CHAPTER 4
Section 4-2
StDev 3.12
4-1. SE Mean 0.99
n 10
Variance StDev 2 3.12 2 9.7344
Variable N Mean SE Mean StDev Variance Minimum Maximum
X 9 19.96 0.99 3.12 9.7344 15.94 27.16
Sum x 399.851
4-2. Mean 24.99
n 16
StDev SE Mean n 0.159 16 0.636
2n
Xi
X 1 E i 1 1 E X 1 2n
2n
2n
4-5. E i
2n i 1 2n
n
Xi
i 1 1 n 1
E X2 E E Xi n ,
n n i 1 n
X1 and X2 are unbiased estimators of .
2 2
The variances are V X1
2n
and V X2
n
; compare the MSE (variance in this case),
1
Engineering Statistics, 5th edition September 15, 2010
MSE( 1 ) 2 / 2 n n 1
2
MSE( 2 ) /n 2n 2
Because both estimators are unbiased, examination of the variances would conclude that X1 is the “better” estimator
with the smaller variance.
4-6. 1 1 1
E ˆ1 E ( X 1 ) E ( X 2 ) E ( X 9 ) (9 E ( X )) (9 )
9 9 9
1 1
E ˆ2 E (3 X 1 ) E ( X 6 ) E (2 X 4 ) [3 2 ] 2
2 2
a) 1 is unbiased estimates and because the expected values of these statistics are equivalent to the true mean, ., but
2 is biased estimates of
X X 2 ... X 9 1
b) V ˆ 1 V 1
9
1 1 2
2 V( X1 ) V (X 2 ) V(X 9 ) 81 (9 ) 9
2
9
2
V(ˆ 1 )
9
3X X 6 2X 4 1
V ˆ 2 V 1
2
2
V(3X1 ) V(X 6 ) V(2X 4 ) 1 (9V(X1 ) V(X 6 ) 4V(X 4 ))
4
2
=
1
4
9 2 2 4 2
1
= (14 2 )
4
7 2
V(ˆ 2 )
2
Because both estimators are not unbiased, the variances cannot be compared to decide which one is the better
estimator.
4-7. Because both 1 and 2 are unbiased, the variances of the estimators can be examined to determine which is the
“better” estimator. The variance of ̂1 is smaller than that of 2 , thus 1 is selected as the better estimator.
MSE (ˆ 1 ) V(ˆ 1 ) 2
Relative Efficiency = 0.5
MSE (ˆ 2 ) V(ˆ 2 ) 4
2
Engineering Statistics, 5th edition September 15, 2010
V ( 1 ) = 10 V ( 2 ) = 4
For unbiasedness, use 1 because it is the only unbiased estimator. As for minimum variance and efficiency we have:
( V( 1 ) Bias2 )1
Relative Efficiency = where the bias for 1 is 0.
( V( ) Bias2 )
2 2
(10 0) 40
Thus, Relative Efficiency =
2
4 16
2
2
40
If the relative efficiency is less than or equal to 1, 1 is the better estimator. Use 1 , when 1
(16 2 )
40 (16 2 )
24 2
4.899 or 4.899
Conclusion: 3 is the most efficient estimator with bias, while 2 is the best “unbiased” estimator.
1
38
E 20S12 E 10S22 E 8S23
1
38
2012 10 22 8 23
1
38
38 2
2
3
Engineering Statistics, 5th edition September 15, 2010
n 2
Xi X
4-13. a) Show that i 1 is a biased estimator of 2 :
n
n 2
Xi X
E i 1
n
2
1 n
E Xi nX
n i 1
1 n
n i 1
2
2
E Xi nE X
1 n 2 2
n i 1
2 n 2
n
1
n
n 2 n 2 n 2 2
1
(( n 1) 2 )
n
2
2
n
Xi X
2
b) Bias = E
2
X i2 nX
2 2
2
2
2
n n n
c) Bias decreases as n increases.
E X2
n
1 n
2 i
E
i 1
X
2
1 n n
V i i
X E X
n2 i 1 i 1
1 2 n
2
n
n2 i 1
1
n2
n 2
n
2
1
n n
n2
2 2 2
E X
2
2 2
n
Therefore, X 2 is a biased estimator of .
4
Engineering Statistics, 5th edition September 15, 2010
b) Bias = E X 2 2
2
n
2 2
2
n
c) Bias decreases as n increases.
Section 4-3
4-15. a) P value ( z0 )
X 11.7 12
P
/ n 0.5 / 5
P( Z 1.34)
0.090123
P-value is 0.090123.
0.5
b) For = 0.05, fail to reject H0 if X 12 1.645 11.63
5
11.63 11.25
=P X 11.63 when 11.25 1 P Z 1 P Z 1.7 0.044565
0.5 / 5
The probability is 0.044565 of failing to reject the null hypothesis if the true mean elongation force is 11.25 kg, with a
level of significance of = 0.05.
c) 1 - = 1 – 0.044565 = 0.955435
X 11.7 12
4-16. a) P value ( z0 ) = P = P(Z 2.4) = 0.
/ n 0.5 / 16
The P-value is approximately 0 when a sample size of 16.
0.5
b) For = 0.05, fail to reject H0 if X 12 1.645 11.79
16
11.79 11.25
=P X 11.79 when 11.25 1 P Z 1 P Z 4.32 0
0.5 / 16
The probability of failing to reject the null hypothesis if the true mean elongation force is 11.25 is approximately 0,
with a level of significance of = 0.05.
c) 1 - = 1 – 0= 1
4-17. Find the boundary of the critical region if = 0.01:
c 14
a) 0.01 = P Z
0.3 / 5
What Z value will give a probability of 0.01? Using Table 1 in the appendix, Z value is 2.33.
c 14
Thus, = 2.33, c = 13.687
0.3 / 5
0.3
b) For = 0.01, fail to reject H0 if X 14 2.33 13.69
5
13.69 13.5
=P X 13.69 when 13.5 1 P Z 1 P Z 1.42 0.077804
0.3 / 5
The probability is 0.077804 of failing to reject the null hypothesis if the true mean elongation force is 13 kg, with a
level of significance of = 0.01.
5
Engineering Statistics, 5th edition September 15, 2010
c) 1 - = 1 – 0.077804= 0.922196
c 12
4-18. a) 0.05 = PZ
0.5 / 16
What Z value will give a probability of 0.05? Using Table 1 in the appendix, Z value is 1.645.
c 12
Thus, = 1.645, c = 11.794
0.5 / 16
0.5
b) For = 0.05, fail to reject H0 if X 12 2.33 11.71
16
11.71 14
=P X 11.71when 14 1 P Z 1 P Z 18.32 1
0.5 / 16
The probability of failing to reject the null hypothesis if the true mean elongation force is 11.25 kg is approximately 0,
with a level of significance of = 0.01.
c) 1 - = 1 – 0 = 1
The probability of failing to reject the null hypothesis when it is actually false is smaller in part c) because the true
mean, = 85, is further from the acceptance region. A larger difference exists.
6
Engineering Statistics, 5th edition September 15, 2010
4-21. a) P( X 185)
X 185 175
P
/ n 20 / 10
P( Z 1.58) 1 P ( Z 1.58)
0.05705
is 0.05705.
b) For = 0.05, fail to reject H0 if X 175 1.645 20 185.4
10
185.4 200
=P X 185.4 when 200 P Z P Z 2.31 0.010444
20 / 10
The probability is 0.0104 of failing to reject the null hypothesis if the true mean foam height is 200 mm, with a level of
significance of = 0.05.
c) 1 - = 1 – 0.0104 = 0.9896
4-22.
a) Because the P-value = 0.057 > α = 0.05, fail to reject the null hypothesis. There is not sufficient evidence to
conclude that the height is greater than 175 mm.
The probability that a value of at least 190 mm would be observed (if the true mean height is 175 mm) is only 0.00889.
Thus, the sample value of x = 190 mm would be an unusual result.
7
Engineering Statistics, 5th edition September 15, 2010
= 1 0.97725
= 0.0228
4-24. n = 16:
c 175
a) 0.05 = P X c when 175 = P Z
20 / 16
= P(Z 1.645)
c 175
Thus, 1.645 = , and c = 183.225
20 / 16
c) For the same level of , with the increased sample size, is reduced. With an increased sample size, the power is
also increased.
b) Power = 1
= P(4.85 X 5.15 when = 5.1)
4.85 5.1 X 5.15 5.1
= P
0.25 / 8 / n 0.25 / 8
= P(2.83 Z 0.57)
= P(Z 0.57) P(Z 2.83)
= 0.715661 0.002327
= 0.7133
c) 1 = 0.2867.
8
Engineering Statistics, 5th edition September 15, 2010
4-27.
= P X cL + P X cH
0.025 = P X c L , 0.025 = P X c H
c 5 = 0.025, thus, 1.96 = cL 5 ; c L = 4.827
P L
0.25 / 8 0.25 / 8
c 5 = 0.025, thus, 1.96 = cH 5 ; c H = 5.173
P H
0.25 / 8 0.25 / 8
P 4.827 X 5.173
Section 4-4
4-29.
a) P value 2 1 z 0 2 1 2.45 2[1 0.992857] 0.014286
b) P value 21 z 0 21 1.53 2[1 0.936992] 0.126016
c) P value 21 z 0 21 2.15 2[1 0.984222] 0.031556
d) P value 21 z 0 21 1.95 2[1 0.974412] 0.051176
e) P value 21 z 0 21 0.25 2[1 0.598706] 0.802588
31.2 20
4-31. a) Z 37.33
1.2 / 16
P value 2[1 (| Z |)] 2[1 ( | 37.33 |)] 2[1 1] 0
9
Engineering Statistics, 5th edition September 15, 2010
One-Sample Z
Test of mu = 20 vs not = 20
The assumed standard deviation = 1.2
b) It is two-sided test.
c) 99% CI, α = 0.01
x z 0.01 / 2 x z 0.01 / 2
n
n
1.2
31.2 2.58 31.2 2.58 1.2
16 16
31.2 0.774 31.2 0.774
30.426 31.974
d) P value 1 ( z ) 1 (4) 1 1 0
2.4
4-32. a) SE Mean 0.48
n 25
P value 1 ( z ) 1 (3.25) 1 0.999423 0.000577
One-Sample Z
95%
Lower
N Mean SE Mean Bound Z P
25 101.560 0.48 100.770 3.25 0.000577
Yes, the null hypothesis can be rejected at the 0.05 level because the P-value = 0.000577 < 0.05.
b) It is a one-sided test.
c) The null hypothesis is rejected at the 0.05 level because μ0 = 99 is not included in the 95% lower CI.
No additional calculations are needed because the lower bound calculation is x z does not depend on the
n
change of μ0 in the hypothesis and one can just check that μ0 = 99 is not contained in the lower CI.
10
Engineering Statistics, 5th edition September 15, 2010
4-33. a) The half-length of the CI is 22.968 – 21.400 = 1.568, and 1.568 1.96 , so SE Mean = / n = 0.80.
n
x 20 21.4 20
z 1.75, P 2[1 (1.75)] 2(0.0401) 0.0802 .
/ 25 0.80
Because the P-value exceeds 0.05 we fail to reject the null hypothesis.
b) z x 20 21.4 20 1.75, P 1 (1.75) 0.0401 , so the null hypothesis is rejected.
/ 25 0.80
(c) Z
0.005 2.5758, so the CI is x 2.5758 21.4 2.5758(0.80), so 19.339 23.461
n
4.34. (a) Z 0.05 1.645, so the lower confidence bound is 105.20-1.645(1.77) = 102.288.
x 100 105.20 100
z 2.9379, P 1 (2.9379) 0.0017 , so the null hypothesis would be rejected.
/ 8 1.77
(b) The P-value is now 2(0.0017) = 0.0034. The null hypothesis is rejected.
(c) No. Z 0.025 1.96, so the lower confidence bound is 105.20-1.96 (1.77) = 101.7308.
4-36. (a) The sample mean is the mid-point of the two intervals, or sample mean = 50.00.
(b) The longest interval is the 95% CI, and that is the first one (32.37, 67.63).
7) Because 1.77>1.645, reject the null hypothesis and conclude that the true average battery life exceeds 4 hours at
= 0.05.
(4.5 4) 50
c) z 0.05 = (1.645 – 17.68) = (-16.035) = 0
0.2
Power = 1- = 1-0 = 1
z z
2
d) n =
2
z0.05 z0.1 2 2
(1.645 1.29) 2 (0.2) 2
1.38,
2 2
(4.5 4) (0.5) 2
n2
e) x z 0.05
n
11
Engineering Statistics, 5th edition September 15, 2010
0. 2
4.05 1.65
50
4.003
Because the lower limit of the CI is just slightly greater than 4, we conclude that the average life is greater than 4 hours
at = 0.05.
Because the P-value > 0.05, we fail to reject H0. There is not sufficient evidence to conclude that the fiber would be
judged acceptable.
b) For = 0.05
(703 690) 9
= z 0.05
14 1.645 3 ( 1.36) 0.086915
The probability is 0.086915 of failing to reject the null hypothesis if the true mean breaking strength is 703 kPa, with a
level of significance of = 0.05.
c) z = z0.05 = 1.645
x z 0.05
n
14
694 1.645
9
686.32
With 95% confidence, the true mean breaking strength is at least 686.32 kPa.
d) Because the value of 100 falls within the confidence interval, we fail to reject the null hypothesis.
( z z ) 2 2 (1.645 1.645) 2142
e) n 58.93 , n = 59
2 62
12
Engineering Statistics, 5th edition September 15, 2010
7) P-value = 2[1 (0.36)] 2[1 0.64058] 0.7188 . Because the P-value > 0.05, we fail to reject H0. There is
not sufficient evidence to conclude that the yield is different from 90%.
92 90 92 90
c) z 0.025 z 0.025
3/ 5 3/ 5
= (1.96 – 1.49) (1.96 – 1.49)
= (0.47) (–3.45)
= 0.680822 0.000280
= 0.680542
With 95% confidence, the true mean yield of the chemical process is between 87.85% and 93.11%.
e) Based on the confidence interval obtained, the null hypothesis cannot be rejected because the hypothesized value of
90% lies within this interval.
7) The P-value = (-2.925) 0.00172. Because the P-value < 0.01, we reject the null hypothesis and conclude that
the exit water meets the federal regulation.
7920 7980
b) z 0.01 4.70 0
80 / 10
c) Set = 1 0.90 = 0.10
( z z ) 2 2 2 2 (2.33 1.29)2 (80) 2
n= = ( z0.01 z0.10 ) = 23.29,
2 (7920 7980) 2 (60)2
n 24.
13
Engineering Statistics, 5th edition September 15, 2010
80
7906 2.33
10
7965
e) The confidence interval constructed does not contain the value 7980. We conclude that the manufacturer’s exit water
meets federal regulation using a 0.01 level of significance.
2.03 2.00
b) z0.01 (0.35) 0.363169
0.05 / 20
e) The confidence interval constructed contains the value 2.00. We fail to reject the null hypothesis that the average
distance of the foil to the edge of the inflator is 2.00 using a 0.01 level of significance.
14
Engineering Statistics, 5th edition September 15, 2010
z z
2
d) n =
2
z0.05 z0.10 2 2
(1.645 1.29) 2 (20)2
8.614,
2 2
(560 540) (20) 2
n 9
e) x z0.050
n
20
551.33 1.645
15
542.83
With 95% confidence, the true mean life is at least 542.83 hrs.
f) Because the value 540 is not within this interval, reject the null hypothesis.
7) Because 26.79 < 2.58, reject the null hypothesis and conclude the mean compressive strength is significantly
different from 3500 at = 0.01.
With 95% confidence, the true mean compressive strength is between 3237.53 psi and 3273.31psi.
With 99% confidence, the true mean compressive strength is between 3231.96 psi and 3278.88 psi.
The 99% confidence interval is wider than the 95% confidence interval. The confidence interval with the greater level of
confidence always results in a wider confidence interval when x , 2, and n are held constant.
15
Engineering Statistics, 5th edition September 15, 2010
2
20(1.96) , n 62.
n 61.4656
5
Section 4-5
4-47. a) P-value = P(t > 2.05): for degrees of freedom of 14 we obtain 0.025 < P-value < 0.05
b) P-value = P(t > |–1.84|): for degrees of freedom of 14 we obtain 0.025 < P-value < 0.05
c) P-value = P(t > 2.00): for degrees of freedom of 14 we obtain 0.025 < P-value < 0.05
d) P-value = P(t > 0.4): for degrees of freedom of 14 we obtain 0.25 < P-value < 0.4
4-49. a) P-value = P(t > 2.05): for degrees of freedom of 24 we obtain 0.025 < P-value < 0.05
b) P-value = P(t > |–1.84|): for degrees of freedom of 24 we obtain 0.025 < P-value < 0.05
c) P-value = P(t > 2.00): for degrees of freedom of 24 we obtain 0.025 < P-value < 0.05
d) P-value = P(t > 0.4): for degrees of freedom of 24 we obtain 0.25 < P-value < 0.40
One-Sample T: X
Test of mu = 91 vs not = 91
The null hypothesis is rejected at the 0.05 level because the P-value = 0.002 < 0.05.
b) It is a two-sided test.
c) The null hypothesis H 0 : 90 is rejected at the 0.05 level because 90 is not included in the 95% CI
d) 99% CI
16
Engineering Statistics, 5th edition September 15, 2010
s
x t 0.005, 24 x t 0.005, 24 s
n n
2.3365
92.5805 2.797 92.5805 2.797 2.3365
25 25
92.5805 1.307 92.5805 1.307
91.2735 93.8875
e) P-value = P(t > 3.38): for degrees of freedom of 24 we obtain 0.001 < P-value < 0.0025
One-Sample T: X
Test of mu = 25 vs > 25
95%
Lower
Variable N Mean StDev SE Mean Bound T P
X 12 25.6818 1.1639 0.3360 25.0783 2.029 0.034
b) The points on the following normal probability plot fall along the line. Therefore, there is no evidence that the
speed-up data is not normally distributed.
17
Engineering Statistics, 5th edition September 15, 2010
Percent
60
50
40
30
20
10
5
1
3.0 3.5 4.0 4.5 5.0 5.5 6.0
Speed-up
c) 95% CI
s s
x t0.025, 24 x t0.025, 24
n n
0.4328 0.4328
4.31322 2.064 4.31322 2.064
13 13
4.31322 0.2478 4.31322 0.2478
4.0654 4.561
One-sided
The power is given to be 1 = 0.8. Using the OC curve in Chart V(c), with d = 1.73 and 0.2 we obtain n = 4.
4-54. a) 1) The parameter of interest is the true mean interior temperature life, .
2) H0: = 22.5
3) H1: 22.5
x
4) Test statistic is t 0
s/ n
5) Reject H0 if t0 < -t/2,n-1 or t0 > t/2,df where t0.025,4 = 2.776
6) x 22.496 , s = 0.378 n = 5
22.496 22.5
t0 0.0237
0.378 / 5
7) P-value > 2* 0.4; P-value > 0.8 which is greater than 0.05, fail to reject the null hypothesis.
b) The points on the normal probability plot fall along the line. Therefore, there is no evidence that the interior
temperature data is not normally distributed.
18
Engineering Statistics, 5th edition September 15, 2010
99 ML Estimates
Mean 22.496
95
StDev 0.338384
90
80
70
Percent
60
50
40
30
20
10
5
c)
| 0 | | 22.75 22.5 |
d= 0.66
0.378
For n = 5, using the OC curve,
Power = 1 – = 1 – 0.7 = 0.3 where 0.7
| 0 | | 22.75 22.5 |
d) d = 0.66
0.378
Using the OC curve, Chart V (a) for = 0.05, d = 0.66, and 0.1 (Power=0.9), n 30
0 61000 60000
b) d = 0.3295
3035
Using the OC curve, Chart V(c) for = 0.05, d = 0.3295, and n = 10, we obtain 0.78 and power of 1 0.78 = 0.22.
Because the power is less than the acceptable level, 10 is not an adequate sample size for detecting a difference with
probability of at least 0.90.
19
Engineering Statistics, 5th edition September 15, 2010
4-56. In order to use t statistics in hypothesis testing, we need to assume that the underlying distribution is normal.
1) The parameter of interest is the true Izod impact strength, .
2) H0: = 1.0
3) H1: 1.0
x
4) t0 =
s/ n
5) Reject H0 if t0 > t,n-1 where t0.01,19 = 2.539
6) x = 1.121 s = 0.328 n = 20
1.121 1.0
t0 = 1.65
0.328 / 20
7) Because 1.65 < 2.539, fail to reject the null hypothesis. There is not sufficient evidence to conclude that the true
Izod impact strength is greater than 1.0 ft-lb/in at = 0.01.
4-57. In order to use t statistics in hypothesis testing, we need to assume that the underlying distribution is normal.
a) 1) The parameter of interest is the mean life in hours, .
2) H0: = 5500
3) H1: > 5500
x
4) Test statistic is t0 =
s/ n
5) Reject H0 if t0 > t,n-1 where t0.05,14 = 1.761
6) x = 5625.1 s = 226.1 n = 15
5625.1 5500
t0 = 2.14
226.1 / 15
7) P-value = P(t > 2.14) and for degrees of freedom of 14 we obtain 0.025 < P-value < 0.05. Because the P-value <
0.05, reject the null hypothesis.
s
b) x t 0.05,14
n
226.1
5625.1 1.761
15
5522.3
c) We are 95% confident that the true mean life is at least 5522.3. Therefore, the confidence interval also implies that
the null hypothesis is rejected (because 5500 is not contained in this interval).
4-58.
Normal Probability Plot
.999
.99
.95
Probability
.80
.50
.20
.05
.01
.001
16.5 16.6 16.7 16.8 16.9 17.0 17.1 17.2 17.3 17.4
Fatty
Av erage: 16.9833 Anderson-Darling Normality Test
StDev : 0.318852 A-Squared: 0.143
N: 6 P-Value: 0.934
The normality assumption appears to be satisfied because the data fall along a straight line.
20
Engineering Statistics, 5th edition September 15, 2010
a) 1) The parameter of interest is the true mean level of polyunsaturated fatty acid, .
2) H0: = 17
3) H1: 17
x
4) Test statistic is t0 =
s/ n
5) Reject H0 if t0 > t/2,n-1 where t0.025,5 = 2.571
6) x = 16.98 s = 0.319 n = 6
16.98 17
t0 = 0154
.
0.319 / 6
7) P-value = 2P(t > 0.154): for degrees of freedom of 5 we obtain 2(0.40) < P-value = 0.80 < P-value. Because the
P-value is greater than 0.05, we fail to reject the null hypothesis.
0.5
b) Using the OC curves on Chart v (b), with d = 1.567, when 0.1, n = 10. Therefore, the current sample
0.319
size of 6 is not adequate.
With 99% confidence, the true mean level of polyunsaturated fatty acid is between 16.455% and 17.505%.
4-59
a) According to the normal probability plot, the data appear to follow a normal distribution. The data fall along a
straight line.
Normal Probability Plot
.999
.99
.95
Probability
.80
.50
.20
.05
.01
.001
21
Engineering Statistics, 5th edition September 15, 2010
x
4) t0 =
s/ n
5) Reject H0 if t0 < t,n-1 where t0.05,11 = 1.796
6) x = 9.117, s = 0.464 n = 12
9.117 9
t0 0.8735
0.464 / 12
7) Because 0.8735 > 1.796, fail to reject the null hypothesis. There is insufficient evidence to conclude that the true
mean breakdown voltage is less than 9 volts at = 0.05.
d) Because 9 volts lies within this confidence interval, we fail to reject the null hypothesis.
8.8 9
e) Using the OC curves on Chart V (c), with d = 0.431, when 0.05, we have n 60.
0.464
4-60.
a) According to the normal probability plot there does not seem to be a severe deviation from normality for this data.
The data appears to fall along a straight line.
Normal Probability Plot
.999
.99
.95
Probability
.80
.50
.20
.05
.01
.001
22
Engineering Statistics, 5th edition September 15, 2010
8.2767 8.2
t0 = 3.18
0.0836 / 12
7) Because 3.18 > 2.201, reject the null hypothesis and conclude the true mean rod diameter is not equal to 8.2 mm
at the 0.05 level of significance.
With 95% confidence, the true mean rod diameter is between 8.226 mm and 8.33 mm. Because this interval does not
contain the hypothesized value of 8.2, we reject the null hypothesis.
Because 3.58 > 1.711, reject the null hypothesis. Conclude that the true mean wall thickness is greater than 4.0 mm at =
0.05.
With 95% confidence, the true mean wall thickness is at least 4.03 mm
23
Engineering Statistics, 5th edition September 15, 2010
4-62. a)
Normal Probability Plot
.999
.99
.95
Probability
.80
.50
.20
.05
.01
.001
The normality assumption appears to be reasonable. The data appear to fall along a straight line.
With 99% confidence, the true mean percent enrichment is between 2.904% and 3.092%. Because the interval contains
the value 2.95%, we conclude that the mean percent enrichment is not significantly different from 2.95%.
24
Engineering Statistics, 5th edition September 15, 2010
32.5 29.6
t0 = 30.851
0.47 / 25
7) Because 30.851 > 2.201, reject the null hypothesis and conclude the mean quantity of syrup is significantly different
from 29.6 ml at = 0.05
b) 2) H0: = 1.0
3) H1: > 1.0
x
4) t0 =
s/ n
5) Reject H0 if t > t/n-1 where t0.05,24 = 1.711
6) t0 = 30.851
7) Because 30.851 > 1.711, reject the null hypothesis. There is sufficient evidence to support the claim that the mean
quantity of syrup dispensed exceeds 29.6 ml at = 0.05.
c) Using the OC curve in Chart V (c), with d = 0.05/0.47 = 0.106 and n = 25 we get 0. Thus the power is
given by 1 = 1. The sample size is adequate for the experiment.
With 99% confidence, the true mean quantity is between 26.39 ml and 38.61 ml.
230.21 233.13
With 90% confidence, the true mean frequency is between 230.21 Hz and 233.13 Hz. There is strong evidence that the
mean natural frequency differs from 235.
b) The data on the normal probability plot fall along the straight line. Therefore, the assumption of normally
distributed data is reasonable.
25
Engineering Statistics, 5th edition September 15, 2010
99 ML Estimates
Mean 26.035
95
StDev 4.66361
90
80
70
Percent
60
50
40
30
20
10
5
20 30 40
Data
| 0 | | 27 25 |
c) d = 0.42
4.78
Using the OC curve, Chart V (d) for = 0.01, d = 0.42, and n = 20, we get 0.7 and power of 10.7 = 0.3.
| 0 | | 27.5 25 |
d) d = 0.52
4.78
Using the OC curve, Chart V (d) for = 0.01, d = 0.52, and 0.1 (Power = 0.9), n 50
Section 4-6
4-67.
a) In order to use 2 statistic in hypothesis testing and confidence interval construction, we need to assume that the
underlying distribution is normal.
1) The parameter of interest is the true standard deviation of the diameter, . However, the answer can be found by
performing a hypothesis test on 2.
2) H0: 2 = 0.0004
3) H1: 2 > 0.0004
( n 1)s2
4) Test statistic is 20 =
2
5) Reject H0 if 02 2 ,n 1 where 02.05,14 23.68
6) n = 15, s = 0.016
20 = ( n 1) s 14(0.032) 8.96
2 2
2
0.0016
26
Engineering Statistics, 5th edition September 15, 2010
7) P-value = P(2 > 8.96) for 14 degrees of freedom: 0.5 < P-value < 0.9. Because the P-value > 0.05, fail to reject
the null hypothesis.
With 95% confidence, the true variance of the hole diameter is greater than 0.0006 mm2. With 95% confidence, the
true standard deviation of the hole diameter is greater than 0.02 mm
c) Based on the lower confidence bound, we fail to reject the null hypothesis.
4-68.
a) In order to use 2 statistic in hypothesis testing and confidence interval construction, we need to assume that the
underlying distribution is normal.
1) The parameter of interest is the true variance of the sugar content, 2.
2) H0: 2 = 18
3) H1: 2 18
( n 1)s2
4) 20 =
2
5) Reject H0 if 20 12 / 2 ,n 1 where 0.975,9 2.70 or 02 2 / 2, n 1 where 0.025, 9 19.02
2
6) n = 10, s2 = 16
( n 1)s 2
9(16)
20 = 8
182
7) Because 2.70 < 8 < 19.02 fail to reject H0. There is not sufficient evidence to conclude that the true variance of
sugar content differs from 18 mg2 at = 0.05.
Take the square root of the endpoints of this interval to find the approximate confidence interval for :
2.75 7.30
With 95% confidence, the true standard deviation of the sugar content is between 2.75 mg and 7.30 mg.
d) Because the hypothesized value lies within this confidence interval, we fail to reject the null hypothesis.
4-69.
a) In order to use 2 statistic in hypothesis testing and confidence interval construction, we need to assume that the
underlying distribution is normal.
1) The parameter of interest is the standard deviation of tire life, . However, the answer can be found from a
hypothesis test on 2.
27
Engineering Statistics, 5th edition September 15, 2010
2) H0: 2 = 30002
3) H1: 2 > 30002
( n 1)s2
4) 20 =
2
5) Reject H0 if 20 2 ,n 1 where 02.05,9 16.919
6) n = 10, s2 = (3035)2
( n 1) s 2 9(3035) 2
20 = 9.2112
2 3000 2
7) Because 9.2112 < 16.919 fail to reject H0. There is not sufficient evidence to conclude that the true standard
deviation of tire life exceeds 3000 km at = 0.05.
b) P-value = P(2 > 9.2112) for 9 degrees of freedom, 0.10 < P-value < 0.50.
9(3035) 2
2
16.919
4,899,877.36 < 2
With 95% confidence, the true variance of tire life is greater than 4,899,877.36 km2. With 95% confidence, the true
standard deviation of tire life is greater than 2213.57 km.
d) Because the hypothesized value falls within this interval, we cannot reject the null hypothesis.
4-70.
a) In order to use 2 statistic in hypothesis testing and confidence interval construction we need to assume that the
underlying distribution is normal.
1) The parameter of interest is the true standard deviation of Izod strength, . However, the answer can be found from
a hypothesis test on 2.
2) H0: 2 = 0.2
3) H1: 2 0.2
( n 1)s2
4) 20 =
2
5) Reject H0 if 20 12 / 2 ,n 1 where 20.995,19 6.84 or 20 2 ,2 ,n 1 where 20.005,19 38.58
6) n = 20, s = 0.328
(n 1) s 2 19(0.328) 2
20 = 20.441
2 0.1
7) Because 6.84 < 20.441 < 38.58 we fail to reject H0. There is not sufficient evidence to conclude that the true
variance of Izod strength is differs from 0.20 m kg/cm at = 0.01.
b) P-value = 2P(2 > 20.441) for 19 degrees of freedom: 0.20 < 2P(2 > 20.441) < 1
19(0.328) 2 19(0.328) 2
2
38.58 6.84
0.053 2 0.299
With 99% confidence, the true variance of Izod strength is between 0.053 (m-kg/cm)2 and 0.299
(m-kg/cm).
28
Engineering Statistics, 5th edition September 15, 2010
d) Because the hypothesized value falls within this confident interval, we fail to reject the null hypothesis.
4-71.
a) In order to use 2 statistic in hypothesis testing and confidence interval construction, we need to assume that the
underlying distribution is normal.
1) The parameter of interest is the true standard deviation of titanium percentage, . However, the answer can be
found from a hypothesis test on 2.
2) H0: 2 = (0.35)2
3) H1: 2 (0.35)2
( n 1)s2
4) 20 =
2
5) Reject H0 if 20 12 / 2 ,n 1 where 02.975,50 32.36 or 02 2 ,2,n 1 where 02.025,50 71.42
6) n = 51, s = 0.37
(n 1) s 2 50(0.37) 2
20 = 55.88
2 (0.35) 2
7) Because 32.36 < 55.88 < 71.42 we fail to reject H0. There is not sufficient evidence to conclude that the true
standard deviation of titanium percentage differs from 0.35 at = 0.01.
b) 2P(2 > 55.88) for 50 degrees of freedom: 0.20 < 2P(2 > 55.88) < 1
With 95% confidence, the true standard deviation of titanium percentage is between 0.31 and 0.46.
d) Because the hypothesized value falls within this confidence interval, we fail to reject the null hypothesis.
Section 4-7
4-72.
a) It is a two-sided test
b) Yes, this test was conducted using the normal approximation to the binomial.
Yes, the normal approximation to the binomial is appropriate because p = 0.3 is not extremely close to 0 or 1 and the
sample size = 250 is relatively large. Also np = 250(0.3) = 75 > 5 and n(1- p) = 250(1 - 0.3) = 175 > 5.
c) Yes, the null hypothesis can be rejected at the 0.05 level because the p-value = 0.006 < 0.05.
d) No, we fail to reject the null hypothesis H 0 : p 0.4 at the 0.05 level because 0.4 is included in 95% CI. No
additional calculations are necessary.
29
Engineering Statistics, 5th edition September 15, 2010
x 95
e) Approximate 90%CI for p: pˆ 0.38
n 250
pˆ (1 pˆ ) pˆ (1 pˆ )
pˆ z 0.05 p pˆ z 0.05
n n
0.38(1 0.38) 0.38(1 0.38)
0.38 1.64 p 0.38 1.64
250 250
0.38 0.0503 p 0.38 0.0503
0.3297 p 0.4303
b) Yes, this test was conducted using the normal approximation to the binomial.
Yes, the normal approximation to the binomial is appropriate because p = 0.65 is not extremely close to 0 or 1 and the
sample size = 800 is relatively large. Also np = 800(0.65) = 520 > 5 and n(1-p) = 800(1-0.65) =280 >5.
x 553
c) Sample p : pˆ 0.69125
n 800
pˆ (1 pˆ )
95% Lower bound : pˆ z 0.05 p
n
0.69125(1 0.69125)
0.69125 1.64 p
800
0.69125 0.0268 p
0.66445 p
P value 2[1 (| Z |)] 2[1 ( | 2.45 |)] 2[1 0.992857] 0.014286
95%
Lower
Sample X N Sample p Bound Z-Value P-Value
1 553 800 0.69125 0.66445 2.45 0.014286
4-74. a) 1) The parameter of interest is the true proportion of deaths from lung cancer, p.
2) H0: p = 0.85
3) H1: p 0.85
x np0 p p0
4) z0 or z0 ; Either approach will yield the same conclusion
np0 1 p0 p0 1 p0
n
5) Reject H0 if z0 < z/2 where z/2 = z0.025 = 1.96 or z0 > z/2 ; where z/2 = z0.025 = 1.96
6) x = 1823 n = 2000 p̂
1823
0.912
2000
x np 0 1823 2000(0.85)
z0 7.7
np 0 1 p0 2000(0.85)(0.15)
7) Because 7.7 > –1.96, fail to reject the null hypothesis and conclude the percentage of deaths resulting from lung
cancer is significantly 85%.
1823
b) p̂ 0.912 n = 2000 For = 0.05, z/2 = z0.025 = 1.96
2000
0.912 1.96 0.912(0.088) p 0.912 + 1.96 0.912(0.088)
2000 2000
0.900 p 0.924
30
Engineering Statistics, 5th edition September 15, 2010
With 95% confidence, the true proportion of deaths resulting from cancer is between 0.900 and 0.924.
c) E = 0.03, = 0.05, z/2 = z0.025 = 1.96 and p = 0.823 as the initial estimate of p,
2
z 1.96
n α / 2 p(1
ˆ p)
ˆ 0.912(1 0.912) 342.57,
E 0.03
n 343.
11
4-75. n = 30, x = 11, pˆ 0.367 , = 0.10, z = z0.10 = 1.28
30
a) 1) The parameter of interest is the true proportion of rollovers, p.
2) H0: p = 0.25
3) H1: p > 0.25
x np0 p p0
4) z0 or z0 ; Either approach yields the same conclusion
np0 1 p0 p 0 1 p0
n
5) Reject H0 if z0 > z where z = z0.10 = 1.28
11
6) x = 11 n = 30 pˆ 0.367
30
pˆ p 0 0.367 0.25
z0 1.48
p 0 1 p 0 (0.25)(0.75)
n 30
7) Because 1.48 > 1.28, reject the null hypothesis and conclude the percentage of rollovers exceeds 25%
2 2
z p 0 (1 p 0 ) z p(1 p ) 1.28 0.25(1 0.25) 1.28 0.35(1 0.35)
c) n
p p0 0.35 0.25
n = 135.67, n 136
e) Because the hypothesized value does not fall within this interval, we can reject the null hypothesis.
2 2
z 1.645
f) n pˆ (1 pˆ ) 0.367(1 0.367) 1571.598 , n 1572
E 0.02
31
Engineering Statistics, 5th edition September 15, 2010
6) x = 22 n = 60 p̂ 22 0.36
60
p̂ p0
0.36 0.30
z0 1.014
p 0 1 p 0 (0.30)(0.70)
n 60
7) Because -1.96 < 1.014 < -1.96, fail to reject the null hypothesis. The percentage of helmets damaged is not
significantly different from 30%.
With 95% confidence, the true proportion of helmets damaged lies within 0.23 and 0.49.
2 2
z 1.96
d) n / 2 pˆ (1 pˆ ) 0.36(1 0.36) 2212.76 , n 2213
E 0.02
2 2
e) n z α / 2 p(1 p) 1.96 0.5(1 0.5) 2401 , n 2401
E 0.02
4-77. 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:
2 2
z 2.58
n α / 2 p(1 p) 0.5(1 0.5) 665.64 , n 666
E 0.05
20
4-78. p̂ 0.02 , n = 1000,
1000
a) 1) The parameter of interest is the true fraction defective, p.
2) H0: p = 0.01
3) H1: p > 0.01
x np0 p p0
4) z0 or z0 ; Either approach will yield the same conclusion
np0 1 p0 p0 1 p0
n
5) Reject H0 if z0 > z where z = z0.05 = 1.645
6) x = 20 n = 1000 p̂ 20 0.02
1000
p̂ p 0
0.02 0.01
z0 3.178
p 0 1 p 0 (0.01)(0.99)
n 1000
7) Because 3.178 > 1.645, reject the null hypothesis. There sufficient evidence to conclude that proportion of
defective calculators does not exceed 0.01.
32
Engineering Statistics, 5th edition September 15, 2010
2 2
z 1.96
n / 2 p(1 p) 0.5(1 0.5) 2401
E 0.02
4-80. a) 1) The parameter of interest is the true fraction of defective integrated circuits, p.
2) H0: p = 0.04
3) H1: p 0.04
x np0 p p0
4) z0 or z0 ; Either approach will yield the same conclusion
np0 1 p0 p0 1 p0
n
5) Reject H0 if z0 < z/2 where z/2 = z0.025 = 1.96 or z0 > z/2 where z/2 = z0.025 = 1.96
6) x = 22 n = 400 p̂ 22 0.06
400
x np 0 22 400(0.04)
z0 1.53
np 0 1 p 0 400(0.04)(0.96)
7) Because 1.96 < 1.53 < 1.96, fail to reject the null hypothesis. There is not sufficient evidence to conclude that the
true fraction of defective integrated circuits differs from 0.04 at = 0.05.
b) P-value = 2(1 - (1.53)) = 0.126. Because the P-value > 0.05, we fail to reject the null hypothesis.
0.04 0.05 1.96 0.04(0.96) 0.04 0.05 1.96 0.04(0.96)
300 300
4-81. a)
0. 05( 0. 95) 0. 05( 0. 95)
300 300
= (0.97) - (-2.56)
= 0.8340 – 0.0052
= 0.8288
2
1.96 0.04(0.96) 1.28 0.05(0.95)
b) n
0. 05 0.04
n = 4396.3, n = 4397
4-82. a) 1) The parameter of interest is the true proportion of engineers who continue their education, p.
2) H0: p = 0.50
3) H1: p 0.50
x np0 p p0
4) z0 or z0 ; Either approach will yield the same conclusion
np0 1 p0 p 0 1 p0
n
5) Reject H0 if z0 < z/2 where z/2 = z0.025 = 1.96 or z0 > z/2 where z/2 = z0.025 = 1.96
117
6) x = 117 n = 484 p 0.242
484
33
Engineering Statistics, 5th edition September 15, 2010
pˆ p0 0.242 0.50
z0 11.352
p0 1 p0 0.5(1 0.5)
n 484
7) Because 11.352 < 1.96, reject the null hypothesis. The data from “Engineering Horizons” yield results
significantly different from the claim reported by “Fortune”, at = 0.05.
4-83. a) 1) The parameter of interest is the true percentage of polished lenses that contain surface defects, p.
2) H0: p = 0.05
3) H1: p < 0.05
x np0 p p0
4) z0 or z0 ; Either approach will yield the same conclusion
np0 1 p0 p 0 1 p0
n
5) Reject H0 if z0 < z where z = z0.05 = 1.645
6) x = 13 n = 400 pˆ
13
0.0325
400
pˆ p0 0.0325 0.05
z0 1.61
p0 1 p0 0.05(1 0.05)
n 400
7) Because –1.61 > 1.645 fail to reject the null hypothesis. The machine cannot be qualified at the 0.05 level of
significance.
0.05(0.95)
c) 0.05 0.02 1.65 400
β 1 1 (1.72) 0.0427
0.02(0.98)
400
2
d) n 1.65 0.05(0.95) 1.65 0.02(0.98) 387.6,
n 388
0.02 0.05
4-84. a) 1) The parameter of interest is the true percentage of football helmets that have flaws, p.
2) H0: p = 0.10
3) H1: p > 0.10
x np0 p p0
4) z0 or z0 ; Either approach will yield the same conclusion
np0 1 p0 p0 1 p0
n
5) Reject H0 if z0 > z where z = z0.01 = 2.33
24
6) x = 24 n = 200 p̂ 0.12
200
x np 0 24 200(0.10)
z0 0.943
np 0 1 p 0 200(0.10)(0.90)
7) Because 0.943 < 2.33 fail to reject the null hypothesis. There is insufficient evidence to support the claim that at least
10% of all football helmets have manufacturing flaws that could potentially cause injury to the wearer, at = 0.01.
34
Engineering Statistics, 5th edition September 15, 2010
4-85. The problem statement implies H0: p = 0.6, H1: p > 0.6 and defines an acceptance region as p̂ 385 0.64 and
600
rejection region as p̂ 0.64
a) = P Pˆ 0.64 when p 0.6 P Z 0.64 0.6 P Z 2 1 P(Z 2) 0.02275.
0.6(0.4)
600
b) = P( P 0.64 when p = 0.75) = P(Z –6.2225) = 0.
4-86. 1) The parameter of interest is the true percentage of batteries that fail during the warranty period, p.
2) H0 : p = 0.002
3) H1 : p < 0.002
x np0 p p0
4) z0 or z0 ; Either approach yields the same conclusion
np0 1 p0 p 0 1 p0
n
5) Reject H0 if z0 < z where z = z0.01 = 2.33
3
6) x = 3 n = 2000 pˆ 0.0015
2000
x np0 3 2000(0.0015)
z0 0.5005
np0 (1 p0 ) 2000(0.0015)(0.9985)
7) Because -0.5005 > 2.33 fail to reject the null hypothesis. There is insufficient evidence to support the claim that less
than 0.2 percent of the company’s batteries will fail during the warranty period with proper charging procedures at =
0.01.
(a) pˆ
25 pˆ (1 pˆ ) 0.6757(0.3243)
4-87. 0.6757 , the CI is pˆ Z / 2 , 0.6757 1.96 , 0.6757 0.1508
37 n 37
(0.5249, 0.8265)
4-88.
823
pˆ 0.823
1000
z2 pˆ (1 pˆ ) z2 / 2 1.96 2 0.823(1 0.823) 1.96 2
pˆ / 2 z / 2 2 0.823 1.96
2n n 4n 2(1000) 1000 4(1000) 2
LCL 2
0.7981
z 1.96 2
1 /2 1
n 1000
2
z / 2 pˆ (1 pˆ ) z / 2
2 1.96 2 0.823(1 0.823) 1.96 2
pˆ z / 2 2 0.823 1.96
2n n 4n 2(1000) 1000 4(1000) 2
UCL 2
0.8454
z 1.96 2
1 /2 1
n 1000
The Agresti-Coull interval is similar to the traditional interval.
35
Engineering Statistics, 5th edition September 15, 2010
4-89.
18
pˆ 0.36
50
2
pˆ (1 pˆ ) z2 / 2 0.36 1.96 1.96 0.36(1 0.36) 1.96
2
z2
pˆ / 2 z / 2 2
2n n 4n 2(50) 50 4(50) 2
LCL 0.2414
z2 / 2 1.96 2
1 1
n 50
2
pˆ (1 pˆ ) z2 / 2 0.36 1.96 1.96 0.36(1 0.36) 1.96
2
z2
pˆ / 2 z / 2 2
2n n 4n 2(50) 50 4(50) 2
UCL 0.4986
z2 / 2 1.96 2
1 1
n 50
The Agresti-Coull interval is similar to the traditional interval.
4-90.
25
pˆ 0.6757
37
2
pˆ (1 pˆ ) z2 / 2 0.675 1.96 1.96 0.6757(1 0.6757) 1.96
2
z2 / 2
pˆ z / 2 2
2n n 4n 2(37) 37 4(37) 2
LCL 0.5147
z2 1.96 2
1 /2 1
n 37
2
pˆ (1 pˆ ) z2 / 2 0.675 1.96 1.96 0.6757(1 0.6757) 1.96
2
z2 / 2
pˆ z / 2 2
2n n 4n 2(37) 37 4(37) 2
UCL 2 2
0.8037
z / 2 1.96
1 1
n 37
The Agresti-Coull interval is shifted to slightly greater proportions than the traditional interval.
Section 4-8
1 1
4-91. a) x t / 2,n 1 s 1 X n1 x t / 2,n1 s 1
n n
1 1
61492 2.262(3035) 1 X n 1 61492 2.262(3035) 1
10 10
(54291.75, 68692.25)
b) x ks, x ks , k = 2.839
61492 – 3035(2.839), 61492 + 3035(2.839)
(52875.64, 70108.37)
4-92. a) A 90% PI is
1 1
x t 0.1 / 2, 201 s 1 = 1.121 (1.729)(0.328) 1 = (0.5399, 1.7021)
n 20
b) x ks, x ks , k = 2.752
1.121 ± (2.752)(0.328) = (0.2183, 2.0237)
36
Engineering Statistics, 5th edition September 15, 2010
1 1
4-93. a) x t / 2,n 1 s 1 X n1 x t / 2,n1 s 1
n n
1 1
5625.1 2.977(226.1) 1 X n 1 5625.1 2.977(226.1) 1
15 15
(4929.93, 6320.27)
b) x ks, x ks , k = 3.562
5625.1 – 226.1(3.562), 5625.1 + 226.1(3.562)
(4819.73, 6430.47)
4-94. a) A 95% PI is
1 1
x t 0.05 / 2,61 s 1 = 16.9833 ( 2.571)(0.3189) 1 = (16.0977, 17.8689)
n 6
b) x ks, x ks , k = 6.345
16.9833 ± (6.345)(0.3189) = (14.9599, 19.0067)
1 1
4-95. a) x t / 2,n 1 s 1 X n1 x t / 2,n1 s 1
n n
1 1
9.117 3.106(0.464) 1 X n 1 9.117 3.106(0.464) 1
12 12
(7.617, 10.617)
b) x ks, x ks , k = 5.079
9.117 – 0.464(5.079), 9.117 + 0.464(5.079)
(6.760, 11.474)
4-96. a) A 90% PI is
1 1
x t 0.05 / 2,121 s 1 = 8.2767 (1.796)(0.0836) 1 = (8.1204, 8.4330)
n 12
b) x ks, x ks , k = 2.404
8.2767 ± (2.404)(0.0836) = (8.0757, 8.4777)
Section 4-10
4-97.
Value 0 1 2 3 4
Observed Frequency 24 30 31 11 4
Expected Frequency 30.12 36.14 6.72 8.67 2.6
37
Engineering Statistics, 5th edition September 15, 2010
02
24 30.12 2 4 2.6 2 91.39
30.12 2.6
7) Because 91.3926 > 9.49 reject H0. There is sufficient evidence to reject the null hypothesis that the distribution of X
is Poisson.
Value 1 2 3 4 5 6 7 8
Observed Frequency 1 11 8 13 11 12 10 9
Expected Frequency 2.721 6.677 10.921 13.398 13.149 10.753 7.538 4.624
Because the first category has an expected frequency less than 3, combine it with the next category:
Value 1-2 3 4 5 6 7 8
Observed Frequency 12 8 13 11 12 10 9
Expected Frequency 9.398 10.921 13.398 13.149 10.753 7.538 4.624
a) 1) The variable of interest is the form of the distribution for the number of flaws.
2) H0: The form of the distribution is Poisson
3) H1: The form of the distribution is not Poisson
4) The test statistic is
k O i E i 2
20 Ei
i 1
0
9.3984 4.6237
7) Because 6.955 < 15.09 fail to reject H0. There is not sufficient evidence to reject the null hypothesis that the
distribution of the number of flaws is Poisson.
Value 5 7 8 9 10 11 12 13 14 15
Rel. Freq 0.133 0.133 0.133 0.167 0.033 0.100 0.100 0.033 0.133 0.033
Observed
4 4 4 5 1 3 3 1 4 1
(Days)
Expected
1.381 3.029 3.635 3.878 3.723 3.249 2.599 1.919 1.316 0.842
(Days)
Because there are several cells with expected frequencies less than 3, the revised table is:
38
Engineering Statistics, 5th edition September 15, 2010
a) 1) The variable of interest is the form of the distribution for the number of calls arriving to a switchboard from noon
to 1pm during business days.
2) H0: The form of the distribution is Poisson
3) H1: The form of the distribution is not Poisson
4) The test statistic is
k O i E i 2
20 Ei
i 1
6) 02 5.79
7) Because 5.79 < 9.49 fail to reject H0. There is not sufficient evidence to reject the null hypothesis that the distribution
for the number of calls is Poisson.
a) 1) The variable of interest is the form of the distribution for the number of cars passing through the intersection.
2) H0: The form of the distribution is Poisson
3) H1: The form of the distribution is not Poisson
4) The test statistic is
k O i E i 2
20 Ei
i 1
Value 0 1 2 3 4
Observed 4 21 10 13 2
Expected 8.899 17.798 14.832 6.592 1.648
The expected frequency for value 4 is less than 3. Combine this cell with value 3:
Value 0 1 2 3-4
Observed 4 21 10 15
Expected 8.899 17.798 14.832 8.24
a) 1) The variable of interest is the form of the distribution for the random variable X.
2) H0: The form of the distribution is binomial with n = 6 and p = 0.25
3) H1: The form of the distribution is not binomial with n = 6 and p = 0.25
4) The test statistic is
k O i E i 2
20 Ei
i 1
39
Engineering Statistics, 5th edition September 15, 2010
02
4 8.8992
15 8.242 10.39
8.899 8.24
7) Because 10.39 > 7.81 reject H0. We conclude that the distribution is not binomial with n = 6 and p = 0.25 at =
0.05.
4-102. The value of p must be estimated. Let the estimate be denoted by p sample
sample mean 1.2
sample mean = [0(21) + 1(30) + 2(22) + 3(6) + 4(1)]/80 = 1.2, p̂ sample 0.10
n 12
Value 0 1 2 3 4
Observed 21 30 22 6 1
Expected 22.59436 30.12582 18.41022 6.818601 1.70465
Because value 4 has an expected frequency less than 3, combine this category with that of value 3:
Value 0 1 2 3-4
Observed 21 30 22 7
Expected 22.59436 30.12582 18.41022 8.523251
a) 1) The variable of interest is the form of the distribution for the number of underfilled cartons, X.
2) H0: The form of the distribution is binomial
3) H1: The form of the distribution is not binomial
4) The test statistic is
k O i E i 2
20 Ei
i 1
2
5) Reject H0 if o 02.1, 2 4.61
2
6) 0 = 1.2026
7) Because 1.2026 < 4.61 fail to reject H0. There is not sufficient evidence to reject the null hypothesis that the
distribution of the number of underfilled cartons is binomial at = 0.10.
Supplemental Exercises
185
P Z = 1
20 / 10
178 P(Z 1.11) = 0.8665 0.1335
181 P(Z 0.63) = 0.7357 0.2643
184 P(Z 0.16) = 0.5636 0.4364
187 P(Z 0.32) = 0.3745 0.6255
40
Engineering Statistics, 5th edition September 15, 2010
0.9
0.8
0.7
0.6
0.5
Beta
0.4
0.3
0.2
0.1
0.0
Mean
4-104.
1.0
0.9
0.8
0.7
Power
0.6
0.5
0.4
0.3
0.2
0.1
180 190 200
Mean
4-105. Symmetric confidence interval: x - 1.96 x 196
. , because z0.025 = 1.96
n n
The length of this interval is 2 196
. 3.92
n n
Asymmetric confidence interval: x 2.33 x 176
. , because z0.01 = 2.325 and z0.04 = 1.75
n n
The length of this interval is (2.33 1.75) 4.08
n n
The symmetric confidence interval is the narrower of the two. An advantage of a symmetric confidence interval is that
in general it is narrower than an asymmetric confidence interval.
41
Engineering Statistics, 5th edition September 15, 2010
4-107. = 60 2 5
Find P s2 7.44 and P s2 2.56
a) n = 20
(n 1) s 2 19(7.44)
P
2
5
P 19 28.27 0.10
2
(n 1)s 2 19(2.56)
P
2
5
2
2
P χ 19 9.73 1 P χ 19 9.73 ; 2
0.05 P χ19 9.73 0.10
b) n = 40
(n 1)s 2 39(7.44)
P
σ 2
5
2
P χ 39 58.03 ; 2
0.025 P χ 39 58.03 0.05
(n 1)s 2 39(2.56)
P
σ2
5
2
2
P χ 39 19.97 1 P χ 39 19.97 ; 2
0.01 P χ 39 19.97 0.025
c) n = 75
(n 1)s 2 75(7.44)
P
σ 2
5
2
P χ 75 111.6 ; 0.005 P χ 75 111.6 0.005
2
(n 1)s 2 75(2.56)
P
σ2
5
2
2
P χ 75 38.4 1 P χ 75 38.4 ;
2
P 75 38.4 0.005
d) As the sample size increases with all other values held constant, the probability P s2 7.44 decreases because the
2
right tail of the distribution becomes relatively shorter.
e) As the sample size increases with all other values held constant, the probability P s2 2.56 decreases because the
left tail of the 2 distribution becomes relatively shorter.
4-108.
a) The data appear to follow a normal distribution based on the normal probability plot because the data fall along a
straight line.
b) It is important to check for normality of the distribution underlying the sample data because the confidence intervals
are based on the assumption of normality (especially for smaller sample sizes were the central limit theorem does
not apply).
c) No, it is not reasonable to infer that the true mean is 14. 5 because 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 because the 95% confidence interval on the variance
contains this value.
42
Engineering Statistics, 5th edition September 15, 2010
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.
g) A 95% PI for is
1 1
x t 0.05 / 2, 201 s 1 = 15.33 ( 2.093)(0.6182) 1 = (14.0042, 16.6558)
n 20
h) x ks, x ks , k = 2.31
15.33 ± (2.31)(0.6182) = (13.9020, 16.7580)
4-109.
a) Based on the normal probability plot the data appear to follow a normal distribution because the data fall along a
straight line.
b) x = 25.12 s = 8.42 n = 9; Use the t-distribution to construct the confidence intervals; = 1 0.99
t , n 1 t 0.01,8 2.896
s
x t ,n 1
n
8.42
25.12 2.896
9
16.99
With 99% confidence, the true mean compressive strength is at least 16.99 Mpa.
c) t / 2 , n 1 t 0.01,8 2.896
s s
x t , n 1 x t , n 1
n
n
8.42 8.42
25.12 2.896 25.12 2.896
9
9
16.99 33.25
With 98% confidence, the true mean compressive strength is between 16.99 and 33.25 Mpa.
The lower endpoint of the one-sided confidence interval is the same as that of the two-sided confidence
interval due to the level of confidence used. In both cases the probability in the left tail is 0.01.
With 99% confidence, the true variance of compressive strength is at most 343.76 (Mpa)2.
43
Engineering Statistics, 5th edition September 15, 2010
(n 1) s 2 (n 1) s 2
2
/ 2 , 8
2
12 ( / 2 ),8
8(8.42) 2 8(8.42) 2
2
20.09 1.65
28.23 2 343.74
With 98% confidence, the true variance of compressive strength is between 28.23 and 343.74 (Mpa)2. The upper
endpoint of the one-sided confidence interval is the same as that of the two-sided confidence interval due to the level of
confidence. In both cases the probability in the right tail is 0.01.
With 98% confidence, the true variance of compressive strength is between 15.86 and 193.12 (Mpa)2.
Comparison of intervals:
Confidence interval on : For the new confidence interval the center is at a lower value and the width of the interval
has decreased.
Confidence interval on 2: The new confidence interval includes lower values and is much narrower than the original
interval.
Effects: We see by correcting the value, the mean and variance have decreased and the length of the intervals have
narrowed. In particular, the sample standard deviation decreased by about 25% causing the confidence intervals to
narrow substantially.
Confidence interval on : There is very little difference when the data value is changed only slightly.
Confidence interval on 2: The confidence interval has changed very little with a slight change in one data value.
Effects: The sample mean, sample variance, and confidence intervals have changed very little.
44
Engineering Statistics, 5th edition September 15, 2010
h) There are two cases that the above exercises illustrate: 1) when a value is changed and the new value lies far from
the sample mean; and 2) when a value is changed and the new value lies near the sample mean. For case 1, the
variance changed dramatically resulting in a narrower confidence interval. The mean also changes noticeably but the
effect is not as large as the change in variance. For case 2, the mean and variance have changed less than in case 1.
The widths of the confidence intervals for the original data, case 1, and case 2 are shown in the table below.
1 1
i) x t / 2,n1 s 1 X n 1 x t / 2,n 1 s 1
n n
1 1
25.12 3.355(8.42) 1 X n 1 25.12 3.355(8.42) 1
9 9
(-4.657, 54.897)
j) x ks, x ks , k = 4.550
25.12 – 8.42(4.550), 25.12 + 8.42(4.550)
(-13.191, 63.431)
4-110. With = 10, the 95% confidence interval on the mean has length of at most 6; the error is then E = 3.1.
2 2
z 1.96
a) n 0.025 102 100 39.98 = 40
3.1 3.1
2 2
z 1.96
b) n 0.025 82 64 25.58
3.1
= 26
3.1
As the standard deviation decreases, with all other values held constant, the sample size necessary to maintain the
confidence level and the length of the interval decreases.
c) We would want to have a relatively large sample size, n 30. With a sample size of at least 30, the central limit
theorem applies so that one can conclude that the sample mean is approximately normally distributed.
p (1 p )
4-111. Sample Mean = p Sample Variance =
n
Sample Size, n Approximate Sample Mean Sample Variance
Sampling Distribution
a) 60 Normal p p̂(1 p̂)
60
b) 70 Normal p p̂(1 p̂)
70
c) 100 Normal p p̂(1 p̂)
100
d) As the sample size increases, the variance of the sampling distribution decreases.
p (1 p )
4-112. z/2 = 1.96 E z / 2
n
a) n = 70 p = 0.1
0.1(1 0.1)
E 1.96 0.0703
70
45
Engineering Statistics, 5th edition September 15, 2010
b) n = 80 p =0.1
0.1(1 0.1)
E 1.96 0.0657
80
c) n = 110 p = 0.1
0.1(1 0.1)
E 1.96 0.0561
110
d) As the sample size increases and all other values held constant, the error decreases.
p (1 p )
e) z/2 = 2.576 E z / 2
n
n = 70
0.1(1 0.1)
E 2.576 0.0924
70
n = 80
0.1(1 0.1)
E 2.576 0.0864
80
n = 110
0.1(1 0.1)
E 2.576 0.0737
110
As the sample size increases and all other values held constant, the error decreases.
f) When the confidence level is increased, the error in estimating the true value of p also increases. This can be seen by
comparing values between parts d) and e). We see that E = 0.0703 with a 95% level of confidence, while E = 0.0924
with a 99% level of confidence for the same sample size.
4-113. = 16, = 305 300 = 5, 0.025, z0.025 = 1.96, using eq. (4-24)
2
5 25 5 25
1.96 1.96 (0.40) (3.52) 0.6554 0.00025 0.655
16 16
a) n = 25:
5 60 5 60
b) n = 60: β 1.96 1.96 (0.46) (4.38) 0.323 0 0.323
16 16
5 100 5 100
β 1.96 1.96 (1.17) (5.09) 0.121 0 0.121
16 16
c) n = 100:
d) , which is the probability of a Type II error, decreases as the sample size increases because the variance of the
sample mean decreases. Consequently, the probability of observing a sample mean in the acceptance region centered at
200 ml/h decreases with larger n.
4-114. = 14, = 305 300 = 5, 0.025, z0.025 = 1.96, using eq. (4-24)
2
5 20 5 20
a) n = 20: 1.96 1.96 (0.36) (3.56) 0.6406 0.0002 0.6404
14 14
5 50 5 50
b) n = 50: 1.96 1.96 (0.565) (4.49) 0.2860 0 0.2860
14 14
5 100 5 100
c) n = 100: 1.96 1.96 (1.61) (5.53) 0.0537 0 0.054
14 14
d) The probability of a Type II error increases with an increase in the standard deviation.
46
Engineering Statistics, 5th edition September 15, 2010
4-115. a) 1) The parameter of interest is the true mean life of a heating element, .
2) H0: = 550
3) H1: > 550
x
4) t0 =
s/ n
5) Reject H0 if t > t,n-1 where t0.05,14 = 1.761
6) x = 598.14 s = 16.93 n = 15
598.14 550
t0 = 11.01
16.93 / 15
7) Because 11.01 > 1.761, reject H0. Conclude the true mean life of a heating element is greater than 550 hours.
b) P-value = P(t > 11.01): for degrees of freedom of 14 we obtain P-value < 0.0005
16.93
c) 598.14 1.761
15
590.44
Because the one-sided CI does not include the value 550, reject the null hypothesis. Conclude that the true mean life of
a heating element is greater than 550 hours.
14(16.93) 2 14(16.93) 2
d) 2
26.12 5.63
153.63 2 712.74
x 0 86 85
b) P-value = P( x 86) P P( z 0.0625 n )
/ n 16 / n
n = 25: P-value = P( z 0.0625 25 ) 0.3783
n = 100: P-value = P( z 0.0625 100 ) 0.2643
n = 400: P-value = P( z 0.0625 400 ) 0.1056
n = 2500: P-value = P( z 0.0625 2500 ) 0.0009 ; only sample that is statistically significant at = 0.01.
c) As the sample size increases, the probability of a error, Type I or Type II, decreases.
4-117.
a) Rejecting a null hypothesis provides strong evidence. Therefore the claim to be proved is stated in the alternative
hypothesis.
47
Engineering Statistics, 5th edition September 15, 2010
At = 0.05, reject the null hypothesis. Conclude that the mean weld strength is greater than 1034 at = 0.05.
n = 300: ( 154
. ) ( 55
. ) (1 0.93822) 0 0.06178
1 1 0.06178 0.938
The power increases as sample size increases, when all other values are held constant.
The power of the test is greater for the larger values of , because a larger value results in a smaller acceptance
region.
48
Engineering Statistics, 5th edition September 15, 2010
p0 p z / 2 p0 (1 p0 ) p0 p z / 2 p (1 p )
n n
p (1 p ) p (1 p )
n n
0.5 0.08 1.96 0.5(1 0.5) 0.5 0.08 1.96 0.08(1 0.08)
100 100
0 .08(1 0 . 08) 0 . 08(1 0 .08)
100 100
( 20.46) (13.52) 0
Power = 1 0 = 1.
As the difference between the hypothesized value of p and the true value of p increases, the power of the test increases.
p = 0.6:
2
2.575( 0.5( 0.5) ) 1645
. ( 0.6( 0.4) )
n = 438.2
0.5 0.6
n = 439
p = 0.8:
2
2.575( 0.5(0.5) ) 1.645( 0.8(0.2) )
n= 42.06
0.5 0. 8
n = 43
A smaller sample size can be used when the true proportion is further from the hypothesized value. It is easier to detect
the difference from the hypothesized value when the difference is greater.
6) n = 15, s = 226.1
14(226.1) 2
02 9.129
78400
7) P-value = P 2 9.129 ; at d.f.=14 0.10 < P-value < 0.50
The P-value is greater than any acceptable significance level, , therefore we fail to reject the null hypothesis. There is
insufficient evidence to support the claim that the standard deviation is less than 280 hours.
b) 7) n = 51, s = 226.1
50(226.1) 2
02 32.603
78400
P-value = P 2 32.603 ; at d.f.=50 0.025 < P-value < 0.05
The P-value is less than 0.05. Therefore we reject the null hypothesis and conclude that the standard deviation is
significantly less than 280 hours.
49
Engineering Statistics, 5th edition September 15, 2010
c) Increasing the sample size increases the test statistic 20 and therefore decreases the P-value, providing more
evidence against the null hypothesis.
7) Because 6.10 > 2.093, reject the null hypothesis. Conclude that the true mean sugar concentration differs from 11.5
at = 0.05.
From Table V the t0 value (in absolute value) is greater than the value corresponding to 0.0005 with 19 degrees of
freedom. Therefore 2(0.0005) = 0.001 > P-value
| 0 | | 11.4 11.5 |
b) d = 4.55
0.022
Using the OC curve, Chart V(a) for = 0.05, d = 4.54, and n 20 we find that 0 and the Power 1.
| 0 | | 11.45 11.5 |
c) d = 2.27
0.022
Using the OC curve, Chart V(a) for = 0.05, d = 2.27, and 1- > 0.9, (<0.1), we find that n should be at least 5.
e) The normality plot below indicates that the normality assumption is reasonable.
95
90
80
70
Percent
60
50
40
30
20
10
1
11.40 11.42 11.44 11.46 11.48 11.50 11.52
Sugar Concentration
50
Engineering Statistics, 5th edition September 15, 2010
2) H0 : = 300
3) H1 : 300
x
4) t 0
s/ n
5) Reject H0 if |t0| > t/2,n-1 where t/2,n-1 = 2.056
6) x 325.50 , s = 198.79 n=27
325.5 300
t0 0.667
198.79 / 27
7) Because 0.667< 2.056, we fail to reject the null hypothesis. There is not sufficient evidence to conclude that the
true mean body weight is different from 300 at = 0.05.
b) From table V the t0 value is found between the values of 0.25 and 0.4 with 26 degrees of freedom. Therefore, 0.5 <
P-value < 0.8. The smallest level of significance at which we reject the null hypothesis is the P-value.
4-122.
a) In order to use the 2 statistic in hypothesis testing and confidence interval construction, we need to assume that the
underlying distribution is normal.
1) The parameter of interest is the true variance of tissue assay, 2.
2) H0 : 2 = 0.6
3) H1 : 2 0.6
( n 1)s2
4) 20 =
2
5) Reject H0 if 20 12 / 2 ,n 1 where 02.995,11 2.60 or 02 2 / 2,n1 where 02.005,11 26.76
6) n = 12, s = 0.759
( n 1) s 2
11(0.759) 2
20 = 10.56
2
0. 6
7) Because 2.6 < 10.56 < 26.77 we fail to reject H0. There is not sufficient evidence to conclude that the true
variance of tissue assay is significantly different from 0.6 at = 0.01.
b) P-value: 0.1 < P-value/2 < 0.5 then 0.2 < P-value < 1
c) 99% confidence interval for , first find the confidence interval for 2
For = 0.05 and n = 12, 02.995,11 2.60 and 02.005,11 26.76
11(0.759) 2 11(0.759) 2
2
26.76 2.60
0.2367 2 2.437
Because 0.6 falls within the 99% confidence interval there is not sufficient evidence to conclude that the population
variance is different from 0.6
51
Engineering Statistics, 5th edition September 15, 2010
7) Because 2.48 > 1.943, reject the null hypothesis. Conclude that the leg strength exceeds 300 watts at = 0.05. The
0.01 < P-value < 0.025
| 0 | | 305 300 |
b) d = 0.3125
16
Using the OC curve, Chart V(c) for = 0.05, d = 0.3125, and n = 7, 0.9 and power = 10.9 = 0.1.
52
Engineering Statistics, 5th edition September 15, 2010
14 5 4.69
15 11 6.10
16 7 7.44
17 3 8.53
18 8 9.25
19 4 9.50
20 4 9.27
21 7 8.62
22 8 7.64
23 4 6.48
24 3 5.27
25 2 4.12
26 4 3.09
27 4 2.23
28 1 1.56
29 1 1.05
30 1 0.68
31 1 0.43
32 2 0.26
34 1 0.09
35 1 0.05
36 2 0.03
39 1 0.00
41 1 0.00
After combining categories with frequencies less than 3, we obtain the following table:
53
Engineering Statistics, 5th edition September 15, 2010
27-28-29-
30-31-32-
34-35-36-
39-41 16 6.38
a) 1) The variable of interest is the form of the distribution for the number of earthquakes per year of magnitude 7.0
and greater.
2) H0: The form of the distribution is Poisson
3) H1: The form of the distribution is not Poisson
4) The test statistic is
k O i E i 2
20 Ei
i 1
5) Reject H0 if o2 02.05,14 23.68
6)
13 4.86 16 6.38
2 2
02 48.94
4.86 6.38
7) Because 48.94 > 23.68 reject H0. Conclude that distribution of the number of earthquakes is not Poisson.
4-126. n=6
a) 1) The parameter of interest is the standard deviation, .
2) H0: 2 = 1.0
3) H1: 2 1.0
( n 1)s2
4) The test statistic is: 20
2
5) Reject H0 if 20 < 12 / 2 ,n 1 where 20.995,5 0.41 or 20 > /
2 2
2 , n1 where 0.005,5 = 16.75
6) n = 6, s = 0.319
5(0.319) 2
20 0.5088
10
.
7) Because 0.41 < 0.5088 < 16.75, fail to reject the null hypothesis. The variance of fatty acid for diet margarine is
not significantly different from 1.0 at = 0.05.
b) n = 51
2) H0 : 2 = 1.0
3) H1 : 2 1.0
( n 1)s2
4) The test statistic is: 20
2
5) Reject H0 if 20 < 12 / 2 ,n 1 where 20.995,50 27.99 or 20 > /
2 2
2 , n1 where 0.005,50 = 79.49
6) n = 51, s = 0.319
50(0.319) 2
20 5.09
10
.
7) Because 5.09 < 27.99, reject the null hypothesis. The true variance of fatty acid for diet margarine is different
from 1.0 at = 0.05.
c) The increased sample size increases the test statistic and also changes the degrees of freedom of the test statistic. The
result is a more sensitive test.
54
Engineering Statistics, 5th edition September 15, 2010
7) Because 1.75 > 1.24, fail to reject the null hypothesis. There is insufficient evidence to conclude the standard
deviation is at most 0.00002 mm.
b) Although the sample standard deviation is less than the hypothesized value of 0.00002, it is not significantly less
(when = 0.01) than 0.00002. The value of 0.00001 could have occurred as a result of sampling variation.
4-128.
Normal Probability Plot
.999
.99
.95
Probability
.80
.50
.20
.05
.01
.001
According to the normal probability plot, the assumption that the underlying distribution is normal is reasonable. The
data fall along a straight line. The assumption of normality should be satisfied in order to perform a hypothesis test
using a 2 test statistic.
P-value = P 2 0.000009 ; P-value < 0.005
55
Engineering Statistics, 5th edition September 15, 2010
7) A P-value less than 0.005 is significant evidence to reject the null hypothesis and conclude the standard deviation is less
than 4 g/l.
7) Because 1.27 > 2.33, fail to reject the null hypothesis. There is insufficient evidence to support the claim that
the true proportion is less than 1%.
P-value = P( Z 1.27) 0.10204
b) Although the sample proportion is less than the hypothesized value of 0.01, it is not significantly less than 0.0. The
value of 0.006 could have occurred as a result of sampling variation.
8
4-130. p 0.005
1600
a) 99% confidence interval: z / 2 2.575
pˆ (1 pˆ ) pˆ (1 pˆ )
pˆ z / 2 p pˆ z / 2
n n
0.00046 p 0.0095
With 99% confidence, the true proportion of aircraft that have wiring errors lies between 0.00046 and 0.0095.
2 2
z 2.575
b) n / 2 p (1 p ) (0.005)(1 0.005) 515.427
E 0.008
n = 516
c) Use p 0.5
2 2
z 2.575
n / 2 pˆ (1 pˆ ) (0.5)(1 0.5) 25,900.88
E 0.008
n 25,901
d) Preliminary information can significantly decrease the sample size required. Without this information, we need to
use p = 0.5 in the computations resulting in the worst case sample size.
4-131. 86
p̂ 0.57
150
a) 90% confidence interval; z / 2 1.645
pˆ (1 pˆ ) pˆ (1 pˆ )
pˆ z / 2 p pˆ z / 2
n n
0.504 p 0.636
With 90% confidence, the true proportion of graduates completing the test within 40 minutes lies between 0.504 and
0.636.
56
Engineering Statistics, 5th edition September 15, 2010
pˆ (1 pˆ ) pˆ (1 pˆ )
pˆ z / 2 p pˆ z / 2
n n
0.491 p 0.649
With 95% confidence, the true proportion of graduates completing the test within 40 minutes lies between 0.491 and
0.649.
The 95% confidence interval is wider than the 90% confidence interval. Higher confidence results in wider intervals,
assuming all other values are held constant.
d) Yes, because neither interval contains the value 0.80, a test with = 0.1 or = 0.05 would reject the null hypothesis that
the proportion equals 0.80.
4-133. X ~ bin(10, 0.3) Implicitly, H0: p = 0.3 and H1: p < 0.3
n = 20
Reject region: p 01
.
57
Engineering Statistics, 5th edition September 15, 2010
. 0.3
01
a) When p =0.3 = P p 01. P Z
0.3(0.7)
20
P( Z 1.95)
0.0256
0.1 0.2
b) When p = 0.2 =1 - Pp̂ 0.1 1 P Z
0.2(0.8)
20
1 P( Z 1.12)
P( Z 1.12)
0.1314
4-134. Create a table for the number of nonconforming coil springs (value) and the observed number of times the
number appeared. One possible table is:
Value 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Obs 0 0 0 0 3 0 2 4 6 9 4 3 1 2 0 1 2 2 0 1
58
Engineering Statistics, 5th edition September 15, 2010
Because several of the expected values are less than 3, some cells are combined resulting in the following table:
a) 1) The variable of interest is the form of the distribution for the number of nonconforming coil springs.
2) H0: The form of the distribution is binomial
3) H1: The form of the distribution is not binomial
4) The test statistic is
k O i E i 2
20 Ei
i 1
5) Reject H0 if 2
0
2
0.05 , 5 11.07
6)
02 3.017
7) Because 3.017 < 11.07 we fail to reject H0. There is not sufficient evidence to conclude that the distribution of
nonconforming springs differs from binomial at = 0.05.
4-135. Create a table for the number of errors in a string of 1000 bits (value) and the observed number of times the number
appeared. One possible table is:
Value 0 1 2 3 4
Obs 3 7 4 5 1
Value 0 1 2 3 4 5
Observed 3 7 4 5 1 0
Expected 3.64839 6.21282 5.28460 2.99371 1.27067 0.43103
Because several of the expected values are less than 3, some cells are combined resulting in the following table:
Value 0 1 2 3
Observed 3 7 4 6
Expected 3.64839 6.21282 5.28460 4.69541
59
Engineering Statistics, 5th edition September 15, 2010
a) 1) The variable of interest is the form of the distribution for the number of errors in a string of 1000 bits.
2) H0: The form of the distribution is binomial
3) H1: The form of the distribution is not binomial
4) The test statistic is
k O i E i 2
20 Ei
i 1
20
3 3.64839 2 6 4.69541 2
0.88971
3.64839 4.69541
7) Because 0.88971 < 9.49 fail to reject H0. There is not sufficient evidence that the distribution of the number of
errors differs from binomial at = 0.05.
s 1.8273
4-137. a) SE Mean
0.456825
n 16
x 45.8971 44.5
T 3.058
s/ n 1.8273 / 16
One-Sample T: X
95%
Lower
Variable N Mean StDev SE Mean Bound T P
X 16 45.8971 1.8273 0.4568 45.0962 3.058 0.004
b) For any level of significance α > 0.004, the null hypothesis is rejected.
c) P-value would be smaller because the difference between the hypothesized value and the sample mean would be
greater.
d) P-value = 2P(t > 3.058) and for degrees of freedom of 15 we obtain
2(0.0025) < P-value < 2(0.005) = (0.005 < P-value < 0.01)
Thus, at the 0.05 level, the null hypothesis H 0 : 44.5 versus H 0 : 44.5 is rejected because the P-value < 0.05.
4-138.
95%
Upper
Sample X N Sample p Bound Z-Value P-Value
1 146 850 0.171765 0.193044 -2.06 0.020
a) It is a one-sided test.
60
Engineering Statistics, 5th edition September 15, 2010
b) Yes, this test was conducted using the normal approximation to the binomial.
Yes, the normal approximation to the binomial is appropriate because p = 0.2 is not extremely close to 0 or 1and the
sample size = 850 is relatively large. Also np = 850(0.2) = 170 > 5 and n(1-p) = 850(1-0.2) =680 >5.
c) The null hypothesis can be rejected at the 0.05 level because the P-value = 0.020 < 0.05, but it cannot be rejected at
the 0.01 level because the P-value = 0.020 > 0.01.
d) P value 2 1 z0 2 1 2.06 2[1 0.980301] 0.039398 0.04 or, more simply, from the
computer output the P-value for the two-sided tests is 2(0.020) = 0.04.
The null hypothesis H 0 : p 0.2 versus H 0 : p 0.2 can be rejected because the P-value < 0.05.
e) Approximate 90% one-sided confidence bound on p
pˆ (1 pˆ )
p pˆ z 0.10
n
0.171765(1 0.171765)
p 0.171765 1.28
850
p 0.171765 0.0166
p 0.188365
4-139.
8
pˆ 0.005
1600
z 2 / 2 pˆ (1 pˆ ) z2 / 2 2.575 2 0.005(1 0.005) 2.575 2
pˆ z / 2 0.005 2.575
2n n 4n 2 2(1600) 1600 4(1600) 2
LCL 0.00207
z2 2.575 2
1 /2 1
n 1600
2
z / 2 2
pˆ (1 pˆ ) z / 2 2.575 2 0.005(1 0.005) 2.575 2
pˆ z / 2 0.005 2.575
2n n 4n 2 2(1600) 1600 4(1600) 2
UCL 0.01201
z 2 / 2 2.575 2
1 1
n 1600
The Agresti-Coull interval is similar to the traditional one.
4-140.
64
pˆ 0.64
100
z 2 / 2 pˆ (1 pˆ ) z 2 / 2 1.645 2 0.64(1 0.64) 1.645 2
pˆ z / 2 0.64 1.645
2n n 4n 2 2(100) 100 4(100) 2
LCL 0.5583
z2 1.645 2
1 /2 1
n 100
2
z / 2 2
pˆ (1 pˆ ) z / 2 1.645 2 0.64(1 0.64) 1.645 2
pˆ z / 2 0.64 1.645
2n n 4n 2 2(100) 100 4(100) 2
UCL 0.7143
z2 / 2 1.645 2
1 1
n 100
The Agresti-Coull interval is similar to the traditional one.
Team Exercises
4-141. a)
61
Engineering Statistics, 5th edition September 15, 2010
.999
.99
.95
Probability
.80
.50
.20
.05
.01
.001
Variable 95.0% CI T P
C1 ( 1.99920, 2.00137) -1.40 0.183
Based on the t-test, we fail to reject the null hypothesis because the P-value > 0.05. There is not sufficient
evidence that the mean differs from 2.001mm.
Based on this result, we fail to reject the null hypothesis because the P-value > 0.05. There is not sufficient evidence to
conclude that the proportion exceeds 0.10.
4-142. a)
62
Engineering Statistics, 5th edition September 15, 2010
.999
.99
.95
Probability
.80
.50
.20
.05
.01
.001
Based on the t-test, we fail to reject the null hypothesis because the P-value > 0.05. There is not sufficient evidence to
conclude that the mean is less than 1000 hrs.
Success = 1
Based on this result, we fail to reject the null hypothesis because the P-value > 0.05. There is not sufficient evidence to
conclude that the proportion less than 0.20.
63
Engineering Statistics, 5th edition September 15, 2010
4-143. a)
Normal Probability Plot
.999
.99
.95
Probability
.80
.50
.20
.05
.01
.001
3 4 5 6 7 8
level
Average: 5.945 Anderson-Darling Normality Test
StDev: 1.65528 A-Squared: 0.178
N: 10 P-Value: 0.892
64
Engineering Statistics, 5th edition September 15, 2010
1000
900
Data
800
700
600
Trail1 Trail2 Trail3 Trail4 Trail5
None of the trials are centered at either of the “true” values of 710.5 and trial 1 has a higher central value.
4-146. For n = 5
a) The coverage of the t CI based on the normal distribution simulation is 94.966
b) The coverage of the t CI based on the chi-square distribution simulation is 82.477
The coverage is close to 95% for the normally distributed data. For the chi-square data the central limit theorem
implies that the sample mean will still tend to be approximately normally distributed so that the confidence interval
coverage with be accurate, but the small sample size of n = 5 in this experiment does not result in a good
approximation.
4-147 For n = 10
a) The coverage of the t CI based on the normal distribution simulation is 95.1960
b) The coverage of the t CI based on the chi-square distribution simulation is 86.1080
For n = 15
a) The coverage of the t CI based on the normal distribution simulation is 95.0440
b) The coverage of the t CI based on the chi-square distribution simulation is 88.0520
For n = 25
a) The coverage of the t CI based on the normal distribution simulation is 94.9670
b) The coverage of the t CI based on the chi-square distribution simulation is 90.2510
In all cases the normally distributed data generates coverage near 95%. For the chi-square data the coverage is closer to
95% for the larger sample sizes than or the small sample sizes. This is the expected result from the central limit
theorem.
65