Sie sind auf Seite 1von 13

4.

3
Analysis of the data:

Factor
Chemical
Bolt

Type
random
random

Levels
4
5

Values
1; 2; 3; 4
1; 2; 3; 4; 5

Analysis of Variance for Tensile Strengths, using Adjusted SS for


Tests
Source
Chemical
Bolt
Error
Total

DF
3
4
12
19

Seq SS
12.950
157.000
21.800
191.750

Adj SS
12.950
157.000
21.800

Adj MS
4.317
39.250
1.817

F
2.38
21.61

P
0.121
0.000

Minitab output for problem 4.3

Conclusion:
The null hypothesis will be that all the treatment means
are equal. The alternative hypothesis is that they are not
all equal. From the Minitab output above, we can conclude
that

there

is

no

difference

between

the

chemical

types

because F = 3.49, which is greater that Fo = 2.38. Thus, do


not reject the null hypothesis. Also, this conclusion can
be

obtained

from

the

p-value

for

treatment

effect

(Chemical), which is 0.1211 > 0.05. The p-value for the


block effect (Bolt), however, is very small (= 0.000), so
the

block

experiment.

effect

dominates

the

variations

in

the

4.8
a)
Analysis of the data:
Factor
Design
Region

Type
random
random

Levels
3
4

Values
1; 2; 3
NE; NW; SE; SW

Analysis of Variance for Response, using Adjusted SS for Tests


Source
Design
Region
Error
Total

DF
2
3
6
11

Seq SS
90755
49036
5429
145220

Adj SS
90755
49036
5429

Adj MS
45378
16345
905

F
50.15
18.06

P
0.000
0.002

ANOVA Table from Minitab for Problem 4.8

From the Minitab output above, the designs do differ from


each other significantly. However, at this point we cannot
decide which design is best for our study.

b)

Level
1
2
3

N
4
4
4

Mean
298.50
473.75
281.25

StDev
75.67
93.75
60.33

Individual 95% CIs For Mean Based on


Pooled StDev
-+---------+---------+---------+-------(--------*--------)
(-------*--------)
(--------*--------)
-+---------+---------+---------+-------200
300
400
500

Pooled StDev = 77.79


Grouping Information Using Fisher Method
Design
2
1
3

N
4
4
4

Mean
473.75
298.50
281.25

Grouping
A
B
B

Means that do not share a letter are significantly different.

Minitab output for Fisher LSD method for comparison

From the Fisher LSD results we can conclude that Design 2


is significantly different in mean than both Design 1 and
3. Hence, Design 2 is the most effective design.
c)
Residual Plots for Response
Versus Fits
40

90

20

Residual

Percent

Normal Probability Plot


99

50
10
1

0
-20
-40

-50

-25

0
Residual

25

50

200

300
400
Fitted Value

Histogram

Versus Order
40

4.5

Residual

Frequency

6.0

3.0
1.5
0.0

500

20
0
-20
-40

-40

-20

0
Residual

20

40

4 5 6 7 8 9
Observation Order

10 11 12

Analysis of the residuals for Problem 4.8

There is violation of the normality assumption. Thus, a


transformation
analysis.

of

the

Therefore,

data
all

is

recommended

conclusions

previous analysis are not valid now.

for

obtained

further
in

the

4.13
a)
Analysis of the voltage data:
Factor
Algorithms
Time Period

Type
random
random

Levels
4
6

Values
1; 2; 3; 4
1; 2; 3; 4; 5; 6

Analysis of Variance for Voltage, using Adjusted SS for Tests


Source
Algorithms
Time Period
Error
Total

DF
3
5
15
23

Seq SS
0.002746
0.017437
0.072179
0.092363

Adj SS
0.002746
0.017437
0.072179

Adj MS
0.000915
0.003487
0.004812

F
0.19
0.72

P
0.901
0.615

The ratio control algorithm does not affect the average


cell voltage at 5% level.

b) and c)
Residual Plots for Pot Noise
Versus Fits
0.02

90

0.01

Residual

Percent

Normal Probability Plot


99

50
10
1

-0.02

-0.01

0.00
0.01
Residual

0.00
-0.01
-0.02
0.00

0.02

0.02

4.5
3.0
1.5

0.01
0.01

Residual

Residual

Frequency

Versus Order

0.02

6.0

.
-0

0.15

Residuals Versus Algorithms

Histogram

0.0

0.05
0.10
Fitted Value

0.00
0.00
-0.01
-0.01

-0.02

0
02

5
0
5
0
5
0
5
01
01
00
00
00
01
01
.
.
.
.
.
.
.
0
0
0
0
-0
-0
-0
Residual

-0.02
1.0

1.5

8 10 12 14 16 18 20 22 24
Observation Order
2.0

2.5
Algorithms

3.0

3.5

4.0

There are some concerns about the normality as well as the


variance of the residuals. It seems that bigger values of
algorithms give wider spread of residuals, while it should
be structure-less pattern. Hence, a log transformation can
be applied.
Residual Plots for ln(Pot Noise)
Normal Probability Plot

Versus Fits

99
0.2

Residual

Percent

90
50
10
1
-0.50

-0.25

0.00
Residual

0.25

0.0
-0.2
-0.4
-4.0

0.50

Histogram

-3.5

-3.0
-2.5
Fitted Value

-2.0

Versus Order

Residuals Versus Algorithms

6.0

0.3

0.2

Residual
Residual

Frequency

0.2

4.5
3.0
1.5

0.1

0.0

0.0
-0.1

-0.2

-0.2
-0.3

-0.4

0.0
-0.4 -0.3 -0.2 -0.1 0.0

0.1

0.2

0.3

-0.4
1.0

Residual

1.5

8 10 12 14 16 18 20 22 24
2.0
2.5
3.0
3.5
4.0
Algorithms
Observation
Order

Analysis of the residuals after applying natural log transformation for Pot Noise

The variance of the residuals now looks better but the


normality plot shows small deviation!

Factor
Algorithms
Time Period

Type
random
random

Levels
4
6

Values
1; 2; 3; 4
1; 2; 3; 4; 5; 6

Analysis of Variance for ln(Pot Noise), using Adjusted SS for Tests


Source
Algorithms
Time Period
Error
Total

DF
3
5
15
23

Seq SS
6.16605
0.94460
0.92704
8.03769

Adj SS
6.16605
0.94460
0.92704

Adj MS
2.05535
0.18892
0.06180

F
33.26
3.06

P
0.000
0.042

From the ANOVA table above, we can conclude that the ratio
control algorithm affects the pot noise.

d)
Individual 95% CIs For Mean Based on
Pooled StDev
Level N
Mean
StDev -----+---------+---------+---------+---1
6 -3.0877 0.2422
(----*----)
2
6 -3.5086 0.3667 (----*----)
3
6 -2.1998 0.2337
(----*----)
4
6 -3.3559 0.3558
(----*----)
-----+---------+---------+---------+----3.50
-3.00
-2.50
-2.00
Pooled StDev = 0.3059
Grouping Information Using Fisher Method
Algorithms
3
1
4
2

N
6
6
6
6

Mean
-2.1998
-3.0877
-3.3559
-3.5086

Grouping
A
B
B C
C

Algorithms verses pot noise analysis

The ratio control algorithm 2 or 4 (no significant


difference between 2 and 4 according to Fisher LCD) will be
selected to minimize the pot noise, since algorithms have
no effect on average cell voltage.

4.21

Factor
Distance
Subject

Type
random
random

Levels
4
5

Values
4; 6; 8; 10
1; 2; 3; 4; 5

Analysis of Variance for Focus Time, using Adjusted SS for Tests


Source
Distance
Subject
Error
Total

DF
3
4
12
19

Seq SS
32.950
36.300
15.300
84.550

Adj SS
32.950
36.300
15.300

Adj MS
10.983
9.075
1.275

F
8.61
7.12

P
0.003
0.004

ANOVA Table for Problem 4.21

Residual Plots for Focus Time


Normal Probability Plot

Versus Fits

99
1

Residual

Percent

90
50

0
-1

10
1

-2
-2

-1

0
Residual

Histogram

10

Versus Order

6.0

4.5

Residual

Frequency

6
Fitted Value

3.0

0
-1

1.5
-2

0.0
-2.0 -1.5 -1.0 -0.5 0.0

0.5

1.0

1.5

Residual

6
8 10 12 14 16
Observation Order

18

20

Analysis of the residuals for problem 4.21

There

seems

observation

to
14.

be

an

Maybe

outlier
it

is

observation,
better

to

which
rerun

is
this

observation. However, from the ANOVA table obtained, there


exists a significant difference in the distance. Blocking
is also significant.

4.22
Analysis of Variance for Reaction Time, using Adjusted SS for Tests
Source
Batch
Day
Ingredients
Error
Total

DF
4
4
4
12
24

Seq SS
15.440
12.240
141.440
37.520
206.640

Adj SS
15.440
12.240
141.440
37.520

Adj MS
3.860
3.060
35.360
3.127

F
1.23
0.98
11.31

P
0.348
0.455
0.000

ANOVA table for problem 4.22


In this experiment, batch and days are the block factors.
The

experimenter

Ingredients

on

is

the

interested
reaction

in

time.

the
Thus,

effect
an

of

the

appropriate

hypothesis would be:


Ho: There is no significant difference in the Ingredients.
H1: There is significant difference in the Ingredients.
Since the p-value of the ingredients 0.000 is less than
0.05, we can conclude that there is significant difference
between

the

Ingredients.

However,

there

is

no

strong

evidence of a difference on batches or days. It seems that


in this experiment we were unnecessarily concerned about
this source of variability.
Analysis of the residuals below shows no evidence of any
violation of any assumption in this experiment.

Residual Plots for Reaction Time


Normal Probability Plot

Versus Fits

99
2

Residual

Percent

90
50
10

1
0
-1
-2

-3.0

-1.5

0.0
Residual

1.5

3.0

Histogram

10

Versus Order

6.0

4.5

Residual

Frequency

6
8
Fitted Value

3.0
1.5

1
0
-1
-2

-2

-1

0
1
Residual

8 10 12 14 16 18 20 22 24
Observation Order

Residuals Versus Day

Residuals Versus Batch


(response is Reaction Time)

-1

-1

-2

-2
1

(response is Reaction Time)

Residual

Residual

0.0

3
Day

3
Batch

Residuals analysis for problem 4.22

Level
A
B
C
D
E

N
5
5
5
5
5

Mean
8.400
5.600
8.800
3.400
3.200

StDev
1.140
2.074
1.643
2.074
1.924

Individual 95% CIs For Mean Based on


Pooled StDev
----+---------+---------+---------+----(------*-----)
(-----*------)
(------*------)
(------*-----)
(------*------)
----+---------+---------+---------+----2.5
5.0
7.5
10.0

Pooled StDev = 1.806


Grouping Information Using Fisher Method
Ingredients
C
A
B
D
E

N
5
5
5
5
5

Mean
8.800
8.400
5.600
3.400
3.200

Grouping
A
A
B
B C
C

Means that do not share a letter are significantly different.

Comparing means of Ingredients for problem 4.22

From the above comparison figure, we may conclude that if


the objective of this study is to minimize the reaction
time then either ingredient D or E should be selected.

4.42

Factor
Concentration
Days

Type
fixed
random

Levels
7
7

Values
2; 4; 6; 8; 10; 12; 14
1; 2; 3; 4; 5; 6; 7

Analysis of Variance for Strength, using Adjusted SS for Tests


Source
Concentration
Days
Error
Total

DF
6
6
8
20

Seq SS
2037.62
394.10
168.57
2600.29

Adj SS
1317.43
394.10
168.57

Adj MS
219.57
65.68
21.07

F
10.42
3.12

P
0.002
0.070

ANOVA Table for problem 4.42

This is a BIBD experiment. The objective is to test if


there is any difference among concentrations.
H0: There is no significant difference in concentrations.
H1: There is significant difference in concentrations.

From the ANOVA table we can conclude we can reject Ho and


say there is a significant difference in concentrations.

There is no indication of any violation of assumptions from


the residual graphs on the next page.

Residual Plots for Strength


Normal Probability Plot

Versus Fits

99
5.0

Residual

Percent

90
50
10
1

2.5
0.0
-2.5
-5.0

-8

-4

0
Residual

110

120

Histogram

130
140
Fitted Value

150

Versus Order

Residual

Frequency

5.0
6
4
2

2.5
0.0
-2.5
-5.0

-6

-4

-2

0
2
Residual

Residuals Versus Concentration

18 20

Residuals Versus Days

(response is Strength)

(response is Strength)

7.5

7.5

5.0

5.0

2.5

2.5

Residual

Residual

6 8 10 12 14 16
Observation Order

0.0

-2.5

0.0

-2.5

-5.0

-5.0
2

8
10
Concentration

12

14

4
Days

Residuals analysis for problem 4.42

Level
2
4
6
8
10
12
14

N
3
3
3
3
3
3
3

Mean
117.00
121.67
129.33
139.67
146.00
120.33
131.00

StDev
3.00
3.79
10.79
10.07
3.61
2.52
4.58

Individual 95% CIs For Mean Based on


Pooled StDev
---------+---------+---------+---------+
(------*-----)
(-----*------)
(------*-----)
(-----*------)
(------*-----)
(-----*------)
(-----*------)
---------+---------+---------+---------+
120
132
144
156

Pooled StDev = 6.34


Grouping Information Using Fisher Method
Concentration
10
8
14
6
4
12
2

N
3
3
3
3
3
3
3

Mean
146.000
139.667
131.000
129.333
121.667
120.333
117.000

Grouping
A
A B
B C
B C
C D
C D
D

Means that do not share a letter are significantly different.

Mean analysis for concentrations.

From the mean comparison graph, we can conclude that if the


objective is to have a maximum strength for the paper
product, then either 10% or 8% hardwood concentration must
be considered.

Das könnte Ihnen auch gefallen