Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,

Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
Sections 12-1
12-1. a)
10 223
XX= 223 5200.9
553 12352
1916.0
Xy = 43550.8
104736.8
171.055
LM1M
553
12352
31729
b) = 3.713 so j) =l71.055+3.714x
1
-1.126x
2
12-3.
-1.126
c) j/ = 171.055+3.714(18) -1.126(43) = 189.49
a) Model l
y = 100+2x
1
+8
y = 108+2x
1
y = 100+ 2x
1
+ 4(8)
y= 132+2x
1
MODEL 1
Model 2
y = 95 + 1.5x
1
+ 3(2)+ 4x
1
j) =l01+5.5x
1
y = 95 + 1.5x
1
+ 3(8)+ 16x
1
j) =l19+17.5x
1

!D
y1JZ"Z
y1"Z
y !
a

D
X
MODEL 2
J !D
y 119
+
11.3X

J
y
!
y I1
+
3.3X
d
b J !.
X
The interaction term umodel 2 afects the slope of the regression equation. That is, it modifes the
amount of change per unit of x
1
on ,
12-1
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
c)
y = 100+ 2x
1
+ 4(5)
y= 120+2x
1
Then, 2 is the expected change on y per unit of x
1

NO, it does not depend on the value ofx2 , because there is no relatonship or interacton between these two
vaables in model 1.
Change per
unit of x
1
y = 95+ 1.5x
1
+ 3(5) + 2x
1
(5)
j) =l l 0+11.5x
1
11.5
y=101+5. 5x
1
j) =l l9+17.5x
1
5. 5 17.5
Yes, result does depend on the value of 7_ because x2 interacts with x, .
12-5. a) The results fom Minitab follow. The model can be expressed as
b)
c)
Satisfaction 144 - 1 . 11 Age - 0 . 585 Severity " 1 . 30 Axiety
n
_

=
1=
1

`_
=
1039.9
= 49.5
nj nj 21
5.9
1
0.13
cov(/) = (') = C, se(/) = C, . =
l
0.13
1.06
fom the Minitab output.

Because the regession coeficients have different standard erors the parameters estors do not have
similar precision of estimation.
Minitab result:
Regression Analysis: Satisfaction versus Age, Severity, Anxiety
The regression eqation is
Satisfaction 144 - 1 . 11 Age 0 . 585 Severity " 1 . 30 Anxiety
Predictor Coef SE Coef T p
Constant 143. 895 5 . 898 2 4. 40 0 . 000
Age -1 . 1135 0 . 1326 -8 . 40 0 . 000
Severity -0 . 5849 0 . 1320 -4 . 43 0 . 000
Anxiety 1 . 296 1 . 056 1 . 23 0 . 233
o 7 . 03710 R-Sq 9 0 . 4% R-Sq(adj) 89 . 0%
Analysis of Variance
Source ! oo MS F p
Regression 3 9738 . 3 3246 . 1 65 . 55 0 . 000
Residual Error 21 1039 . 9 49 . 5
Total 24 1077 8 . 2
Source ! Seq SS
Age 1 8756 . 7
Severity 1 907 . 0
Anxiety 1 74 . 6
12-2
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
Unusual Observations
Obs
9
Age
27 . 0
Satisfaction
75 . 00
Fit
93 . 28
SE Fit
2 . 98
Residual
-18 . 28
St Resid
-2 . 87R
12-7. The regression eqation is
12-9.
mpg 49 . 9 - 0 . 0104 cid - 0 . 0012 rhp - 0 . 00324 etw 0 . 29 cp - 3 . 86 axle
" 0 . 190 n/v
Predictor Coef SE Coef T p
Constant 49 . 90 19 . 67 2 . 54 0 . 024
cid -0 . 01045 0 . 02338 -0 . 45 0 . 662
rhp -0 . 00120 0 . 01631 -0 . 07 0 . 942
etw -0 . 0032364 0 . 0009459 -3 . 42 0 . 004
'
0 . 292 1 . 765 0 . 17 0 . 871
axle -3 . 855 1 . 329 -2 . 90 0 . 012
n/v 0 . 1897 0 . 2730 0 . 69 0 . 498
o 2 . 22830 R-Sq 89 . 3 % R-Sq(adj) 84 . 8%
Analysis of Variance
Source DF oo MS F p
Regression 6 581 . 898 9 6 . 983 19 . 53 0 . 000
Residual Error 14 69 . 514 4 . 965
Total 20 651 . 412
a) y = 49.90-0.01045x
1
-0.0012x
2
-0.00324x
3
+ 0.292x
4
-3.855x
5
+ 0.1897x
6
where x1=cid x
2
=rhp x3=et x4=cm x5=axle x6=nlv
b) 8
2
= 4.965
se<0) = 19.67 , se(1) = 0. 02338 , se(
2
) = 0. 01631 , se(3) = 0. 0009459 ,
se(4) = 1.765 , se(5) = 1.329 8d se(6) = 0.273
c) y = 49.90-0.01045(215) -0.00 12(253) -0.0032( 4500) + 0.292(9.9) -3.855(3.07) + 0.1897(30.9)
= 29.867
The regression eqation is
y 47. 2 - 9 . 74 x1 " 0 . 428 x2 " 1 8 . 2 x3
Predictor Coef SE Coef T p
Constant 47 . 17 49 . 58 0 . 95 0 . 356
x1 -9 . 73 5 3 . 692 -2 . 64 0 . 018
x2 0 . 4283 0 . 2239 1. 91 0 . 074
x3 18 . 237 1 . 312 13 . 90 0 . 000
o 3 . 47963 R-Sq 99 . 4% R-Sq(adj) 99. 3%
Analysis of Variance
Source DF oo MS F p
Regression 3 30532 10177 840 . 55 0 . 000
Residual Error 16 194 12
Total 19 30725
12-3
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
12-11.
a) = 47.2 -9.74 x1 * 0.428 X*18.2 x3
b) 8
2
= 12
c) The estimated standard erors ofthe coefcient estitors are provided mthe above table fom Mtb (SE
Coe).
Because the regression coeficients have different standard erors the parameters estitors do not have
similar precision of estimation.
d) y = 47.17-9.735(14.5) + 0.4283(220) + 18.237(5) = 91.42
The regression equation is
rads = - 440 19 . 1 ms 68 . 1 exosure time
Predictor Coef SE Coef T p
Constant -440 . 3 9 94 . 20 -4 . 68 0 . 000
mAmps 19 . 147 3 . 460 5 . 53 0 . 000
exposure time 68 . 080 5 . 241 12 . 99 0 . 000
o = 23 5. 718 R-Sq = 84 . 3% R-Sq(adj) = 83 . 5%
Analysis of Variance
Source DF oo MS F p
Regression 2 11076473 5538237 99 . 67 0 . 000
Residual Error 37 2055837
Total 39 13132310
a) y = -440.39+ l 9.147x
1
+ 68.080x
2
55563
where x1 = mms x
2
= ExposureTe
b) 8
2
= 55563
se(f) = 94.20 , sec1) = 3.460 , and se(/
2
) = 5.241
c) .Y = -440.93+19.147(15)+68.080(5) = 186.675
12-13. The regression equation is
density = - 0 . 110 " 0 . 407 dielectric constant 2 . 11 loss factor
Predictor Coef SE Coef T p
Constant -0 . 1105 0 . 2501 -0 . 44 0 . 670
dielectric constant 0 . 4072 0 . 1682 2 . 42 0 . 042
loss factor 2 . 108 5 . 834 0 . 36 0 . 727
o = 0 . 00883422 R-Sq = 9 9 . 7% R-Sq(adj) = 99 . 7%
Analysis of Variance
Source DF SS
Regression 2 0 . 23563
Residual Error 8 0 . 00062
Total 10 0 . 23626
MS
0 . 11782
0 . 00008
a) y = -0.1105 + 0.4072x
1
+ 2.108x
2
where x1 = DielectcConst x
2
= LossFactor
12-4
F
1509 . 64
p
0 . 000
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
12-15.
b) 8
2
= 0.00008
se(/0) = 0.2501, se(/1) = 0.1682 , ad se(/
2
) = 5.834
c) y = -0.1105+0.4072(2.5)+ 2.108(0.03) = 0.97074
The regression eqation is
Useful range (ng) 239 " 0 . 334 Brightness (%)
Predictor Coef SE Coef T p
Constant 238 . 56 45. 23 5 . 27 0 . 002
Brightness (%) 0 . 3339 0 . 6763 0 . 49 0 . 639
Contrast (%) -2 . 7167 0 . 6887 -3. 94 0 . 008
o 36 . 3493 R-Sq 7 5 . 6% R-Sq(adj ) 6 7 . 4%
Analysis of Variance
Source DF oo MS F p
Regression 2 24518 12259 9 . 28 0 . 015
Residual Error 6 7928 1321
Total 8 32446
a) y= 238.56+0.3339x1-2.7167x
2
where x1 = %Brghtness x
2
= %Contast
b) 8
2
= 1321
c) se(/0) = 45.23, se(/1) = 0.6763, and se(/3
2
) = 0.6887
9
y = 238.56+ 0.3339(80)-2.7167(75) = 61.5195
- 2 . 72 Contrast
12-17. a) The results fom Minitab follow. The model ca be expressed as:
(%)
Rating Pts 2 . 99 1 . 2 0 Pet Comp 4 . 60 Pet TD - 3 . 81 Pet Int
Regression Analysis: Rating Pts versus Pet Comp, Pet TD, Pet lnt
The regression eqation is
Rating Pts 2 . 99 1 . 20 Pet Comp 4 . 60 Pet TD - 3 . 81 Pet Int
Predictor Coef SE Coef T p
Constant 2 . 986 5 . 877 0 . 51 0 . 615
Pet Comp 1 . 19857 0 . 09743 12 . 30 0 . 000
Pet TD 4 . 5956 0 . 3848 11 . 94 0 . 000
Pet Int -3. 8125 0 . 4861 -7 . 84 0 . 000
o 2 . 03479 R-Sq 9 5 . 3% R-Sq(adj ) 94 . 8%
Analysis of Variance
Source DF oo MS F p
Regression 3 2373. 59 791 . 20 191 . 09 0 . 000
Residual Error 28 115 . 93 4 . 14
Total 31 2489 . 52
Source DF Seq SS
Pet Com 1 1614 . 43
Pet TD 1 504 . 49
Pet Int 1 254 . 67
12-5
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
12-19.
n
b) 8
2
=

=
SS
E
=
115.93
= 4_14
nj nj 28
5.88
c)
A 2

2
A
2 0.097
cov(() = O (X A) = O C, se({) = C . =
l
0.38
fom the SE Coef column uthe Minitab outut.
0.48

Rg Pts = 2.99 1.20*60 4.60*4-3.81 *3 = 81.96

Predictor
Constant
Xl
X2
Coef
383 . 80
-3 . 6381
-0 . 11168
SE Coef 1
36 . 22 1 0 . 60
0 . 5665 -6 . 42
0 . 04338 -2 . 57
I
0 . 002
0 . 008
0 . 082
S 12 . 35 R-Sq 9 8 . 5%
R-Sq( adj) 97 . 5 %
Variance Analysis of
Source
Regression
Residual Error
Total
DF
2
3
5
SS
29787
458
30245
a) j=383.80-3.6381x1-0.1117x
2
MS
14894
153
F
9 7 . 59
b)
A
2 A A A
= 153.0, se((0) = 36.22, se((1) = 0.5665 , and se({
2
) = .04338
c)
9
y = 350.9943 -1.272(25)-0.1539(1000) = 165.29
Predictor Coef SE Coef 1
Constant 484 . 0 101.3 4 . 78
Xl -7 . 656 3 . 84 6 - 1. 9 9
X2 - 0 . 2221 0 . 1129 - 1 . 9 7
X1*X2 0 . 004087 0 . 003871 1 . 06
S 12. 12 R-Sq 9 9 . 0% R-Sq(adj)
Analysis of Variance
Source DF SS MS
I
0 . 041
0 . 185
0 . 188
0 . 402
97 . 6%
F
Regression 3 29951 . 4 9983 . 8 67 . 92
Residual Error 2 294 . 0 147 . 0
Total 5 30245 . 3
y = 484.0-7.656x1 -0.2221x
2
+ 0.004087x1x
2
I
0 . 002
I
0 . 015
e)
A
2 A A A A
= 147.0, se((0) = 101.3, se((1) = 3.846, se((2) = 0.113 and se((1
2
) = 0.0039

= 484.0 -7.656(25)-0.222(1000)-o.oo41(25X10oo) = 173.1

The predicted value is smaller.
12-6
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
Sections 12-2
12-21. a)
/
.
-(
.

=
' '
, null hyothesis (. = (

is rejected at l level if I

I
`

1
2
.
se(/ . )
! ! :P1
!
SS !k M
:

=
R
=
R
, regression is sigificant at l level U
/
`'
k
_
SSE /(n-) ME
B
The missing quantities are as follows:
Predictor Coef SE Coef T 1
Constant 253.81 4.781 53.0872 0
xi 2.7738 0.1846 15.02 0
X -3.5753 0.1526 -23.4292 0
Source DF bb MS F
Regression 2 22784 11392 445.2899
Residual Error 12 307 25.5833
Total 14 23091
R-Squared = 22784/23091 = 0.9867
b) From the P-value fom the F test (F = 445.2899) for regression is signifcant.
c) Each individual regressor is significant to the model that contains the other regessors.
12-23. a) n = lO,k = 2, Q= 3,
|
.

=

= (
k
= 0
|,
(
j
"0 for at least one j
(= 0.05
2
s = 371595.6-
(1916)
= 4490

10
1916
/'X'y=i71.055 3.713 -1.126 43550.8 =371511.9
104736.8
SS
R
= 371511.9-
1916
2
=4406.3
10
ssE= s

-ss
R
= 4490-4406.3 = 83.7
55_
_
= k =
4406.3/ 2
= 184.25

55
_
83.7/7
B1
'.. -
= 4

74
'. `'. .
i
0
Rejet H0 and conclude that the regression model is signifcant at a = 0.05. P-value = 0.000
12-7
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
b)
s
s
5 = M
E
=
E
= 11.957
,

s-(
,
)= -

.=

11.957(0.00439) = 0.229
s-|

)= '

.=

11.957(0.00087) = 0.10199
|
.
,
= 0
|
.
,
"0

.-(
)
=
3

713
= 16.21
0.229
,,-
=

-
= 2
3
65
Rect H 0, P-value ~0.00 1.

.-(

)
=
-1.126
= -11.04
0.10199
Rejet H0, P-value ~0.001.
Both regession coeficients sigifcant.
12-25. a)
,
= 4.82
/

= 8.21
,
= 0.98
b)
| ,
= 0
|
.
,
"0
a =0.05

= 0.98

,
= 2

074

.'

..
P-value = 2(4.08 E-5) = 8.16 E-5
P-value = 291.91 E-8) = 3.82 E-8
P-value = 2 (0.1689) = 0.3378
Do not reject
|
Here .
,
does not contibute significantly to the model.
12-27. a)
|
.!
=

=
,
=

.
= 0
|,
: at least one

" 0
/
= 19.53
/
, ..
=
/

.,,

= 2

848
'. `'.. .
,

Rejet H0 and conclude regession model is sigicant at a = 0.05.

b) The t-test statistics for through 6 are -0.45,-0.07,-3.42, 0.17, -2.90, 0.69. Because l: 02514= -2.14,
the regessors that contibute to the model at a= 0.05 are -and oIe.
12-29. a)
| /,
=
/

= /
,
= 0
|
.

"0
/
= 828.31
'.
, ,
.
= 3

34
'. `'.. ,
.
for at least one j
Rejet H0 and conclude regression is signifcant at a = 0.05.
12-8
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
b)

O 12.2856
a =0.05
,, .
-
,

..
212
|
.
,
0

0
,
0
|, ,
"0

"0
,
"0

-2.58

1.84

13.82

.`..
,
.
.
..
,
.

.`..
,
.
Rejet H0 Do not reject H 0. Reject H0.
12-31. a)
| ,

0
b)
|
for at least one

"0
a = 0.05

99.67

,-
3

252
'. `'..
,-
The regression equation is
rads - 440 19 . 1 mAmps 68 . 1 exposure time
Predictor Coef SE Coef T p
Constant -440 . 3 9 94 . 20 -4 . 68 0 . 000
mAmps 19 . 147 3 . 460 5 . 53 0 . 000
exosure time 68. 080 5 . 241 12 . 99 0 . 000
o 235 . 718 R-Sq 84 . 3% R-Sq(adj) 83 . 5%
Analysis of Variance
Source DF SS
Regression 2 11076473
Residual Error 37 2055837
Total 39 13132310
MS
5538237
55563
F
99 . 67
p
0 . 000
Reject H0 and conclude regession model is sigificant at a = 0.05. P-value ~0.000001
o

M 55563

s-(
,
)
o_ 3.460
|
.
,
0
|, ,
"0
a = 0.05

s-(
,
)

19.147
5.539
3.460

-
,

,-
20262

`,
,
,-
,Reject
|
and conclude that
,
is significant in the model at a = 0.05.

s-(

)
_ 5.241
12-9
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
12-33.
|
.

= 0
|
.

0
a= 0.05
,

=
---

.-(
'

)
=
68.080
= 12.99
5.241
'

-
,
=
'

-
= 20262
I

I
> ,-
,Reject
|
conclude that
'

is significant uthe model at a = 0.05.

Useful range (ng) 239 " 0 . 334 Brightness (%)
Predictor Coef SE Coef T p
Constant 238 . 56 45 . 23 5 . 2 7 0 . 002
Brightness (%) 0 . 3339 0 . 6763 0 . 49 0 . 639
Contrast (%) -2 . 7167 0 . 6887 -3 . 94 0 . 008
o 3 6 . 3493 R-Sq 75 . 6% R-Sq(adj) 67 . 4%
Analysis of Variance
Source
Regression
Residual Error
Total
a)
|
.!
=

= 0
DF SS
2 24518
6 7928
8 32446
|,
: for at least one

~ 0
a= 0.05
55_

=
k
=
2451 8
/
2
=9.28

55
_
7928/6
n]

.
= 5

14
'.
>
'...
MS
12259
1321
F
9 . 2 8
p
0 . 015
- 2 . 72 Contrast (%)
b)
Reject H0 and conclude the regession model is sigcant at a = 0.05 P-value = 0.015.
,

= M = 1321

.-(
,
)=

_ = 0.6763
|
.
,
=
"
|
.
,

0
a =0.05
,

=
---

.-(
,
)
=
0.3339
= 0.49
0.6763
'

-
,
=
'

.
= 2.447
12-10
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
12-35. a)
b)
I

<

-
_ , Fail to reject
|

there is no enough evidence to conclude that

,
is significant in the model at O
= 0.05.

s-('

)=

_ = 0.6887
|
.

= 0
|

"0
D = 0.05
=

.
.
s-('

)
=
-2.7167
=-3.94
0.6887

-
,
=

,
= 2.447

-
_ , Reject
|
. conclude U

is significant in the model at O 0.05.

The Minitab output follows. The test statistic is F 191.09. Because the P-value is near zero, the regressionis
significant at a = 0.05 .

'
.
.

=
' , ' _ null hypothesis '= '

is rejected at ! level if I
`
,

or the P-value ~ a
s-|') ' '
1 -n]
!
The P-values of 8regessors are less th.an 0.05. Therefore, 8 individual variables in the model are
sigifcant.
a) The Minitab output for three regressors is foiJowed by the Minitab outut for two regressors. From the
regression sum of squares in each model the F test for _ is
:
.
=
SS
_
|,'
)/r
=
2373.59-1782.96
= 142.66
M
E
4.14
The F-test P-value is near zero. Therefore the regressor (TD percentage) is significant to the model. This is
the equivalent to the t test on the coeficient of _ The F statistic 142.66 11.94
_
except for some round
offerror.
Minitab results of regression on three variables and on two variables are shown below.
Regression Analysis: Rating Pts versus Pet Comp, Pet TO, Pet lnt
The regression equation is
Rating Pts 2 . 99 " 1 . 20 Pet Com " 4 . 60 Pet TD - 3 . 81 Pet Int
Predictor Coef SE Coef T p
Constant 2 . 986 5 . 877 0 . 51 0 . 615
Pet Comp 1 . 19857 0 . 09743 12 . 3 0 0 . 000
Pet TD 4 . 5956 0 . 3848 11 . 94 0 . 000
Pet Int -3 . 8125 0 . 4861 -7 . 84 0 . 000
o 2 . 03479 R-Sq 9 5 . 3 % R-Sq(adj) 94 . 8%
Analysis of Variance
Source DF oo MS F p
Regression 3 2373 . 59 791 . 20 191 . 09 0 . 000
Residual Error 28 115 . 93 4 . 14
Total 31 2489 . 52
12-ll
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
Source DF Seq SS
Pet Com 1 1614 . 43
Pet TD 1 504 . 49
Pet Int 1 254 . 67
Unusual Observations
Obs Pet Comp Rating Pts Fit SE Fit Residual St Resid
11 61 . 1 87 . 700 83 . 691 0 . 371 4 . 009 2 . 00R
18 59 . 4 84 . 700 79 . 668 0 . 430 5 . 032 2 . 53R
21 65 . 7 81 . 000 8 5 . 020 1 . 028 -4 . 020 -2 . 29R
31 59 . 4 7 0 . 000 7 5 . 141 0 . 719 -5 . 141 -2 . 70R
R denotes an observation with a large standardized residual.
Regession Analysis: Rating Pts versus Pet Comp, Pet lot
The regression equation is
Rating Pts - 9 . 1 " 1 . 66 Pet Comp - 3 . 08 Pet Int
Predictor Coef SE Coef
Constant -9 . 11 14 . 04
Pet Com 1 . 6622 0 . 2168
Pet Int -3 . 076 1 . 170
s 4 . 93600 R-Sq 71 . 6%
Analysis of Variance
Source
Regression
Residual Error
Total
Source DF
Pet Com 1
Pet Int 1
DF ss
2 1782 . 96
29 706 . 56
31 2489 . 52
Seq SS
1614 . 43
168 . 53
Unusual Observations
Obs Pet Comp Rating Pts
1 65 . 3 105 . 500
28 59 . 1 73 . 700
31 59 . 4 7 0 . 000
T p
-0 . 65 0 . 522
7 . 67 0 . 000
-2 . 63 0 . 014
R-Sq( adj) 69 . 7%
M F p
891 . 48 3 6 . 59 0 . 000
2 4. 3 6
Fit SE Fit Residual
92 . 355 1 . 247 13 . 145
73 . 744 2 . 977 -0 . 044
82 . 241 0 . 980 -12 . 241
St Resid
2 . 75R
-0 . 01 X
-2 . 53R
R denotes an observation with a large standardized residual.
X denotes an observation whose X value
12-37. )=
+
j
,
.
,
+

Assume no interaction model.

a) H0: P
1
P
2
0
H
1
at least one '

~ 0

= 97.59
'..
,
= 9.55
'. `'..
,
Rejet H0 P-value 0.002
12-12
gives it large influence.
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
b)
|
=0
|, ,

= -6.42

,
=

,
= 3
182
'.`

. .,
Reject H0 for regressor
,

c)
,
|
,

)= 1012
|
.

= 0
|
,

0
(= 0.05
'

= 6.629
',
=
'.,
= 1 0

13
'.'...
Do not reject H0
|
.
,
=

=
,
= 0
Hat least one

~ 0
!
a= 0.05
/
= 7.714
'
=
'.
.,
= 19 . l 6
'. '..,
Do not reject H0.
e)
| ,
= 0
|
, ,

0
a= 0.05
|
.

= 0
|,

= -2.57

,
=

,
= 3
182
I
.
'
. .,
Do not reject H0 for regressor

(
,
,
)= 29951.4-29787 = 163. 9
_
=

=
163.9
= 1 . 1 1

M
E
147
/

.
= 1851
'.'..
Do not reject H0
f -

= 147.0
,

(no interaction term) = 153.0

M
E
( -

) was reduced in the model with the interacton ter.

12-13
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
12-39. a) The Mini tab outut follows. The P-value for the F-test is near zero. Therefore, the regession is
b)
significant at both C0.05 or C0.01.
'_

. Because the P-values for Age and Severity are ~0.05 both regessors are significant to the
s-
|)
!
model. Because the P-value for Aiety is 0.233, it is not sigifcant to the model at level C 0.05.
Minitb result:
Regression Analysis: Satisfaction versus Age, Severity, Anxiety
The regression equation is
Satisfaction 144 - 1 . 11 Age - 0 . 585 Severity 1 . 30 Aiety
Predictor Coef SE Coef T p
Constant 143 . 895 5 . 898 2 4. 40 0 . 000
Age -1 . 1135 0 . 1326 -8 . 40 0 . 000
Severity -0 . 5849 0 . 1320 -4 . 43 0 . 000
Anxiety 1 . 296 1 . 056 1 . 23 0 . 233
s 7 . 03710 R-Sq 9 0 . 4% R-Sq(adj) 89 . 0%
Analysis of Variance
Source DF ss M F p
Regression 3 973 8 . 3 3246 . 1 65 . 55 0 . 000
Residual Error 21 103 9 . 9
Total 24 10778 . 2
Source
Age
Severity
Anxiety
DF
1
1
1
Seq SS
8756 . 7
907 . 0
74 . 6
Unusual Observations
Obs Age Satisfaction
9 27 . 0 7 5 . 00
Fit
93 . 28
49 . 5
SE Fit
2 . 98
Residual
-18 . 28
St Resid
-2 . 87R
Sections 12-3 and 12-4
.
|
12-41. a)

,.
-

171.055

_ s-(,
)
171.055 (2.365X5L217)
171.055 121.128
49.927 _

_292.183

, .
-
\
.:.+

-
s-(+
,
)
3.713 (2.365Xl.556)
3.7133.680
0.033::
i
::7.393

|
'

, .
-
\
12-14
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
-1.126

_ se(/_)
-1.126 (2.365X0.693)
-1.126 1.639
-2.7655/
2
50.513
b) .
,
= 18
.

= 43
;

= 189.471
x(xx)
-
,
x

= 0.305065
189 .471 (2.365)5-0- .7- 87-5- (0- .3-0 - 50-6-5)
158.815 5
..
5 220.127
c) O 0.05
.
,
= 18
.

= 43
;

= 189.471
x(xx)
-
,
x

= 0.305065
;

, ,.
-
,
'

(
'
+ x(x x,
,
x
)
189.411 (2.365N 550. 7875(1.305065)
126.064 ,

252.878
12-43. a) -20.477
/
,
51.269
-0.245
/

51.076
14.428
/
,
522.159
b)
..
= 91.372 s-(;
)
= 4.721
91.372 2.921(4.721)
c)
77.582 ,

5105.162
'r
=91.372 .-(,
.
)=3.163

91.372 (2.921X3.163)
82.133 5
..
5100.611
12-45. a) -6.9467
/
,
-0.3295
-0.36515/3
2
50.1417
b) -45.8276
/
,
530.5156
-1.3426
/

0.8984
-0.03433
/
,
50.04251
These part b. intervals are much wider.

.,
= 2.921
Yes, the addition of this term increased the standard eror of the regression coeficient estimators.
12-15
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
12-47. a)
b)
c)
12-49. a)
b)
c)
.

/
1

a/
2
,n
-p '
19.147
i_
,

se(6

)
19.147 (2.0262X3.460)
19.147 7.014458
12.1363:/
1
: 26.1577

0,
i
a/
2
,n
-p
68.080
i_
,

se(,
,
)
68.080 (2.0262X5.241)
68.080 7.014458
57.4607:0
,
:78.6993
'r.
=-85.1

!
i
". a/
2
,n
-p
se
||,.
)
-85.1 (2.7154X54.6)
-233.4:
.
:63.2
)_= -85.1
-85.1 2.7154(241.95)
-742.09:,__571.89

.
_

= 1.860
-0.576 ::0::0.355
0.0943: /
1
:0.7201
-8.743: 0,: 12.959
'r.
=0.8787
!
i se
||
)
. a/
2
,n
-p ,
.
0.8787 (l .860X0.00926)
0.86148:
.
:0.89592
,_= 0.8787
0.8787 1.86(0.0134)
0.85490: ,_:0.90250
12-51. The regression equation is
'.
_
.
,

= 2

7154
se(,_)= 241.95
se(

)= 0.00926
.

.
_
.
,

= 1.860
Useful range (ng) 239 " 0 . 334 Brightness (%) - 2 . 72 Contrast ( %)
Predictor Coef SE Coef T P
Constant 238 . 56 45 . 23 5 . 27 0 . 002
Brightness (%)
Contrast (%)
s 3 6 . 3493
0 . 3339 0 . 6763 0 . 49 0 . 639
-2 . 7167 0 . 6887 -3 . 94 0 . 008
R-Sq 75 . 6% R-Sq( adj) 6 7 . 4%
Analysis of Variance
Source DF ss MS F p
Regression 2 24518 12259 9 . 28 0 . 015
Residual Error 6 7928 1321
Total 8 32446
12-16
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
a)
i_

__
,s
= 3.707
-2.173: [__2.841
-5.270: (,
_-0.164

Predicted Values for New Observations

New
Obs Fit
1 44 . 6
SE Fit
21 . 9
99% CI
(-3 6 . 7 , 125 . 8)
99% PI
(-112 . 8 , 202 . 0)
Values of Predictors for New Observations
New Contrast
Obs Brightness (%) (%)
1 70 . 0 80 . 0
'.

i
-
,
-]
se
|
,
44.6 (3.707X21.9)
-36.7:;.
,.
: 125.8
.
c) )_= 44.6
44.6 3.707(42.44)
-112.8: ,__202.0
i
_

__
,s
= 3.707
d) Predicted Values for New Obserations
New
Obs Fit
1 187 . 3
SE Fit
2 1 . 6
99% CI
(107 . 4 , 267 . 2)
99% PI
(30 . 7 , 344 . 0)
Values of Predictors for New Observations
New Contrast
Obs Brightness (%) (%)
1 5 0 . 0 2 5 . 0
CI: 107.4:
.
_267.2
PI: 30.7: ,__344.0
These intervals are wider because the regessors are set at exteme values in the x space and the standard
errors are geater.
12-53. a) The Minitab outut follows. The outut is used to obtain estimates of the coeficients and standard errors. The
confdence intervals for the coeficients are computed from 6
i
_

_
,
,
,
se
|
6):, _6+
i
__
,
,
,
se(6).
From the t table, t
00228
2. 048. The condence intervals for the , 's are
-9.052 15.024
0.999 ___ 1.398
3.807 5.384
-4.808 -2.817
12-17
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
81. 85-2.048(0.43) : ]_
j
_ : 81. 85 + 2.048(0.43)

and this approxately equals (because of round of the confidence interval u( 8 0 . 9 6 5 , 82 . 72 5) the
Mtb outut.
Regression Analysis: Rating Pts versus Pet Comp, Pet TD, Pet lnt
The regression equation is
Rating Pts 2 . 99 1 . 20 Pet Cor " 4 . 60 Pet TD - 3 . 81 Pet Int
Predictor Coef SE Coef
Constant 2 . 986 5 . 877
Pet Cor 1 . 19857 0 . 09743
Pet TD 4 . 5956 0 . 3848
Pet Int -3 . 8125 0 . 4861
o 2 . 03479 R-Sq 9 5 . 3 %
Analysis of Variance
Source
Regression
Residual Error
Total
Source DF
Pet Cor 1
Pet TD 1
Pet Int 1
DF oo
3 2373 . 59
28 115 . 93
31 2489 . 52
Seq SS
1614 . 43
504. 49
254 . 67
Unusual Observations
Obs Pet Comp Rating Pts
11 61 . 1 87 . 700
18 59 . 4 84 . 700
21 65 . 7 81 . 000
31 59 . 4 70 . 000
T p
0 . 51 0 . 615
12 . 3 0 0 . 000
11. 94 0 . 000
-7 . 84 0 . 000
R-Sq(adj) 94 . 8%
MS F p
791 . 20 191 . 09 0 . 000
4 . 14
Fit SE Fit Residual St Resid
83 . 691 0 . 371 4 . 009 2 . 00R
7 9 . 668 0 . 430 5 . 032 2 . 53R
85 . 020 1 . 028 -4 . 020 -2 . 29R
7 5 . 141 0 . 719 -5 . 141 -2 . 70R
R denotes an observation with a large standardized residual.
Predicted Values for New Observations
New
Obs Fit
1 81 . 845
SE Fit
0 . 430
95% CI
( 80 . 965, 82 . 725)
12-18
95% PI
(77 . 585, 86. 104)
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
12-55. a) t0514=2.977
-8.658 : /3
0
: 108.458
-0.08 : /3
2
: 0.059
-0.05 : /
3
: 0.047
-0.006 : /
7
: 0
-4.962 : /
8
: 5.546
-7.811 : /9 : 0.101
1.102 : /
1
0
: -0.523
''
|,
.
= 29.71
',.

...
se(
|r.
29.71 (2.977XI.395)
25.557 :
,

.
: 33.863
c) y = 61.001 -0.0208x
2
-0.0035x7-3.457x9
t
00
5
,
1
7
= 2.898
50.855 : /
0
: 71.147
-0.036 : /
2
: -0.006
-0.006 : /7 : -0.001
-6.485 : /39 : -0.429
a) The intervals upart c) are narrower. Al of the regressors used upart c) are signifcant, but not 8 of those
used mpart a) are significant. The model used in part c) is preferable.
Section 12-5
12-57. a) The regression eqation is
rg 49 . 9 - 0 . 0104 cid - 0 . 0012 rhp - 0 . 00324 etw " 0 . 29 c - 3 . 86 axle
0 . 190 n/v
Predictor Coef SE Coef T p
Constant 49 . 90 19 . 67 2 . 54 0 . 024
cid -0 . 01045 0 . 02338 -0 . 45 0 . 662
rhp -0 . 00120 0 . 01631 -0 . 07 0 . 942
etw -0 . 0032364 0 . 0009459 -3 . 42 0 . 004
cr 0 . 292 1. 765 0 . 17 0 . 871
axle -3 . 855 1 . 329 -2 . 90 0 . 012
n/v 0 . 1897 0 . 2730 0 . 69 0 . 498
S 2 . 22830 R-Sq 8 9 . 3 % R-Sq(adj) 84 . 8%
12-19
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
b) There appears to be a outlier. Otherwise, the normality assumption is not violated.
Normal Probability Plot of the Residuals
(reis!
.
-

u
-!

-3 -2 -1
Stndardized Residual
c) The plots do not show mviolations of the assumptions.
Residuals Versus the Fitted Values Residuals Versus cid
(reis! (re is m

-1

-1

2 -2
-3

a 2 3 3 10 20 D
Fitd Value d
Residuals Versus etw Residuals Versus cmp
(ree is m (reism

..

-1

-1

2
-3

3 3 9 H qo q> q qJ 1QO
et cm
Residuals Versus axle
Residuals Versus n/v
(.is m
{re is m

-1

-!

-2
-2
-3
3

29 Z7 30 32 3! 37 40 42
a 2 3 3 4
axle
n/v
12-20
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
d) 0.036216, 0.000627, 0.041684, 0.008518, 0. 026788, 0.040384, 0.003136,
0.196794, 0.267746, 0.000659, 0.075126, 0. 000690, 0. 041624, 0.070352,
0.008565, 0.051335, 0.001813, 0. 019352, 0.000812, 0.098405, 0.574353
None of the values is greater t I so none of the observations are influential.
12-59. a) The Mntab outut follows. The proportion oftotal variability explained by this model is:
R
2
=

_
=
2373.59
= 0.95

I
2489.52
b) Normal Probability Plot: Some moderate, but not severe, departures fom normality are indicated.

5.0
Normal Probabil ityPlot of V6 Residuals
(reeisRatng Pz)
.z.5 0.0
Residual

/.5
c) Plot the residuas versus fitted value and versus each regressor.
0
There is no obvious model failure in the plot of fitted values versus residuals. There is a modest increase in
variabilit in the middle range of fitted values. The residual versus PctTD shows some non-random patters.
Possibly a non-linear ter would benefit the model.
^
?

?
^
7
se
,s
eesccasVersusthe Fitted Values
{pqisRatrg[)

=
o
.
=
o

d 9u
RtedValue
Residuals Versus Pet TO
{~pis Ratg v)
.
.
! 1!

.
o
.
o o
.
,s
se

Pet
12-21
=
eesduasverscsPet Corr
(restska gj)

_
.
eo
-----
.

.
--
.

.
^
9 5
s

,s
.o

,s
se{
% 3
.
o
S
PctC
Residuals Versus Pet I nt
{~pis Ragv)

oo
o
o
.
.

o o
Pctlnt
G % 6 O
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.

A plot of Cook's distance measures follows. Although no points exceed the usual criterion of distance greater
than 1, two points are different and might be fer studied for influence.
Time Series Plot of COOKl
.1
.
_
.Z
.1
+Z +O + Z+ Z Z! 3
Index
Regression Analysis: Rating Pts versus Pet Comp, Pet ,Pet lnt
The regression eqation is
Rating Pts 2 . 99 1 . 2 0 Pet Cor " 4 . 60 Pet TD - 3 . 81 Pet Int
Predictor Coef SE Coef T p
Constant 2 . 986 5 . 877 0 . 51 0 . 615
Pet Comp 1 . 19857 0 . 09743 12 . 3 0 0 . 000
Pet TD 4 . 5956 0 . 3848 11 . 94 0 . 000
Pet Int -3 . 8125 0 . 4861 -7 . 84 0 . 000
o 2 . 03479 R-Sq 9 5 . 3 % R-Sq(adj) 9 4. 8%
Analysis of Variance
Source DF oo MS F p
Regression 3 2373 . 59 791 . 20 191 . 09 0 . 000
Residual Error 28 115 . 93 4 . 14
Total 31 2489 . 52
Source DF Seq SS
Pet Cor 1 1614 . 43
Pet TD 1 504 . 49
Pet Int 1 254 . 67
Unusual Obserations
Obs Pet Cor Rating Pts Fit SE Fit Residual St Resid
11 61 . 1 87. 700 83 . 691 0 . 371 4 . 009 2 . 00R
18 5 9 . 4 84 . 700 79. 668 0 . 430 5 . 032 2 . 53R
21 6 5 . 7 81. 000 85 . 020 1 . 028 -4 . 020 -2 . 29R
31 5 9 . 4 7 0 . 000 75 . 141 0 . 719 -5 . 141 -2 . 70R
R denotes an obseration with a large standardized residual.
12-22
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
12-61. a) R
,
= 84.3%
b) Assumption of noralit appears adequate.
Normal Probability Plot of the Residuals
(responsis rad)

9
.
1 0

-5u
J

^
-z9

u ?
Residual

5u 75u
c) There are funnel shapes in the graphs, so the assumption of constant variance is violated. The model is inadequate.
79
9
z9
.
u
.z9
.Su
d)
Residuals Versus the Fitted Values
(resis rads)
7
9

u

"

"
.
e

"

@
+
e +

.z9
.9
-U
Residuals Versus exposure time
(resis rads)

*+

~+

" +

m+

v
+

u 1u 15
exposure time
Cook's distance values
0.032728 0.029489
0.077299 0.3436
0.001985 0.002068
0.000643 0.000375
0.006095 0.005442
0.015557 0.077846
0.020725 0.021539
u

5u

1u
ftted Value
l
7U

u

-Z

-U
2u 1u
0.023724 0.014663
0. 008489 0.007592
0.021386 0.105059
0.0002 0.000209
0.0043 0.002564
0.07828 0.070853
0.177299 0.731526

No, none of the observations Da Cook's distance geater U1.

12-23
15u
f

1S
Residuals Versus mAmps
(resis rads)

2u
0.008279
0.006018
0.000926
0.002467
0.0014
0.057512
2S
mm s

3
0.008611
0.003612
0.000823
0.013062
0.001459
0.036157
35

~
4u
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
12-63. a) R
,
= 99. 7 %
U.U1U
U.U5
G

-U.U5
-U.U1U
b) Assumption of noralit appears adequate.
Normal Probability Plot of the Residuals
(resposis det)
!
9

>

+-

-
'
10

1
-U.UZ -U.U1 U.U U.U1
Residual
U.UZ
c) There is a non-constant variance shown ugraphs. Therefore, the model is inadequate.
U.U1U
U.U5
Residual s Versus the Fitted Values
(resis o-vp)

-U.U5
-U.U1U
U.7

U.8 U.S 1.U 1.1
Ftted Value

1.Z
Residuals Versus dielectric constant Residuals Versus loss factor
(resois det)
U.U1U

U.U5

_.

-U.U5

-U.U1U

Z.U Z.Z Z.4 Z.t Z. 3.U U.U1o U.UU

dielectric constant
d) Cook's distance values
0.255007
0.10971
0.692448 0.008618 0.011784 0.058551 0.077203
0.287682 0.001337 0.054084 0.485253
No, none of the observations Da Cook's distance geater t 1.
12-24
(reois t)

U.U5 U.UU U.U3o

loss factor

U.UU U.Oo
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
12-65. a) R
2
= 75.6 %
b) Assumption of noralit appears adequate.
Normal Probability Plot of the Residuals
(resis Usefl rang(ng))

9
.

-8D

-tU -4

W -

-ZD U ZD
Residual

A 6 U
c) Assumption of constant variance is a possible concer. One point is a concer as a possible outlier.

R
"
7
`
Z

-Z
-R
D

"
R
..
"
7
`
m
" Z

-Z

-R
Z O
Residuals Versus the Fitted Values
(response is Useful range (ng))

1D 1 1ZD
Fitted Value
Residuals Versus Contrast (%)
(response is Useful range (ng))

R D
Contrast (%)
12-25

1D 1!D Z

!
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.

z
-
7
v
"
o
Z

-Z

--
D
d Cook's distance values
0.006827
0.154825

Residuals Versus Brightness (%)

(respse is Useful range (ng))

0.032075
0.220637

1
Brightness (%)
0.045342 0.213024 0. 000075
0.030276 0.859916

1
No, none of the observations has a Cook's distance greater than 1.
12-67. a) R
2
= 0.985
b) R
2
= 0.99
increases with addition of interaction term. No, adding additional regressor will always increase r2
12-69. a) There is some indication of nonconstant variance since the residuals appear to "fan ouf' with increasing
values ofy.
Residual Plotfor y
8

<
m

-!

4
~H
7

0 40 80 !20 160 200 240
Predicted
12-26
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
b) Source
Model
Error
Sum of Squares
30531 . 5
193. 725
30725 . 2
DF
3
16
19
Mean Square
1017 7 . 2
12. 1078
F-Ratio
840. 546
Pvalue
. 0000
c)
d
Total (Corr. )
R-squared = 0 . 993695 Stnd. error of est .
R-squared (Adj . for d . f . ) = 0 . 992513 Durbin-Watson statistic
3. 47963
1 . 77758
R
2
= 0.9937 or 99.37 %;

= 0.9925 or 99.25%;
Independent variable
CONSTANT
xl
x2
x3
Model fitting results for : log( y)
coefficient
6 . 22489
-0 . 16647
- 0 . 000228
0 . 157312
std. error
1 . 124522
0 . 083727
0 . 005079
0 . 029752
t-value
5 . 5356
- 1 . 9882
- 0 . 0448
5 . 2875
sig. level
0 . 0000
0 . 0642
0 . 9648
0 . 0001
R-SQ . (ADJ. ) = 0 . 9574 SE= 0 . 078919 MAE= 0 . 053775 DurbWat= 2 . 031
Previously: 0 . 0000 0 . 000000 0 . 000000 0 . 000
20 observations fitted, forecast ( s) computed for 0 missing val . of dep . var .

= 6.22489 -0.16647x
1
-0.000228x
2
+0.157312x
3
. 1
. b

U
O

0
O
. b
. 1

. 1 b
4
8es|dua' |'orfo'og(y)

4. O 4. o 4. J
|ed|cred
Y1o1 cXb1Dt8 cutVa1utc

b. Z

b. b
There is curvature uthe plot. The plot does not give much more infortion as to which model is preferable.
12-27
useroame. George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
e)

Residual |'ot or|og(y)

U. 1

U.Ub

U
m
I
7
v
0
o
U.Ub

U.T

U. T b

3. 3 b. 3 7. 3 9. 3 T T . 3
x3
Y1o1 cXb1Dt8 cutVa1utc
Variance does not appear constant. Crvature is evident.
Independent variable
CONSTANT
x1
x2
1/x3
Model fitting results for : log( y)
coefficient std. error
6 . 222045 0 . 547157
- 0 . 198597 0 . 034022
0 . 009724 0 . 001864
-4 . 436229 0 . 351293
t-value
11 . 3 716
-5 . 8374
5 . 2180
-12 . 6283
sig . level
0 . 0000
0 . 0000
0 . 0001
0 . 0000
R-SQ . (ADJ . ) 0 . 9893 SE= 0 . 039499 MAE= 0 . 028896 DurbWat= 1 . 869
Source
Model
Error
Analysis of Variance
Sum of Squares
2 . 75054
0 . 0249631
2 . 77550
for the
DF
3
16
19
Full Regression
Mean Square F-Ratio
0 . 916847 587 . 649
0 . 00156020
P-value
. 0000
Total (Corr. )
R-squared = 0 . 991006
R-squared (Adj . for d . f . ) = 0 . 98932
Stnd. error of est . = 0 . 0394993
Durbin-Watson statistic = 1 . 86891
L.LU
!
L.L
D
T

L.L<
U
D
~L.LT
~L.L
~L.L
S. U
LSIrg T7S
HeSIOOO| IIC1 1Cr ICg(j)
. T . .
IreOIC1eO
. S
The residual plot indicates better conforance to assumptions.
Crvature is removed when using llx3 as the regressor instead of and the log of the response data.
12-28
useroame. George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
12-71. a) p = k
+
1 = 2 + 1 = 3
Average size = pin= 3/25 = 0.12
b) Leverage point criteria:
Sections 12-6
hu >2(p/n)
h
ii
>2(0.12)
h
ii
>0.24
h
l
7
,
1
7 = 0.2593
h
1
8,
1
8
= 0.2929
Points 17 ad 18 are leverage point.
12-73. a) y = -1.633 + 1 .232x -1 .495x
2
b) /0 = 1858613, reject H0
c) t0 = -601.64, reject
|
d Model is acceptable. Observation nuber 10 has large leverage.
Normal |rObaDi\j O\ O! the Residuals
(response |s y)
1

c
L
U
C
L

.
c
2
1
-
u.uu4
u.uu
u.uu4

_ u.uuz
`V

u.uuz

u.uuu u.uuz
KeS|OuaI
Residuals Versus 7
(response s y)

u.uu4

u.uu6

.U@

u.uuz|

u.uu
u.u u. 1 . 1 .o z.u z.
7
12-29
useroame. George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
12-75.
.
o.oo.

o.ooz-
`
C
C
.U
-o.ooz-
-u.uu
a) Predictor
Constant
xl
x2
x3
x1x2
x1x3
x2x3
x1A2
x2A2
x3A3
S
I
0 . 06092
Residuals Versus the Fitted Values

1 1
-8 -7
Coef
- 1 . 769
0 . 4208
0 . 2225
-0 . 12800
- 0 . 01988
0 . 009151
0 . 002576
- 0 . 01932
- 0 . 00745
0 . 000824
R-Sq

(response is y)

1
-6
1
-5

1
-4
Fitted Value
SE Coef T
1 . 287 - 1. 3 7
0 . 2942 1. 43
0 . 1307 1. 70
0 . 07025 -1 . 82
0 . 01204 -1 . 65
0 . 007621 1 . 2 0
0 . 007039 0 . 37
0 . 01680 -1 . 15
0 . 01205 -0 . 62
0 . 001441 0 . 57
9 1 . 7 % R-Sq( adj )
I

1
-3

I
0 . 188
0 . 172
0 . 108
0 . 087
0 . 118
0 . 247
0 . 719
0 . 267
0 . 545
0 . 575
87 . 0%

1
-2
Analysis of
Source
Regression
Residual Error
Total
Variance
1
9
16
25
SS
0 . 655671
0 . 059386
0 . 715057
MS
0 . 072852
0 . 003712

19 . 63
y -1.769+0.42lx
1
+ 0.222x
2
-0. 128x
3
-0.02x
12
+ 0.009x
13
+0.003x
23
-
0.019x-0.007 x+ 0.00 1x
b) H
o /,
/
2
/
3

/
23
O
H, at least one ["0
'

1 9.628
!.9
1
n
2

54
'. `'....
Rect H0 and conclude the model is significant at D 0.05
12-30

1
-1

0 . 000
useroame. George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
c) Model is acceptable.

L
u

L
e
C
2

-Z
U.1U

:e
V
C
O
NC|| |CDD|||\j ||C1 C!\C KC5|Ou|5
(response is y)
+
-U.U5
J
1
+

1
+
+

+
U.UU U.U5
ReS|dua
j
KC5|dU|5 VC|5U5 \C|\\CO V|uC5
(response is y)
-+
U. 1 U
:eW
-U.U5
U.U U.1
+ -

d
|
.
,
.
=

=
,
=
,
=
,
=

= 0
H, at least one

"0
..
,
.
= 2

74
'. '..
,
.
Do not reject H0
ss ,
|'
.
, ,,)
U.Z C.3
F|ttedVa
j
ue
U.A

--= 1 . 612
0.003712
U.5
=
,
|
,

)
,
|'
.'
)
= 0.65567068 -0.619763
= 0.0359
Reduced Model: =

+
,
.
,
+

+
,
.
,
12-77. a) Use indicator variable for tansmission type.
There are tree possible tansmission types: 1,L5 and M6. So, two indicator variables could be used where
x3 1 ifts L5, 0 otherwise and x4 1 ifts M6, 0 otherwise.
b) y = 56.677 -0.1457x
1
-0.00525x
2
-0. 138x
3
-4. 179x
4
12-31
useroame. George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
c) The P-value for testing
|
.[
+
0 is 0.919 which is not significant. However, the P-value for testing
|
.[

0 is 0.02 which is significant for values of D`0.02. Thus, it appears that whether or not the
transmission is manual affects mpg, but there is not a significant diference between the types of automatic
transmission.
12-79. a) The u Cpequation is: x1, x2
C_ 3.0 and M
E
55563.92
y
-440.39+ 19.147x
1
+ 68.080x
2
The u M
E
equation is the same as the D Cp.
b) Same as equation in par (a).
c) Same as equation in part (a).
9
Same as equation in par (a).
e) All methods give the same equation with either D Cp or D^p_.
12-81. a) The u Cpequation is: x1
C_ 1 . 1 and M
E
0.0000705
y
-0.20052 + 0.467864x
1
The u M
E
equation is the same as the D Cp.
b) Same as equation in par (a).
c) Same as equation in part (a).
9
Same as equation in par (a).
e) All methods give the same equation with either D Cp or D^.
12-83. a) The u Cpequation is: x2
C_ 1 .2 and M
E
1 178.55
y 253.06-2.5453x
2
The u M
E
equation is the same as the D Cp.
b) Same as equation in par (a).
c) Same as equation in part (a).
9
Same as equation in par (a).
e) All methods give the same equation with either D Cp or D^.
12-85. a) The u Cpequationis: x1, x2
C_ 2
.
9 and M
E
10.49
y
-50.4+0.67lx
1
+ 1.30x
2
The u M
E
equation is the same as the D Cp.
b) Same as equation in par (a).
c) Same as equation in part (a).
9
Same as equation in par (a).
e) All methods give the same equation with either D Cp or D^p_.

There are no observatons with a Cook's distance geater than 1 so the results wil be the same.
12-32
useroame. George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
12-87. a) The Mintab outut follows. The frst model mthe table with seven variables mzes MS
1
and Cp.
Best Subsets Regression: Pts versus Att, Comp, ...
PctCor, YdsPerAtt, PctTD, Pctint
Response is Pts
Y
d
p s
c p p
t e p c
' ' r c t
A o o Y A t L I I
t m m d t T T n n n
Vars R-Sq R-Sq(adj) Mallows C-p o t p p s t D D g t t
1 67 . 2 66 . 1 3863 5 . 3 5 . 2187 X
1 64 . 8 63 . 7 413 81 . 3 5 . 4009 X
2 85 . 1 8 4. 1 17511 . 0 3 . 5748 X X
2 83 . 9 82 . 8 18943 . 5 3 . 7180 X X
3 9 5 . 3 9 4. 8 5461 . 8 2 . 0348 X X X
3 93 . 1 9 2 . 4 8064 . 5 2 . 4708 X X X
4 100 . 0 100 . 0 5 . 3 0 . 14621 X X X X
4 98 . 5 98 . 3 173 0 . 7 1 . 1713 X X X X
5 100 . 0 100 . 0 6 . 1 0 . 14558 X X X X X
5 100 . 0 100 . 0 6 . 2 0 . 14600 X X X X X
6 100 . 0 100 . 0 5 . 2 0 . 13990 X X X X X X
6 100 . 0 100 . 0 6 . 4 0 . 14362 X X X X X X
7 100 . 0 100 . 0 5 . 1 0 . 13632 X X X X X X X
7 100 . 0 100 . 0 7 . 1 0 . 14277 X X X X X X X
8 100 . 0 100 . 0 7 . 0 0 . 13892 X X X X X X X X
8 100 . 0 100 . 0 7 . 1 0 . 13924 x x x x x x x X
9 100 . 0 100 . 0 9 . 0 0 . 14203 X x x x x x x x x
9 100 . 0 100 . 0 9 . 0 0 . 14204 x x x x x x x X X
10 100 . 0 100 . 0 11 . 0 0 . 14537 x x x x x x x x x x
The Minitab output for this model follows.
Regression Analysis: RatingPts versus Att, PctComp, .. a
The regression equation is
RatingPts - 0 . 69
+
0 . 00738 Att
+
0 . 827 PctComp - 0 . 00150 Yds
4 . 82 YdsPerAtt
0 . 0702 TD
+
3 . 04 PctTD - 4 . 19 Pctint
Predictor Coef SE Coef T p
Constant -0 . 693 1 . 264 -0 . 55 0 . 589
Att 0 . 007381 0 . 003258 2 . 27 0 . 033
PctCor 0 . 826817 0 . 008980 92 . 07 0 . 000
Yds -0 . 0015027 0 . 0006166 -2 . 44 0 . 023
YdsPerAtt 4 . 8206 0 . 2555 18 . 87 0 . 000
TD 0 . 07025 0 . 04601 1 . 53 0 . 140
PctTD 3 . 0386 0. 2013 15 . 10 0 . 000
Pctint -4 . 19493 0 . 03545 -118 . 34 0 . 000
o 0 . 136317 R-Sq 100 . 0% R-Sq(adj ) 100 . 0%
Analysis of Variance
Source DF oo MS F p
Regression 7 2489 . 08 355 . 58 1913 5 . 65 0 . 000
Residual Error 24 0 . 45 0 . 02
Total 31 2489 . 52
12-33
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
b) Stepwise regression with the default Minitab parameters shown below selects the four-variable model
YdsPerAtt, Pctint, PctTD, PctComp.
Stepwise Regression: Pts versus Att, Comp, ...
Alpha-to-Eter: 0 . 15 Alpha-to-Remove : 0 . 15
Response is Pts on 10 predictors, with N 32
Step
Constant
YdsperAtt
T-Value
P-Value
Pctlnt
T-Value
P-Value
PctTD
T-Value
P-Value
PctComp
T-Value
P-Value
s
R-Sq
R-Sq(adj)
Mallows C-p
1
14 . 195
10 . 092
7 . 84
0 . 000
5 . 22
67 . 18
66 . 09
3863 5 . 3
2
24 . 924
1 0 . 296
1 1 . 21
0 . 000
3
32 . 751
7 . 432
8 . 7 8
0 . 000
-4 . 786 -4 . 932
-5 . 49 -7 . 90
0 . 000 0 . 000
3 . 72
83 . 90
82 . 79
18943 . 5
3 . 187
5 . 3 6
0 . 000
2 . 66
92 . 06
91. 21
9332 . 6
The Minitab output for this model follows.
4
2 . 128
4 . 214
73 . 46
0 . 000
-4 . 164
-118 . 11
0 . 000
3 . 310
101 . 17
0 . 000
0 . 8284
96 . 04
0 . 000
0 . 146
99. 98
9 9 . 97
5 . 3
Regression Analysis: RatingPts versus YdsPerAtt, Pctlnt, PctTD, PctComp
The regression equation is
RatingPts 2 . 13 4 . 21 YdsPerAtt - 4 . 16 Pctlnt 3 . 31 PctTD
+
0 . 828 PctCom
Predictor Coef SE Coef T p
Constant 2 . 1277 0 . 4224 5 . 04 0 . 000
YdsPerAtt 4 . 21364 0 . 05736 73 . 46 0 . 000
Pctlnt -4 . 16391 0 . 03526 -118 . 11 0 . 000
PctTD 3 . 31029 0 . 03272 101 . 17 0 . 000
PctComp 0 . 828385 0 . 008626 96 . 04 0 . 000
s 0 . 146206 R-Sq 100 . 0% R-Sq(adj) 100 . 0%
Analysis of Variance
Source DF ss M F p
Regression 4 2488 . 95 622 . 24 29108 . 69 0 . 000
Residual Error 27 0 . 58 0 . 02
Total 31 2489 . 52
The model is
y 2. 13 + 4.21
Y
dsPerAtt -4. 16Pctlnt + 3. 31PctT + 0
.
828PctComp
12-34
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
c) Forward selection with the default Minitab parameters shown below selects the same four-variable model
YdsPerAtt, Pctlnt, PctTD, PctComp upart (b).
Stepwise Regression: Pts versus Att, Comp, ...
Forward selection. Alpha-to-Enter : 0 . 25
Response is Pts on 10 predictors, with N 32
Step
Constant
YdsperAtt
T-Value
P-Value
Pctlnt
T-Value
P-Value
PctTD
T-Value
P-Value
PctCom
T-Value
P-Value
o
R-Sq
R-Sq(adj)
Mallows C-p
1
14 . 195
10 . 092
7 . 84
0 . 000
5 . 22
67 . 18
66 . 09
3863 5 . 3
2 3
24 . 924 32 . 751
1 0 . 296 7 . 432
1 1 . 21 8 . 78
0 . 000 0 . 000
-4 . 786
-5 . 49
0 . 000
3 . 72
83 . 90
82 . 79
18943 . 5
-4 . 932
-7 . 90
0 . 000
3 . 187
5 . 3 6
0 . 000
2 . 66
92 . 06
9 1 . 21
9332 . 6
4
2 . 128
4 . 214
73 . 46
0 . 000
-4 . 164
-118 . 11
0 . 000
3 . 310
101 . 17
0 . 000
0 . 8284
96 . 04
0 . 000
0 . 146
99 . 98
99 . 97
5 . 3
d Backward eltion with the default Mini tab parameters shown below selects the six-variable model.
Att, PctComp, Yds, YdsPerAtt, PctTD, Pctlnt. Itis similarto the model withminimumMS
E
except variable TD is excluded.
Stepwise Regression: Pts versus Att, Comp, . . .
Backard elimination. Alpha-to-Remove: 0 . 1
Response is Pts on 10 predictors, with N 32
Step 1 2 3
Constant
Att
T-Value
P-Value
Com
T-Value
P-Value
PctComp
T-Value
P-Value
Yds
T-Value
P-Value
YdsperAtt
-0 . 6871
0 . 0074
1 . 56
0 . 133
0 . 000
0 . 02
0 . 982
0 . 8253
1 6 . 92
0 . 000
-0 . 00159
-1 . 65
0 . 115
4 . 86
-0 . 7140
0 . 0074
2 . 02
0 . 056
0 . 8264
8 6 . 99
0 . 000
-0 . 00158
-2 . 20
0 . 038
4 . 85
12-35
-0 . 6751
0 . 0074
2 . 22
0 . 036
0 . 8264
89 . 3 0
0 . 000
-0 . 00156
-2 . 3 9
0 . 026
4 . 85
4
-0 . 6928
0 . 0074
2 . 27
0 . 033
0 . 8268
92 . 07
0 . 000
-0 . 00150
-2 . 44
0 . 023
4 . 82
5
-0 . 1704
0 . 0055
1 . 78
0 . 088
0 . 8289
91 . 04
0 . 000
-0 . 00083
-1 . 88
0 . 072
4 . 55
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
T-Value 11. 78 1 6 . 28 17 . 69 18 . 87
P-Value 0 . 000 0 . 000 0 . 000 0 . 000
TD 0 . 075 0 . 075 0 . 074 0 . 070
T-Value 1 . 46 1 . 50 1 . 53 1 . 53
P-Value 0 . 158 0 . 148 0 . 139 0 . 140
PctTD 3 . 018 3 . 019 3 . 019 3 . 039
T-Value 13 . 47 13 . 82 14 . 15 15 . 10
P-Value 0 . 000 0 . 000 0 . 000 0 . 000
Lng 0 . 0001 0 . 0001
T-Value 0 . 05 0 . 05
P-Value 0 . 960 0 . 961
Int 0 . 011 0 . 011 0 . 010
T-Value 0 . 3 1 0 . 33 0 . 3 3
P-Value 0 . 756 0 . 747 0 . 743
Pctlnt -4 . 241 -4 . 240 -4 . 238 -4 . 195
T-Value -27 . 73 -29 . 74 -31 . 63 -118 . 34
P-Value 0 . 000 0 . 000 0 . 000 0 . 000
s 0 . 145 0 . 142 0 . 139 0 . 136
R-Sq 99 . 98 9 9 . 98 99 . 98 99 . 98
R-Sq{adj) 99 . 97 9 9 . 97 99 . 98 99 . 98
Mallows C-p 11 . 0 9 . 0 7 . 0 5 . 1
The Minitab outut for this model follows.
Regression Analysis: RatingPts versus Att, PctComp, ...
The regression eqation is
2 4. 46
0 . 000
3 . 341
9 6 . 41
0 . 000
-4 . 177
-121 . 82
0 . 000
0 . 140
99 . 98
99 . 98
5 . 2
RatingPts

- 0 . 17 0 . 00550 Att 0 . 829 PctCom - 0 . 000826 Yds

" 4 . 55 YdsPerAtt " 3 . 34 PctTD - 4. 1 8 Pctlnt
Predictor Coef SE Coef T p
Constant -0 . 170 1 . 248 -0 . 14 0 . 893
Att 0 . 005499 0 . 003095 1 . 78 0 . 088
PctComp 0 . 828933 0 . 009105 91 . 04 0 . 000
Yds -0 . 0008259 0 . 0004398 -1 . 88 0 . 072
YdsPerAtt 4 . 5455 0 . 1858 24 . 46 0 . 000
PctTD 3 . 34144 0 . 03466 9 6 . 41 0 . 000
Pctlnt -4 . 17685 0 . 03429 -121 . 82 0 . 000
s

0 . 139897 R-Sq

100 . 0% R-Sq{adj)

100 . 0%
Analysis of Variance
Source DF ss MS F p
Regression 6 2489 . 03 414 . 84 21196 . 51 0 . 000
Residual Error 25 0 . 49 0 . 02
Total 31 2489 . 52
The model is
y -0. 17 + 0.0055Att + 0
.
8289PctComp-0.0008
Y
ds + 4.5455
Y
dsPet + 3. 3414PctTd
~
4+ 1 769PcUnt
e) The four variable model (PctComp, YdsPerAtt, PctTD, Pctlnt) has the second mu Cp and
also has small MS
E
and large adjusted R-squared. 1is also a model with the few regressors so it is prefered.
12-36
Username: George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
12-89. This analysis includes the emissions variables he, co, and co2. Hwould be reasonable to model without these
variables as regessors.
Best Subsets Regression: mpg versus cid, rhp, ...
Response is ]Q
a
c r e c x n c
Mallows i h t 1 / h c
Vars R-SQ R-Sq(adj) C-p o d p w p e c 2
1 66 . 0 64 . 2 2 6 . 5 3 . 4137 X
1 59 . 1 57 . 0 3 5 . 3 3 . 7433 X
2 81 . 6 79 . 5 8 . 6 2 . 5813 X X
2 78 . 1 7 5 . 7 13 . 0 2 . 8151 X X
3 88 . 8 86 . 8 1 . 3 2 . 0718 X X X
3 88 . 8 86 . 8 1 . 4 2 . 0755 X X X
4 9 0 . 3 87 . 8 1 . 5 1 . 9917 X X X X
4 89 . 9 8 7 . 3 2 . 0 2 . 0302 X X X X
5 90 . 7 87 . 6 2 . 9 2 . 0057 X X X X X
5 90 . 7 87 . 6 2 . 9 2 . 0064 X X X X X
6 91 . 0 8 7 . 2 4 . 5 2 . 0442 X X X X X X
6 91 . 0 8 7 . 1 4 . 5 2 . 0487 X X X X X X
7 9 1 . 3 86 . 6 6 . 2 2 . 0927 X X X X X X X
7 9 1 . 2 8 6 . 4 6 . 3 2 . 1039 X X X X X X X
8 9 1 . 4 85 . 6 8 . 0 2 . 1651 x x x x x x x x
8 9 1 . 4 85 . 6 8 . 1 2 . 1654 X X X X X X X X
9 9 1 . 4 84 . 4 10 . 0 2 . 2562 x x x x x x x x x
a) The minimum L_(1. 3) model is: y = 61 .001 -0.02076x
ci
-0.00354x
e
l -3.457x
a/e
The minimum MSE (4.0228) model is:
y =49.5 -0.017547x
i
-0.0034252x +1 .29x -3. 184x
1
-0.0096x
0
2
c
e
t C
ae
c
b) y = 63.31 -0.0178x
ci
-
0.00375x
e
l -3.3x
al
e
-0.0084x
cOl
c) Same as equation as the min M

equation mpart (a)

9
y = 45. 18-
0.00321x
e
l -4.4x
a/e
+ 0.385x
n
lv
e) The minimum C
P
model is preferred because it has a very low MSE as well ( 4.29)

Only one indicator variable is used for tansmission to distinguish the automatic fom manual types and two
indicator variables are used for U.
7 0 for automatic (I, LS) and 1 for manual (M6) and
7
|
0 if d4 orR and 1 if d F; 70 if d4 or F and 1 if dR.
The minimum C
P
(4
.
0) model is the same as the minimum MSE (2.267) model:
y = l0-0.0038023x +3.936x +15. 216x -
O. Ol l l l8x
0
2
-7.40lx +
e
t cm c tns
3. 613lx
diI
+2.342x
di
2
Stepwise:
y = 39.12-
0.0044x
e
i
+0.27lx
n
lv
-4.5x
t
s
+3.2x
d1l
+ 1.7x
di
2
Forward selection:
y = 41 . 12 -0.00377x
e
l +
0.336x
n
lv -2. lx
a/e -3.4x
tns
+
2

1x
dil
+ 2x
di
2
Backward selection: same minimum C
P
and minimum MSE.
Prefer the model giving the minimum C_and minimum MSE.
12-37
useroame. George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
12-91 . The default settings for F-to-enter and F-to-remove for Minitab were used. Different settngs can change the
models generated by the method.
a) y = -0.304 + aas..
,
-0.031x
3
+ aaa+.
c

= 4.04 M
E
= o.oo4
b) y = -0.256 + aa:s.
,
+ a.a22.

-aa42.
,
+ aaaas.
c

= 4.66 M
E
= o.oo4
c) The forward selection model in part (a) is more parsimonious with a lower cand equivalent MS
E
. Therefore,
we prefer the model in part (a).
12-93. n 25 ] 7 p 8 M
E(
f
/)
= 10
a) p 4
SS
E .aa
M ==
-
-= 14.29
E
n-p 25-4
SS
E
C = -n+2p

M
E(
f)
=
.aa
-25+2(4)
10
=5 +8 =13
YES, c >p
b) p =5
SS
E
=275
M =
SS
E
=

=1 1
E
n-p 30-5
c =
275
-25+2(5) = 12.5

ia
Yes, both MSE and care reduced.
Suplemental Exercises
12-95. a) Because the matix is 3 7 3 two regressors are i.n the regression model. The intercept is also in the model.
b) cov(/) = o
|
xx,
,
= -

c , therefore the variances of the two variables regession coeficients are:

50(0.0013329) = 0.066645 and 50(0
.
0009108) = 0.04554.
c) s-(,
)=
,

c
, ,
= 6.6849
12-97. a)
|

,
=

=
:
= a
|,

,
a
D 0.01
'

= 1321.39
!..,.
= 438
'.``',,,
for at least one j
Rejet H0 and conclude that regression is sigifcant. P-value ~aaaaai
12-38
useroame. George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
b) D 0.01

.
=2

72
|
.

,
= a
|
.

=a
|
.

=0
|, ,
= a
|,

=a
|,

=0

=19.95

=2.53

.
,,.
.` ,,
.
.
,,.
Do not reject H0 Rejet H0 Do not reject H0
c) Curvature is evident in the residuals plots fom this model.
NCfm HlODBDl t|llO1O 1h6 K65t Ou5
(rep is))

c z

a
..

-0.U -0.CZ -D.D| C.O

Residual
WB8!OUB!GNBfGUGM11BO NUBG
-
{.) !\]
+
*
.

.*
.
12-99. a) y =4.87 +6.12x-6.53x-3.56x -L44x
b)
|
.!
=

=
,
=

=a
|,
: at least one

.a
O 0.05
/
=21. 79
'...,

=2-87
'. `'..

D.C| D.0
.-
Rejet H0 and conclude regession is sigificant at D 0.05.
D 0.05

=2.086
| ,
=0
|
.

=a
|,,
=0
|,
=a

=5.76

=-5.96
|
.
,
=a
|
.

=a
|, ,
=a
|,

=a

=:a

=-4.99

003
WBG1OUGNBfGUGY<
\^1
.
J
M
*

.`
...
.`
...

.`
...

.`
...
Rejet H0 RejectH0 Rejet H0 Rejet H0
12-39
M .
.
.
M
useroame. George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
c) The residual plots are more satisfactory than those mExercise 12-90.

o.>

9.o

o.>
.
1P

.1.>

KB5OUl5BI5U5B!LBClUB5
(->/)

.z

.z -: 0
ftted Value
KB5OUl5BI5U5 X1'
(r->/)
:.C

o.>

o.

.

.

:s
-!(
-:.s

-zo
o.z o.s o.( o.> o.s o.
*
KB5CU5BI5U5Xd'
(r-s/)
:.1

o>

9.o

o>
.:.
1
.:.

.z
o. o.( o.>
-
us u?
12-101. a)
)
= 21 .068 -1 .404x+0.0055x
4
+0.000418x
5
M
E
= 0. 013156
C =4.0
1
b) Same as model in part (a)
c) x4, x, withC
P
= 4. 1 and M
E
= 0.0134
1P

o.>

9.o

o.>
.
1P

.1.S
-z
1
!
o.

o.

:
-1!
-1S
-z

o.z

:.s
KB5OUl5BI5U5 XZ'
(->/)

.
..

o.s u(
z*
KB5CU5BI5U5X'
(r-s/)

z9

z>
(-

u>
I

I
o.s o.?

3C s.>
d) The model mpart (c) is simpler with values for ^_ and Cp similar to those in part (a) and (b). The part model
is preferable.
12-40
useroame. George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
e) V(f;) = 52.4
V(f
4
) = 9.3
V(f5 ) =29. 1
Yes, Vs for x; and x5 exceed 10.
12-103. a) y = 300.0+ 0.85x1 + 10.4x2
b)
c)
y = 300 + 0.85(100) + 10.4(2) = 405.8
= 1230.5
.

,
=120.3

,
=
.

,
=1230.5-120.3 = 1 1 10.2
_
=

,
=
1 1 10.2
= 555.1
,
k 2
_
=

,
=
120.3
= 10.025
,
nj 1 5-3
'.
=
_
,
=
555.1
= 55.37
_
,
10.025
|
.
,
=

= O
|,
: at least one

" 0
O =0.05
/
= 55.37
!..
= 3

89
'. `'. . ,
Rejet H0 and conclude that the regession model is significant at O 0.05

,
=
1 1 10

2
= 0.9022 or 90.22
%
1230.5
.
d) ]3 p =4

I
_ I
=

,

=
1 17.2
= 10.65
,
nj 1 1
No, +
,
increased with the addition of _ because the reduction u,was not enough to compensate for the
loss uone degree of feedom in the error sum of squares. This is why +,can be used a model selection
criterion.
e)
,
=
.

,
=1230.5-1 17.20 = 1 1 13.30

,
|
, ,
.
)=
,
|
'
,.
)
,
|
.
)
= 1 1 13.30-1 1 10.20
= 3. 1
|
.
,
= 0
|
.
,
" 0
O 0.05
12-41
useroame. George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
12-105. a)
f
.
=

,|, ,
.
)

,
/j
!.
,
..
= 4

84
'. '. . .
, ,
Do not reject H0.

,
_
3. 1 / 1
1 17.2 / 1 1

,
=

|
_)= 0
.
94(0.50) = 0.47

L
-_
R
= o.s-0.47 = o.o3
|
.
,
=

`
=

.
= 0
|,
: at least one

" 0
D = 0.05
'.
=

,
/k
=
0.471 6
= 18
_
28

,
/ 0.031 7
'..
-
= 3.87
'. `'...-
Rejet H0
0.291
b) k=5 n = 14 p =6

=0.92
c)
12-107. a)
b)
c)

,
I =

| _) = 0.92(0.50) = 0.46

L
' - _
R
= o.5 -0.46 = o.o4

,
|
.,
.
)=
,
|
'
)
,
|
-1.-1,
= 0.47-0.46
= 0.01
_

,| ,,., +.
)

_ 0.01 1 1 _
2
'.

,
/( ,
-
0.0418
-
!.
,

= 5.32
'.'...
Do not reject H0. There is not sufcient evidence that the removed vaable is sigifcant at D = 0.05.

0 04
_
,|
-1.-1,=
,
=
`
= 0.005
8
_
,|
;)=
0

03
= 0.004
7
No, the +,is larger for the reduced model, although not by much. GeneraUy, if adding a variable to a model
reduces the +,it is an indication that the variable Dbe useful in explg the response variable. Here
the decrease in +s,is not very great because the added variable had no real explanatory power.
Because the Dis 4 by 4 there are 3 regessors in the model (plus the intercept).
Because (XX)
,
is diagonal each element can be inverted to obtain (XX) and fom the norl equations the
(1, 1) element (the top-lef element) of (XX) = n. Therefore, n = 1/0.125 = 8.
The original columns are orthogonal to each other.
12-42
useroame. George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
Mind-Expanding Exercises
12-109. Using n = 20, k = 4,
/os,4
15
= 3.06 . Reject Ho i f
R
2
/ 4
> 3.06
(1 -R
2
) / 15 -
> 0.816
(1 -R
2
) -
Then, R
2
2 0.449 results in a signifcant regression.
12-1 1 1. From Exercise 12-76, e; is ith element of(I-H)Y is,
e
;
= -h
i,l
-h
i,
2
y
2
-.
.
. -
\HI;
-1
+(1 -h
;)I;
-h
i,i+II;+I
-.
.
. -
\n
Y
n
and
V(e.) = (h
2
1
+ h
2
2
+ . . . + h
2
.
I
+ (1 -h
2
) + h
2
.
+
1
+ . . . + h
2
)cr
2
t t, t, t,J- t,J t, t,n
The expression in parentheses is recognized to be the ith diagonal element of (I-H(!-H)= I-H Dymatix
multiplication. Consequently, V(e ) = (1 -h. .)cr
2
Assume l~j. Now,
^
e
i
= -h
i,li
-
\
2
Y
2
-.
.
. -
\HI;
-1
+(1 -
\)
-h
i,i+l+l
-.
.
. -h
i,nyn
e
j
= -h
j,l
-h
j,
2
Y
2
-.
.
. -h
j
,
j
-
lj
-
1
+ (1 -h
j)
Y
j
-h
j
,
j
+
lj+l
-.
.
. -h
j
,nyn
Because they; 's are independent,
Cov(e , e . ) =(h.
1
h .
1
+h.
2
h .
2
+ . . . +h. .
1
h . .
1
+(1 -h. .)h .
} }, 5 }, ^ },l ; },l
+ h .
+
1
h . .
+
1
+ . . . +h (1 -h ) +h. .
+
1
h . .
+1
+ . . . +h. h . )cr
2
^ J,J t,J J,J t,J J,J t,n J,n
The expression in parentheses is recognized as the ijth element of (1-H)(I-H) =I-H. Therefore,
Cov(e. , e . ) = -h..cr
2
.
} 9
12-113. a) Min L = (y-X/)'(y-X/) subjectto :T =c
This is equivalent to Min Z= (y-X/)'(y-X/) + 2A'(Tf3-c)
where A' = [A ,A , . . . , A ] is a vector ofLanguage multipliers.
1 2 p
g/
= -2X' y + 2(X' X)/ + 2T' A
8/
_= 2T-c, . bC
g/
=O
g

Then we get
(X'X)/ +T' A=X' y

T/ = c

8z
=O .
8A
where , is the constained estimator. From the fi rst of these equations,

/ = (X' X)
-1
(X'y -T'A) = /-(X' X)
-1
T' A

12-43
useroame. George Mason University Book: Student Solutions Manual Applied Statistics and Probability for Engineers,
Fifth Edition . No part of any chapter or book may be reproduced or transmitted in any form by any means without the prior
written permission for reprints and excerpts from the publisher of the book or chapter. Redistribution or other use that
violates the fair use privilege under U.S. copyright laws (see 17 USC107) or that otherwise violates these Terms of Serice
is strictly prohibited. Violators will be prosecuted to the full extent of U.S. Federal and Massachusetts laws.
Frm the second, T-T(X' X)
-1
T' `= c or `= [T(X' X)
-1
TT
1
( -c)
Then
= -(X' X)
-1
T'[T(X'Xr
1
TT
1
(T-c)

= +(X' Xr
1
T'[T(X'X)
-1
TT
1
(c-T)
b) This solution would be appropriate in situations where you have tested the hypothesis that T c and
concluded that this hypothesis cannot be rejected.
12-44