t = 0.92 and P-value =0.1797= 0.1797
    3. (C)z=0.92z = 0.92 and P-value =0.3588= 0.3588
    4. (D)z=0.92z = 0.92 and P-value =0.8206= 0.8206
  13. A plan for early voting on weekends will go on the state ballot only if more than 50% of registered voters support it. The campaign director for the weekend voting plan will test H0:p=0.50H_0: p = 0.50 against Ha:p>0.50H_a: p > 0.50, where pp is the proportion of all registered voters in the state who support a plan for early voting on weekends. Which of the following describes a Type II error in this context?
    1. (A)Not concluding that the proportion of all registered voters in the state who support a plan for early voting on weekends is greater than 50% when in fact it is 50%.
    2. (B)Not concluding that the proportion of all registered voters in the state who support a plan for early voting on weekends is greater than 50% when in fact it is greater than 50%.
    3. (C)Not concluding that the proportion of the sampled voters who support a plan for early voting on weekends is greater than 50% when in fact it is greater than 50%.
    4. (D)Concluding that the proportion of all registered voters in the state who support a plan for early voting on weekends is greater than 50% when in fact it is 50%.
  14. A clothing retailer selected independent random samples of shoppers at its outlet stores and at its mall stores (90 from outlet stores and 60 from mall stores) and asked each shopper how they would rate their visit. The outlet stores have about 400,000 shoppers a year, and the mall stores about 900,000. The table below shows the results. Which of the following are appropriate hypotheses for a chi-square test?
    {"rowVar":"Store type","colVar":"Rating","rows":["Outlet","Mall"],"cols":["Excellent","Good","Fair or poor"],"counts":[[25,41,24],[21,25,14]]}Store typeExcellentGoodFair or poorTotalOutlet25412490Mall21251460Total466638150
    1. (A)H0H_0: The counts in the table show no association. HaH_a: The counts in the table show an association.
    2. (B)H0H_0: The distribution of visit rating is not the same for outlet and mall shoppers. HaH_a: The distribution of visit rating is the same for outlet and mall shoppers.
    3. (C)H0H_0: The distribution of visit rating is the same for outlet and mall shoppers. HaH_a: The distribution of visit rating is not the same for outlet and mall shoppers.
    4. (D)H0H_0: Each category of visit rating is equally likely. HaH_a: The categories of visit rating are not equally likely.
  15. An engineer is comparing three methods, A, B and C, of estimating the mean fuel economy of a model of car. To compare them, a simulation took 30 random samples from a population in which the mean fuel economy of all cars of the model is 28.6 miles per gallon, and used each method on each sample. The dotplots below show the 30 estimates from each method; the dashed line marks 28.6. Based on the dotplots, which method is the best choice for estimating the mean fuel economy of all cars of the model?
    {"groups":[{"name":"Estimator A","values":[26.2,25.8,25.8,25.6,26.6,25.4,26,26.4,25.8,26.2,26,26.2,25.8,25.4,25.6,26,25.2,26.4,25.8,26.6,26.4,25.6,26.4,25.8,25.2,26.2,26.2,26,25.4,26.2]},{"name":"Estimator B","values":[26,29.2,27.4,30.4,27.6,29.4,28,32.8,27.8,32,30.2,31.2,32.8,30.4,29.6,31,30.2,29,24.2,27.6,28.4,31,27.6,26.8,28.2,26,28.2,25,30.4,25.2]},{"name":"Estimator C","values":[30.8,30.2,29.6,27.8,28.8,28.4,28.4,27.8,27.4,29.8,27.8,30.6,29.4,29.2,28.2,28.6,28.6,28.2,30,29,29.8,29.2,29.2,27.6,28.6,27.4,29.4,27.6,27.8,27.4]}],"unit":0.2,"axis":{"lo":24,"hi":33,"step":1,"dp":0},"observed":28.6}Estimator AEstimator BEstimator C24252627282930313233Estimate (miles per gallon)
    1. (A)Estimator A, because its values are closest together, so it is unbiased.
    2. (B)Estimator C, because it is unbiased and has less variability than Estimator B.
    3. (C)Estimator A, because it has the least variability.
    4. (D)Estimator B, because it is unbiased.
  16. A large university's health center set a goal that 35% of its students get a flu shot this year. The health center director suspects that the proportion of all students at the university who got a flu shot this year is less than 35%. In a random sample of 40 students at the university, 11 got a flu shot this year. To see what could happen by chance, the health center director simulated 100 random samples of 40 assuming that 35% of all students at the university got a flu shot this year, and recorded the number in each sample who got a flu shot this year. The dotplot below shows the results; the dashed line marks 11. A dot on the dashed line counts as 11 or fewer. Based on the simulation, what is the estimated P-value of a test of H0:p=0.35H_0: p = 0.35 against Ha:p<0.35H_a: p < 0.35?
    {"groups":[{"name":"Simulated","values":[11,20,12,11,15,11,17,13,10,14,17,12,14,16,15,15,15,15,13,19,14,15,16,20,15,18,13,16,15,13,8,13,14,15,11,14,12,15,17,15,12,15,14,13,12,14,15,15,14,14,15,17,12,15,14,12,12,14,13,14,10,17,20,13,14,12,21,14,9,15,15,14,13,16,17,18,12,14,12,12,17,13,12,20,10,13,13,14,12,19,12,15,16,13,14,16,15,10,9,18]}],"unit":1,"axis":{"lo":5,"hi":25,"step":5,"dp":0},"observed":11}510152025Number of students out of 40 who got a flu shot this year
    1. (A)0.040.04
    2. (B)0.070.07
    3. (C)0.110.11
    4. (D)0.930.93
  17. Last year, 35% of the shoppers at a grocery chain brought their own bags. The manager of the grocery chain will test H0:p=0.35H_0: p = 0.35 against Ha:p>0.35H_a: p > 0.35, where pp is the proportion of all shoppers at the grocery chain who bring their own bags. Which of the following describes a Type I error in this context?
    1. (A)Concluding that the proportion of all shoppers at the grocery chain who bring their own bags is greater than 35% when in fact it is 35%.
    2. (B)Not concluding that the proportion of all shoppers at the grocery chain who bring their own bags is greater than 35% when in fact it is greater than 35%.
    3. (C)Not concluding that the proportion of all shoppers at the grocery chain who bring their own bags is greater than 35% when in fact it is 35%.
    4. (D)Concluding that the proportion of the sampled shoppers who bring their own bags is greater than 35% when in fact it is 35%.
  18. A cafeteria tested whether a calorie label on the menu board changes how many diners choose a salad. A total of 230 diners who agreed to take part were randomly assigned to the menu with calorie labels (80) or the menu without labels (150). Of the 80 diners who saw the label, 26 chose a salad; of the 150 diners who did not see the label, 36 chose a salad. The cafeteria manager wants to know whether the proportion of diners like these who would choose a salad with the label is different from the proportion of diners like these who would choose a salad without the label. Let p1p_1 be the proportion of diners like these who would choose a salad with the label and p2p_2 be the proportion of diners like these who would choose a salad without the label. Which of the following are appropriate hypotheses?
    1. (A)H0:p1−p2=0.09H_0: p_1 - p_2 = 0.09 and Ha:p1−p2≠0.09H_a: p_1 - p_2 \neq 0.09
    2. (B)H0:p1−p2=0H_0: p_1 - p_2 = 0 and Ha:p1−p2≠0H_a: p_1 - p_2 \neq 0
    3. (C)H0:p^1−p^2=0H_0: \hat{p}_1 - \hat{p}_2 = 0 and Ha:p^1−p^2≠0H_a: \hat{p}_1 - \hat{p}_2 \neq 0
    4. (D)H0:p1−p2=0H_0: p_1 - p_2 = 0 and Ha:p1−p2>0H_a: p_1 - p_2 > 0
  19. A physical therapist tested whether a daily stretching routine changes how many office workers report back pain after six weeks. A total of 201 office workers who volunteered were randomly assigned to the stretching routine (78) or no routine (123). Of the 78 workers given the routine, 19 reported back pain; of the 123 workers not given the routine, 57 reported back pain. The therapist tested H0:p1−p2=0H_0: p_1 - p_2 = 0 against Ha:p1−p2<0H_a: p_1 - p_2 < 0, where p1p_1 is the proportion of office workers like these who would report back pain with the routine and p2p_2 is the proportion of office workers like these who would report back pain without it. Assume the conditions for inference are met. Which of the following are the test statistic and the P-value of the test?
    1. (A)z=−3.13z = -3.13 and P-value =0.0009= 0.0009
    2. (B)z=−3.13z = -3.13 and P-value =0.0017= 0.0017
    3. (C)z=−3.32z = -3.32 and P-value =0.0004= 0.0004
    4. (D)t=−3.13t = -3.13 and P-value =0.0012= 0.0012
  20. A pet supply company selected a random sample of 206 of its online customers and recorded whether each customer lives in an apartment or a house and which kind of pet the customer owns. The company has about 250,000 online customers. The company carried out a chi-square test of the null hypothesis that there is no association between home type and kind of pet owned for the company's online customers. The table and the computer output from the test are shown below. Which of the following is an appropriate conclusion at the α=0.10\alpha = 0.10 level?
    {"rowVar":"Home type","colVar":"Pet owned","rows":["Apartment","House"],"cols":["Cat","Dog","Other"],"counts":[[23,19,9],[46,97,12]]}Home typeCatDogOtherTotalApartment2319951House469712155Total6911621206{"test":"chisq"}Chi-Square TestChi-SqDFP-Value10.7920.0045
    1. (A)The P-value of 0.0045 is less than α=0.10\alpha = 0.10, so the null hypothesis is rejected. There is not convincing statistical evidence that there is an association between home type and kind of pet owned for the company's online customers.
    2. (B)The P-value of 0.0045 is less than α=0.10\alpha = 0.10, so the null hypothesis is rejected. There is convincing statistical evidence that home type causes differences in kind of pet owned for the company's online customers.
    3. (C)The P-value of 0.0045 is less than α=0.10\alpha = 0.10, so the null hypothesis is rejected. There is convincing statistical evidence that there is an association between home type and kind of pet owned for the company's online customers.
    4. (D)The P-value of 0.0045 is less than α=0.10\alpha = 0.10, so the null hypothesis is not rejected. There is not convincing statistical evidence that there is an association between home type and kind of pet owned for the company's online customers.
  21. To estimate the proportion of all users of a fitness app who rate the app five stars, the design team of the fitness app selected a random sample of 410 users of the fitness app; 161 of them rate the app five stars. A 95% confidence interval for the proportion of all users of the fitness app who rate the app five stars is (0.345,0.440)(0.345, 0.440). The design team has heard a claim that less than 37% of all users of the fitness app rate the app five stars. Which of the following is the best conclusion about the claim?
    1. (A)Because 0.37 is inside the interval, the interval shows that exactly 37% of all users of the fitness app rate the app five stars.
    2. (B)Because 0.37 is inside the interval, the interval gives convincing evidence that less than 37% of all users of the fitness app rate the app five stars.
    3. (C)Because most of the interval is below 0.37, the interval gives convincing evidence that less than 37% of all users of the fitness app rate the app five stars.
    4. (D)Because 0.37 is inside the interval and so is a plausible value of the proportion, the interval does not give convincing evidence that less than 37% of all users of the fitness app rate the app five stars.
  22. A district counselor selected a random sample of 357 high school students in a large school district and recorded each student's grade level and usual amount of sleep on school nights. The district has about 26,000 high school students. The table below shows the results. The counselor wants to know whether there is an association between grade level and amount of sleep on school nights for high school students in the district. Which of the following is the appropriate test?
    {"rowVar":"Grade","colVar":"Sleep","rows":["Grade 9","Grade 10","Grade 11","Grade 12"],"cols":["Under 7 hours","7 to 8 hours","Over 8 hours"],"counts":[[30,49,19],[24,22,15],[38,27,11],[54,43,25]]}GradeUnder 7 hours7 to 8 hoursOver 8 hoursTotalGrade 930491998Grade 1024221561Grade 1138271176Grade 12544325122Total14614170357
    1. (A)A one-sample zz-test for a population proportion
    2. (B)A chi-square test for homogeneity
    3. (C)A chi-square test for independence
    4. (D)A two-sample zz-test for a difference between two population proportions

    End of the multiple-choice questions

Answer key

12345678910
DDBADACDBA
3.133.83.103.83.23.33.33.153.153.7
11121314151617181920
BCBCABACDC
3.83.143.13.63.83.123.133.153.43.14

The second row is the correct letter; the third is the course topic each question tests.

Worked solutions

1. Answer (D) Topic 3.13

  1. Since 0.0028 is less than α=0.05\alpha = 0.05, reject H0H_0.
  2. The conclusion is about the population proportions, in the direction of HaH_a, as evidence and not proof.
  3. The answer is: The P-value of 0.0028 is less than α=0.05\alpha = 0.05, so the null hypothesis is rejected. There is convincing statistical evidence that the proportion of all rural households in the state that have a backup generator is greater than the proportion of all urban households in the state that have a backup generator.

2. Answer (D) Topic 3.8

  1. Power is greater (and the probability of a Type II error smaller) when the true value of pp is farther from 0.35 in the direction of HaH_a.
  2. Of the values in the direction of HaH_a, 0.45 is farthest from 0.35.
  3. The answer is: p=0.45p = 0.45.

3. Answer (B) Topic 3.10

  1. Independent random samples make the two groups independent, and the 10% condition makes the responses within each sample nearly independent.
  2. Each sample is less than 10% of its population.
  3. The answer is: The two samples were selected independently and at random, and 300 is less than 10% of the 210,000 adults under 40, and 300 is less than 10% of the 260,000 adults 40 and over, so the responses can be treated as independent.

4. Answer (A) Topic 3.8

  1. Statement I is false: a smaller sample makes p^\hat{p} more variable, which lowers power.
  2. Statement II is false: a smaller α\alpha makes H0H_0 harder to reject, which lowers power.
  3. Statement III is true: 0.23 is farther from 0.15 in the direction of HaH_a, and a larger difference is easier to detect.
  4. So only statement III is true. The answer is III only.

5. Answer (D) Topic 3.2

  1. The sampling distribution of p^\hat{p} has standard deviation p(1−p)n\sqrt{\frac{p(1 - p)}{n}} (the 10% condition holds: 60 is less than 10% of the population).
  2. Compute 0.78(0.22)60=0.0535\sqrt{\frac{0.78(0.22)}{60}} = 0.0535.
  3. So the standard deviation is 0.0535.

6. Answer (A) Topic 3.3

  1. The sample proportion is p^=36220=0.164\hat{p} = \frac{36}{220} = 0.164, and for 99% confidence z∗=2.576z^* = 2.576.
  2. The standard error is 0.164(0.836)220=0.0249\sqrt{\frac{0.164(0.836)}{220}} = 0.0249, so the margin of error is 2.576(0.0249)=0.0642.576(0.0249) = 0.064.
  3. The interval is 0.164±0.0640.164 \pm 0.064; with the unrounded values, 0.1636±0.06420.1636 \pm 0.0642 gives 0.0990.099 to 0.2280.228: (0.099,0.228)(0.099, 0.228).

7. Answer (C) Topic 3.3

  1. Statement I is true: both 36 and 184 are at least 10.
  2. Statement II is true: the city has about 900,000 smartphone owners, and 10% of that is more than 220.
  3. Statement III is false: the 10% condition justifies independence; normality comes from the large counts condition.
  4. So the true statements are I and II. The answer is I and II only.

8. Answer (D) Topic 3.15

  1. If H0H_0 is true, a cell's expected count is (row total)(column total)table total\frac{(\text{row total})(\text{column total})}{\text{table total}}.
  2. For City and Transit: 100(42)150=28.00\frac{100(42)}{150} = 28.00.
  3. So the expected count is 28.00.

9. Answer (B) Topic 3.15

  1. A chi-square test for a two-way table has (rows−1)(columns−1)(\text{rows} - 1)(\text{columns} - 1) degrees of freedom; the total row and column are not counted.
  2. The table has 2 rows of home location and 3 columns of way to work.
  3. The answer is: df=2df = 2, because the table has 2 rows and 3 columns: (2−1)(3−1)=2(2 - 1)(3 - 1) = 2.

10. Answer (A) Topic 3.7

  1. The sample proportion is p^=24400=0.06\hat{p} = \frac{24}{400} = 0.06, and the test's standard error uses the null value: 0.05(0.95)400=0.0109\sqrt{\frac{0.05(0.95)}{400}} = 0.0109.
  2. z=0.06−0.050.0109=0.92z = \frac{0.06 - 0.05}{0.0109} = 0.92.
  3. The P-value is the area above z=0.92z = 0.92 under the standard normal curve: 0.1794.
  4. The answer is: z=0.92z = 0.92 and P-value =0.1794= 0.1794.

11. Answer (B) Topic 3.8

  1. A Type II error fails to reject H0H_0 when H0H_0 is false (the alternative is true).
  2. Here that is not concluding that p>0.50p > 0.50 when in fact p>0.50p > 0.50.
  3. The answer is: Not concluding that the proportion of all registered voters in the state who support a plan for early voting on weekends is greater than 50% when in fact it is greater than 50%.

12. Answer (C) Topic 3.14

  1. Separate groups were compared on one variable, so the test is for homogeneity: H0H_0 says the distribution is the same in every group.
  2. HaH_a says the opposite, without claiming every group or category differs.
  3. The answer is: H0H_0: The distribution of visit rating is the same for outlet and mall shoppers. HaH_a: The distribution of visit rating is not the same for outlet and mall shoppers.

13. Answer (B) Topic 3.1

  1. Estimator C and Estimator B are centered at 28.6, so they appear unbiased; Estimator A is centered below it, so it is biased.
  2. Of the two unbiased estimators, Estimator C has the smaller spread, so its estimates tend to be closer to 28.6.
  3. The answer is: Estimator C, because it is unbiased and has less variability than Estimator B.

14. Answer (C) Topic 3.6

  1. Each dot is one simulated sample of 40 under H0H_0 (p=0.35p = 0.35); the P-value is estimated by the share of dots at 11 or fewer, the direction of HaH_a.
  2. Of the 100 dots, 11 are at 11 or fewer, so the estimate is 11100=0.11\frac{11}{100} = 0.11.
  3. So the estimated P-value is 0.11.

15. Answer (A) Topic 3.8

  1. A Type I error rejects H0H_0 when H0H_0 is true.
  2. Here that is concluding that p>0.35p > 0.35 when in fact p=0.35p = 0.35.
  3. The answer is: Concluding that the proportion of all shoppers at the grocery chain who bring their own bags is greater than 35% when in fact it is 35%.

16. Answer (B) Topic 3.12

  1. H0H_0 says there is no difference: p1−p2=0p_1 - p_2 = 0.
  2. The cafeteria manager asks whether the proportion of diners like these who would choose a salad with the label is different from the proportion of diners like these who would choose a salad without the label, so Ha:p1−p2≠0H_a: p_1 - p_2 \neq 0.
  3. The answer is: H0:p1−p2=0H_0: p_1 - p_2 = 0 and Ha:p1−p2≠0H_a: p_1 - p_2 \neq 0.

17. Answer (A) Topic 3.13

  1. p^1=0.2436\hat{p}_1 = 0.2436 and p^2=0.4634\hat{p}_2 = 0.4634. Under H0H_0 the proportions are equal, so the test pools them: p^c=19+5778+123=0.3781\hat{p}_c = \frac{19 + 57}{78 + 123} = 0.3781.
  2. SE=0.3781(0.6219)(178+1123)=0.0702SE = \sqrt{0.3781(0.6219)\left(\frac{1}{78} + \frac{1}{123}\right)} = 0.0702, so z=0.2436−0.46340.0702=−3.13z = \frac{0.2436 - 0.4634}{0.0702} = -3.13.
  3. The P-value is the area below z=−3.13z = -3.13 under the standard normal curve: 0.0009.
  4. The answer is: z=−3.13z = -3.13 and P-value =0.0009= 0.0009.

18. Answer (C) Topic 3.15

  1. From the output, the P-value is 0.0045, which is less than α=0.10\alpha = 0.10.
  2. The conclusion is stated about all the company's online customers, in terms of HaH_a.
  3. The answer is: The P-value of 0.0045 is less than α=0.10\alpha = 0.10, so the null hypothesis is rejected. There is convincing statistical evidence that there is an association between home type and kind of pet owned for the company's online customers.

19. Answer (D) Topic 3.4

  1. A value inside a confidence interval is plausible; values outside it are not.
  2. The value 0.37 is inside the interval and so is a plausible value of the proportion, so 37% itself is still a plausible value.
  3. The answer is: Because 0.37 is inside the interval and so is a plausible value of the proportion, the interval does not give convincing evidence that less than 37% of all users of the fitness app rate the app five stars.

20. Answer (C) Topic 3.14

  1. One random sample was taken and each student was classified by two categorical variables, grade level and amount of sleep on school nights.
  2. A question about association between two variables in one population calls for a test for independence.
  3. The answer is: A chi-square test for independence.

Free AP Statistics worksheets at sigmaprep.io/worksheets/ap-statisticsOriginal questions written by Sigma Prep. AP is a registered trademark of the College Board, which is not affiliated with this sheet.

What does the AP Statistics exam ask on Unit 3?

Unit 3 is about a fifth of the AP Statistics multiple choice: 86 of the 406 official multiple-choice questions on today's course (21.2%), and 5 of the 24 sample questions in the 2026 course description, the only ones in the 2027 format.

  • One proportion is most of the unit. Confidence intervals for one proportion (topics 3.3 and 3.4) hold 24 of the 86 Unit 3 questions and tests for one proportion (topics 3.5 to 3.8) hold 28; two proportions (topics 3.9 to 3.13) hold 12 and chi-square (topics 3.14 and 3.15) 14. Topics 3.3 and 3.8 are the biggest, with 14 each.
  • Errors and power are asked in both sections. Power has 8 official multiple-choice questions on 8 of the 13 sources and Type I and Type II errors have 4, and 4 free-response questions from 2012 to 2026 had a part on the errors.
  • Comparing two proportions is a free-response favorite. A two-proportion z-test was part of 4 official free-response questions from 2012 to 2026, 2 of them since 2019, and of the 2026 sample Question 3, but it has only 2 official multiple-choice questions.
  • Questions come in sets on the 2027 exam. 6 of the 24 sample questions in the 2026 course description are in 2 sets that share one stimulus. This sheet has two sets of two questions.
  • Topic 3.11 has no official question. No official question, multiple choice or free response, has asked about using a confidence interval for p₁ − p₂ to evaluate a claim or interpreting a confidence interval for p₁ − p₂, so they are left off this sheet.

The most-tested kind here is constructing a one-proportion z-interval, with 9 official questions; every question on this sheet is original, and every answer was checked by an independent computer check.

Counted from the official AP Statistics questions: multiple choice from 13 College Board sources, 1997–2026, and free response 2012–2026. No official question is reproduced here.

How often does the AP Statistics exam ask each Unit 3 question type?

Official AP Statistics questions by question type, Unit 3 (Inference for Categorical Data: Proportions), and how many are on this worksheet
Question typeOn this worksheetOfficial multiple choice, 1997–2026Sources it was on (of 13)On the 2012–2019 practice examsFree-response questions, 2012–2026
Bias and variability of estimatorsTopic 3.113321
Sampling distribution of p̂: shape, mean, SDTopic 3.213320
Construct a one-proportion z-intervalTopic 3.329683
Use a confidence interval for p to evaluate a claimTopic 3.412222
p-value from a simulated null distributionTopic 3.611112
Carry out a one-proportion z-testTopic 3.713332
Identify Type I/II errors and consequencesTopic 3.824424
Power: definition and what increases itTopic 3.828841
Conditions for two-proportion inferenceTopic 3.1013314
Hypotheses for a two-proportion testTopic 3.1210004
Carry out a two-proportion z-testTopic 3.1312214
Conclusion of a two-proportion testTopic 3.1310004
Chi-square hypotheses, conditions and statistic behaviorTopic 3.1411114
Choose a chi-square test (homogeneity vs independence)Topic 3.1412212
Chi-square expected counts, df, statistic and p-valueTopic 3.1529853
Conclusion of a chi-square testTopic 3.1512214

Multiple-choice counts are of the 406 official questions on today’s course whose main topic is that kind, 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. Free-response counts are of the 84 operational questions, 2012–2026 (no 2020 exam was released), with a part on that kind. Not on this sheet (asked less often): sample size for a margin of error (proportion); interpret a p-value / statistical significance; conclusion of a one-proportion test; hypotheses for a one-proportion test; CI width vs sample size, confidence level and p̂; one-sided vs two-sided p-values; identify a one-proportion z procedure; interpret a confidence interval or margin of error (proportion); sampling distribution of p̂₁ − p̂₂; conditions for one-proportion inference; identify a two-proportion z procedure; probability of type i error (alpha) in repeated testing; interpret the confidence level (proportion); construct a two-proportion z-interval; probability about a sample proportion. Never asked by an official question yet (topic 3.11, in the course before 2026): use a CI for p₁ − p₂ to evaluate a claim; interpret a confidence interval for p₁ − p₂.

Questions about this worksheet

Unit 3, Inference for Categorical Data: Proportions, has 15 topics in the College Board's 2026 course and exam description: estimators and bias; the sampling distribution of a sample proportion; constructing and interpreting a confidence interval for a population proportion and using it to judge a claim; setting up and carrying out a significance test for a proportion, with P-values (including P-values from a simulation); Type I and Type II errors and power; the same for the difference of two proportions (the sampling distribution of the difference, the two-proportion z-interval and z-test); and chi-square tests for homogeneity and independence in a two-way table. The course description weights the unit at 15–25% of the multiple-choice section.

Yes, on every question. The AP Statistics exam allows a graphing calculator with statistical capabilities throughout both sections (on the digital exam, the Desmos calculator in Bluebook is also available), and it supplies a formula sheet and tables. Most Unit 3 questions read a value from a calculator test or interval (1-PropZTest, 2-PropZTest, 1-PropZInt, 2-PropZInt, χ²-Test) or compute a standard error and a z statistic. Each worked solution shows the formula and the numbers, so the work can be checked by hand or with the formula sheet.

We tagged 406 official multiple-choice questions on today's course, from 13 College Board sources (released exams, practice exams and course-description samples, 1997–2026), and the official free-response questions from 2012 to 2026, then sorted the Unit 3 questions into 33 kinds and gave the kinds the exam asks most the most questions (2 on “Construct a one-proportion z-interval”, 2 on “Identify Type I/II errors and consequences”, 2 on “Power: definition and what increases it”, 2 on “Chi-square expected counts, df, statistic and p-value”). With more kinds than questions, the kinds the exam asks least are left out of this sheet, as are the 2 kinds on topic 3.11 (a confidence interval for a difference of proportions used to judge a claim) that no official question has asked about yet. Two sets of questions share one stimulus, as on the 2027 exam.

As a timed unit practice exam: 20 original multiple-choice questions with four choices, in the 2027 format. Allow about 2 minutes per question, as the exam does, check the answer key, read the worked solution of every question you missed, then redo those kinds of question. For the free-response part, use the Unit 3 FRQ set, which has a scoring guide for every point.

No. Every question here is original, written by Sigma Prep in the style of the exam. We studied which kinds of questions the official exams ask and how often; no official question is copied or reworded. Each answer, and every number in the displays, was checked by an independent computer check before the sheet was published.

Yes. Every question has a worked solution that reaches the correct answer step by step. The answer key and the solutions are shown with the sheet, and the PDF prints them after the questions. There is also a PDF without them to hand out. For free-response practice, the Unit 3 FRQ set has a scoring guide with a model answer for every point.

AP is a registered trademark of the College Board, which was not involved in making this worksheet and does not endorse it. All questions are original to Sigma Prep. Frozen 2026-11-19.