CHAPTER 8
Section 8-1
8-1
value of z0 which is 2.14. From Table III, (2.14) = P(Z<2.14) = 0.9838 and the
confidence level is 2(0.9838-0.5) = 96.76%.
b) The confidence level for
by the value of z0 which is 2.14. From Table III, (2.49) = P(Z<2.49) = 0.9936 and the
confidence level is is 2(0.9936-0.5) = 98.72%.
c) The confidence level for
by the value of z0 which is 2.14. From Table III, (1.85) = P(Z<1.85) = 0.9678 and the
confidence level is 93.56%.
8-2
8-3
8-4
a) 95% CI for
x z / n x + z / n
1000 1.96(20 / 10 ) 1000 + 1.96(20 / 10 )
987.6 1012.4
b) .95% CI for ,
n = 25, = 20 x = 1000, z = 1.96
x z / n x + z / n
1000 1.96(20 / 25 ) 1000 + 1.96(20 / 25 )
992.2 1007.8
c) 99% CI for ,
n = 10, = 20 x = 1000, z = 2.58
x z / n x + z / n
1000 2.58(20 / 10 ) 1000 + 2.58(20 / 10 )
983.7 1016.3
d) 99% CI for ,
n = 25, = 20 x = 1000, z = 2.58
x z / n x + z / n
1000 2.58(20 / 25 ) 1000 + 2.58(20 / 25 )
989.7 1010.3
e) When n is larger, the CI is narrower. The higher the confidence level, the wider the CI.
8-1
8-5
8-6
a) Sample mean from the first confidence interval =37.53 + (49.87-37.53)/2 = 43.7
Sample mean from the second confidence interval =35.59 + (51.81-35.59)/2 = 43.7
b) The 99% CI is (35.59, 51.81) and the 95% CI is (37.53, 49.87). The higher the confidence
level, the wider the CI.
8-7
39.2 = 20 n
2
39.2
n =
= 3.84
20
Therefore, n = 4.
b) Find n for the length of the 99% CI to be 40. Za/2 = 2.58
1/2 length = (2.58)(20) / n = 20
51.6 = 20 n
2
51.6
n =
= 6.66
20
Therefore, n = 7.
8-8
a)
b) Interval (1): 3124.9 3215.7 was calculated with 95% Confidence because it has a smaller
half-length, and therefore a smaller confidence interval. The 99% confidence level will make the interval
larger.
8-9
8-10
95% Two-sided CI on the breaking strength of yarn: where x = 98, = 2 , n=9 and z0.025 = 1.96
x z 0.025 / n x + z 0.025 / n
98 1.96(2) / 9 98 + 1.96(2) / 9
96.7 99.3
8-11
95% Two-sided CI on the true mean yield: where x = 90.480, = 3 , n=5 and z0.025 = 1.96
x z 0.025 / n x + z 0.025 / n
90.480 1.96(3) / 5 90.480 + 1.96(3) / 5
87.85 93.11
8-2
8-12
99% Two-sided CI on the diameter cable harness holes: where x =1.5045 , = 0.01 , n=10 and
z0.005 = 2.58
x z 0.005 / n x + z 0.005 / n
1.5045 2.58(0.01) / 10 1.5045 + 2.58(0.01) / 10
1.4963 1.5127
8-13
n
n
0.001
0.001
74.036 2.58
74.036 + 2.58
15
15
74.0353 74.0367
b) 99% One-sided CI on the true mean piston ring diameter
For = 0.01, z = z0.01 =2.33 and x = 74.036, = 0.001, n=15
x z 0.01
0.001
74.036 2.33
15
74.0354
The lower bound of the one sided confidence interval is less than the lower bound of the two-sided
confidence. This is because the Type I probability of 99% one sided confidence interval (or = 0.01) in the
left tail (or in the lower bound) is greater than Type I probability of 99% two-sided confidence interval (or
/2 = 0.005) in the left tail.
8-14
x z 0.025
x + z 0.025
n
n
25
25
1014 1.96
1014 + 1.96
20
20
1003 1025
b) 95% One-sided CI on the true mean piston ring diameter
For = 0.05, z = z0.05 =1.65 and x = 1014, =25 , n=20
x z0.05
25
1014 1.65
20
1005
8-3
The lower bound of the one sided confidence interval is lower than the lower bound of the two-sided
confidence interval even though the level of significance is the same. This is because all of the Type I
probability (or ) is in the left tail (or in the lower bound).
8-15
n
n
31.62
31.62
3250 1.96
3250 + 1.96
12
12
3232.11 3267.89
b) 99% Two-sided CI on the true mean compressive strength
z /2 = z0.005 = 2.58
x z0.005
x + z0.005
n
n
31.62
31.62
3250 2.58
3250 + 2.58
12
12
3226.4 3273.6
The 99% CI is wider than the 95% CI
8-16
95% Confident that the error of estimating the true mean life of a 75-watt light bulb is less than 5
hours.
For = 0.05, z /2 = z0.025 = 1.96 , and =25 , E=5
z
1.96(25)
n = a / 2 =
= 96.04
5
Set the width to 6 hours with = 25, z0.025 = 1.96 solve for n.
1/2 width = (1.96)(25) / n = 3
49 = 3 n
2
49
n = = 266.78
3
Therefore, n = 267.
8-18
99% Confident that the error of estimating the true compressive strength is less than 15 psi
For = 0.01, z /2 = z0.005 = 2.58 , and =31.62 , E=15
z
2.58(31.62)
n = a / 2 =
= 29.6 30
15
8-4
Therefore, n=30
8-19
To decrease the length of the CI by one half, the sample size must be increased by 4 times (22).
z / 2 / n = 0.5l
Now, to decrease by half, divide both sides by 2.
( z / 2 / n ) / 2 = (l / 2) / 2
( z / 2 / 2 n ) = l / 4
( z / 2 / 22 n ) = l / 4
Therefore, the sample size must be increased by 22.
8-20
x z / 2
x + z / 2
n
n
z / 2
z
z
1 z / 2
= /2
= /2
=
2n 1.414 n 1.414 n 1.414 n
If n is doubled in Eq 8-7:
z / 2
4n
z / 2
2 n
z / 2
1 z
= / 2
2 n
2 n
x z 0.005
x + z 0.005
n
n
0.5
0.5
13.77 2.57
13.77 + 2.57
11
11
13.383 14.157
b) 95% lower-confidence bound on the mean temperature
For = 0.05, z = z0.05 =1.65 and x = 13.77, = 0.5, n =11
x z 0.05
0.5
13.77 1.65
11
13.521
c) 95% confidence that the error of estimating the mean temperature for wheat grown is
less than 2 degrees Celsius.
For = 0.05, z /2 = z0.025 = 1.96, and = 0.5, E = 2
z
1.96(0.5)
n = a / 2 =
= 0.2401
2
8-5
d) Set the width to 1.5 degrees Celsius with = 0.5, z0.025 = 1.96 solve for n.
1/2 width = (1.96)(0.5) / n = 0.75
0.98 = 0.75 n
2
0.98
n =
= 1.707
0.75
Therefore, n = 2.
Section 8-2
8-22
t 0.025,15 = 2.131
t 0.05,10 = 1.812
t 0.005, 25 = 2.787
t 0.001,30 = 3.385
a)
t 0.025,12 = 2.179
d)
8-24
a)
t 0.05,14 = 1.761
8-25
8-23
t 0.10, 20 = 1.325
b)
t 0.025, 24 = 2.064
c)
t 0.005,13 = 3.012
b)
t 0.01,19 = 2.539
c)
t 0.001, 24 = 3.467
10
n
n
1.605
1.605
25.1848 2.262
25.1848 + 2.262
10
10
24.037 26.333
8-26
SE Mean =
stDev
6.11
= 1.58 , therefore N = 15
N
N
sum 751.40
Mean =
=
= 50.0933
N
15
Variance = ( stDev) 2 = 6.112 = 37.3321
=
n
n
6.11
6.11
50.0933 2.145
50.0933 + 2.145
15
15
46.709 53.477
8-6
8-27
n
n
3645.94
3645.94
60139.7 2.131
60139.7 + 2.131
16
16
58197.33 62082.07
8-28
x = 1.10 s = 0.015 n = 25
95% CI on the mean volume of syrup dispensed
For = 0.05 and n = 25, t /2,n-1 = t0.025,24 = 2.064
s
s
x t 0.025, 24
x + t 0.025, 24
n
n
0.015
0.015
1.10 2.064
1.10 + 2.064
25
25
1.094 1.106
8-30
n
n
16
315
315 2.447
315 + 2.447
7
7
300.202 329.798
8-31
8-7
n
9.9
118.3 + 2.650
14
125.312
x + t 0.005,13
8-32
n
n
1.53
1.53
231.67 2.132
231.67 2.132
5
5
230.2 233.1
By examining the normal probability plot, it appears that the data are normally distributed. There does not
appear to be enough evidence to reject the hypothesis that the frequencies are normally distributed.
The data appear to be normally distributed based on examination of the normal probability plot below.
Normal Probability Plot for frequencies
ML Estimates - 95% CI
99
ML Estimates
95
90
Mean
231.67
StDev
1.36944
80
Percent
8-33
70
60
50
40
30
20
10
5
1
226
231
236
Data
Therefore, there is evidence to support that the annual rainfall is normally distributed.
8-8
99
Mean
StDev
N
AD
P-Value
95
90
485.9
90.30
20
0.288
0.581
Percent
80
70
60
50
40
30
20
10
5
200
300
400
500
Rainfall
600
700
800
n
n
90.34
90.34
485.8 2.093
485.8 + 2.093
20
20
443.520 528.080
The data appear to be normally distributed based on examination of the normal probability plot below.
Therefore, there is evidence to support that the solar energy is normally distributed.
Probability Plot of Solar
Normal - 95% CI
99
Mean
StDev
N
AD
P-Value
95
90
80
Percent
8-34
70
60
50
40
30
20
10
5
50
55
60
65
Solar
70
8-9
75
80
65.58
4.225
16
0.386
0.349
s
s
x t 0.025,15
x + t 0.025,15
n
n
4.225
4.225
65.58 2.131
65.58 + 2.131
16
16
63.329 67.831
8-35
n
n
15.7
15.7
317.2 3.250
317.2 + 3.250
10
10
301.06 333.34
a) The data appear to be normally distributed based on examination of the normal probability plot below.
Therefore, there is evidence to support that the level of polyunsaturated fatty acid is normally distributed.
Normal Probability Plot for 8-25
ML Estimates - 95% CI
99
95
90
80
Percent
8-36
70
60
50
40
30
20
10
5
1
16
17
Data
s
s
x + t 0.005,5
x t 0.005,5
n
n
0.319
0.319
16.98 4.032
16.98 + 4.032
6
6
16.455 17.505
8-10
18
The 99% confidence for the mean polyunsaturated fat is (16.455, 17.505). There is high confidence that
the true mean is in this interval
8-37
a) The data appear to be normally distributed based on examination of the normal probability plot below.
b) 95% two-sided confidence interval on mean comprehensive strength
Normal Probability Plot for Strength
ML Estimates - 95% CI
99
ML Estimates
95
90
Mean
2259.92
StDev
34.0550
Percent
80
70
60
50
40
30
20
10
5
1
2150
2250
2350
Data
n
n
35.6
35.6
2259.9 2.201
2259.9 + 2.201
12
12
2237.3 2282.5
c) 95% lower-confidence bound on mean strength
s
x t 0.05,11
n
35.6
2259.9 1.796
12
2241.4
8-38
a) According to the normal probability plot there does not seem to be a severe deviation from normality for
this data. This is due to the fact that the data appears to fall along a straight line.
8-11
99
95
90
Percent
80
70
60
50
40
30
20
10
5
1
8.15
8.20
8.25
8.30
Data
s
s
x + t 0.025,14
x t 0.025,14
n
n
0.025
0.025
8.23 2.145
8.23 + 2.145
15
15
8.216 8.244
c) 95% upper confidence bound on mean rod diameter t0.05,14 = 1.761
n
0.025
8.23 + 1.761
15
8.241
x + t 0.025,14
8-39
a) The data appear to be normally distributed based on examination of the normal probability plot below.
Therefore, there is evidence to support that the speed-up of CNN is normally distributed.
8-12
99
Mean
StDev
N
AD
P-Value
95
90
4.313
0.4328
13
0.233
0.745
Percent
80
70
60
50
40
30
20
10
5
3.0
3.5
4.0
4.5
Speed
5.0
5.5
6.0
n
n
0.4328
0.4328
4.313 2.179
4.313 + 2.179
13
13
4.051 4.575
c) 95% lower confidence bound on mean speed-up
s
x t 0.05, 24
n
0.08
4.05 1.711
25
4.023
There is high confidence that the true mean wall thickness is greater than 4.023 mm.
8-13
99
ML Estimates
95
90
Mean
2.90167
StDev
0.0951169
Percent
80
70
60
50
40
30
20
10
5
1
2.6
2.7
2.8
2.9
3.0
3.1
3.2
Data
8-41
a) The data appear to be normally distributed. There is not strong evidence that the percentage of
enrichment deviates from normality.
b) 99% two-sided confidence interval on mean percentage enrichment
For = 0.01 and n = 12, t /2,n-1 = t0.005,11 = 3.106, x = 2.9017 s
= 0.0993
s
s
x t 0.005,11
x + t 0.005,11
n
n
0.0993
0.0993
2.902 3.106
2.902 + 3.106
12
12
2.813 2.991
8-14
Section 8-3
8-42
02.05,10 = 18.31
02.025,15 = 27.49
02.01,12 = 26.22
02.95, 20 = 10.85
02.99,18 = 7.01
02.995,16 = 5.14
02.005, 25 = 46.93
8-43
c) 90% CI and df = 19
and
14(0.008) 2
2
29.14
0.00003075 2
8-45
0.00003075 2
0.005545
One may take the square root of the variance bound to obtain the confidence bound for the standard
deviation.
8-46
n = 10 s = 4.8
2 / 2,n1 = 02.025,9 = 19.02
and
9( 4.8) 2
9( 4.8) 2
2
19.02
2.70
2
10.90 76.80
3.30 < < 8.76
8-47
50(0.37) 2
50(0.37) 2
2
71.42
32.36
0.096 2 0.2115
Taking the square root of the endpoints of this interval we obtain,
0.31 < < 0.46
8-48
n = 17 s = 0.09
8-15
and
16(0.09) 2
16(0.09) 2
2
28.85
6.91
2
0.0045 0.0188
0.067 < < 0.137
8-49
The data appear to be normally distributed based on examination of the normal probability plot below.
Therefore, there is evidence to support that the mean temperature is normally distributed.
99
Mean
StDev
N
AD
P-Value
95
90
Percent
80
70
60
50
40
30
20
10
5
17
18
19
20
21
22
Mean Temp
23
24
25
n = 8 s = 0.9463
2 / 2,n1 = 02.025,7 = 16.01 and 12 / 2,n1 = 02.975,7 = 1.69
7(0.9463) 2
7(0.9463) 2
2
16.01
1.69
2
0.392 3.709
0.626 < < 1.926
8-50
n = 41 s = 15.99
2 / 2,n1 = 02.025, 40 = 59.34 and 12 / 2,n1 = 02.975, 40 = 24.43
40(15.99) 2
40(15.99) 2
2
59.34
24.43
2
172.35 418.633
13.13 < < 20.46
8-16
21.41
0.9463
8
0.352
0.367
The data dont appear to be normally distributed based on examination of the normal probability plot
below. Therefore, there is not enough evidence to support that the time of tumor appearance is normally
distributed. So the 95% confidence interval for is invalid.
99
Mean
StDev
N
AD
P-Value
95
90
88.78
15.99
41
1.631
<0.005
Percent
80
70
60
50
40
30
20
10
5
60
80
100
TimeOfTumor
120
140
n = 15 s = 0.00831
2 / 2,n1 = 02.025,14 = 26.12
and
14(0.00831) 2
6.53
2
0.000148
0.0122
The data do not appear to be normally distributed based on an examination of the normal probability plot
below. Therefore, the 95% confidence interval for is not valid.
99
Mean
StDev
N
AD
P-Value
95
90
80
Percent
8-51
40
70
60
50
40
30
20
10
5
3.44
3.45
3.46
3.47
3.48
Gauge Cap
3.49
8-17
3.50
3.472
0.008307
15
0.878
0.018
8-52
n = 10
and
02.995,9 = 1.73
9(1.913) 2
9(1.913) 2
2
23.59
1.73
2
1.396 19.038
b) 99% lower confidence bound for 2
For = 0.01 and n = 10,
9(1.913) 2
2
21.67
1.5199 2
c) 90% lower confidence bound for 2
For = 0.1 and n = 10,
9(1.913) 2
2
14.68
2.2436 2
1.498
d) The lower confidence bound of the 99% two-sided interval is less than the one-sided interval. The lower
confidence bound for 2 is in part (c) is greater because the confidence is lower.
Section 8-4
8-53
a) 95% Confidence Interval on the fraction defective produced with this tool.
p =
13
z / 2 = 1.96
= 0.04333 n = 300
300
p (1 p )
p (1 p )
p z / 2
p p + z / 2
n
n
0.04333(0.95667)
0.04333(0.95667)
p 0.04333 + 1.96
300
300
0.02029 p 0.06637
b) 95% upper confidence bound z = z 0.05 = 1.65
0.04333 1.96
p p + z / 2
p (1 p )
n
p 0.04333 + 1.650
0.04333(0.95667)
300
p 0.06273
8-18
8-54
a) 95% Confidence Interval on the proportion of such tears that will heal.
p = 0.676 n = 37 z / 2 = 1.96
p (1 p )
p p + z / 2
n
p z / 2
0.676 1.96
p (1 p )
n
0.676(0.324)
0.676(0.324)
p 0.676 + 1.96
37
37
0.5245 p 0.827
b) 95% lower confidence bound on the proportion of such tears that will heal.
p z
0.676 1.64
8-55
p (1 p )
p
n
0.676(0.33)
p
37
0.549 p
a) 95% confidence interval for the proportion of college graduates in Ohio that voted for George Bush.
p =
412
= 0.536
768
n = 768 z / 2 = 1.96
p z / 2
p (1 p )
p p + z / 2
n
0.536 1.96
p (1 p )
n
0.536(0.464)
0.536(0.464)
p 0.536 + 1.96
768
768
0.501 p 0.571
b) 95% lower confidence bound on the proportion of college graduates in Ohio that voted for George Bush.
p z
0.536 1.64
8-56
p (1 p )
p
n
0.536(0.464)
p
768
0.506 p
p =
823
= 0.823 n = 1000 z / 2 = 1.96
1000
8-19
p z / 2
0.823 1.96
p (1 p )
p p + z / 2
n
p (1 p )
n
0.823(0.177)
0.823(0.177)
p 0.823 + 1.96
1000
1000
0.7993 p 0.8467
z
1.96
n = / 2 p (1 p ) =
0.823(1 0.823) = 621.79 ,
0.03
E
n 622.
c) E = 0.03, = 0.05, z /2 = z0.025 = 1.96 at least 95% confident
z
1.96
n = / 2 (0.25) =
(0.25) = 1067.11 ,
0.03
E
n 1068.
8-57
12
= 0.4 n = 30 z / 2 = 1.96
30
p (1 p )
p z / 2
p p + z / 2
n
p =
0.4 1.96
p (1 p )
n
0.4(0.6)
0.4(0.6)
p 0.4 + 1.96
30
30
0.225 p 0.575
z
1.96
n = / 2 p (1 p ) =
0.4(1 0.4) = 2304.96 ,
0.02
E
n 2305.
c) E = 0.02, = 0.05, z /2 = z0.025 = 1.96 at least 95% confident
z
1.96
n = / 2 (0.25) =
(0.25) = 2401.
0.02
E
8-58
p =
18
= 0.36
50
z / 2 = 1.96
n = 50
8-20
p z / 2
p (1 p )
p p + z / 2
n
p (1 p )
n
0.36(0.64)
0.36(0.64)
p 0.36 + 1.96
50
50
0.227 p 0.493
0.36 1.96
b)
z
1.96
n = / 2 p(1 p) =
0.36(1 0.36) = 2212.76
0.02
E
n 2213
2
z
1.96
c) n = / 2 p(1 p) =
0.5(1 0.5) = 2401
0.02
E
8-59
The worst case would be for p = 0.5, thus with E = 0.05 and = 0.01, z /2 = z0.005 = 2.58 we obtain a
sample size of:
z
2.58
n = / 2 p(1 p) =
0.5(1 0.5) = 665.64 , n 666
E
0.05
8-60
z
2.58
n = / 2 p (1 p ) =
0.5(1 0.5) = 5758.13 , n 5759
E
0.017
Section 8-6
8-61
x t 0.025,15 s 1 +
1
1
x n +1 x + t 0.025,15 s 1 +
n
n
1
1
x n +1 60139.7 + 2.131(3645.94) 1 +
16
16
52131.1 x n +1 68148.3
60139.7 2.131(3645.94) 1 +
The prediction interval is considerably wider than the 95% confidence interval (58,197.3 62,082.07).
This is expected because the prediction interval needs to include the variability in the parameter estimates
as well as the variability in a future observation.
8-62
8-21
1
1
x n +1 x + t 0.005,19 s 1 +
n
n
1
1
x n +1 1.25 + 2.861(0.25) 1 +
20
20
0.517 x n +1 1.983
1.25 2.861(0.25) 1 +
The lower bound of the 99% prediction interval is considerably lower than the 99% confidence interval
(1.108 ). This is expected because the prediction interval needs to include the variability in the
parameter estimates as well as the variability in a future observation.
8-63
95% Prediction Interval on the volume of syrup of the next beverage dispensed
x = 1.10 s = 0.015 n = 25 t /2,n-1 = t0.025,24 = 2.064
x t 0.025, 24 s 1 +
1
1
x n +1 x + t 0.025, 24 s 1 +
n
n
1
1
x n +1 1.10 2.064(0.015) 1 +
25
25
1.068 x n +1 1.13
1.10 2.064(0.015) 1 +
1.106
90% prediction interval the value of the natural frequency of the next beam of this type that will be tested.
given x = 231.67, s =1.53 For = 0.10 and n = 5, t /2,n-1 = t0.05,4 = 2.132
x t 0.05, 4 s 1 +
1
1
x n +1 x + t 0.05, 4 s 1 +
n
n
1
1
x n +1 231.67 2.132(1.53) 1 +
5
5
228.1 x n +1 235.2
231.67 2.132(1.53) 1 +
95% Prediction Interval on the volume of syrup of the next beverage dispensed
t /2,n-1 = t0.025,19 = 2.093
n = 20 x = 485.8 s = 90.34
x t 0.025,19 s 1 +
1
1
x n +1 x + t 0.025,19 s 1 +
n
n
1
1
x n +1 485.8 2.093(90.34) 1 +
20
20
292.049 x n +1 679.551
485.8 2.093(90.34) 1 +
8-66
The 95% prediction interval is wider than the 95% confidence interval.
99% prediction interval on the polyunsaturated fat
8-22
1
1
x n +1 x + t 0.005,5 s 1 +
n
n
1
1
x n +1 16.98 + 4.032(0.319) 1 +
6
6
15.59 x n +1 18.37
16.98 4.032(0.319) 1 +
The length of the prediction interval is much longer than the width of the confidence interval
16.455 17.505 .
8-67
x t 0.005,9 s 1 +
1
1
x n +1 x + t 0.005,9 s 1 +
n
n
1
1
x n +1 317.2 3.250(15.7) 1 +
10
10
263.7 x n +1 370.7
317.2 3.250(15.7) 1 +
1
1
x n +1 x + t 0.025,14 s 1 +
n
n
1
1
x n +1 8.23 2.145(0.025) 1 +
15
15
8.17 x n +1 8.29
8.23 2.145(0.025) 1 +
x t 0.05,11 s 1 +
1
1
x n +1 x + t 0.05,11 s 1 +
n
n
1
1
x n +1 2260 + 1.796(35.57 ) 1 +
12
12
2193.5 x n +1 2326.5
2260 1.796(35.57) 1 +
8-70
8-23
x t 0.05, 24 s 1 +
1
1
x n +1 x + t 0.05, 24 s 1 +
n
n
1
1
x n +1 4.05 1.711(0.08) 1 +
25
25
3.91 x n +1 4.19
4.05 1.711(0.08) 1 +
8-71
90% prediction interval for enrichment data given x = 2.9 s = 0.099 n = 12 for = 0.10
and n = 12, t /2,n-1 = t0.05,11 = 1.796
x t 0.05,12 s 1 +
1
1
x n +1 x + t 0.05,12 s 1 +
n
n
1
1
x n +1 2.9 + 1.796(0.099) 1 +
12
12
2.71 x n +1 3.09
2.9 1.796(0.099) 1 +
x t 0.05,12 s
2.9 1.796(0.099)
1
1
x + t 0.05,12 s
n
n
1
1
2.9 1.796(0.099)
12
12
2.85 2.95
The prediction interval is wider than the CI on the population mean with the same confidence.
The 99% confidence interval is
x t 0.005,12 s
2.9 3.106(0.099)
1
1
x + t 0.005,12 s
n
n
1
1
2.9 + 3.106(0.099)
12
12
2.81 2.99
The prediction interval is even wider than the CI on the population mean with greater confidence.
8-72
x t 0.05, 24 s 1 +
1
x n +1
n
1
x n +1
25
3.91 x n +1
4.05 1.711(0.08) 1 +
The prediction interval bound is much lower than the confidence interval bound of
8-24
4.023 mm
8-73
95% tolerance interval on the life of the tires that has a 95% CL
given x = 60139.7 s = 3645.94 n = 16 we find k=2.903
x ks , x + ks
60139.7 2.903(3645.94 ), 60139.7 + 2.903(3645.94 )
(49555.54, 70723.86)
95% confidence interval (58,197.3 62,082.07) is shorter than the 95%tolerance interval.
8-74
99% tolerance interval on the Izod impact strength PVC pipe that has a 90% CL
given x=1.25, s=0.25 and n=20 we find k=3.368
x ks , x + ks
1.25 3.368(0.25 ), 1.25 + 3.368(0.25 )
(0.408, 2.092)
The 99% tolerance interval is much wider than the 99% confidence interval on the population mean (1.090
1.410).
8-75
95% tolerance interval on the syrup volume that has 90% confidence level
x = 1.10 s = 0.015 n = 25 and k=2.474
x ks , x + ks
1.10 2.474(0.015), 1.10 + 2.474(0.015)
(1.06, 1.14)
8-76
99% tolerance interval on the polyunsaturated fatty acid in this type of margarine that has a
confidence level of 95% x = 16.98 s = 0.319 n=6 and k = 5.775
x ks , x + ks
95% tolerance interval on the rainfall that has a confidence level of 95%
95% tolerance interval on the diameter of the rods in exercise 8-27 that has a 90% confidence level
x = 8.23 s = 0.0.25 n=15 and k=2.713
8-25
x ks , x + ks
8.23 2.713(0.025), 8.23 + 2.713(0.025)
(8.16, 8.30)
The 95% tolerance interval is wider than the 95% confidence interval on the population mean
(8.216 8.244).
8-79
99% tolerance interval on the brightness of television tubes that has a 95% CL
given x = 317.2 s = 15.7 n = 10 we find k=4.433
x ks , x + ks
317.2 4.433(15.7 ), 317.2 + 4.433(15.7 )
(247.60, 386.80)
The 99% tolerance interval is much wider than the 95% confidence interval on the population mean
301.06 333.34 .
8-80
90% tolerance interval on the comprehensive strength of concrete that has a 90% CL
given x = 2260 s = 35.57 n = 12 we find k=2.404
x ks , x + ks
2260 2.404(35.57 ), 2260 + 2.404(35.57 )
(2174.5, 2345.5)
The 90% tolerance interval is much wider than the 95% confidence interval on the population mean
2237.3 2282.5.
8-81
x ks , x + ks
2.9 4.150(0.099), 2.9 + 4.150(0.099)
(2.49, 3.31)
The 99% tolerance interval is much wider than the 95% CI on the population mean (2.84 2.96)
8-82
x ks , x + ks
4.05 2.077(0.08 ), 4.05 + 2.077(0.08 )
(3.88, 4.22)
The lower bound of the 90% tolerance interval is much lower than the lower bound on the 95% confidence
interval on the population mean (4.023 )
8-26
b) 90% lower tolerance bound on bottle wall thickness that has confidence level 90%.
given x = 4.05 s = 0.08 n = 25 and k = 1.702
x ks
4.05 1.702(0.08 )
3.91
The lower tolerance bound is of interest if we want to make sure the wall thickness is at least a certain
value so that the bottle will not break.
8-27
Supplemental Exercises
8-83
1 + 2 = . Let = 0.05
Interval for 1 = 2 = / 2 = 0.025
The confidence level for x 1.96 / n x + 1.96 / n is determined by the by the value of z0
Where
which is 1.96. From Table III, we find (1.96) = P(Z<1.96) = 0.975 and the confidence level is 95%.
Interval for 1 = 0.01, 2 = 0.04
The confidence interval is x
because = 0.05 . The symmetric interval does not affect the level of significance; however, it does
affect the length. The symmetric interval is shorter in length.
8-84
= 50
unknown
a) n = 16 x = 52 s = 1.5
52 50
to =
8 / 16
=1
The P-value for t0 = 1, degrees of freedom = 15, is between 0.1 and 0.25. Thus we would conclude that the
results are not very unusual.
b) n = 30
to =
52 50
= 1.37
8 / 30
The P-value for t0 = 1.37, degrees of freedom = 29, is between 0.05 and 0.1. Thus we conclude that the
results are somewhat unusual.
c) n = 100 (with n > 30, the standard normal table can be used for this problem)
zo =
52 50
= 2.5
8 / 100
The P-value for z0 = 2.5, is 0.00621. Thus we conclude that the results are very unusual.
d) For constant values of x and s, increasing only the sample size, we see that the standard error of X
decreases and consequently a sample mean value of 52 when the true mean is 50 is more unusual for the
larger sample sizes.
8-85
= 50, 2 = 5
n = 16 find P( s 2 7.44) or P( s 2 2.56)
15(7.44)
2
P( S 2 7.44) = P 152
= 0.05 P 15 22.32 0.10
2
5
2
Using Minitab P( S 7.44) =0.0997
15(2.56)
2
P( S 2 2.56) = P 152
= 0.05 P 15 7.68 0.10
5
2
Using Minitab P( S 2.56) =0.064
a) For
b) For
8-28
2 29(7.44)
2
P( S 2 7.44) = P 29
2
Using Minitab P( S 7.44) = 0.044
2 29(2.56)
2
P( S 2 2.56) = P 29
2
Using Minitab P( S 2.56) = 0.014.
n = 71 P( s 2 7.44) or P( s 2 2.56)
70(7.44)
2
P( S 2 7.44) = P 702
= 0.005 P 70 104.16 0.01
5
2
Using Minitab P( S 7.44) =0.0051
70(2.56)
2
P( S 2 2.56) = P 702
= P 70 35.84 0.005
5
2
Using Minitab P( S 2.56) < 0.001
c) For
d) The probabilities get smaller as n increases. As n increases, the sample variance should approach the
population variance; therefore, the likelihood of obtaining a sample variance much larger than the
population variance will decrease.
e) The probabilities get smaller as n increases. As n increases, the sample variance should approach the
population variance; therefore, the likelihood of obtaining a sample variance much smaller than the
population variance will decrease.
8-86
a) The data appear to follow a normal distribution based on the normal probability plot since the data fall
along a straight line.
b) It is important to check for normality of the distribution underlying the sample data since the confidence
intervals to be constructed should have the assumption of normality for the results to be reliable (especially
since the sample size is less than 30 and the central limit theorem does not apply).
c) No, with 95% confidence, we can not infer that the true mean could be 14.05 since this value is not
contained within the given 95% confidence interval.
d) As with part b, to construct a confidence interval on the variance, the normality assumption must hold
for the results to be reliable.
e) Yes, it is reasonable to infer that the variance could be 0.35 since the 95% confidence interval on the
variance contains this value.
f) i) & ii) No, doctors and children would represent two completely different populations not represented by
the population of Canadian Olympic hockey players. Because neither doctors nor children were the target
of this study or part of the sample taken, the results should not be extended to these groups.
8-87
a) The probability plot shows that the data appear to be normally distributed. Therefore, there is no
evidence conclude that the comprehensive strength data are normally distributed.
8-29
s
x t 0.01,8
n
8.42
25.12 2.896
9
16.99
The lower bound on the 99% confidence interval shows that the mean comprehensive strength is most
likely be greater than 16.99 Megapascals.
c) 98% two-sided confidence interval on the mean
s
s
x t0.01,8
x + t0.01,8
n
n
8.42
8.42
25.12 2.896
25.12 + 2.896
9
9
16.99 33.25
The bounds on the 98% two-sided confidence interval shows that the mean comprehensive strength will
most likely be greater than 16.99 Megapascals and less than 33.25 Megapascals. The lower bound of the
99% one sided CI is the same as the lower bound of the 98% two-sided CI (this is because of the value of
)
d) 99% one-sided upper bound on the confidence interval on 2 comprehensive strength
s = 8.42, s 2 = 70.90
02.99,8 = 1.65
8(8.42) 2
1.65
2
343.74
The upper bound on the 99% confidence interval on the variance shows that the variance of the
comprehensive strength is most likely less than 343.74 Megapascals2.
e) 98% two-sided confidence interval on 2 of comprehensive strength
8-30
s
s
x t 0.01,8
x + t 0.01,8
n
n
6.31
6.31
23 2.896
23 + 2.896
9
9
16.91 29.09
98% two-sided confidence interval on 2 comprehensive strength
Fixing the mistake decreased the values of the sample mean and the sample standard deviation. Because
the sample standard deviation was decreased the widths of the confidence intervals were also decreased.
g) The exercise provides s = 8.41 (instead of the sample variance). A 98% two-sided confidence interval
s
s
x t0.01,8
x + t0.01,8
n
n
8.41
8.41
25 2.896
25 + 2.896
9
9
16.88 33.12
98% two-sided confidence interval on 2 of comprehensive strength
Fixing the mistake did not affect the sample mean or the sample standard deviation. They are very close to
the original values. The widths of the confidence intervals are also very similar.
h) When a mistaken value is near the sample mean, the mistake will not affect the sample mean, standard
deviation or confidence intervals greatly. However, when the mistake is not near the sample mean, the
value can greatly affect the sample mean, standard deviation and confidence intervals. The farther from the
mean, the greater is the effect.
8-88
With = 8, the 95% confidence interval on the mean has length of at most 5; the error is then E = 2.5.
2
z
2 1.96
a) n = 0.025 8 =
64 = 39.34 = 40
2.5
2.5
z
2 1.96
b) n = 0.025 6 =
36 = 22.13 = 23
2.5
2.5
8-31
As the standard deviation decreases, with all other values held constant, the sample size necessary to
maintain the acceptable level of confidence and the length of the interval, decreases.
8-89
x ks , x + ks
15.33 2.564(0.62 ), 15.33 + 2.564(0.62 )
(13.74,
16.92)
= 3.368
x ks , x + ks
15.33 3.368(0.62 ), 15.33 + 3.368(0.62 )
(13.24, 17.42)
8-90
95% prediction interval for the next sample of concrete that will be tested.
given x = 25.12 s = 8.42 n = 9 for = 0.05 and n = 9, t /2,n-1 = t0.025,8 = 2.306
x t 0.025,8 s 1 +
1
1
x n +1 x + t 0.025,8 s 1 +
n
n
1
1
x n +1 25.12 + 2.306(8.42) 1 +
9
9
4.65 x n +1 45.59
25.12 2.306(8.42) 1 +
a) There is no evidence to reject the assumption that the data are normally distributed.
Normal Probability Plot for foam height
ML Estimates - 95% CI
99
ML Estimates
95
90
Mean
203.2
StDev
7.11056
80
Percent
8-91
70
60
50
40
30
20
10
5
1
178
188
198
208
218
228
Data
8-32
s
s
x t 0.025,9
x t 0.025,9
n
n
7.50
7.50
203.2 2.262
203.2 + 2.262
10
10
197.84 208.56
c) 95% prediction interval on a future sample
1
1
x t 0.025,9 s 1 +
n
n
x t 0.025,9 s 1 +
1
1
203.2 + 2.262(7.50) 1 +
10
10
185.41 220.99
d) 95% tolerance interval on foam height with 99% confidence k = 4.265
x ks , x + ks
203.2 2.262(7.50) 1 +
99
95
90
Percent
80
70
60
50
40
30
20
10
5
1
0.59
0.60
0.61
0.62
0.63
0.64
0.65
0.66
Data
8-33
x t 0.005,39
0.624 2.7079
c)
n
0.013
x + t 0.005,39
s
n
0.624 + 2.7079
40
0.618 0.630
0.013
40
99% prediction interval on the coefficient of restitution for the next baseball that will be tested.
x t 0.005,39 s 1 +
1
1
x n +1 x + t 0.005,39 s 1 +
n
n
1
1
x n +1 0.624 + 2.7079(0.013) 1 +
40
40
0.588 x n +1 0.660
0.624 2.7079(0.013) 1 +
d) 99% tolerance interval on the coefficient of restitution with a 95% level of confidence
( x ks , x + ks )
(0.624 3.213(0.013), 0.624 + 3.213(0.013))
(0.582, 0.666)
e) The confidence interval in part (b) is for the population mean and we may interpret this to imply that
99% of such intervals will cover the true population mean. For the prediction interval, 99% of such
intervals will cover a future baseballs coefficient of restitution. For the tolerance interval, 95% of such
intervals will cover 99% of the true distribution.
8-93
95% Confidence Interval on the proportion of baseballs with a coefficient of restitution that
exceeds 0.635.
8
= 0.2 n = 40
40
p (1 p )
p z
p
n
p =
0.2 1.65
8-94
z = 1.65
0.2(0.8)
p
40
0.0956 p
a) The normal probability shows that the data are mostly follow the straight line, however, there are some
points that deviate from the line near the middle. It is probably safe to assume that the data are normal.
8-34
99
95
90
Percent
80
70
60
50
40
30
20
10
5
1
0
10
Data
x t 0.025,19
3.265 2.093
c)
n
2.127
x + t 0.025,19
s
n
3.265 + 2.093
20
2.270 4.260
2.127
20
95% prediction interval on the oxygen concentration for the next stream in the system that will be
tested..
x t 0.025,19 s 1 +
1
1
x n +1 x + t 0.025,19 s 1 +
n
n
1
1
x n +1 3.265 + 2.093(2.127) 1 +
20
20
1.297 x n +1 7.827
3.265 2.093(2.127) 1 +
d) 95% tolerance interval on the values of the dissolved oxygen concentration with a 99% level of
confidence
( x ks , x + ks )
(3.265 3.168(2.127), 3.265 + 3.168(2.127))
(3.473, 10.003)
e) The confidence interval in part (b) is for the population mean and we may interpret this to imply that
95% of such intervals will cover the true population mean. For the prediction interval, 95% of such
intervals will cover a future oxygen concentration. For the tolerance interval, 99% of such intervals will
cover 95% of the true distribution
8-95
a) There is no evidence to support that the data are not normally distributed. The data points appear to fall
along the normal probability line.
8-35
99
ML Estimates
95
90
Mean
1.529
StDev
0.0556117
Percent
80
70
60
50
40
30
20
10
5
1
1.4
1.5
1.6
1.7
Data
x t 0.005, 29
1.529 2.756
c)
n
0.0566
x + t 0.005, 29
s
n
1.529 + 2.756
30
1.501 1.557
0.0566
30
99% prediction interval on the tar content for the next sample that will be tested..
x t 0.005,19 s 1 +
1
1
x n +1 x + t 0.005,19 s 1 +
n
n
1
1
x n +1 1.529 + 2.756(0.0566) 1 +
30
30
1.370 x n +1 1.688
1.529 2.756(0.0566) 1 +
d) 99% tolerance interval on the values of the tar content with a 95% level of confidence
( x ks , x + ks )
(1.529 3.350(0.0566), 1.529 + 3.350(0.0566))
(1.339, 1.719)
e) The confidence interval in part (b) is for the population mean and we may interpret this to imply that
95% of such intervals will cover the true population mean. For the prediction interval, 95% of such
intervals will cover a future observed tar content. For the tolerance interval, 99% of such intervals will
cover 95% of the true distribution
8-96
8-36
p z a / 2
0.0067 1.96
p (1 p )
p p + z a / 2
n
p (1 p )
n
0.0067(1 0.0067)
0.0067(1 0.0067)
p 0.0067 + 1.96
1200
1200
0.0021 p 0.0113
b) No, there is not sufficient evidence to support the claim that the fraction of defective units produced is
one percent or less at = 0.05. This is because the upper limit of the control limit is greater than 0.01.
8-97
p (1 p )
p p + z a / 2
n
p z a / 2
0.005 2.58
p (1 p )
n
0.005(1 0.005)
0.005(1 0.005)
p 0.005 + 2.58
1600
1600
0.0004505 p 0.009549
z
2.58
n = / 2 p(1 p) =
0.005(1 0.005) = 517.43 , n 518
0.008
E
c) E = 0.008, = 0.01, z /2 = z0.005 = 2.58
z
2.58
n = / 2 p(1 p) =
0.5(1 0.5) = 26001.56 , n 26002
0.008
E
d) A bound on the true population proportion reduces the required sample size by a substantial amount. A
sample size of 518 is much more reasonable than a sample size of over 26,000.
8-98
p =
117
= 0.242
484
p z / 2
z / 2 = 1.645
p (1 p )
p p + z / 2
n
0.210 p 0.274
p (1 p )
n
With 90% confidence, the true proportion of new engineering graduates who were planning to continue
studying for an advanced degree is between 0.210 and 0.274.
b) 95% confidence interval; z/ 2 = 196
.
p z / 2
p (1 p )
p p + z / 2
n
0.204 p 0.280
p (1 p )
n
With 95% confidence, we believe the true proportion of new engineering graduates who were planning to
continue studying for an advanced degree lies between 0.204 and 0.280.
c) Comparison of parts (a) and (b):
8-37
The 95% confidence interval is larger than the 90% confidence interval. Higher confidence always yields
larger intervals, all other values held constant.
d) Yes, since both intervals contain the value 0.25, thus there in not enough evidence to determine that the
true proportion is not actually 0.25.
8-99
a) The data appear to follow a normal distribution based on the normal probability plot
since the data fall along a straight line.
b) It is important to check for normality of the distribution underlying the sample data
since the confidence intervals to be constructed should have the assumption of
normality for the results to be reliable (especially since the sample size is less than 30
and the central limit theorem does not apply).
c) 95% confidence interval for the mean
n
n
6.33
6.33
22.73 2.228
22.73 + 2.228
11
11
18.478 26.982
d) As with part b, to construct a confidence interval on the variance, the normality
assumption must hold for the results to be reliable.
e) 95% confidence interval for variance
n = 11 s = 6.33
2 / 2,n1 = 02.025,10 = 20.48
and
10(6.33) 2
10(6.33) 2
2
20.48
3.25
2
19.565 123.289
8-100
a) The data appear to be normally distributed based on examination of the normal probability plot below.
Therefore, there is evidence to support that the energy intake is normally distributed.
8-38
99
Mean
StDev
N
AD
P-Value
95
90
Percent
80
70
60
50
40
30
20
10
5
6
Energy
n
n
0.5645
0.5645
5.884 3.250
5.884 + 3.250
10
10
5.304 6.464
8-39
5.884
0.5645
10
0.928
0.011
a.)
,2r
1 , 2 r
= P 2 < < 2
2Tr
2Tr
is
2T
r , 2Tr
2 2
2 , 2 r 1 2 , 2 r
8-102
z 1
1 =
Therefore,
x2
is
2(489) 2(489)
,
= (28.62,101.98)
34.17 9.59
z 1
1 x22
dx = 1
e dx
2
1 1 = ( z 1 ) .
1 (1 1 ) + (1 2 )
1 (1 1 ) + (1 + 1 ) .
1 (1 1 ) = 2 e
1
subject to
z2
1
2
1 (1 + 1 ) = 2 e
1
z2
Upon setting the sum of the two derivatives equal to zero, we obtain
z1 = z 1 . Consequently,
8.103
1 = 1 , 2 1 =
8-104
1 + 2 = . Therefore, we
a)
P( X i ~) = 1 / 2
P(allX ~) = (1 / 2) n
i
P(allX i ~) = (1 / 2) n
8-40
and
z2
2
1 = 2 =
=e
z2
1
2
. This is solved by
P( A B) = P( A) + P( B) P( A B)
n
1 1
1
1
= + = 2 =
2 2
2
2
n 1
1
1 P( A B) = P(min( X i ) < ~ < max( X i )) = 1
2
~
b) P(min( X i ) < < max( X i )) = 1
We would expect that 950 of the confidence intervals would include the value of . This is due to 9the
definition of a confidence interval.
Let X bet the number of intervals that contain the true mean (). We can use the large sample
approximation to determine the probability that P(930 < X < 970).
Let
p =
950
930
970
= 0.950 p1 =
= 0.930 and p 2 =
= 0.970
1000
1000
1000
p(1 p) 0.950(0.050)
=
n
1000
(0.970 0.950)
(0.930 0.950)
P(0.930 < p < 0.970) = P Z <
P Z <
0.950(0.050)
0.950(0.050)
1000
1000
0.02
0.02
= P Z <
P Z <
= P( Z < 2.90) P( Z < 2.90) = 0.9963
0.006892
0.006892
8-41