AP STATISTICS AP EXAM REVIEW

Name: _______________________

Choose the correct hypothesis test for the following situations and complete the requested information. 1. A random sample of 49 medical doctors in LA showed that they worked an average of 53.1 hours/week with a standard deviation of 7.2 hours/week. If the California average is 60 hours/week, does this give evidence that the LA doctors work significantly less than the rest of California? Type of test: one sample t-test Ho: = 60 hours/week Ha: < 60 hours / week where is the average hours per week worked by LA doctors Requirements: random sample The population standard deviation is unknown The sample size is large enough (n = 49) for the CLT to account for any skewness. Test statistic: t=-6.71 P value: 1.03*10-8 or ~0 Reject/fail to reject: reject Conclusion:

2. White blood cell counts are normally distributed with mean 7500 and variance 500. If a patient has taken 50 laboratory blood tests that have a mean of 6899.75 and a standard deviation of 393.44, does this give evidence that his white blood cell count is significantly different than normal? Type of test: one sample z-test Ho: = 7500 Ha: 7500 where is the average white blood cell count in this patient. Requirements: representative sample I know the population standard deviation ( = The population is normally distributed. Test statistic: z=-189.81 P value: 0 Conclusion:

500 = 22.36)
Reject/fail to reject: reject

3. USA Today reported that in 1992, 39% of all elementary school children claimed that when they grow up they want to do something to help other people. However, in 1995, 128 of a random sample of 317 of these same children claimed that when they grow up they want to do something to help other people. Does this information indicate that there has been an attitude change either way? Type of test: one proportion z-test Ho: = .39 Ha: .39 where is the true proportion of elementary students in 1995 who want to help others when they grow up. Requirements: Simple random sample There are more than 10 successes and failures (128 and 89 respectively). There are more than 317*10=3170 elementary students nationwide. Test statistic: z = .503 P value: .615 Reject/fail to reject: fail to reject Conclusion: 4. The manager of a sporting goods store offered a bonus commission to his salespeople when they sold more goods. A new manager dropped the bonus system. For a random sample of six sales people, the weekly sales (in thousands of dollars) are shown in the following table with and without the bonus system Salesperson 1 2 3 4 5 6 with bonus 2.9 3.0 5.8 4.4 5.3 5.6 w/o bonus 2.8 2.5 5.9 3.5 4.6 4.6 difference 0.1 0.5 -0.1 0.9 0.7 1.0 Does this imply that sales dropped when the bonus system was discontinued? bvandenberg Page 1 4/24/2013

Type of test: matched pair t-test Ho : = 0 Ha : > 0 where is the average difference in weekly sales in thousands between with bonus and without bonus weeks (with without) Requirements: Simple random sample with 2 matched groups. The sample size is small, but reveals no gross outliers or skewness. Test statistic: t = 2.876 P value: .017 Reject/fail to reject: reject Conclusion: 5. In 1975, a random sample of 1484 adult U.S. citizens was surveyed, and 193 strongly agreed with the statement, People should take care of themselves. Then, in 1991, a survey of 1013 adult U.S. citizens showed that only 61 strongly agreed with the statement. Does this indicate that the proportion of U.S. adults who strongly agree with the given statement has dropped? Type of test: 2 proportion z-test H o : 1 = 2 Ha : 1 > 2 where 1 is the proportion of US citizens in 1975 who think people should take care of themselves and 2 is the same for US citizens in 1991. Requirements: Both samples are an SRS. Each group has at least 5 successes and failures (193, 1291, 61, and 952 respectively). There were more than 1484*10=14840 adults in 1975 and more than 1013*10 = 10130 adults in 1991. Test statistic: z = 5.67 P value: 7.2*10-9 or .0000 Reject/fail to reject: reject Conclusion: 6. Five small cities in California gave information about the percentage of 16-19 year olds not in school and the death rate per 1000 residents. Given the information below, can the percentage of 16-19 year olds not in school be used to predict the death rate? Give statistical evidence of your belief. City 1 2 3 4 5 % of 16-19 year olds 16.2 9.9 19.5 19.7 9.8 death rate/1000 7.7 8.8 7.0 8.1 8.4 Type of test: Linear Regression t-test Ho : = 0 Ha : 0 where is the true slope of the regression line between % of 16-19 year olds and the death rate per 1000. ( I used 0 since I am not sure if there are more young people NOT IN SCHOOL whether that will be associated with greater or fewer deaths per 1000.) Requirements: There is a true linear relationship between the % of 16-19 year olds and death rate. (Check residual plot for a pattern.) The standard deviation of the line is constant throughout the domain of x values (teen percentages). (Check residual plot for increasing or decreasing spread). The residuals are normally distributed about the line. (A histogram of the residuals should be approximately normal.) Test statistic: t = -2.14 P value: .122 Reject/fail to reject: fail to reject Conclusion: 7. Reading Nook Bookstore has 750 retail outlets across the country. The sales director wanted to see if Christmas music affects book sales in December. She randomly assigned some of the outlets to pipe in music and others not to. Then sales records for the month of December were kept. The results are shown in the Table. Test the hypothesis that sales and Christmas music are independent.

bvandenberg

Page 2

4/24/2013

With music Without music

<10,000 5 10

10-20,000 18 7

>20,000 7 3

Type of test: 2 test for independence Ho: Sales are independent of music being played. Ha: Sales are not independent of music being played. Requirements: The stores were an SRS. None of the expected values are 0 and only 1 is less than 5 (which is 16.7%). Expected matrix:

9 15 6 6 10 4
P value: .042 Reject/fail to reject: At =0.05, reject

Test statistic: 2 = 6.36 Conclusion:

8. The following is based on information taken from Winter Wind Studies in Rocky Mountain National Park, by Glidden. At five weather stations on Trail Ridge Road in Rocky Mountain National Park, the peak wind gusts (mi/hr) in January and April are recorded below Weather station 1 2 3 4 5 January 139 122 126 64 78 April 104 113 100 88 61 Difference 35 9 26 -24 17 Does this information indicate that the peak wind gusts are higher in January than April? Type of test: matched pair t-test Ho : = 0 Ha : > 0 Where is the average difference in peak wind gusts (mph) in Rocky Mountain National Park between January and April. Requirements: This isnt really an SRS, but it seems reasonable for the situation. The groups are matched. The sample is small, but it reveals no outliers or extreme skewness. Test statistic: t = 1.24 P value: 0.14 Reject/fail to reject: fail to reject Conclusion:

9. To compare 2 dog-training programs, an obedience school trained 43 dogs using Program A and 41 dogs using Program B. For Program A, the average number of training hours required was 24.8 with a standard deviation of 3.1 hours. For Program B, the mean was 22.9 hours with a standard deviation of 3.3 hours. Is there a significant difference between the two programs? Type of test: Two sample t-test Ho: A = B Ha: A B where A = pop avg. number of training hours required for dog obedience Program A and B = pop avg. number of training hours required for dog obedience Program B Requirements: Test statistic: t = 2.72 Conclusion: P value: .0081 Reject/fail to reject: fail to reject

10. Snoop Incorporated is a firm that does market surveys. The Rollum Sound Company hired Snoop to study the age distribution of people who buy CDs. To check the Snoop report, Rollum used a random sample of 519 customers and obtained the following data: bvandenberg Page 3 4/24/2013

Customer age (yrs)

<14 14.18 19-23 24-28 29-33 >33 Type of test: 2 goodness of fit test Ho: The distribution of customers in the Snoop report is the same as the customers in the sample. Ha: The distribution of customers in the Snoop report is different than the customers in the sample. Requirements: The customers in the sample should be an SRS of all customers. No more than 20% of the expected numbers should be less than 5. In this case, the smallest is 51.9, so I am OK. None of the expected counts should be zero. I am ok here also. Test statistic: 2 = 15.65 df=5 P value: .0079 Reject/fail to reject: reject Conclusion: 11. Five hundred adults participated in a comparison of the effectiveness of 3 arthritic pain relievers. Each participant used 1 of the 3 medications for 1 month and then was asked if the product was effective. The results were as follows: Pain Reliever Effective A B C Yes 115 78 140 No 60 72 35 Do the sample proportions differ significantly at the 1 percent level? Type of test: chi-sq test for homogeneity Ho: The proportion of adults for which the pain medication was effective was the same for each brand of pain reliever Ha: The proportion of adults for which the pain medication was effective was different for each brand of pain reliever Requirements: Test statistic: 28.56 Conclusion: P value: ~0 Reject/fail to reject: reject

% of customers In Snoop report 12% 29% 11% 10% 14% 24%

number of customers Expected in sample 88 62.28 135 150.51 52 57.09 40 51.9 76 72.66 128 124.56

12. The manufacturer of a particular brand of microwave popcorn claims that only 2 percent of its kernels of corn fail to pop. A competitor, believing that the actual percentage is larger, tests 2,000 kernels and finds that 44 failed to pop. Do these results provide sufficient evidence to support the competitors belief? Type of test: 1-proportion z-test Ho: = .02 Ha: > .02 where is the true proportion of corn kernels that fail to pop Requirements: Simple random sample There are more than 10 successes and failures (1956 and 44 respectively). There are more than 2000*10=20000 corn kernels Test statistic: z = .64 P value: 0.26 Reject/fail to reject: fail to reject Conclusion:

bvandenberg

Page 4

4/24/2013

13. Dannon is supposed to put 29.6 grams of fruit in their fruit-flavored yogurt. To see if enough fruit is being placed in cups of fruit-flavored yogurt, a quality inspector measured the amount in 27 containers. She found an average of 28.9 grams of fruit and standard deviation of 2.8 grams. Does this show that there is probably enough yogurt in the cups? Type of test: one sample t-test Ho: = 29.6 Ha: < 29.6 g where is the population average amount of fruit in Dannons fruit-flavored yogurt Requirements: representative sample The population standard deviation is unknown The sample size is almost large enough (n = 27) for the CLT to account for any skewness. Test statistic: t = - 1.3 P value: .1027 Reject/fail to reject: fail to reject Conclusion: 14. A shoe manufacturer has developed a new running shoe that purportedly enables one to run faster. Eight adults participated in an experiment in which each ran a mile with regular track shoes and then ran a mile the next day with the new shoes. Their running times in seconds are below runner track shoe new shoe differences 1 321 318 3 2 307 299 8 3 397 401 -4 4 269 260 9 5 285 285 0 6 364 363 1 7 295 289 6 8 302 296 6 Is there evidence that the new shoe works better? Type of test: matched pairs t-test Ho: d = 0 Ha: d < 0 where mu is the population average difference in running times for a mile for adults wearing track shoes and the new shoes (new shoe time track shoe time) Requirements: shoes not randomly assigned to days The groups are matched. The sample is small, but it reveals no outliers or extreme skewness. Test statistic: t = -2.31 P value: 0.0271 Reject/fail to reject: reject Conclusion: 15. The following comes from a study of small cities in Illinois and the habits of their residents with regard to eating and drinking (noteits more expensive to eat out than at home). City population Eating/drinking Naperville 88.5 95.9 Hoffman Estates 46.6 28.1 Wheaton 53.4 23.0 Oak Park 53.7 37.1 Mount Prospect 53.2 39.7 Elmhurst 43.6 45.8 Is there a significant association between the population of a city and how much they spend on eating/drinking? Type of test: Linear Regression t-test Ho : = 0 Ha : 0 where is the true slope of the regression line between population and amount spent on eating/drinking bvandenberg Page 5 4/24/2013

Requirements: There is a true linear relationship between the population size and amount spent on eating/drinking. (Check residual plot for a pattern.) The standard deviation of the line is constant. (Check residual plot for increasing or decreasing spread). The residuals are normally distributed about the line. (A histogram of the residuals should be approximately normal.) Test statistic: t = P value: Reject/fail to reject: Conclusion: 16. A leading auto manufacturer claimed that a popular version of its minivan was available in the Midwest for an average price or $16,000 with a standard deviation of $800. A consumer group doubted that report and surveyed 50 recent purchasers of the minivan to dispute the manufacturers claim. They came up with an average of $16,277. Does this show that the manufacturer is making a false claim? Type of test: z-test for means (given population SD) Ho: = $16,000 Ha: > $16,000 where is the population average price of a minivan available in the Midwest from a leading auto manufacturer Requirements: representative sample I know the population standard deviation ( = 800.) The population is normally distributed. Test statistic: z = 2.45 P value: 0.0144 Reject/fail to reject: reject Conclusion:

Thanks to Betsy VandenBerg, Lincoln-Way High School, New Lenox, IL bvandenberg Page 6 4/24/2013

bvandenberg

Page 7

4/24/2013