CHAPTER 9 ESTIMATION AND CONFIDENCE INTERVALS 1. See definition on page 362 of the text. 3. See definition on page 363 of the text. 5. (a) Standard error of the mean = 8 ; (1 - ) = (1 – 0.1) = 0.9. Since the population distribution is approximately normal, the distribution of X is approximately normal. Hence, margin of error = z 2 ( / n ) z0.05 ( / n ) 1.645 confidence interval = X 1.645 (b) Standard error of the mean = 8 8 ; 90% 50 ; (1 - ) = (1 – 0.025) = 0.975. Since sample size is large enough (n = 50 > 30), the distribution of X is approximately normal. Hence, margin of error = z 2 ( / n ) z0.0125 ( / n ) 2.24 interval = X 2.24 (c) 50 Standard error of the mean = 50 ; 97.5% confidence 60 ; (1 - ) = (1 – 0.01) = 0.99. Since sample size is large enough (n = 60), the distribution of X is approximately normal. Hence, margin of error = z 2 ( / n ) z0.005 ( / n ) 2.575 X 2.575 60 Since the population is approximately normally distributed, the distribution of X is approximately normal. Hence, a 90 percent confidence interval estimate for is 7. = X z 2 n X z0.05 = 26 1.645 6 9. 60 ; 99% confidence interval = n 16 23.5325, 28.4675 (a) It is the number in the table of Student’s t Distribution in row corresponding to “18” and 2 column corresponding to “ 0.010 ” . It equals 34.8053. (b) Since the required left tail probability is 0.025, the right tail probability is 1 – 0.025 = 0.955. It is the number in the table of Student’s t Distribution in row corresponding to “30” 2 and column corresponding to “ 0.975 ”. So, L = 16.7908. (c) 2 2 Total area to the right of 0.990 is 0.99. We want area between 0.990 and U to be 0.98. 2 Hence, area to the right of U should be (0.99 – 0.98) = 0.01. Thus, U 0.01 . It is the 9-1 2 number in the table of Student’s t Distribution in row “5” and column “ 0.01 ” and equals 15.0863. (d) 2 2 We want area between L and 0.05 to be 0.925. Area to the left of 0.05 is 0.05. Hence, 2 total area to the right of L should be 0.925 + 0.05 = 0.975. Thus, L 0.975 . It is the 2 number in the table of Student’s t Distribution in row “15” and column “ 0.975 ” and equals 6.26214. 11. (a) Since the population is normal, n 1 S 2 2 follows chi-square distribution with df = (n – 1) 9. A 99 percent confidence interval estimate for 2 is 2 2 (n 1) s 2 , (n 1) s 2 0.005 0.995 9(5) , 9(5) (1.908, 25.938) 23.5893 1.734926 (b) An 80 percent confidence interval estimate for 2 is 2 2 (n 1) s 2 , (n 1) s 2 0.1 0.9 9(5) , 9(5) (3.065, 10.8) 14.6837 4.16816 13. 15. (a) (1 - ) = 0.8, so = (1 – 0.8) = 0.2. Hence, t = t/2 = t0.1 (for df = 30) = 1.31 (from the t-table) (b) From the t-table, we find that t0.005 (for df = 6) = 3.707. (c) By symmetry of t-distribution it follows that t = -t0.1. From the t-table, we get t0.1 (for df = 24) = 1.318. Hence, t = -1.318. (d) The right tail area corresponding to t = 1 – 0.9 = 0.1. Thus, t = t0.1 (for df = 12) = 1.356 (from the t-table). (e) The left tail area corresponding to t0.995 is (1 – 0.995) = 0.005. Thus, t0.995 = -t0.005. From the t-table, we get t0.005 (for df = 14) = 2.977. Hence, t0.995 = -2.977. (a) Since the population distribution is approximately normal, the distribution of X is S n approximately t-distribution with df = n – 1. Calculate x and s. (1 - ) = 0.95. So, = (1 – 0.95) = 0.05. For df, t/2 = t0.025 = 2.201. 95% confidence interval estimate for is x 2.201( s / 12 ) . 9-2 (b) Calculate x and s. (1 - ) = 0.9. So, = (1 – 0.9) = 0.1. For df = 19, t/2 = t0.05 = 1.729. 90% confidence interval estimate for is x 1.729 s (c) 20 . Calculate x and s. (1 - ) = 0.99. So, = (1 – 0.99) = 0.01. For df = 7, t/2 = t0.005 = 3.499. 99% confidence interval estimate for is x 3.499 s 8 . 17. x 48.16 42.22 20 61.46 49.353 (48.16 49.353) 2 (42.22 49.353) 2 19 9.013 s (61.46 49.353) 2 Since the population distribution is approximately normal, the distribution of X is S n approximately t-distribution with df = n – 1 = 19. (1 - ) = 0.99. So, = (1 – 0.99) = 0.01. For df = 19, t/2 = t0.005 = 2.861 99% confidence interval estimate for is x 2.861 s 20 49.353 2.861 9.013 20 43.587,55.119 The value 50 lies in the 99% confidence interval estimate. Hence, the analysis does not provide reason to rule out the probability that the value of = 50. The value 60 does not lie in the interval estimate. Hence it will be unreasonable to assume that = 60. 19. (a) The best point estimate of is the sample mean x = 21.9 eggs/month. (b) (1 - ) = 0.98. So, = (1 – 0.98) = 0.02. For df = 19, t/2 = t0.01 = 2.539 Hence, a 98% confidence interval estimate for is = x t0.01 s n = 21.9 2.539 2.1 (c) The population is normally distributed, but population standard deviation is unknown. In this case (d) 20 = (20.708,23.092) X has t-distribution with (n - 1) degrees of freedom. S n Value 22 lies in the 98% confidence interval estimate. Hence, the analysis does not provide evidence against the probability that the value of = 22. Value 24 does not lies within confidence interval estimate. Hence it will be reasonable to assume that the value of is not 24. 9-3 21. (a) The best estimate of the value of the population proportion is pˆ (b) An estimate of the standard error of the proportion is pˆ 1 pˆ n (c) 300 0.75 . 400 0.75 0.25 0.022 400 (400)(0.75) > 5 and (400)(0.25) > 5. Hence, we can assume normal approximation. An approximate 99% confidence interval estimate of population proportion is = pˆ z0.005 pˆ 1 pˆ n = 0.75 2.575 0.022 0.693,0.807 . 23. (d) The estimator used in (c) for finding the confidence interval estimate produces an interval containing the population proportion, p, 99 percent times in the long run. Given the high probability, it would be reasonable to expect the interval currently obtained to contain the value of p. (a) The best point estimate of the population proportion is pˆ (b) (300)(0.05) > 5 and (300)(0.95) > 5. Hence, an approximate 95% confidence interval estimate of p is = pˆ z0.025 pˆ 1 pˆ = 0.05 1.96 (c) 25. 15 0.05 . 300 n 0.05 0.95 300 0.025,0.075 The entire 95% confidence interval estimate obtained in (b) lies to the left of 0.1. Hence, it is reasonable to infer that p is less than 0.1. He should not return the lot. Let us hope that the sample size, n, will be large enough for the distribution of X to be approximately normal. Then, the sample size n should be at least z 2 2.57515 = 59.676 or 60. E 5 2 2 This value of n is large enough for assumption of normality. So the bound is valid. 27. Let us hope that the sample size, n, will be large enough for normal approximation. We want (1 - ) = 0.99, so = 0.01. z = z0.005 = 2.575. 9-4 An approximate lower bound on the required sample size is 2 2.575 0.45 0.55 164.108 or 165. 0.1 This value of n is large enough for assumption of normality. So the bound is valid. 29. We want the margin of error to be no more than + 0.05. Let us hope that the sample size will be large enough for distribution of X to be approximately normal. An approximate lower bound on the required sample size is z0.025 s 1.96 0.2 61.466 or 62. E = 0.05 2 2 This value of n is large enough for assumption of normality. So the bound is valid. 31. (a) Let us hope that the sample size will be large enough for applying normal approximation. An approximate lower bound on the required sample size is 2 2 z0.05 p 1 p 1.645 0.3 0.7 = 5682.65 or 5683. E 0.01 This value of n is large enough for assumption of normality. So the bound is valid. (b) One way is to first obtain a better estimate of p by first choosing a pilot sample of moderate size, say 100. If the sample size based on new estimate is still large, then we shall have to increase the allowable margin of error and/or decrease the confidence level. For example, if the allowable margin of error is changed to 0.05, then an approximate lower 2 1.645 0.3 0.7 = 227.3 or 228. bound on sample size n will be 0.05 33. The ratio (n/N) = (49/500) = 0.098 is greater than 0.05, but not too large. The population size is moderately large, and the population distribution is approximately normal. Hence, we shall use Formula 9-18 in the textbook. (1 - ) = 0.99. So = (1 – 0.99) = 0.01. t/2 = t0.005 (for df = 48) is approximately 2.68. An approximate 99% confidence interval estimate for is: = x t0.005 s N n 9 451 = 40 2.68 n N 1 49 499 = (36.724, 43.276). 35. The ratio (n/N) = (30/300) = 0.1 is greater than 0.05, but not too large. pˆ 18 0.6 . 30 9-5 npˆ =30 (0.6) > 5 and n 1 pˆ = 30 (40) > 5. Hence n is large enough for normality assumption. An approximate 95% confidence interval estimate for p is: ( pˆ )(1 pˆ ) N n 0.6 1.96 N 1 n = pˆ z0.025 0.6 0.4 30 270 299 = (0.433, 0.767). 37. Let us hope that the sample size will be large enough for assumption of normality. An approximate lower bound for the sample size is 2 2 z0.05 p 1 p 1.96 0.35 0.65 = 2184.91 , or 2185. E 0.02 This value of n is large enough for normality assumption. So the bound is valid. 39. (a) The best point estimate 54. That is, the best choice of a single value as an estimate of is 54. (b) n = 49 is large enough for assumption of normality of distribution of X . For df = 48, t0.025 = approximately 2.01. A 95% confidence interval estimate 10 51.13,56.87 . 49 is 54 2.01 41. Let us hope that the sample size will be large enough for assumption of normality. 2 z s An approximate lower bound on the required sample size is 0.025 = E 1.96 500 96.04 or 97. 100 2 This value of n is large enough for assumption of normality. Hence the bound is valid. 43. (a) A point estimate of population mean is 1.01 kg. (b) The sample size (n = 36) is large enough for us to assume that distribution of approximately student’s t-distribution For df = 35, t0.025 = approximately 2.031. A 95% confidence interval estimate 0.02 1.003,1.017 . 36 is 1.01 2.031 9-6 X is S/ n 45. (a) The sample size (n = 50) is large enough for assumption of normality. For df = 49, t0.005 = approximately 2.68. A 99% confidence interval estimate is x t0.005 s = n 0.01 0.64 2.68 0.636, 0.644 . 50 47. (b) The result, thus provides significant evidence that value of is not 0.63. It will therefore be reasonable to conclude that the value of is not 0.63. (a) The value of sample proportion is p̂ = 0.63. n p̂ = (1000) (0.63) > 5 and n (1 - p̂ ) = (1000) (0.37) > 5. Hence, the sample size is large enough for normality assumption. A 95% confidence interval estimate is pˆ z0.025 (b) 49. 0.63 0.37 1000 0.6, 0.66 . The entire interval in (a) lies above 0.6. The analysis thus provides sufficient evidence that the population proportion p is greater than or equal 0.6. It is therefore reasonable to conclude that the claim, that p is less than 0.6 is false. The sample size is large enough to assume normality. For df = 35, t0.05 = approximately 1.69. A 90% confidence interval estimate of is x t0.05 51. pˆ (1 pˆ ) = 0.63 1.96 n (a) s 8200 = 32000 1.69 29690.33,34309.67 . n 36 Let us hope that the sample size will be large enough for assumption of normality. p = 0.21. 2 z0.025 p 1 p = An approximate lower bound for n is E 2 1.96 0.21 0.79 708.13 or 709. 0.03 This is large enough for assumption of normality. Hence the bound is valid. z 2 0.5 An approximate lower bound on the required sample size n is = E 2 (b) 1.96 0.5 1067.11 or 1068. 0.03 2 9-7 Hence, 1068 accountants should be contacted. Let us hope that the sample size will be large enough for assumption of normality. p = 5/50 =0.1. 53. 2 z0.025 p 1 p = An approximate lower bound for n is E 2 1.96 0.1 0.9 864.36 or 865. This is An approximate lower bound for n is 0.02 large enough for assumption of normality. Hence the bound is valid. 55. (a) x 94 78 83 15 84 89.4667 . Thus the best point estimate of is 89.4667. (b) The sample mean is s (94 89.4667) 2 (78 89.4667) 2 14 (84 89.4667) 2 8.08 Since the population distribution is approximately normal, the distribution of X is S n approximately t-distribution with df = n – 1 = 14. For df = 14, t0.025 = 2.145. Hence, 95% confidence interval estimate is 8.08 89.4667 2.145 84.99,93.99 . 15 (c) 57. The confidence interval estimate in (b) lies entirely above 80. Hence it is reasonable to conclude that the mean stress level is in the dangerous level. The sample size (n = 60) is large enough for assumption of normality. = (1 – 0.98) = 0.02. For df = 59, t/2 = t0.01 = approximately 2.39. Hence, an approximate 0.75 2.53, 2.99 . 60 98% confidence interval estimate of is 2.76 2.39 59. (a) The best point estimate of the population mean, , we can obtain is x (b) 2698 2028 10 2379 2408.8 . Since the population distribution is approximately normal, the distribution of approximately t-distribution with df = n – 1 = 9. The sample standard deviation is 9-8 X is S n s (2698 2408.8) 2 (2474 2408.8) 2 9 (2379 2408.8) 2 = 304.4276. For df = 9, t0.025 = 2.262. A 95% confidence interval estimate for is 304.4276 2408.8 2.262 2191.04, 2626.56 . 10 61. (a) (b) x 64 66 12 66 62.583 Since the population distribution is approximately normal, the distribution of X is S n approximately t-distribution with df = n – 1 = 11. The sample standard deviation is s (64 62.583) 2 (66 62.583) 2 11 (66 62.583) 2 3.942 For df = 11, t0.05 =1.796. Hence, 90% confidence interval estimate 3.942 60.539, 64.627 . 12 is 62.583 1.796 (c) 63. The value 60 lies in the interval estimate obtained in part (b). Thus, our analysis does not provide significant evidence against the claim. We have no reason to doubt the claim. A point estimate of proportion, p, of the entire population (Ontario) who blame the provincial government for the tragedy is p̂ = 0.22. n p̂ = (1001) (0.22) > 5 and n (1 - p̂ ) = (1001) (0.78) > 5. Hence, the sample size is large enough for normality assumption. An approximate 95% confidence interval estimate for p is pˆ z0.025 65. (a) pˆ (1 pˆ ) = 0.22 1.96 n 0.22 0.78 = (0.194, 0.246). 1001 A point estimate of proportion, p, of Canadians who favour user fees and charges is p̂ = 0.54. n p̂ = (1400) (0.54) > 5 and n (1 - p̂ ) = (1400) (0.46) > 5. Hence, the sample size is large enough for normality assumption. 9-9 An approximate 95% confidence interval estimate is pˆ z0.025 0.54 1.96 (b) 0.54 0.46 1400 pˆ (1 pˆ ) = n 0.514, 0.566 . The entire interval lies above 0.5. Hence, it is reasonable to conclude that p is greater than 0.5, that is, majority of Canadians favour moderate user fees and charges. Select a sample of size 10; calculate x and s. For df = 9, t0.025 = 2.262. A 95% confidence 67. s . The answer will vary according to the sample 10 interval estimate is x 2.262 chosen. 69. The Minitab and Excel outputs are given below. (See the Minitab and Excel instructions given in the chapter.) It should be noted that Minitab output directly gives the confidence interval estimates, while the Excel output gives in the row labeled “Confidence level”, the margin of error. (a) Using the Excel output, we get for selling price of the homes: 95 percent confidence interval estimate for population mean = ( sample mean margin of error) = (221.1029 9.116045) = (211.9868122, 230.2189021). Minitab Output One-Sample T: Selling price Variable Price N 105 Mean 221.10 StDev 47.11 SE Mean 4.60 EXCEL OUTPUT Selling price Mean Standard Error Median Mode Standard Deviation Sample Variance Kurtosis Skewness Range Minimum Maximum Sum Count Confidence Level(95.0%) 9-10 221.1029 4.597017 213.6 188.3 47.1054 2218.919 -0.2768 0.474013 220.3 125 345.3 23215.8 105 9.116045 95.0% CI (211.99, 230.22) (b) A 90 percent confidence interval estimate for the fraction of home that were sold more than $240000 = ( pˆ z ( pˆ )(1 pˆ ) ) (0.249,0.399) . n 2 MINITAB OUTPUT Test and CI for One Proportion: Sold more than $240000 Variable Prices X 34 N 105 Sample p 0.323810 90.0% CI (0.248697, 0.398922) Z-Value -3.61 P-Value 0.000 MEGASTAT OUTPUT Confidence interval - proportion 90% confidence level 0.32381 proportion 105 n 1.645 z 0.324 half-width 0.249 upper confidence limit 0.399 lower confidence limit (c) For number of bedrooms: 90 percent confidence interval estimate for the population mean = (mean margin of error) = (3.8 0.346460433) = (3.45354, 4.14646). Minitab Output One-Sample T: Bedrooms Variable Bedrooms N 105 Mean 3.800 9-11 StDev 1.503 SE Mean 0.147 98.0% CI (3.454, 4.146) Excel Output # of bedrooms Mean Standard Error Median Mode Standard Deviation Sample Variance Kurtosis Skewness Range Minimum Maximum Sum Count Confidence Level(98.0%) (d) 3.8 0.146635028 4 4 1.502561915 2.257692308 -0.199838427 0.660884776 6 2 8 399 105 0.346460433 A 90 percent confidence interval estimate for the proportion of homes sold that have a garage = = ( pˆ z 2 ( pˆ )(1 pˆ ) ) (0.601,0.751) . n Minitab Output Test and CI for One Proportion: Garage Variable Garage X 71 N 105 Sample p 0.676190 90.0% CI (0.601078, 0.751303) MEGASTAT OUTPUT Confidence interval - proportion 90% confidence level 0.67619 proportion 105 n 1.645 z 0.676 half-width 0.601 upper confidence limit 0.751 lower confidence limit 9-12 Z-Value 3.61 P-Value 0.000 (e) A 95 percent confidence interval estimate for the proportion of homes sold that have two or more bathrooms = (0.779, 0.916) Minitab Output Test and CI for One Proportion: Baths Variable Baths X 89 N 105 Sample p 0.847619 95.0% CI (0.778878, 0.916361) Z-Value 7.12 P-Value 0.000 MEGASTAT OUTPUT Confidence interval - proportion 95% confidence level 0.847619 proportion 105 n 1.960 z 0.848 half-width 0.779 upper confidence limit 0.916 lower confidence limit 71. The Minitab and Excel outputs are given below. (See the Minitab and Excel instructions given in the chapter.) It should be noted that Minitab output directly gives the confidence interval estimates, while the Excel output gives in the row labeled “Confidence level”, the margin of error. Thus, using Excel, we get : For TSE 300 data : 92 percent confidence interval = (mean margin of error) (9776.126 272.6599) (9503.4661,10048.7859) For BCE data : 95 percent confidence interval = (mean margin of error) (38.199 2.117026) (36.081974, 40.316026) . For Air Canada data : 90 percent confidence interval = (mean margin of error) (17.3875 0.960303) (16.427197,18.347803). These are the same (except for rounding) as the ones obtained using Minitab. Minitab Output: One-Sample T: TSE 300 VARIABLE TSE 300 N 20 MEAN 9776 STDEV SE MEAN 659 147 92.0% CI (9503, 10049) One-Sample T: BCE VARIABLE BCE N 20 MEAN 38.20 One-Sample T: Air Canada 9-13 STDEV SE MEAN 4.52 1.01 95.0% CI (36.08, 40.32) EXCEL OUTPUT EXCEL OUTPUT BCE TSE 300 Mean Standard Error Median Mode Standard Deviation Sample Variance Kurtosis Skewness Range Minimum Maximum Sum Count Confidence Level(92.0%) Mean Standard Error Median Mode Standard Deviation Sample Variance Kurtosis Skewness Range Minimum Maximum Sum Count Confidence Level(95.0%) 9776.126 147.4212 9517.935 10778.8 659.2875 434660 -1.15449 0.608497 1821.2 9020.88 10842.08 195522.5 20 272.6599 EXCEL OUTPUT Air Canada Mean Standard Error Median Mode Standard Deviation Sample Variance Kurtosis Skewness Range Minimum Maximum Sum Count Confidence Level(90.0%) 9-14 17.3875 0.555367 17.475 19.4 2.483677 6.168651 -0.24516 -0.53549 9.25 11.5 20.75 347.75 20 0.960303 38.199 1.011467 37.475 37 4.523419 20.46132 -0.94464 0.308152 15.92 31.75 47.67 763.98 20 2.117026