252y0612 10/18/06 (Open in ‘Print Layout’ format) ECO252 QBA2 FIRST HOUR EXAM October 17-18 2005 Version 2 Name _____KEY___________ Hour of class registered _____ Class attended if different ____ Show your work! Make Diagrams! Exam is normed on 50 points. Answers without reasons are not usually acceptable. I. (8 points) Do all the following. x ~ N 7, 7.5 7 7 6 7 z P 0.13 z 0 .0517 1. P6.00 x 7.00 P 7.5 7.5 For z make a diagram. Draw a Normal curve with a mean at 0. Indicate the mean by a vertical line! Shade the area between zero and 0.15. Because this is completely on the left of zero and touches zero, we can simply look up our answer on the standardized Normal table. If you wish, make a completely separate diagram for x . Draw a Normal curve with a mean at 7. Indicate the mean by a vertical line! Shade the area between 6 and 7. This area includes the mean (7), but does not include any points to the right of the mean, so that we neither add nor subtract. 10 .22 7 Pz 0.43 Pz 0 P0 z 0.43 .5 .1664 .6664 2. Px 10.22 P z 7.5 For z make a diagram. Draw a Normal curve with a mean at 0. Shade the entire area below 0.43. Because this is on both sides of zero, we must add. If you wish, make a completely separate diagram for x . Draw a Normal curve with a mean at 7. Shade the entire area below 10.22. Since the area is on both sides of the mean, we add. 0 7 17 .5 7 z P 3.27 z 0.93 3. P17.5 x 0 P 7.5 7.5 P3.27 z 0 P0.93 z 0 .4995 .3238 .1757 For z make a diagram. Draw a Normal curve with a mean at 0. Shade the area between -3.27 and -0.93. Because this is completely on the left of zero, we must subtract. If you wish, make a completely separate diagram for x . Draw a Normal curve with a mean at 7. Shade the area between -17.5 and 0. Note that zero is below the mean. These numbers are both to the left of the mean (7), so we subtract. 4. x.04 (Do not try to use the t table to get this.) For z make a diagram. Draw a Normal curve with a mean at 0. z.04 is the value of z with 4% of the distribution above it. Since 100 – 4 = 96, it is also the 96th percentile. Since 50% of the standardized Normal distribution is below zero, your diagram should show that the probability between z.04 and zero is 96% - 50% = 46% or P0 z z.04 .4600 . The closest we can come to this is P0 z 1.75 .4599 . (1.76 is also acceptable here.) So z.04 1.75 . To get from z.04 x . x 7 1.757.5 20.125 . If you wish, make a completely separate diagram for x . Draw a Normal curve with a mean at 7. Show that 50% of the distribution is below the mean (7). If 4% of the distribution is above z.04 , it must be above the mean and have 46% of the distribution between it and the mean. to x.04 , use the formula x z , which is the opposite of z 20 .125 7 Check: Px 20.125 P z Pz 1.75 Pz 0 P0 z 1.75 7.5 .5 .4599 .0401 4% 1 252y0612 10/18/06 (Open in ‘Print Layout’ format) II. (5 points-2 point penalty for not trying part a.) In order not to violate the truth in labeling law a teabag must contain at least 5.5 oz of tea. A sample of 9 items is taken from a large number of tea bags. The data below is found. (Recomputing what I’ve done for you is a great way to waste time.) b) and d) require statistical tests. a. Compute the sample standard deviation, s , of the waiting times. Show your work! (2) b. Is the population mean significantly below 5.5 (Use a 95% confidence level)? Show your work! (3) [13] c. (Extra Credit) Find an approximate p value for your null hypothesis. (2) d. Assume that the population standard deviation is 0.10 and create a 94% confidence interval for the mean. (2) e. (Extra Credit) Given the data, is 0.10 a reasonable value for the population standard deviation? Original data Solution: x 2 266 .4394 , n 9 x 48 .96 , x Row a) x2 1 2 3 4 5 6 7 8 9 x 48.96 5.4400 x 5.35 28.6225 5.68 32.2624 5.35 28.6225 5.47 29.9209 5.31 28.1961 5.49 30.1401 5.41 29.2681 5.48 30.0304 5.42 29.3764 48.96 266.4394 n sx2 x 9 2 nx 2 n 1 266 .4394 95.4400 2 8 0.0965 0.0120625 8 sx .0120625 0.109828 . b) This material is edited from the solution to problem 9.54. Given: 0 5.5, s 0.109828 n 9, s df n 1 8, x 5.4400 and .05. So s x n 0.109828 9 0.03661 . Note that tn 1 t..805 1.860 . Note that ‘Below 5.5’ does not contain an equality, so that it must be an alternative hypothesis. Our hypotheses are H 0 : 5.5 and H1 : 5.5 . There are three ways to do this. You should have used one. x 0 5.4400 5.5 1.639 . The smaller the sample mean is, the more sx 0.03661 negative will be this ratio. We will reject the null hypothesis if the ratio is smaller than tn1 t..805 1.860 . Make a diagram showing a Normal curve with a mean at 0 and a shaded 'reject' zone below -1.860. Since the test ratio is not below -1.860, we do not reject H 0 . (ii) Critical value: We need a critical value for x below 5.5. Common sense says that if the sample mean is too far below 5.5, we will not believe H 0 : 5.5 . The formula for a critical value for the sample mean is x t n1 s , but we want a single value below 5.5, so use (i) Test Ratio: t cv 0 2 x xcv 0 tn 1sx 5.5 1.860 0.03661 5.4319 . Make a diagram showing an almost Normal curve with a mean at 5.5 and a shaded 'reject' zone below 5.4319. Since x 5.4400 is not below 5.4319, we do not reject H 0 . (iii) Confidence interval: x t sx is the formula for a two sided interval. The rule for a one2 sided confidence interval is that it should always go in the same direction as the alternate hypothesis. Since the alternative hypothesis is H1 : 5.5 , the confidence interval is x t n1s 5.4400 1.860 0.03661 5.5081 . Make a diagram showing an almost x Normal curve with a mean at x 5.4400 and, to represent the confidence interval, shade the area below 5.5081 in one direction. Then, on the same diagram, to represent the null hypothesis, H 0 : 5.5 , shade the area above 5.5 in the opposite direction. Notice that these do overlap. 2 252y0612 10/18/06 (Open in ‘Print Layout’ format) What the diagram is telling you is that it is possible for 5.5081 and H 0 : 5.5 to both be true. (If you follow my more recent suggestions, it is actually enough to show that 5.5 is on the interval.) So we do not reject H 0 . c) (Extra credit) If you wish to use a p-value approach, since this is a left-side test, the p-value is the probability of getting a value below 5.4067 when the null hypothesis is true. If we use the test ratio above we get. pval Px 5.4400 Pt 1.639 . To find this approximately, go to the df 8 line of the t table. 1.639 is between t 8 1.860 and t 8 1.397 . This means that P t 8 1.860 .05 and ..05 .. 10 P t 8 1.397 .10 . Since -1.639 is between these values, we can conclude that 05 pval .10. If we want to use this in b), we can say that since the p-value is not below .05, we do not reject the null hypothesis. Note that this is confirmed by Minitab, with more accurate computations. The column that you are using is t1, its mean is 5.40667, a confidence interval is 5.48239 , the t-ratio is equal to -2.29 and it gives a pvalue of .026. One-Sample T: t2 Test of mu = 5.5 vs < 5.5 Variable t1 N 9 Mean 5.44000 StDev 0.11011 SE Mean 0.03670 95% Upper Bound 5.50825 T -1.63 P 0.070 d) In this section, we assume that the population standard deviation is 0.10 and create a 92% confidence interval for the mean. Here 1 .92 .08. We do this by observing that the formula for a 2-sided confidence interval when the population standard deviation is known is x z x . Fortunately, we 2 0.10 0.0333 . The confidence interval is thus n 9 5.4400 1.750.0333 5.4400 0.0583 or 5.38 to 5.50. e) (Extra Credit) Given the data, we want to see if 0.10 a reasonable value for the population standard deviation? H 0 : 0.10 and H1 : 0.10 . We will assume .05 and use the chi – squared formula, found out on Page 1 that z .04 1.75 . x 2 n 1s 2 02 80.0120625 2 8 8 9.684 . We test this against 2.975 2.1797 and 2..025 17.5346. .10 Since it is between them we cannot reject the null hypothesis. 3 252y0612 10/18/06 (Open in ‘Print Layout’ format) III. Do as many of the following problems as you can. (2 points each unless marked otherwise adding to 18+ points). Show your work except in multiple choice questions. (Actually – it doesn’t hurt there either.) If the answer is ‘None of the above,’ put in the correct answer. 1. Which of the following is a Type 1 error? a) Rejecting the null hypothesis when the null hypothesis is false. b) *Rejecting the null hypothesis when the null hypothesis is true. c) Not rejecting the null hypothesis when the null hypothesis is true. d) Not rejecting the null hypothesis when the null hypothesis is false. e) All of the above f) None of the above. 2. A librarian provides a confidence interval estimate for the mean number of books checked out daily. The estimate is 330 to 740. The point estimate that this interval is based on is. a) 1070. b) 740 c) *535 d) 330 e) None of the above f) There is not enough information to tell. Explanation: Since the sample mean sits in the middle of a confidence interval, it must be an average of the two end points. 3. (BLK8.30) We want a 99% confidence interval for the average income of in a town, and the population standard deviation is known to be $1000. We have taken a sample of size 50 earlier and found that the sample mean is $20000. What sample size should we take if the width of the interval is to be no more than $100? a) *2655 b) 2654 c) 665 d) 664 e) 50 f) 40 g) 20 h) None of the above. z 2 2 2.576 2 1000 2 Explanation: The formula in the outline is n 2 2654 .3. This is always e 50 2 rounded up – in this case to 2655. 4. An entrepreneur is considering the purchase of a coin-operated laundry. The present owned claims that over the past 6 years the average daily income was $700. A sample is taken of daily revenue over a period of 28 days. Statistics are computed from the sample. If we want to test the statement that the mean is $700, which of the following tests is most appropriate? (1 point changed to 2) a) z -test of a population mean. b) z -test of a population proportion. c) * t -test of a population mean. d) 2 -test of a population variance. e) F -test. f) All of the above could be used. g) We do not have enough information. [7] 4 252y0612 10/18/06 (Open in ‘Print Layout’ format) 5. We are trying to estimate the median income in a region. We wish to test if the median is over $30000. We do not know the population variance. We can compute a statistic or statistics from a sample of 27 incomes. To do this test, the statistic or statistics we need is (are) a) The proportion of incomes that are above the sample mean, x . b) The proportion of incomes that are above the sample median x.50 c) *The number of incomes in the sample that are above $30000. d) The sample mean, x . e) The sample mean, x and the sample variance s 2 . f) The sample median x.50 . g) The sample variance s 2 . [9] 6. A restaurant company only opens restaurants in areas with a mean family income above $45000. The company takes a sample of 144 families in Hotzeplotz and finds a sample mean income of $46396 and a sample standard deviation of $5400. Should they open? .05 a) State your null and alternative hypotheses. (1) b) Test your hypothesis in a) by finding a critical value for the sample mean. Can we say that the result of the sample is above $45000? Why? (3) c) Do a 2-sided confidence interval for the mean. (2) d) Do a 2-sided confidence interval for the variance. (2) [17] e) (Extra credit) Explain in as much detail as reasonable how you would find a confidence interval for the median. (2) Solution: This material is Part II all over again. a) Note that ‘above $45000’ does not contain an equality, so that it must be an alternative hypothesis. Our hypotheses are H 0 : 45000 and H1 : 45000 . Given: 0 45000 s 5400 , n 144 , df n 1 143 , x 46396 and .05 . So 5400 450 . tn 1 t..143 05 1.656 . 12 n 144 b) We need a critical value for x above 45000. Common sense says that if the sample mean is too far above 45000, we will not believe H 0 : 45000 . The formula for a critical value for the sample mean is x t n1 s , but we want a single value above 45000, so use x t n1s s sx cv 5400 0 x 2 cv 0 x 45000 1.656 450 45745 .20 . Make a diagram showing an almost Normal curve with a mean at 45000 and a shaded 'reject' zone above 45745.20. Since x 46396 is above 45745.20, we reject H 0 . c) Confidence interval for the mean: x t sx is the formula for a two sided interval. We need 2 143 t.025 1.977 . 45000 1.977 450 45000 899 .65 . We can say that P44110 .35 45899 .65 .95 . d) Confidence interval for the variance: The outline says that for small samples the formula is n 1s 2 2 2 2 s 2DF z 2 2DF Since n 1s 2 12 2 , but if the degrees of freedom are too large for the chi-square table use s 2DF z 2 2DF . Using the second formula 2143 286 16 .912 , we have 5400 2143 1.96 2143 5400 2143 1.96 2143 . 5400 16.912 5400 16 .912 , which becomes 1.96 16.912 1.96 16.912 5400 16 .912 5400 16 .912 and finally 4839 .18 6107 .87 . 18 .872 14 .952 5 252y0612 10/18/06 (Open in ‘Print Layout’ format) As I have said in class, no one really wants a confidence interval for the variance, but if you square the above you get 23417558 2 37306017 e) Confidence interval for the median: If we remember n 144 , n 1 z . 2 n we can use the formula given in the 144 1 1.960 144 60 .74 . To be conservative, round 2 2 this down to 60. Put the numbers in order by size as x1 to x144 . The interval will be x60 to x85 . (Note that the index of the second number is 144 + 1 – 60 = 85. outline for large samples. k 7. Return to the problem in Question 6. Let’s say that the sample mean payment of $46396 and the sample standard deviation of $5400 come from a sample of 81. Assume that your null hypothesis is correct and that you get a t-ratio of 2.327. What should you say that the p-value is? (3 points). Note that showing your calculations here could get you partial credit. a) Exactly .01 b) Exactly .02 c) *Between .025 and .01 d) ***Between .05 and .02 e) Between .01 and .005 f) Between .02 and .01 g) Exactly .99 h) Exactly .98 i) **Between .975 and .99 j) Between .95 and .98 k) Between .98 and .99 l) None of the above – Show your answer! [20] 80 Solution: If you have 80 degrees of freedom, the t-table says t..80 025 1.990 , t..01 2.374 , 80 80 t..005 2.639 , and t..001 3.195 . This means that Pt 1.990 .025 and Pt 2.374 .01 . Since 2.327 lies between these values, .01 Pt 2.327 .025 *If you said that your hypotheses were H 0 : 45000 and H1 : 45000 , you have a right sided test and pvalue Pt 2.327 , which will be between .01 and .025. **However, if you said that your hypotheses were H 0 : 45000 and H1 : 45000 , you have a left-sided test and pvalue Pt 2.327 , which will be between .975 and .99. ***If you said that your hypotheses were H 0 : 45000 and H1 : 45000 , you have a twosided test and pvalue 2Pt 2.327 , which will be between .02 and .05. I can only promise to look at your reasoning if you have any other answer. 6 252y0612 10/18/06 (Open in ‘Print Layout’ format) 8. (Mann) According to a 1992 survey, 45% of workers say that they would change careers if they could. You wish to show that the proportion of workers in your company that want to change careers is below the national figure. Find the correct set of hypotheses below. a) H 0: p .45 H1 : p .45 b) H 0: p .45 H1 : p .45 c) H 0: p .45 H1 : p .45 d) H : p .45 * 0 H1 : p .45 e) H 0: p .45 * H1 : p .45 f) H 0: p .45 ** H1 : p .45 g) H 0: p .45 ** H1 : p .45 h) None of the above. Put in your answer! 9. [22] (Mann) According to a 1992 survey, 45% of workers say that they would change careers if they could. You wish to show that the proportion of workers in your union that want to change careers is below the national figure. Assume that your null hypothesis in 8 is correct, that you take a sample of 350 workers and that you compute the ratio z 90%, you should do the following. a) b) c) d) e) f) g) h) p .45 . If your confidence level is .45.55 350 Reject the null hypothesis if the ratio is not between -1.96 and 1.96. Reject the null hypothesis if the ratio is not between -1.645 and 1.645 Reject the null hypothesis if the ratio is above 1.645 Reject the null hypothesis if the ratio below -1.645 Reject the null hypothesis if the ratio is below 1.282 **Reject the null hypothesis if the ratio is above 1.282 *Reject the null hypothesis if the ratio is below -1.282 None of the above. Put in your answer! [24] Solution: This is clearly a one-sided test and .10 . The table says z .10 1.282 . H 0: p .45 **If you said (wrongly) , you have a right-sided test and you reject the null hypothesis if H1 : p .45 z 1.282 . H 0: p .45 *However, if you said , you have a left-sided test and you would reject the null hypothesis if H1 : p .45 z 1.282 . Note that for a t or z ratio, zero is never in the reject region, so that e) could never be true. 7 252y0612 10/18/06 (Open in ‘Print Layout’ format) 10. We wish to use a 2-sided confidence interval to test a proportion. The following things may influence the size of a confidence interval: (.5 points for good ones, .5 off for bad ones, at worst zero) 1. Decreasing x to make it closer to .5n 2. Increasing x from .85 n to .9n 3. Changing the null hypothesis to make p 0 closer to .5 4. Increasing the sample size. (Assume a large population) 5. Decreasing the sample size. . (Assume a large population) 6. Increasing the confidence level 7. Increasing the significance level 8. Using a continuity correction with a relatively small sample. 9. Decreasing the population size from 15 n to 14 n 10. Increasing the population size from 5n to 6 n Put down the numbers of the things that make the confidence interval smaller. ________________ Solution: The confidence interval has the following form: p p z s p , where p 2 x and, if we include n N n pq . N 1 n 1. Decreasing x to make it closer to .5n . This makes p closer to .5. It was explained in class that pq becomes larger as p approaches .5. This makes the interval larger. 2. Increasing x from .85 n to .9n . This makes p farther from .5 and makes the interval smaller. a finite sample correction, the standard error is s p 3. Changing the null hypothesis to make p 0 closer to .5. p 0 does not appear in the formula for a confidence interval, so it has no effect. 4. Increasing the sample size. (Assume a large population.) Since n , the sample size, is in the pq , it makes the interval smaller. n 5. Decreasing the sample size. (Assume a large population.) This makes the interval larger. 6. Increasing the confidence level. If we raise the confidence level 1 , the significance level denominator of s p gets smaller. This gives us a larger value of z 2 , so it makes the interval larger. 7. Increasing significance level . This makes the interval smaller. 8. Using a continuity correction. This has the effect of making all intervals slightly larger. It has little effect with a large sample. 9. Decreasing the population size from 15 n to 14 n . If we use a finite population correction, it has the general effect of decreasing the standard error. Decreasing the population size makes its effects stronger, so it makes the interval smaller. 10. Increasing the population size from 5n to 6 n . Let the size of the population be an , where a is a value from 1 to 20. Then the finite population correction is N n N 1 an n a 1 . This n 1 1 1 n obviously grows as a grows. This makes the interval larger. Summarizing: Larger: 1, 5, 6, 8, 10 Smaller: 2, 4, 7, 9 No effect: 3 8 252y0612 10/18/06 (Open in ‘Print Layout’ format) ECO252 QBA2 FIRST EXAM October 11-12 2006 TAKE HOME SECTION Name: _________________________ Student Number and class: _________________________ IV. Do at least 3 problems (at least 7 each) (or do sections adding to at least 20 points - Anything extra you do helps, and grades wrap around) . Show your work! State H 0 and H 1 where appropriate. You have not done a hypothesis test unless you have stated your hypotheses, run the numbers and stated your conclusion.. (Use a 95% confidence level unless another level is specified.) Answers without reasons usually are not acceptable. Neatness and clarity of explanation are expected. This must be turned in when you take the in-class exam. Note that from now on neatness means paper neatly trimmed on the left side if it has been torn and paper written on only one side. 1. (Moore, Notz) You are thinking that it may be desirable to start a wellness program for your (large) company. You are told that the company will only start such a program if you can show that the blood pressure of a group of mid-level executives is above normal. The individuals are all between 35 and 44 years old and US statistics show that mean systolic blood pressure for men in that age range is 128. You take a sample of 72 executives and get the following results. x 139.60 136.21 114.18 128.45 120.68 127.51 128.07 161.43 161.09 130.61 130.15 154.74 126.29 124.48 117.51 138.22 169.05 116.14 182.53 141.96 122.14 163.18 124.61 100.03 130.77 140.35 158.74 120.02 137.23 127.22 141.54 105.67 149.55 109.52 131.40 126.54 118.77 141.15 150.30 126.93 144.71 127.32 136.69 125.06 135.21 149.44 133.89 118.37 124.80 133.00 131.74 135.69 169.61 126.71 107.30 122.73 125.35 152.64 109.62 116.59 132.00 117.84 120.01 117.47 145.25 159.94 112.34 145.10 119.39 127.67 117.97 112.40 To personalize the data below take the last digit of your student number, divide it by 10 and add it to the numbers below. If the last digit of your student number is zero, add 1.00. Label the problem ‘Version 1,’ ‘Version 2,’ … ‘Version 10’ according to the number that you used. (For example, Seymour Butz’s student number is 976502, so he will add 0.20 and change the data to 139.80, 130.81, 182.73 etc. – but see the hint below, you do not need to write down all the numbers that you are using, just your computations.) x 9528 .41 Hint - if you use the computational formula: For the original numbers n 72 , and x 2 1280763 .7 . If you add a quantity a to a column of numbers, x a x na, x a x 2a x na 2 2 2 Assume that the Normal distribution applies to the data and use a 99% confidence level. a. Find the sample mean and sample standard deviation of the incomes in your data, showing your work. (1) (Your mean should be fairly near 132 and your sample standard deviation should be near 16 or 17.) b. State your null and alternative hypotheses (1) c. Test the hypothesis using a test ratio (1) d. Test the hypothesis using a critical value for a sample mean. (1) e. Test the hypothesis using a confidence interval (1) f. Find an approximate p-value for the null hypothesis. (1) g. On the basis of your tests, will you get a wellness program? Why? (1) h. How do your conclusions change if the sample of 72 is taken from a population of 200? (2) i. Assume that the Normal distribution does not apply and, using your data, test that the median is above 128. (3) [12] j. (Extra credit) Use your data to an approximate 90% 2-sided confidence interval for the median. 9 252y0612 10/18/06 (Open in ‘Print Layout’ format) 2. Once again, assume that the Normal distribution applies, but assume a population standard deviation of 16 and that we are testing whether the mean is above 128. (90% confidence level) a. State your null and alternative hypotheses(1) b. Find a p-value for the null hypothesis using the mean that you found in a. On the basis of your p-value, would you reject the null hypothesis? Why? (1) c. Create a power curve for the test. (6) [20] 3. (Moore, Notz) A recent survey said that nationwide 73% of all freshman students identified being well-off financially as an important lifetime goal. You believe that the proportion of freshman business majors with that goal is higher than the national figure. You take a survey of a random sample of 200 students and find that 152 a have being well-off as an important goal, where a is the second to last digit of your student number. If the second to last digit of your student number is zero, a 10 . (For example, Seymour Butz’s student number is 976502, so he will add 10 and say that x 152 10 162 ) a. Formulate your null and alternative hypotheses and do a hypothesis test with a 95% confidence level. (2) b. Find a p-value for the null hypothesis. (1) c. Find the p-value for the null hypothesis if x 162 a (1) d(c). (Extra credit) How would your answer to a) change if your sample of 200 came from a population of 300? (1) e(d). (Extra credit) Using a critical value of the proportion for testing your null hypothesis, create a power curve for the test by using the alternate hypothesis and finding the power for values of p1 .73 . (Up to 6 points) f(d). Assume that p .73 , how large a sample would you need to estimate the proportion above that have being well-off with an error of .005? (2) g(e). Use the proportion that you found in a) to create a 2-sided 95% confidence interval for the proportion. Does it differ significantly from .73? Why? (2) [28] 4. Standard deviation is often a measure of reliability. A manufacturer is providing a connector with a mean length of 2.5 mm and is getting complaints that the connector is often too large or too small for the intended use. The previous standard deviation to the length of the part was 0.025mm, but the manufacturer introduces a process that should make the standard deviation smaller. A sample of 25 items is taken which yields a sample standard deviation of 0.030 a . To get a take the third to last digit of your student number and multiply it by 0.001. (For example, Seymour Butz’s student number is 976502, so he will subtract .005 and say that s 0.030 0.005 0.025 . ) a. Formulate the null and alternative hypotheses necessary to see if the goal has been achieved and test the hypothesis using a 95% confidence level and a test ratio. (2) b. What assumptions are necessary to perform this test? (1) c. Try to get a rough p-value. Interpret its meaning (1.5) d(c). Do a 95% two- sided confidence interval for the standard deviation (1) e(d). (Extra credit) Redo 4a) using an appropriate confidence interval. (2) f(e). (Extra credit) Find a critical value for s in 4a). (1) g(f). The number of claims for missing baggage in a large metropolitan airport supposedly follows a Poisson distribution with a mean of 72 per week. Assume that in a given week 92 are lost. Test this hypothesis using a test ratio and a 95% confidence level. (2) h(g). Find approximate critical values for the number of bags that could be lost in 4f. (2) i(h). (Extra credit) Find the power of the test in 4g) if the average number of lost bags per week is really 87. (3) j(i). I claim that x is binomially distributed with p .01 . Test this assertion using a 2-sided test if there are 4 successes in 10 trials. (2) [39.5] 10 252y0612 10/18/06 (Open in ‘Print Layout’ format) 1. (Moore, Notz) You are thinking that it may be desirable to start a wellness program for your (large) company. You are told that the company will only start such a program if you can show that the blood pressure of a group of mid-level executives is above normal. The individuals are all between 35 and 44 years old and US statistics show that mean systolic blood pressure for men in that age range is 128. You take a sample of 72 executives and get the following results. x 139.60 136.21 114.18 128.45 120.68 127.51 128.07 161.43 161.09 130.61 130.15 154.74 126.29 124.48 117.51 138.22 169.05 116.14 182.53 141.96 122.14 163.18 124.61 100.03 130.77 140.35 158.74 120.02 137.23 127.22 141.54 105.67 149.55 109.52 131.40 126.54 118.77 141.15 150.30 126.93 144.71 127.32 136.69 125.06 135.21 149.44 133.89 118.37 124.80 133.00 131.74 135.69 169.61 126.71 107.30 122.73 125.35 152.64 109.62 116.59 132.00 117.84 120.01 117.47 145.25 159.94 112.34 145.10 119.39 127.67 117.97 112.40 To personalize the data below take the last digit of your student number, divide it by 10 and add it to the numbers below. If the last digit of your student number is zero, add 1.00. Label the problem ‘Version 1,’ ‘Version 2,’ … ‘Version 10’ according to the number that you used. (For example, Seymour Butz’s student number is 976502, so he will add 0.20 and change the data to 139.80, 130.81, 182.73 etc. – but see the hint below, you do not need to write down all the numbers that you are using, just your computations.) x 9528 .41 Hint - if you use the computational formula: For the original numbers n 72 , and x 1280763 .7 . If you add a quantity a to a column of numbers, 2 x a x na, x a x 2a x na 2 2 2 Assume that the Normal distribution applies to the data and use a 99% confidence level. a. Find the sample mean and sample standard deviation of the incomes in your data, showing your work. (1) (Your mean should be fairly near 132 and your sample standard deviation should be near 16 or 17.) Solution: For version 0, the mean is 132.339. For version 10, the mean is 133.339. All computations for the sample variance and the sample standard deviation should give identical answers. The solution for version 0 follows. .01 . x 9528 .41 , x 2 1280763 .7 , n 72 . This means x x 9528 .41 132 .339 . n 72 Using the computational formula, we get the following. x 2 nx 2 1280763 .7 72132 .339 2 19783 .71 278 .6439 . So sx2 71 n 1 71 sx 278 .6439 16 .6926 . If you used the definitional formula, you should get x x 2 sx 19783.71 . Computations are available in an appendix. s2 n 278 .6439 3.870054 1.9672 72 b. State your null and alternative hypotheses (1) Note that ‘Above normal (128)’ does not contain an equality, so that it must be an alternative hypothesis. Our hypotheses are H 0 : 128 and H 1 : 128 . c. Test the hypothesis using a test ratio (1) 11 252y0612 10/18/06 (Open in ‘Print Layout’ format) x 0 132 .339 128 133 .339 128 2.7140 2.2057 . Version 10: t 1.9672 sx 1.9672 The larger the sample mean is, the more positive will be this ratio. We will reject the null hypothesis if the ratio is larger than tn 1 t ..71 01 2.380 . Make a diagram showing a Normal curve with a mean at 0 and a shaded 'reject' zone above 2.380. Since the test ratio is below 2.2057 for Version 0, we do not reject H 0 . Since the test ratio is above 2.380 for Version 10, we reject H . Note: t n 1 t 71 2.647 . Version 0: t 0 2 ..005 d. Test the hypothesis using a critical value for a sample mean. (1) Critical value: We need a critical value for x above 128. Common sense says that if the sample mean is too far above 128, we will not believe H 0 : 128 . The formula for a critical value for the sample mean is x t n1 s , but we want a single value above 128, so use cv 0 2 x xcv 0 tn1 s x 128 2.380 1.9672 132 .682 . Make a diagram showing an almost Normal curve with a mean at 128 and a shaded 'reject' zone above 132.682. For Version 0, since x 132 .339 is below 132.682, we do not reject H 0 . For Version 10, since x 133 .339 is above 132.682, we reject H 0 . e. Test the hypothesis using a confidence interval (1) Confidence interval: x t sx is the formula for a two sided interval. The rule 2 for a one-sided confidence interval is that it should always go in the same direction as the alternate hypothesis. Since the alternative hypothesis is H 1 : 128 , the confidence interval for Version 0 is x t n1 s 132 .339 2.380 1.9672 127 .657 . Make a diagram x showing an almost Normal curve with a mean at x 132 .339 and, to represent the confidence interval, shade the area below above 127.657 in one direction. Then, on the same diagram represent H 0 : 128 by shading the area below 128 or simply showing 128. Since there are values below or equal to 128 that are also above or equal to 127.657, we do not reject H 0 . The confidence interval for Version 10 is 133 .339 2.380 1.9672 128 .657 . Make a diagram showing an almost Normal curve with a mean at x 133 .339 and, to represent the confidence interval, shade the area below above 128.657 in one direction. Then, on the same diagram represent H 0 : 128 by shading the area below 128 or simply showing 128. Since there are no values below or equal to 128 that are also above or equal to 127.657, we reject H 0 . f. Find an approximate p-value for the null hypothesis. (1) x 0 132 .339 128 133 .339 128 2.7140 2.2057 . Version 10: t Version 0: t 1.9672 sx 1.9672 71 71 71 71 1.994 , t .01 2.647 and t .001 2.380 , t .005 3.209 . So There are 71 degrees of freedom. t .025 for version 0, .01 pvalue .025 and for version 10, .001 pvalue .005 . g. On the basis of your tests, will you get a wellness program? Why? (1) . If you rejected the null hypothesis, you have shown above normal blood pressure and you get your program. Otherwise you do not. h. How do your conclusions change if the sample of 72 is taken from a population of 200? (2) sx N n N 1 s2 n 200 72 (1.9672 ) 0.6432 1.9672 .8020 1.9672 1.5777 200 1 12 252y0612 10/18/06 (Open in ‘Print Layout’ format) Version 0: t x 0 132 .339 128 133 .339 128 3.3840 2.7502 . Version 10: t 1.5777 sx 1.5777 tn 1 t ..71 01 2.380 . Make a diagram showing a Normal curve with a mean at 0 and a shaded 'reject' zone above 2.380. Since both test ratios are now above 2.2057, we reject H 0 in all cases. The Minitab output for this problem reads as follows. One-Sample T: s0 Test of mu = 128 vs > 128 Variable s0 N 72 Mean 132.339 99% Lower StDev SE Mean 16.692 1.967 Bound 127.657 T 2.21 P 0.015 Bound 128.657 T 2.71 P 0.004 One-Sample T: s10 Test of mu = 128 vs > 128 Variable s10 N 72 Mean 133.339 99% Lower StDev SE Mean 16.692 1.967 i. Assume that the Normal distribution does not apply and, using your data, test that the median is above 128. (3) [12] Our alternative hypothesis is H 1 : 128 , so that our null hypothesis is H 0 : 128 . See appendix for the values in order. Let x be the number of values above 128. x nx x p n Version Version Version Version Version Version Version Version Version Version Version 0 1 2 3 4 5 6 7 8 9 10 below 128 35 35 35 35 34 33 33 32 31 31 31 above 128 37 37 37 37 38 39 39 40 41 41 42 .5138 .5138 .5138 .5138 .5278 .5417 .5417 .5555 .5694 .5694 .5833 In the outline section on the sign test we have the following. Hypotheses about Hypotheses about a proportion a median If p is the proportion If p is the proportion above 0 below 0 H 0 : 0 H 1 : 0 H 0 : p .5 H 1 : p .5 H 0 : p .5 H 1 : p .5 H 0 : 0 H 1 : 0 H 0 : 0 H 1 : 0 H 0 : p .5 H 1 : p .5 H 0 : p .5 H 1 : p .5 H 0 : p .5 H 1 : p .5 H 0 : p .5 H 1 : p .5 13 252y0612 10/18/06 (Open in ‘Print Layout’ format) H : p .5 H : 128 So our hypotheses 0 correspond to 0 H 1 : p .5 H 1 : 128 x .5 p p0 To compute a test ratio, we use z n p0 q0 .25 n n x n .5 .5 2x n n . Note that z.01 2.327 n .25 .25 .00347222 .05893 and n 72 8.48528 . n 72 So compute the test ratio and reject the null hypothesis if the test ratio is above 2.327. With two different formulas, it is not surprising that we get slightly different answers because of rounding error. x .5 x (number 2x n x n p above128) n n .25 and that n 72 , so that n above 128 Version 0 37 .5138 Version 1 Version 2 Version 3 37 37 37 .5138 .5138 .5138 Version 4 38 .5278 Version 5 39 .5417 Version 6 39 .5417 Version 7 40 .5555 Version 8 41 .5694 Version 9 41 .5694 z 237 72 .5138 .5 .2342 or z .2357 .05893 8.48532 .2342 .2342 .2342 .2357 .2357 .2357 .7076 .7071 1.17777 1.1785 .5278 .5 238 72 .4717 or z .4714 .05893 8.48532 .5417 .5 239 72 z .7076 or z .7071 .05893 8.48532 z .5555 .5 240 72 z .9418 or z .9428 .05893 8.48532 .5694 .5 241 72 z 1.1777 or z 1.1785 .05893 8.48532 .5833 .5 242 72 .1.4220 or z 1.4142 Version 10 42 .5833 z .05893 8.48532 Though these values are all in the ‘do not reject’ zone, note that as x increases, , so does z . However, it never reaches the ‘reject’ zone. j. (Extra credit) Use your data to get an approximate 90% 2-sided confidence interval for the median. Note that z.025 1.645 . If the index of the lower number is k , the index of the upper number will be n 1 k . The large sample formula for the index of the lower number in a confidence interval for the median is n 1 z . 2 n 72 1 1.645 72 k =29.52, and n 1 k 72 1 29 44 , so use x29 and 2 2 x44 . If we put the numbers in order we have the following limits. 14 252y0612 10/18/06 (Open in ‘Print Layout’ format) x29 Version Version Version Version Version Version Version Version Version Version Version 2. 0 1 2 3 4 5 6 7 8 9 10 126.54 126.64 126.74 126.84 126.94 127.04 127.14 127.24 127.34 127.44 127.54 x44 133.00 133.10 133.20 133.30 133.40 133.50 133.60 133.70 133.80 133.90 134.00 Once again, assume that the Normal distribution applies, but assume a population standard deviation of 16 and that we are testing whether the mean is above 128. (90% confidence level) a. State your null and alternative hypotheses(1) b. Find a p-value for the null hypothesis using the mean that you found in a. On the basis of your p-value, would you reject the null hypothesis? Why? (1) c. Create a power curve for the test. (6) Solution: The problem says that 16 , n 72 , H 1 : 128 and .10 H : 128 16 16 2 3.55555 1.8856 and a. Our hypotheses are 0 and we need x 72 72 H 1 : 128 z z.10 1.282 . 132 .24 128 b. Assume that the sample mean is 132.34. The p-value is Px 132 .4 P z 1.8856 Pz 2.25 .5 P0 z 2.25 .5 .4878 .0122 . Since the p-value is below .10 , we reject the null hypothesis. c. First we need a critical value. This is a 1-sided test and since the alternate hypothesis is H 1 : 128 , we need a critical value above 128. The two-sided formula is xcv 0 z x , and we will use 2 x cv 0 z x 128 1.282 1.8856 130 .42 . This is a right-sided test, so we will use points to the right of 128. Make a diagram: Show a Normal curve with a mean at 128 and a critical value at 130.42. The ‘reject’ zone is above 130.42. If the population mean is, in fact, above 128, and we do not reject the null hypothesis because the sample mean is below 130.42, we make a Type II error. The probability of such 130 .42 1 an error for a population mean of 1 128 is Px 130.42 1 P z . The points that 1.8856 I will use are 128, 129.21, 130.42, 131.63 and 132.84. I picked these because 129.21 is midway between 128 and the critical value of the sample mean. Because the first three points are apart by 1.21, I picked the last two points by adding 1.21 to the previous point. 130 .42 128 Px 130 .42 128 P z Pz 1.28 .5 .3997 .8997 1 1 128 1.8856 Power 1 .8997 .1003 130 .42 129 .21 Px 130 .42 129 .21 P z 1 129.21 Pz 0.64 .5 .2389 .7389 1.8856 Power 1 .7389 .2611 130 .42 130 .42 Px 130 .42 130 .42 P z 1 130.42 Pz 0 .5 1.8856 Power 1 .5 .5 (Power is always .5 at the critical value.) 15 252y0612 10/18/06 (Open in ‘Print Layout’ format) 130 .42 131 .63 Px 130 .42 131 .63 P z Pz 0.64 .5 .2389 .2611 1.8856 Power 1 .2611 .7389 130 .42 132 .84 Px 130 .42 132 .84 P z 1 132.84 Pz 1.28 .5 .3997 .8997 1.8856 Power 1 .8997 .1003 If we want to try for higher power, add 1.21 to 132.84 and get 134.05 130 .42 134 .05 1 134.05 Px 130 .42 134 .05 P z Pz 1.92 .5 .4726 .9726 1.8856 1 131.63 3. (Moore, Notz) A recent survey said that nationwide 73% of all freshman students identified being well-off financially as an important lifetime goal. You believe that the proportion of freshman business majors with that goal is higher than the national figure. You take a survey of a random sample of 200 students and find that 152 a have being well-off as an important goal, where a is the second to last digit of your student number. If the second to last digit of your student number is zero, a 10 . (For example, Seymour Butz’s student number is 976502, so he will add 10 and say that x 152 10 162 ) a. Formulate your null and alternative hypotheses and do a hypothesis test with a 95% confidence level. (2) b. Find a p-value for the null hypothesis. (1) c. Find the p-value for the null hypothesis if x 162 a (1) d. (Extra credit) How would your answer to a) change if your sample of 200 came from a population of 300? (1) e. (Extra credit) Using a critical value of the proportion for testing your null hypothesis, create a power curve for the test by using the alternate hypothesis and finding the power for values of p1 .73 . (Up to 6 points) f. Assume that p .73 , how large a sample would you need to estimate the proportion above that have being well-off with an error of .005? (2) g. Use the proportion that you found in a) to create a 2-sided 95% confidence interval for the proportion. Does it differ significantly from .73? Why? (2) [28] Solution: The problem states that we are testing p .73 , when n 200 and .05 In Seymour’s case x 162 or p 162 .81 and we can compute 200 p0 q0 .73.27 .009855 0.031393 , where p0 comes from our null hypothesis and n 200 q0 1 p0 . a. Since p .73 does not contain an equality, it must be an alternative hypothesis our hypotheses are p H 0 : p .73 . This is a right- sided test since values of the observed proportion above .73 are the only H 1 : p .73 values that could lead to a rejection of the null hypotheses. It is most expedient to use a test ratio, p p0 .81 .73 z 2.55 . Make a diagram. Show a normal curve with a mean at zero and an area of p .031393 5% above z.05 1.645 . The area above 1.645 is the ‘rejection zone.’ Since our value of z is above 1.645, it is in the ‘rejection zone,’ and we reject the null hypothesis. 16 252y0612 10/18/06 (Open in ‘Print Layout’ format) b. We find a p-value for the null hypothesis. pvalue P p .81 Pz 2.55 .5 P0 z 2.55 .5 .4946 .0054 . Since this is below .05 , we reject the null hypothesis. 162 a 152 .76 . (This should have 200 200 p p0 .76 .73 been x 152 a , and it should have asked for a conclusion.) z 0.96 p .031393 c. We find the p-value for the null hypothesis if x 162 a . p pvalue P p .76 Pz 0.96 .5 P0 z 0.96 .5 .3315 .1685 . Since this is not below .05 , we do not reject the null hypothesis. d. We show how our answer to a) would change if our sample of 200 came from a population of 300. If our sample of 200 came from a population of 300, the standard error would be much smaller. .73.27 .33445 .009855 .5783149 0.031393 .018155 200 p p0 .81 .73 This would lead to a much larger value of z 4.41 . The p-value would be much p .018155 p N n N 1 p0 q0 300 200 n 300 1 smaller, but we would still reject the null hypothesis. e. (Extra credit) We use a critical value of the proportion for testing the null hypothesis and create a power curve for the test by using the alternate hypothesis and finding the power for values of p1 .73 . To do this, remember p H 0 : p .73 . The critical H 1 : p .73 becomes pcv p0 z p .73 1.645.031393 .7816 p0 q0 0.031393 , .05 , z.05 1.645 and n value is above .73, so that pcv p0 z p 2 This is a right-sided test, so we will use points to the right of 0.73. Make a diagram: Show a Normal curve with a mean at 0.73 and a critical value at .7816. The ‘reject’ zone is above .7816. If the population proportion is, in fact, above .73, and we do not reject the null hypothesis because the sample mean is below .7816, we make a Type II error. The probability of such an error for a population proportion of p 1 .73 is .7816 p1 p1q1 , where p . The points that I will use are .73, .7558, P p .7816 p p1 P z 200 p .7816, .8074 and .8332. I picked these because .7558 is midway between .73 and the critical value of the sample proportion. Because the first three points are apart by .0258, I picked the last two points by adding .0258 to the previous point. p1q1 .031393 p1 .73 , p 200 .7816 .73 P z Pz 1.644 .5 .4500 .95 1 p Power 1 .95 ..05 p1 .7558, p .7558 .2442 .0009228 .0303781 200 .7816 .7558 P z Pz 0.85 .5 .3032 .8032 p Power 1 .8032 .1968 17 252y0612 10/18/06 (Open in ‘Print Layout’ format) .7816 .2184 .0008535 .0292148 200 p1 .7816, p Power 1 .5 .5 p1 .8074, p .7816 .7816 P z p Pz 0 .5 .8074 .1926 .00077753 .0278842 200 .7816 ..8074 P z Pz 0.93 .5 .3238 .1762 p Power 1 .1762 .8238 p1 .8332, p .8332 .1668 .00069489 .026361 200 .7816 .8332 P z Pz 1.97 .5 .4756 .0244 p Power 1 .0244 .9756 f. We assume that p .73 and find how large a sample we need to estimate the proportion above that have being well-off as a goal with an error of .005. The formula for sample size for a proportion is in the outline. n p .73 , q 1 p 1 .73 .27 . So n .73.27 1.960 2 .005 2 pqz 2 e2 . If .05 , z .025 1.960 . If 30287 .1 . Round this up to 30288. g. We use the proportion that we found in a) to create a 2-sided 95% confidence interval for the proportion and ask if it differs significantly from .73. 162 .81 and the formula is p p z s p , where Seymour used p 2 200 pq .81.19 .0007695 .027740 . So p .81 1.960 .027740 .81 .0544 or .7556 to n 200 .8644. Since this interval does not include .73, the proportion differs significantly from .73. sp 4. Standard deviation is often a measure of reliability. A manufacturer is providing a connector with a mean length of 2.5 mm and is getting complaints that the connector is often too large or too small for the intended use. The previous standard deviation to the length of the part was 0.025mm, but the manufacturer introduces a process that should make the standard deviation smaller. A sample of 25 items is taken which yields a sample standard deviation of 0.030 a . To get a take the third to last digit of your student number and multiply it by 0.001. (For example, Seymour Butz’s student number is 976502, so he will subtract .005 and say that s 0.030 0.005 0.025 . ) a. Formulate the null and alternative hypotheses necessary to see if the goal has been achieved and test the hypothesis using a 95% confidence level and a test ratio. (2) b. What assumptions are necessary to perform this test? (1) c. Try to get a rough p-value. Interpret its meaning (1.5) d. Do a 95% two- sided confidence interval for the standard deviation (1) e. (Extra credit) Redo 4a) using an appropriate confidence interval. (2) f. (Extra credit) Find a critical value for s in 4a). (1) g. The number of claims for missing baggage in a large metropolitan airport supposedly follows a Poisson distribution with a mean of 72 per week. Assume that in a given week 92 are lost. Test this hypothesis using a test ratio and a 95% confidence level. (2) h. Find approximate critical values for the number of bags that could be lost in 4g. (2) 18 252y0612 10/18/06 (Open in ‘Print Layout’ format) i. (Extra credit) Find the power of the test in 4g) if the average number of lost bags per week is really 87. (3) j. I claim that x is binomially distributed with p .01 . Test this assertion using a 2-sided test if there are 4 successes in 10 trials. (2) [39.5] Solution: The problem says 2.5 , 0.025 , n 25 and Seymour uses s 0.025 . This indicated that I had not played with these results enough to give you a ‘good’ number for s . The manufacturer wants 0.025 , which is an alternate hypothesis because it does not contain an equality. a. We formulate the null and alternative hypotheses necessary to see if the goal has been achieved and test the hypothesis using a 95% confidence level and a test ratio. H : 0.025 . DF n 1 24 . The test ratio for a variance or a standard deviation and a .05 and 0 H 1 : 0.025 small sample is 2 n 1s 2 02 . Our test is left sided, so we need a value of 2 with 5% of the 24 distribution below it. We therefore say that our ‘reject’ region will be below 2 .95 13.8484 . Make a diagram. Draw a 2 curve with a mean at 24 and shade the area below 13.8484. Since 2 24 .025 2 .025 2 24 , we cannot reject the null hypothesis. Of course, it should be obvious that there is no improvement. b. We list assumptions necessary for the test in a). The underlying distribution should be Normal. c. We try to get a p-value and interpret it. Our table says that the two closest values to 24 are . pvalue Ps 0.025 P 2 24 .90 24 15.6587 and 2 .10 33.1962 . So all we can say is .10 pvalue .90 . Actually 24 and we can say .P 2 15.6587 .10 and .P 2 33.1962 .90 hence .10 pvalue .90 . This means that we cannot hope to reject the null hypothesis unless we use a significance level above 10%. d. We find a 95% confidence interval for the standard deviation. The interval given on Page 1 of the supplement is n 1s 2 2 2 2 2 2 1 2 24 n 1s 2 2 .975 12.4012 and 24 .025 or n 1s 2 .2025 2 39.3641so we have 2 n 1s 2 .2975 . Table 1 says that 24 .025 2 24 .025 2 or 2 39 .3641 12 .4012 .0003811 2 .0012096 or 0.0195 .0348 . 19 252y0612 10/18/06 (Open in ‘Print Layout’ format) e. We find a 95% confidence interval appropriate for testing a). If we use an appropriate confidence interval, it should be of the type 2 ? or ? . We want 5% of the distribution above the point that we pick. The interval that we did in the last section implies that 2.5% of the distribution is above , so 5% must be above n 1s 2 .295 24 . The table says 2 .95 13.8484 , so out limit is 2 n 1s 2 .2975 24 .025 2 or 13 .8484 2 0.001083 or .0329 . f. We find a critical value for the test in 4a). Table 3 implies that the critical values for a 2-sided hypothesis 2 are s cv2 2 .025 0 n 1 2 and s cv2 2 .975 0 n 1 . We need one value below 0.025, so use 13 .8484 0,025 2 .0003606 . The square root is s cv .01899 . We reject the null n 1 24 hypothesis if the sample standard deviation is below .01899. 2 s cv2 2 .95 0 At this point we are starting a completely new problem. The number of claims for missing baggage in a large metropolitan airport supposedly follows a Poisson distribution with a mean of 72 per week. Assume that in a given week 92 are lost. g. We test the hypothesis that the mean is 72 using a test ratio and a 95% confidence level. The Poisson distribution with a mean of 72 is approximately the Normal distribution with a mean of 72 72 . We know that x 92 . From what we know about the Normal H 0 : 72 Poisson x 92 72 20 distribution z . To do a 2.36 . Our hypotheses are 8.48528 72 H 1 : 72 Poisson 95% test, make a diagram. Show a Normal curve with a mean at zero and ‘reject’ zones below z.025 1.960 and above z.025 1.960 . Our value of z is in the upper reject zone, so reject the null hypothesis. and a standard deviation of h. We find approximate critical values to do the test in 4g). 95% of the values of z are between -1.960 and 1.960. Use the formula x z to get xcv 72 1.960 72 72 16 .63 or 55.37 to 88.37. We reject the null hypothesis if x is below 56 or above 88. (Note that since x 92 , we reject the null hypothesis, but that this was not required.) i. We are supposed to find the power of the test in 4g) if the average number of lost bags per week is really 87. I’ll try this if you will! This is another totally new problem. j. I claim that x is binomially distributed with p .01 . We test this assertion using a 2-sided test if there are 4 successes in 10 trials. H 0 : p .01 . n 10 and x 4 Since np 10.01 0.1 , 4 is above the expected value and , if we H 1 : p .01 use the binomial table, pvalue 2Px 4 21 Px 3 21 1 0 . Since this is below any reasonable significance level, reject the null hypothesis. 20