Chapter Four

Engineering Statistics, 5th edition September 15, 2010
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
Variable N Mean SE Mean StDev Sum

X 16 24.99 0.159 0.636 399.851
4-3. Sum x 131.869

Mean    10.9891
n 12
 n

 ( X ) 2 
1 n
1  (131.869) 2 
Variance    X 2  i 1  (1658.89)    19.07
n  1  i 1 n  12  1  12 
 
 
Variance = 209.77
4-4. Sum x 2977.70

Mean    198.5133
n 15
 n

 ( X ) 2 
1 n
1  ( 2977.70) 2 
Variance   X 2  i 1   (592589.64)    105.4634
n  1  i 1 n  15  1  15 
 
StDev 105.4634
SE Mean    2.6516
n 15
Variance = 1476.4873
SE Mean = 9.9213
 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
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
4-8. Because both estimators are unbiased:

MSE (ˆ 1 ) V(ˆ 1 ) 2 / 9 2
Relative Efficiency =   
MSE (ˆ 2 ) V(ˆ 2 ) 7 2 / 2 63
MSE (ˆ 1 ) V(ˆ 1 ) 2

4-9.    0.5
MSE (ˆ 2 ) V(ˆ 2 ) 4
4-10. E( 1 )   E( 2 )   / 2

Bias  E ( 2 )  

= 
2

= 
2
2
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
If 4.899    4.899 then use  2 .

For unbiasedness, use  1 . For efficiency, use  1 when   4.899 or   4.899 and use  2 when –4.899 <  < 4.899.
4-11. E( 1 )   No bias V (ˆ1 )  16  MSE (ˆ1 )

E( 2 )   No bias V (ˆ2 )  11  MSE (ˆ2 )
E( 3 )   Bias V( 3 )  6 includes (bias2)
To compare the three estimators, calculate the relative efficiencies:

MSE (ˆ1 ) 16
  1.45 , because rel. eff. > 1 use  2 as the estimator for 
MSE (ˆ2 ) 11
MSE (ˆ1 ) 16
MSE (ˆ3 ) 6
MSE (ˆ2 ) 11
MSE (ˆ3 ) 6
Conclusion:  3 is the most efficient estimator with bias, while  2 is the best “unbiased” estimator.
4-12. n1 = 20, n2 = 10, n3 = 8

Show that S2 is unbiased:
 20S12  10S22  8S23 
 
E S2  E
 38




1
38       
E 20S12  E 10S22  E 8S23

1
38

2012  10 22  8 23 

1
38

38 2 
 2
Therefore, S2 is an unbiased estimator of 2 .
3
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
Therefore, is a biased estimator of 2 .

n
b) Bias = E

 
2
X i2  nX  
 2  2 
2
 2  
2 
 n  n n
 
c) Bias decreases as n increases.
4-14. Show that X 2 is a biased estimator of .

 
Using E X2  V( X)   E ( X)
2
 
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
 
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 
 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
5
c) 1 -  = 1 – 0.077804= 0.922196
 c  12 
4-18. a) 0.05 = PZ  
 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 
 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
4-19. a)  = P( X  78.5) + P( X  81.5)

 X   78.5  80   X  81.5  80 
= P  + P 
2 / 9 

 / n 2/ 9   / n
= P(Z  2.25) + P(Z  2.25)
= P(Z  2.25) + (1  P(Z  2.25))
= 0.01222 + 0.01222 = 0.0244
b)  = P(78.5  X  81.5 when  = 83)

 78.5  83 X   81.5  83 
= P  
 2/ 9 / n 2 / 9 
= P(6.75  Z  2.25)
= P(Z  2.25)  P(Z  6.75)
= 0.01222  0 = 0.0122
c)  = P(78.5  X  81.5 when  = 85)

 78.5  85 X   81.5  85 
= P  
 2/ 9 / n 2 / 9 
= P(9.75 Z  5.25)
= P(Z  5.25)  P(Z  9.75)
= 0.
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.
4-20. Use n = 5, everything else held constant:

a) P( X  78.5) + P( X 81.5)
 X   78.5  80   X  81.5  80 
= P  + P 
2 / 5 

/ n 2/ 5   / n
= P(Z  1.68) + P(Z  1.68)
= 0.093
6
b)  = P(78.5  X  81.5 when  = 83)

 78.5  83 X   81.5  83 
= P  
 2/ 5 / n 2 / 5 
= P(5.03  Z  1.68)
= P(Z  1.68)  P(Z  5.03)
= 0.04648  0
= 0.04648
c)  = P(78.5  x  81.5 when  = 85)

 78.5  85 X   81.5  85 
= P  
 2/ 5 / n 2 / 5 
= P(7.27 Z  3.91)
= P(Z  3.91)  P(Z  7.27)
= 0.00005  0
= 0.00005
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.
b) P( X > 190 when  = 175)

 X   190  175 
= P   
 / n 20 / 10 
= P(Z > 2.37) = 1  P(Z  2.37)
= 1  0.99111
= 0.0089.
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.
4-23. Using n = 16:

a)  = P( X > 185 when  = 175)
 X   185  175 
= P   
 / n 20 / 16 
= P(Z > 2)
= 1  P(Z  2)
7
= 1  0.97725
= 0.0228

 
 16 
 183.23  200 
 =P  X  183.23when   200   P  Z    P  Z  3.35   0.000404
 20 / 16 
The probability is 0.000404 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) Power = 1 –  = 1 – 0.000404 = 0.999596
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

 
 16 
 183.23  195 
 =P  X  183.23 when   195  P  Z    P  Z  2.35  0.009387
 20 / 16 
The probability is 0.009387 of failing to reject the null hypothesis if the true mean foam height is 195 mm, with a level
of significance of  = 0.05.
c) For the same level of , with the increased sample size,  is reduced. With an increased sample size, the power is
also increased.
4-25. a)  = P( X  4.85 when  = 5) + P( X  5.15 when  = 5)

 X   4.85  5   X   5.15  5 
= P   + P  
  / n 0.25 / 8    / n 0.25 / 8 
= P(Z  1.7) + P(Z  1.7)
= P(Z  –1.7) + (1  P(Z  1.7))
= 0.044565 + 0.044565
= 0.8913.
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.
4-26. Using n = 16:

a)  = P( X  4.85 when  = 5) + P( X > 5.15 when  = 5)
 X  4.85  5   X  5.15  5 
= P   + P  
  / n 0.25 / 16    / n 0.25 / 16 
8
= P(Z  2.4) + P(Z > 2.4)

= P(Z  –2.4) +(1  P(Z  2.4))
= 0.0082 + 0.0082
= 0.0164.
b)  = P(8.85  X  9.15 when  = 9.1)

 8.85  9.1 X   9.15  9.1 
= P    
 0.25 / 16  / n 0.25 / 16 
= P(4  Z  0.8) = P(Z  0.8)  P(Z  4)
= 0.7881  0
= 0.7881
c) 1   = 0.2119
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
4-28. a) P  value  1   ( z0 )  1   (2.05)  1  0.979818  0.020182

b) P  value  1   ( z 0 )  1   (1.53)  1  0.936992  0.063008
c) P  value  1   ( z0 )  1   (0.4)  1  0.515953  0.484047
d) P  value  1   ( z 0 )  1   (1.85)  1  0.967843  0.032157
e) P  value  1  ( z0 )  1  (1.84)  1  0.967116  0.032884
4-29.      
a) P  value  2 1   z 0  2 1   2.45  2[1  0.992857]  0.014286
b) P  value  21    z 0   21    1.53   2[1  0.936992]  0.126016
c) P  value  21    z 0   21   2.15   2[1  0.984222]  0.031556
d) P  value  21    z 0   21  1.95   2[1  0.974412]  0.051176
e) P  value  21    z 0   21    0.25   2[1  0.598706]  0.802588
4-30. a) P  value   ( z 0 )   (2.15)  0.015778

b) P  value   ( z 0 )   (1.80)  0.035930
c) P  value   ( z 0 )   (2.50)  0.006210
d) P  value   ( z 0 )   (1.60)  0.054799
e) P  value   ( z 0 )   (0.35)  0.636831
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
One-Sample Z
Test of mu = 20 vs not = 20
The assumed standard deviation = 1.2
N Mean SE Mean 95% CI Z P

16 31.2000 0.3000 (30.6120, 31.7880) 37.33 0
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
Test of mu = 100 vs > 100

The assumed standard deviation = 2.4
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.
d) 95% two-sided CI on the mean

     
x  z 0.05 / 2      x  z 0.05 / 2 
  n

 n  
 2.4   
101.560  1.96     101.560  1.96 2.4 
  
 25   25 
101.560  0.9408    101.560  .9408
100.6192    102.5008
e) P  value  2[1  (| Z |)]  2[1  ( | 3.25 |)]  2[1  0.999423]  0.001154
10
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.
4-35. (a) Because(2.14) = 0.9838, the upper probability = 1 - 0.9838 = 0.0162.

Because it is two sided, the confidence level = 2(0.0162) = 0.0324
(b) Because(2.49) = 0.9936, the upper probability = 1 - 0.9838 = 0.0064.
(c) Because(1.85) = 0.9678, the upper probability = 1 - 0.9678 = 0.0322.
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).
4-37. a) 1) The parameter of interest is the true average battery life, .

2) H0 :  = 4
3) H1 :  > 4
x
4) z0 
/ n
5) Reject H0 if z0 > z where z0.05 = 1.645
6) x  4.05 ,  = 0.2
4.05  4
z0   1.77
0.2 / 50
7) Because 1.77>1.645, reject the null hypothesis and conclude that the true average battery life exceeds 4 hours at
 = 0.05.
b) P-value=1-  ( Z 0 ) =1- (1.77)  0.04
 (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
n2
  
e) x  z 0.05  

 n
11
 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.
4-38. a) 1) The parameter of interest is the true mean breaking strength, .

2) H0:  = 100
3) H1:   100
x
4) Test statistic is z0 
/ n
5) Reject H0 if z0 > z where z 0.05 = 1.645
6) x = 694,  = 2
694  690
z0   0.9
14 / 9
7) P-value = P(Z  0.9) = 1  P(Z < 0.9) = 0.18406.
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
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
4-39. a) 1) The parameter of interest is the true mean yield, .

2) H0:  = 90
3) H1:   90
x
/ n
5) Reject H0 if z0 <  z / 2 or z0 > z / 2 where z0.025 = 1.96
6) x  90.48 ,  = 3
90.48  90
z0   0.36
3/ 5
12
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%.
 z / 2  z  2   z0.025  z0.05 2 32  196

2
.  1.65 9
2
b) n =  4.67 , therefore n5
2 85  90 2  5 2
 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
d) For  = 0.05, z/2 = z0.025 = 1.96

     
x  z0.025      x  z0.025  
 n  n
 3   3 
90.48  196
.      90.48  196
.  
 5  5
87.85    93.11
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.
4-40. a) 1) The parameter of interest is the true mean ppm of benzene, .

2) H0:  = 7980
3) H1:  < 7980
x
/ n
5) Reject H0 if z0 <  z where z0.01 = 2.33
6) x = 7906,  = 80
7906  7980
z0   2.925
80 / 10
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.
d) For  = 0.01, z = z0.01 = 2.33

  
  x  z 0.01  

 n
13
 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.
4-41. a) 1) The parameter of interest is the true mean distance, .

2) H0:  = 2.00
3) H1:  > 2.00
x
/ n
6) x = 2.02,  = 0.05
2.02  2.00
z0   1.79
0.05 / 20
7) P-value = 1   (1.79) = 1  0.96327 = 0.03673. Because the P-value > 0.01, fail to reject the null hypothesis.
 2.03  2.00 
b)    z0.01     (0.35)  0.363169
 0.05 / 20 
c) Set  = 1  0.90 = 0.10

( z  z ) 2  2 2 2 (2.33  1.29)2 (0.05)2
n= = ( z0.01  z0.10 ) 0.05  = 36.4
2 ( 2.03  2.00) 2 (0.03)2
n  37.
d) For  = 0.01, z = z0.01 = 2.33

  
x  z 0.01  

 n
 0.05 
2.02  2.33 

 20 
1.994  
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.
4-42. a) 1) The parameter of interest is the true mean life, .

2) H0:  = 540
3) H1:   540
x
4) z0 
/ n
6) x = 551.33,  = 20
551.33  540
z0   2.19
20 / 15
7) Because 2.19 > 1.645, reject the null hypothesis and conclude that life exceeds 540 hrs at  = 0.05.
b) P-value = P(Z > 2.19) = 1  P(Z  2.19) = 1   (2.19) = 0.0143.

 (560  540) 15 
c)  =  z 0.05  = (1.645  3.873) = (2.23) = 0.012874
 20 
 
14
 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.
4-43. a) 1) The parameter of interest is the true mean compressive strength, .

2) H0 :  = 3500
3) H1 :   3500
x
4) z0 
/ n
5) Reject H0 if z0 < z/2 where z0.005 = 2.58 or z0 > z/2 where z0.005 = 2.58
6) x  3255.42 ,  = 31.62
3255.42  3500
z0   26.79
31.62 / 12
7) Because 26.79 < 2.58, reject the null hypothesis and conclude the mean compressive strength is significantly
different from 3500 at  = 0.01.
b) Smallest level of significance = P-value = 2[1   (26.84) ]= 2[1  1] = 0

The smallest level of significance at which we are willing to reject the null hypothesis is approximately 0.
c) z/2 = z0.025 = 1.96

     
x  z0.025      x  z0.025  
 n  n
 31.62   31.62 
3255.42  1.96     3255.42  1.96
  12 

 12   
3237.53    3273.31
With 95% confidence, the true mean compressive strength is between 3237.53 psi and 3273.31psi.
d) z/2 = z0.005 = 2.58

     
x  z0.005      x  z0.005  
 n  n
 31.62   31.62 
3255.42  2.58     3255.42  2.58
  12 

 12   
3231.96    3278.88
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.
4-44. z/2 = z0.025 = 1.96 , E=5
15
2
 20(1.96)  , n  62.
n   61.4656
 5 
4-45. z/2 = z0.025 = 1.96 , E = 0.01

2
 0.05(1.96)  , n  97.
n   96.04
 0.01 
4-46. z/2 = z0.005 = 2.58, E = 15

2
 3162
. (2.58) 
n   29.58 , n  30.
 15 
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
4-48. a) P-value = 2*P(t > 2.48): for degrees of freedom of 9 we obtain

2(0.01) < P-value < 2(0.025) = (0.02 < P-value < 0.05)
b) P-value = 2*P(t > |-3.95|): for degrees of freedom of 9 we obtain
2(0.001) < P-value < 2(0.0025) = (0.002 < P-value < 0.005)
c) P-value = 2*P(t > 2.69): for degrees of freedom of 9 we obtain
2(0.01) < P-value < 2(0.025) = (0.02 < P-value < 0.05)
d) P-value = 2*P(t > 1.88): for degrees of freedom of 9 we obtain
2(0.025) < P-value < (0.05) = (0.05 < P-value < 0.10)
e) P-value = 2*P(t > |-1.25|): for degrees of freedom of 9 we obtain
2(0.10) < P-value < 2(0.25) = (0.20 < P-value < 0.50)
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
4-50. a) StDev  SE Mean  n  0.4673  25  2.3365

 s 
Upper Limit 95%CI  x  t 0.025, 24  

 n
 2.3365 
 92.5805  2.064 

 25 
 93.5450
One-Sample T: X
Test of mu = 91 vs not = 91
Variable N Mean StDev SE Mean 95% CI T P

X 25 92.5805 2.3365 0.4673 (91.6160, 93.5450) 3.38 0.002
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
 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
4-51. a) There are 11 degrees of freedom on the t-test statistic.

b) StDev  SE Mean  n  0.3360  12  1.1639
 s 
95% Lower bound : x  t0.05,11  
 n
 1.1639 
25.6818  1.796  
 12 
25.0783  
x 25.6818  25
T   2.029
s/ n 1.1639 / 12
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
4-52. a) There are 9 degrees of freedom on the t-test statistic.

b) SE Mean  StDev  3.62  1.1447
n 10
x   49.30  50
T   0.6115
s / n 3.62 / 10
P-value = 2P(t > |-0.6115|) and for 9 degrees of freedom we obtain 2(0.25) < P-value < 2(0.4) = 0.5 < P-value < 0.8
4-53. a) 1) The parameter of interest is the true speed-up, .

2) H0:  = 4
3) H1:  > 4
x
4) Test statistic is t 0 
s/ n
5) Reject H0 if t0 > t,n-1 where t0.05,12 = 1.782
6) x  4.31322 , s = 0.4328, and n = 13
4.31322  4
t0   2.6094
0.4328 / 13
7) P-value = P(t > 2.6094) and for 12 degrees of freedom we obtain 0.01 < P-value < 0.025
Because the P-value < 0.025, and therefore less than 0.05, we reject the null hypothesis.
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
Probability Plot of Speed-up

Normal - 95% CI
99
Mean 4.313
StDev 0.4328
95 N 13
AD 0.233
90
P-Value 0.745
80
70
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
d) d =   |    0 |  | 4.75  4.0 |  1.73

  0.4328
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
Normal Probability Plot for temp

ML Estimates - 95% CI
99 ML Estimates
Mean 22.496
95
StDev 0.338384
90
80
70
Percent
60
50
40
30
20
10
5
21.5 22.5 23.5

Data
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
4-55. a) 1) The parameter of interest is the true mean life, .

2) H0:  = 60000
3) H1:   60000
x
4) Test statistic is t0 =
s/ n
6) x = 61492 s = 3035 n = 10
61492  60000
t0 =  1.55
3035 / 10
7) P-value = P(t > 1.55) and for degrees of freedom of 9 we obtain 0.05 < P-value < 0.10. Because the P-value >
0.05, fail to reject the null hypothesis.
    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.
c) t,n-1 = t0.05,9 = 1.833

 s 
x  t 0.05,9  

 n
 3035 
61492  1.833 

 10 
59732.78  
d) Because 60,000 km falls within this confidence interval, we fail to reject the null hypothesis.
19
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
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
s/ n
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
a) 1) The parameter of interest is the true mean level of polyunsaturated fatty acid, .
2) H0:  = 17
3) H1:   17
x
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.
c) For  = 0.01, t/2,n-1 = t0.005,5 = 4.032

 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
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.
.999
.99
.95
Probability
.80
.50
.20
.05
.01
.001
8.4 8.9 9.4 9.9

voltage
Average: 9.117 Anderson-Darling Normality Test
StDev: 0.463810 A-Squared: 0.136
N: 12 P-Value: 0.968
b) 1) The parameter of interest is the true mean breakdown voltage, .

2) H0 :  = 9
3) H1 :  < 9
21
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.
c) For  = 0.05 and n = 12, t,n-1 = t0.05,11 = 1.796

 s 
  x  t 0.05,11  

 n
 0.464 
  9.117  1.796 

 12 
  9.358
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.
.999
.99
.95
Probability
.80
.50
.20
.05
.01
.001
8.2 8.3 8.4

diameter
StDev: 0.0835935 A-Squared: 0.250
N: 12 P-Value: 0.680
b) 1) The parameter of interest is the true mean rod diameter, .

2) H0:  = 8.2
3) H1:   8.2
x
4) t0 =
s/ n
5) Reject H0 if t0 < -t/2,n-1 t0 > t/2,n-1 where t0.025,11 = 2.201
6) x = 8.2767 s = 0.0836 n = 12
22
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.
c) P-value = 2P(t > 3.18) for degrees of freedom of 11:

0.005 < P-value < 0.01
d) For  = 0.05 and n = 12, t/2,n-1 = t0.025,11 = 2.201

 s   s 
x  t 0.025,11      x  t 0.025,11 
 


 n  n
 0.0836   0.0836 
8.2767  2.201     8.2767  2.201
 


 12   12 
8.226    8.33
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.
4-61. a) 1) The parameter of interest is the true mean wall thickness, .

2) H0:  = 4.0
3) H1:   4.0
x
4) t0 =
s/ n
6) x = 4.058 s = 0.081 n = 25
4.058  4.0
t0 =  3.58
0.081 / 25
7) P-value = P(t > 3.58); 0.0005 < P-value < 0.001
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.
b) t/2,n-1 = t0.05,24 = 1.711

 s 
x  t 0.05,24  

 n
 0.081 
4.058  1.711 

 25 
4.030  
With 95% confidence, the true mean wall thickness is at least 4.03 mm
23
4-62. a)
.999
.99
.95
Probability
.80
.50
.20
.05
.01
.001
2.9 3.0 3.1

Enrich
StDev : 0.104953 A-Squared: 0.599
N: 12 P-Value: 0.092
The normality assumption appears to be reasonable. The data appear to fall along a straight line.
b) 1) The parameter of interest is the true mean percent enrichment, .

2) H0:  = 2.95
3) H1:   2.95
x
s/ n
5) Reject H0 if t0 > t/2,n-1 where t0.025,11 = 2.201
6) x = 2.998 s = 0.105 n = 12
2.998  2.95
t0 =  1.58
0.105 / 12
7) P-value = 2P(t > 1.58) for degrees of freedom of 11: 0.1 < P-value < 0.2. Because the P-value > 0.05, fail to
reject the null hypothesis.
c) For  = 0.01 and n = 12, t/2,n-1 = t0.005,11 = 3.106

 s   s 
x  t 0.005,11     x  t 0.005,11 
 n  n
 0.105   0.105 
2.998  3.106     2.998  3.106
  12 

 12   
2.904    3.092
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%.
4-63. a) 1) The parameter of interest is the true mean quantity of syrup, .

2) H0:  = 1.0
3) H1:   1.0
x
4) t0 =
s/ n
5) Reject H0 if t0 <  t/2,n-1 where -t0.025,24 = 65 or t0 > t/2,n-1, where t0.025,24 = 65
6) x = 32.5 s = 0.47 n = 25
24
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.
d) For  = 0.05 and n = 25, t/2,n-1 = t0.025,24 = 65

 s   s 
x  t 0.025,24      x  t 0.025,24  
 n  n
 0.47   0.47 
32.5  65     32.5  65 
 25   25 
26.39    38.61
With 99% confidence, the true mean quantity is between 26.39 ml and 38.61 ml.
4-64. The parameter of interest is the true mean natural frequency, .

For  = 0.10 and n = 5, t/2,n-1 = t0.05,4 = 2.132
 s   s 
x  t0.05,4      x  t0.05,4 
 n  n 
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.
4-65. a) 1) The parameter of interest is the true mean rainfall, .

2) H0:  = 25
3) H1:   25
x
4) t0 =
s/ n
6) x = 26.04 s = 4.78 n = 20
26.04  25
t0 =  0.97
4.78 / 20
7) Because 0.97 < 2.539, fail to reject the null hypothesis. There is insufficient evidence to indicate that the true mean
rainfall is greater than 25 acre-feet at  = 0.01. 0.10 < P-value < 0.25
b) The data on the normal probability plot fall along the straight line. Therefore, the assumption of normally
distributed data is reasonable.
25
Normal Probability Plot for rainfall

ML Estimates - 95% CI
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 10.7 = 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
e) 99% lower confidence bound on the mean diameter

 s 
x  t0.01,19  

 n
 4.78 
26.04  2.539 

 20 
23.326  
Because the lower limit of the CI is less than 25, there is insufficient evidence to conclude that the true mean rainfall is
greater than 25 acre-feet at = 0.01
Section 4-6
a) P-value = P(  0 > 22.35):

2
4-66. for degrees of freedom of 14 we obtain 0.05 < P-value < 0.1
b) P-value = P(  > 23.50):

2
0 for degrees of freedom of 14 we obtain 0.05 < P-value < 0.1
c) P-value = P(  > 25.00):

2
d) P-value = P(  > 28.55):

2
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
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.
b) 95% lower confidence interval on 2:

For  = 0.05 and n = 15, 2 ,n 1   02.05,14  23.68
14(0.032) 2 < 2
23.68
0.0006 < 2
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.
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.
b) P-value = 2P(2 > 8)  1 for 9 degrees of freedom
c) 95% confidence interval for :

First find a confidence interval for 2:
For  = 0.05 and n = 10, 2 / 2 , n 1   20.025,9  19.02 and 12  / 2 ,n 1   20.975,9  2.70
9(16) 9(16)
 2 
19.02 2.70
7.57  2  53.33
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.
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
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.
c) 95% lower confidence interval for 2:

For  = 0.05 and n = 10, 2 ,n 1   02.05,9  16.919
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
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
c) 99% confidence interval for 2 :

For  = 0.01 and n = 20, 2 / 2 , n 1   20.005,19  38.58 and 12  / 2 ,n 1   20.995,19  6.84
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
d) Because the hypothesized value falls within this confident interval, we fail to reject the null hypothesis.
4-71.
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
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
c) 95% confidence interval for :

First find the confidence interval for 2:
For  = 0.05 and n = 51, 2 / 2 , n 1   20.025,50  71.42 and 12  / 2 ,n 1   20.975,50  32.36
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
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.
Test and CI for One Proportion
Test of p = 0.3 vs p not = 0.3
Sample X N Sample p 95% CI Z-Value P-Value

1 95 250 0.380000 (0.319832, 0.440168) 2.76 0.006
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
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
4-73. a) It is a one-sided test
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
Test of p = 0.65 vs p > 0.65
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
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%
 p  p  z  p 0 (1  p 0 ) / n   0.25  0.35  1.28 0.25(1  0.25) / 30 

b)     0      0.0137
 p(1  p ) / n   0.35(1  0.35) / 30 
  0.5055
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
d) 0.367  1.28 0.367(0.633)  p

30
0.254  p
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 
4-76. n = 50, x = 18, pˆ  22  0.36 ,  = 0.05, z/2 = z0.025 = 1.96

60
a) 1) The parameter of interest is the true proportion of damage, p.
2) H0: p = 0.30
3) H1: p  0.30
x  np0 p  p0
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
31
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%.
b) P-value = 2P(Z > 1.01) = 2(0.156248) = 0.312496
c) 0.36  1.96 0.36(0.64)  p  0.36 + 1.96 0.36(0.64)

60 60
0.24  p  0.48
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
np0 1  p0  p0 1  p0 
n
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.
b)  0.01  0.02  Z 0.05 (0.01)(0.99) /1000 

      (1.09)  0.137857
 (0.02)(0.98) /1000 
c) Use equation in part b) to solve for n when  = 0.10,  = 0.05, n = 1000, z 0.05 = 1.645, p = 0.02, and p0 = 0.01.
Using MINITAB, n = 1178 with power = 0.9002.
4-79. E = 0.02,  = 0.05, z/2 = z0.025 = 1.96
32
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
np0 1  p0  p0 1  p0 
n
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.
c) p̂ = 22/400 = 0.06,  = 0.05, n = 400, Z 0.025 = 1.96

ˆ ˆ
p(1-p) ˆ ˆ
p(1-p)
pˆ - z α/2 ˆ α/2
<p<p+z
n n
0.06  0.94 0.06  0.94
0.06  1.96  p  0.06  1.96
400 400
0.0367 < p < 0.0833
d) Because p0 = 0.04 is in the CI, we fail to reject H0. The fraction of the defective circuits is not significantly different
from 0.04.
   
 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
np0 1  p0  p 0 1  p0 
n
117
6) x = 117 n = 484 p   0.242
484
33
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.
b) P-value = 2(1  (11.352)) = 2(1  1) = 0
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
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.
b) P-value = (–1.61) = 0.0537
 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
np0 1  p0  p0 1  p0 
n
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.
b) P-value = 1  (0.943) = 0.1728.
34
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
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)
(b) A lower confidence bound is most appropriate here. The bound is

pˆ (1  pˆ ) 0.6757(0.3243)
pˆ  Z  p, 0.6757  1.645  p, 0.5491  p
n 37
4-88.
823
pˆ   0.823
1000
z2 pˆ (1  pˆ ) z2 / 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
4-89.
18
pˆ   0.36
50
2
pˆ (1  pˆ ) z2 / 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
z2 / 2 1.96 2
1 1
n 50
2
pˆ (1  pˆ ) z2 / 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
z2 / 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 ) z2 / 2 0.675  1.96  1.96 0.6757(1  0.6757)  1.96
2
z2 / 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 ) z2 / 2 0.675  1.96  1.96 0.6757(1  0.6757)  1.96
2
z2 / 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 n1  x  t  / 2,n1 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, 201  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
1 1
4-93. a) x  t  / 2,n 1 s 1   X n1  x  t  / 2,n1 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 / 2,61  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 n1  x  t  / 2,n1 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 / 2,121  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
The degrees of freedom are k  p  1 = 5  1  1 = 3
a) 1) The variable of interest is the form of the distribution for X.

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, 4  9.49

6)
37
 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.
b) P-value = 0.414 (found using Minitab)
4-98. Estimated mean = 4.907
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.
k  O i  E i 2
 20   Ei
i 1
5) Reject H0 if  2o   20.01,5  15.09
6)  2  12  9.3984     9  4.6237   6.955

2 2
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:
Value 5-7 8 9 10 11 12-15

Observed
8 4 5 1 3 9
(Days)
Expected
4.41 3.635 3.878 3.723 3.249 6.676
(Days)
38
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.
k  O i  E i 2
 20   Ei
i 1
5) Reject H0 if  o2   02.05,4  9.49
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.

All expected frequencies are greater than 3.
a) 1) The variable of interest is the form of the distribution for the number of cars passing through the intersection.
k  O i  E i 2
 20   Ei
i 1
5) Reject H0 if  2o  20.05,24  36.42

6) Estimated mean = 49.6741
 20  769.57
7) Because 769.57 >>> 36.42, reject H0. We conclude that the distribution is not Poisson at  = 0.05.
b) P-value = 0 (found using Minitab)
4-101. Mean = np = 6(0.25) = 1.5
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
k  O i  E i 2
 20   Ei
i 1
39
5) Reject H0 if  2o   20.05,3  7.81

6)
 02 
4  8.8992 
15  8.242  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
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.
b) P-value = 0.5481 with d.f. = 2 (found using Minitab)
Supplemental Exercises
4-103. Operating characteristic curve:

x  185
 x   185   
  P Z    P Z  
 20 / 10   20 / 10 
The probabilities given in the following table were found using Minitab.
 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
190 P(Z  0.79) = 0.2148 0.7852

193 P(Z  1.26) = 0.1038 0.8962
196 P(Z  1.74) = 0.0409 0.9491
199 P(Z  2.21) = 0.0136 0.9864
0.9
0.8
0.7
0.6
0.5
Beta
0.4
0.3
0.2
0.1
0.0
180 190 200
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.
4-106. a) H0: p = 0.9995, H1: p < 0.9995
41
b) H0:  = 45, H1:  > 45

c) H0: p = 0.05, H1: p < 0.05
d) H0:  = 90, H1:  < 90
e) H0:  = 10, H1:  < 10
f) H0:  = 2160, H1:  > 2160
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
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, 201  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.
d) 12  , n 1   20.99 ,8 =1.65

(n  1) s 2
2 
 21 .n 1
8(8.42) 2
2 
1.65
  343.74
2
With 99% confidence, the true variance of compressive strength is at most 343.76 (Mpa)2.
e)  2 / 2,8  20.09 12( / 2),8  1.65 with  = 0.02
43
(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.
f) Change 40.2 to 20.4

x  22.9 and s2 = 39.83
98% confidence on :
 6.3   6. 3 
22.9  2.896     22.9  2.896
  9

 9  
16.82    28.98
With 98% confidence, the true mean compressive strength is between 16.82 and 28.98 Mpa.
98% confidence interval on 2 :

8(39.83) 8(39.83)
2 
20.09 1.65
15.86   2  193.12
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.
g) Change 25.8 to 24.8

x  25 and s2 = 70.84
98% confidence on :
 8.42   8.42 
25  2.896     25  2.896
  9 

 9   
16.87    33.13
With 98% confidence, the true mean compressive strength is between 16.87 and 33.13.
98% confidence interval on 2 :

8(70.84) 8(70.84)
2 
20.09 1.65
28.21   2  343.47
With 98% confidence, the true variance of compressive strength is between 28.21 and 343.47 (Mpa)2.
Comparison of intervals parts f) and g):
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
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.
Width of Confidence Interval

Parameter of Interest Original Data Case 1 Case 2
Mean,  16.26 12.16 16.26
Variance, 2 315.51 177.26 315.26
1 1
i) x  t  / 2,n1 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
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
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
4-116. H0:  = 85  = 16 the true mean is 86

 x 85  86 
a)   P( x  85 when   86)  P    P( z  0.0625 n )
  / n 16 / n 
n = 25:   P( z  0.0625 25 )  0.3783
n = 100:   P( z  0.0625 100 )  0.2643
n = 400:   P( z  0.0625 400 )  01056
.
n = 2500:   P( z  0.0625 2500 )  0.0009
 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.
b) Assume the data follow a normal distribution.
1) The parameter of interest is the mean weld strength, .

2) H0:  = 1034
3) H1:   1034
47
4) The test statistic is: t  x  0  1087  1034  2.79

0
s/ n 85 / 20
5) Reject H0 if t > t,n-1 where t0.05,19 = 1.729
6) x  1087 and t0 =2.79
7) P-value = P (t  2.79)  0.005  p  value  0.10
At  = 0.05, reject the null hypothesis. Conclude that the mean weld strength is greater than 1034 at  = 0.05.
4-118. H0: p = 0.5 versus H1 : p  0.5
a) Find the Power, 1   for 95% confidence

 p (1  p0 )   p(1  p) 
 p 0  p  z / 2 0   p 0  p  z / 2 
   n    n 
 p(1  p)   p(1  p) 
   
 n   n 
 0.5(1  0.5)   0.6(1  0.6) 
 0.5  0.6  196
.   0.5  0.6  196
. 
  n    n 
 0.6(1  0.6)   0.6(1  0.6) 
   
 n   n 
n = 100:    (0.041)   (4.04)  0.4836  0  0.4836

1    1  0.4836  0.5164
n = 150:    ( 0.50)  ( 4.5)  (1  0.69146)  0  0.30856

1    1  0.30856  0.691
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.
b) Find the Power, 1   for 99% confidence
n = 100:   (0.59)  ( 4.67)  0.7224  0  0.7224

1    1  0.7224  0.278
n = 150:   (013. )  ( 512

. )  0.55172  0  0.55172
1    1  0.55172  0.448
n = 300:   ( 0.91)  ( 616

. )  1  0.81859  0  018141
.
1    1  0.18141  0.819
The power of the test is greater for the larger values of , because a larger  value results in a smaller acceptance
region.
c) n = 100 p = 0.08  = 0.05
48
   
 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.
d)  = 0.01,  = 0.05, z/2 =2.575, z = 1.645
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.
4-119. a) 1) the parameter of interest is the standard deviation, 

2) H0: 2 = 78400
3) H1: 2 < 78400
( n  1)s2
4) The test statistic is:  20 
2
5) Because no critical value is given, we will calculate the p-value
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
c) Increasing the sample size increases the test statistic  20 and therefore decreases the P-value, providing more
evidence against the null hypothesis.
4-120. a) 1) The parameter of interest is the true mean sugar concentration, .

2) H0 :  = 11.5
3) H1 :   11.5
x
4) t 0 
s/ n
5) Reject H0 if |t0| > t/2,n-1 where t/2,n-1 = 2.093
6) x  11.47 , s = 0.022 n=20
11.47  11.5
t0   6.10
0.022 / 20
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.
d) 95% two sided confidence interval

 s   
x  t 0.025,19      x  t 0.025,19  s 
  
 n  n
 0.022   
11.47  2.093     11.47  2.093 0.022 
  
 20   20 
11.46    11.48
We can conclude that the mean sugar concentration content is not equal to 11.5 because that value is not within the
e) The normality plot below indicates that the normality assumption is reasonable.
Probability Plot of Sugar Concentration

Normal
99
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
4-121. a) 1) The parameter of interest is the true mean body weight, .
50
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.
c) 95% two sided confidence interval

 s   
x  t 0.025,19      x  t 0.025,19  s 
  
 n  n
 198.79   
325.5  2.056     325.5  2.056 198.79 
  
 27   27 
246.84    404.16
There is not enough evidence to conclude that the mean body weight differs from 300 because 300 is within the 95%
4-122.
a) In order to use the 2 statistic in hypothesis testing and confidence interval construction, we need to assume that the
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,n1 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
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
4-123. a) 1) The parameter of interest is the true mean sodium content, .

2) H0 :  = 300
3) H1 :  > 300
x
4) t 0 
s/ n
51
5) Reject H0 if t0 > t,n-1 where t,n-1 = 1.943

6) x  315 , s = 16 n=7
315  300
t0   2.48
16 / 7
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 = 10.9 = 0.1.
c) if 1 -  > 0.9 then  < 0.1 and n is approximately 100

 s 
d) Lower confidence limit is x  t , n 1    303.2 because 300 is not contained in the CI reject the null hypothesis

 n
4-124. a) 1) The parameter of interest is the true success rate

2) H0: p = 0.78
3) H1: p > 0.78
x  np 0 pˆ  p 0
4) z 0  or z 0  ; Either approach yields the same conclusion
np 0 1  p 0  p 0 1  p 0 
n
289
6) x = 289 n = 350 pˆ   0.83
350
x  np 0 289  350(0.78)
z0    2.06
np 0 1  p 0  350(0.78)(0.22)
7) Because 2.06 > 1.645, reject the null hypothesis. Conclude that the true success rate is greater than 0.78, at  =
0.05. P-value= 1-0.9803 = 0.0197
b) The 95% lower confidence interval:

pˆ (1  pˆ )
pˆ  z  p
n
0.83(0.17)
0.83  1.645  p
350
0.7970  p
Because the hypothesized value of 0.78 is not contained in the CI reject the null hypothesis
4-125. The Expected Frequency is found by using the Poisson distribution

e  x
P( X  x)  where   [6(1)  7(1)    39(1)  41(1)] / 105  19.514
x!
Estimated mean = 19.514
Number of Observed Expected

Earthquakes Frequency Frequency
6 1 0.03
7 1 0.08
8 4 0.18
10 3 0.78
11 2 1.38
12 2 2.24
13 6 3.36
52
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:
Number of Observed Expected

Earthquakes Frequency Frequency
6-7-8-10-11-
12 13 4.68
13 6 3.36
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
53
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.
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.
b) P-value < 0.005
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
2
5) Reject H0 if  20 < 12  / 2 ,n 1 where  20.995,5  0.41 or  20 > /
2 2
2 , n1 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
2
5) Reject H0 if  20 < 12  / 2 ,n 1 where  20.995,50  27.99 or  20 > /
2 2
2 , n1 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.
4-127. Assume the data follow a normal distribution.
54
a) 1) The parameter of interest is the standard deviation, .

2) H0: 2 = (0.00002)2
3) H1: 2 < (0.00002)2
( n  1)s2
2
5)  02.99,7  1.24 reject H0 if  20  124
.
6) s = 0.00001 and  = 0.01
7(0.00001) 2
 20   175
.
( 0.00002) 2
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.
.999
.99
.95
Probability
.80
.50
.20
.05
.01
.001
16.620 16.625 16.630

Concentration
StDev : 0.0040401 A-Squared: 0.173
N: 10 P-Value: 0.900
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.
1) The parameter of interest is the standard deviation, .

2) H0: 2 = 16
3) H1: 2 < 16
( n  1)s2
2
5) Because no critical value is given, we calculate the P-value
6) n  10
s2  0.004
9(0.004) 2
 20   0.000009
16

P-value = P  2  0.000009 ; P-value < 0.005
55
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.
4-129. a) 1) The parameter of interest is the true proportion, p.

2) H0: p = 0.01
3) H1: p < 0.01
p  p 0
4) The test statistic is: z0 
p0 (1  p0 )
n
5) z0.01 = 2.33, reject H0 if z0 < 2.33
6) p̂  6  0.006
1000
0.006  0.01
z0   1.27
0.01(1  0.01)
1000
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.
b) 95% confidence interval; z / 2  196

.
56
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.
c) Comparison of parts a) and b):
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.
4-132. X ~ bin(20, 0.4) H0: p = 0.4 and H1: p  0.4
4/20 = 0.20 8/20 = 0.4
Rejection region: p̂  0.20 or p̂  0.4
Use the normal approximations for parts a) and b)
a) When p = 0.4, α  P(pˆ  0.20)  P(pˆ  0.4)

   
 0.2  0.4   0.4  0.4 
 P Z    PZ  
 0.4(0.6)   0.4(0.6) 
   
20  20 
 P(Z  1.83)  P(Z  0)
 P(Z  1.83)  (1  P(Z  0))
 0.034  0.5
 0.534
 
 0.2  0.2 0.4  0.2 
b) When p = 0.2, β  P(0.2  p
ˆ  0.4)  P  Z 
 0.2(0.8) 0.2(0.8) 
 
20 20 
 P(0  Z  2.24)
 P(Z  2.24)  P(Z  0)
 0.9875  0.5
 0.4875
4-133. X ~ bin(10, 0.3) Implicitly, H0: p = 0.3 and H1: p < 0.3
n = 20
Reject region: p  01
.
Use the normal approximation for parts a), b) and c):
57
 
 
.  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 - Pp̂  0.1  1  P Z  
 0.2(0.8) 
 20 
 
 1  P( Z  1.12)
 P( Z  1.12)
 0.1314
c) Power = 1   = 1  0.1314 = 0.8686
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
The value of p must be estimated. Let the estimate be denoted by p sample

0(0)  1(0)  2(0)    19(1)
sample mean   9.8
40
sample mean 9.8
pˆ sample    0.196
n 50
Value Observed Expected

0 0 0.00132
1 0 0.01511
2 0 0.08486
3 0 0.31128
4 3 0.83853
5 0 1.76859
6 2 3.04094
7 4 4.38212
8 6 5.39988
9 9 5.77713
10 4 5.43022
11 3 4.52695
12 1 3.37296
13 2 2.26033
14 0 1.36952
15 1 0.75353
16 2 0.37789
17 2 0.17327
18 0 0.07283
19 1 0.02812
58
Because several of the expected values are less than 3, some cells are combined resulting in the following table:
Value Observed Expected

0-6 5 6.06063
7 4 4.38212
8 6 5.39988
9 9 5.77713
10 4 5.43022
11 3 4.52695
12 9 8.40844
a) 1) The variable of interest is the form of the distribution for the number of nonconforming coil springs.
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
The value of p must be estimated. Let the estimate be denoted by p sample

0(3)  1(7)  2( 4)  3(5)  4(1)  5( 0)
sample mean   17
.
20
sample mean .
17
p sample    0.0017
n 1000
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
a) 1) The variable of interest is the form of the distribution for the number of errors in a string of 1000 bits.
k  O i  E i 2
 20   Ei
i 1
5) Reject H0 if 20  20.05,2  5.99

6)
 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.
4-136. a) H0:  = 5000, H1:  > 5000 Upper-tailed

b) H0:  = 60,000, H1:  > 60,000 Upper-tailed
c) H0:  = 2, H1:  < 2, Lower-tailed
d) H0: p = 0.60, H1: p > 0.60 Upper-tailed
e) H0:  = 42,000, H1:  > 42,000 Upper-tailed
f) H0:  = 0.02, H1:  < 0.02, Lower-tailed
g) H0: 2 = 0.05, H1: 2 < 0.05, Lower-tailed
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
Test of mu = 44.5 vs > 44.5
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.
Test of p = 0.2 vs p < 0.2
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
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 ) z2 / 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
z2 / 2 1.645 2
1 1
n 100
The Agresti-Coull interval is similar to the traditional one.
Team Exercises
4-141. a)
61
.999
.99
.95
Probability
.80
.50
.20
.05
.01
.001
1.997 1.998 1.999 2.000 2.001 2.002 2.003

thickness
StDev: 0.0020402 A-Squared: 0.368
N: 16 P-Value: 0.386
The data appear to be normally distributed.
b) Using Minitab, the results are

Test of mu = 2.001 vs mu not = 2.001
Variable N Mean StDev SE Mean

C1 16 2.00029 0.00204 0.00051
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.
c) Using Minitab, the results are

Test of p = 0.1 vs p > 0.1
Sample X N Sample p 95.0% Lower Bound Z-Value P-Value

1 2 16 0.125000 0.000000 0.33 0.369
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
.999
.99
.95
Probability
.80
.50
.20
.05
.01
.001
940 990 1040 1090

Ex4-102
N: 20 P-Value: 0.630
b) Using Minitab, the results are
Test of mu = 1000 vs mu < 1000
Variable N Mean StDev SE Mean

Ex4-102 20 990.9 49.9 11.2
Variable 95.0% Upper Bound T P

Ex4-102 1010.2 -0.82 0.211
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.
c) Using Minitab, the results are
Test of p = 0.2 vs p < 0.2
Success = 1
Variable X N Sample p 95.0% Upper Bound Z-Value P-Value

Ex4-102c 3 20 0.150000 0.281331 -0.56 0.288
* NOTE * The normal approximation may be inaccurate for small samples.
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
4-143. a)
.999
.99
.95
Probability
.80
.50
.20
.05
.01
.001
3 4 5 6 7 8
level
N: 10 P-Value: 0.892
b) 1) The parameter of interest is the standard deviation of CO level, .

2) H0:  = 2.0
3) H1:  > 2.0
( n  1)s2
4)  20 =
2
5) Reject H0 if 20  2 ,n 1 where  02.05,9  16.919
6) n = 10, s2 = (1.6553)2
(n  1) s 2 9(1.6553) 2
 20 =   6.165
2 4.0
7) Because 6.165 < 16.919 fail to reject H0. There is not sufficient evidence to conclude that the true standard
deviation of CO level exceeds 2.0 ppm at  = 0.05.
4-144. Student report on an example they find for a hypothesis test
4-145. The Minitab box plots are as follows:
64
Boxplot of Trail1, Trail2, Trail3, Trail4, Trail5

1100
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
For n = 15
For n = 25
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

Chapter Four

Diunggah oleh

Informasi Dokumen

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Chapter Four

Diunggah oleh

Hak Cipta:

Format Tersedia

Engineering Statistics, 5th edition September 15, 2010

Variable N Mean SE Mean StDev Sum

4-3. Sum x 131.869

4-4. Sum x 2977.70

4-8. Because both estimators are unbiased:

MSE (ˆ 1 ) V(ˆ 1 ) 2

4-10. E( 1 )   E( 2 )   / 2

If 4.899    4.899 then use  2 .

4-11. E( 1 )   No bias V (ˆ1 )  16  MSE (ˆ1 )

To compare the three estimators, calculate the relative efficiencies:

4-12. n1 = 20, n2 = 10, n3 = 8

Therefore, S2 is an unbiased estimator of 2 .

Therefore, is a biased estimator of 2 .

4-14. Show that X 2 is a biased estimator of .

4-19. a)  = P( X  78.5) + P( X  81.5)

b)  = P(78.5  X  81.5 when  = 83)

c)  = P(78.5  X  81.5 when  = 85)

4-20. Use n = 5, everything else held constant:

b)  = P(78.5  X  81.5 when  = 83)

c)  = P(78.5  x  81.5 when  = 85)

b) P( X > 190 when  = 175)

4-23. Using n = 16:

b) For  = 0.05, fail to reject H0 if X  175  1.645 20   183.23

c) Power = 1 –  = 1 – 0.000404 = 0.999596

b) For  = 0.05, fail to reject H0 if X  175  1.645 20   183.23

4-25. a)  = P( X  4.85 when  = 5) + P( X  5.15 when  = 5)

4-26. Using n = 16:

= P(Z  2.4) + P(Z > 2.4)

b)  = P(8.85  X  9.15 when  = 9.1)

4-28. a) P  value  1   ( z0 )  1   (2.05)  1  0.979818  0.020182

4-30. a) P  value   ( z 0 )   (2.15)  0.015778

N Mean SE Mean 95% CI Z P

Test of mu = 100 vs > 100

d) 95% two-sided CI on the mean

4-35. (a) Because(2.14) = 0.9838, the upper probability = 1 - 0.9838 = 0.0162.

4-37. a) 1) The parameter of interest is the true average battery life, .

b) P-value=1-  ( Z 0 ) =1- (1.77)  0.04

4-38. a) 1) The parameter of interest is the true mean breaking strength, .

7) P-value = P(Z  0.9) = 1  P(Z < 0.9) = 0.18406.

4-39. a) 1) The parameter of interest is the true mean yield, .

 z / 2  z  2   z0.025  z0.05 2 32  196

d) For  = 0.05, z/2 = z0.025 = 1.96

4-40. a) 1) The parameter of interest is the true mean ppm of benzene, .

d) For  = 0.01, z = z0.01 = 2.33

4-41. a) 1) The parameter of interest is the true mean distance, .

c) Set  = 1  0.90 = 0.10

d) For  = 0.01, z = z0.01 = 2.33

4-42. a) 1) The parameter of interest is the true mean life, .

b) P-value = P(Z > 2.19) = 1  P(Z  2.19) = 1   (2.19) = 0.0143.

4-43. a) 1) The parameter of interest is the true mean compressive strength, .

b) Smallest level of significance = P-value = 2[1   (26.84) ]= 2[1  1] = 0

c) z/2 = z0.025 = 1.96

d) z/2 = z0.005 = 2.58

4-44. z/2 = z0.025 = 1.96 , E=5

4-45. z/2 = z0.025 = 1.96 , E = 0.01

4-46. z/2 = z0.005 = 2.58, E = 15

4-48. a) P-value = 2*P(t > 2.48): for degrees of freedom of 9 we obtain

4-50. a) StDev  SE Mean  n  0.4673  25  2.3365

Variable N Mean StDev SE Mean 95% CI T P

4-51. a) There are 11 degrees of freedom on the t-test statistic.

4-52. a) There are 9 degrees of freedom on the t-test statistic.

4-53. a) 1) The parameter of interest is the true speed-up, .

Probability Plot of Speed-up

d) d =   |    0 |  | 4.75  4.0 |  1.73