Unit-II Design of Experiments

UNIT-II DESIGN OF EXPERIMENT

The sequence of steps taken to ensure a scientific analysis leading to valid inferences about
the hypothesis is called design of experiment. The main aim of the design of experiments is to
control the extraneous variables and hence to minimize the experimental error so that the
results of the experiments could be attributed only to the experimental variables.
The basic principles of design of experiments:
(i) Randomization
(ii) Replication
(iii) Local Control

(i). Randomisation:

The random allocation of treatments to the experimental units. Randomize to avoid confounding
between treatment effects and other unknown effects. The purpose of randomization is to
remove bias and other sources of extraneous variation which are not controllable.

(ii)Replication:

The repetition of a treatment within an experiment allows. To quantify the natural variation
between experimental units. To increase accuracy of estimated effects.

(iii) Local Control

It has been observed that all extraneous source of variation is not removed by randomization
and replication, i.e. unable to control the extraneous source of variation. Thus, we need to a
refinement in the experimental technique. In other words, we need to choose a design in such a
way that all extraneous source of variation is brought under control. For this purpose, we make
use of local control, a term referring to the amount of (i) balancing, (ii) blocking and (iii) grouping
of experimental units.
Basic design of Experiments:
Depending on the number of extraneous variables whose effects are to be controlled, various
design procedures are developed in the study of experimental design. We shall consider here
three important designs.
(1) Completely randomized Design (C.R.D)
(2) Randomized Block Design (R.B.D)
(3) Latin Square Design (LSD)

1
 Unit-II Design of Experiments

ANOVA:
Analysis of Variance is a technique that will enable us to test for the significance of the
difference among more than two sample means.
Assumptions of analysis of variance:
(i) The sample observations are independent
(ii) The environmental effects are additive in nature
(iii) The samples have been randomly selected from the population.
(iv) Parent population from which observations are taken in normal.
One Way Classification (or) Completely randomized Design (C.R.D)
The C.R.D is the simplest of all the designs, based on principles of randomization and
replication. In this design, treatments are allocated at random to the experimental units over the
entire experimental materials.
Advantages of completely randomized block design:
The advantages of completely randomized experimental design as follows:
(i) Easy to lay out.
(ii) Allow flexibility
(iii) Simple statistical analysis
(iv) lots of information due to missing data is smaller than with any other design
Working Procedure (One – Way classification)

Null Hypothesis H 0 : There is no significance difference between the treatments.

Alternate Hypothesis H1 : There is a significance difference between the treatments.

Analysis:
Step 1: Find N= number of observations
Step 2: Find T = The total value of observations

T2
Step 3: Find the correction Factor = C.F =
N
Step 4: Calculate the total sum of squares = SST = (  X12 +  X 22 +  X 32 + ...) − C.F

( X ) + ( X ) + ( X )
2 2 2

Step 5: Column Sum of Square SSC = + ... − C.F.

1 2 3

c1 c2 c3
Where ci = Total number of observations in each column ( i = 1, 2, 3,... )
Step 6: Find SSE=SST-SSC
Step 7: Prepare the ANOVA TABLE to calculate F-ratio.

2
 Unit-II Design of Experiments

Degree
Source of Sum of Mean
of F- Ratio
Variation Squares Square
freedom
MSC
Between SSC FC = if MSC  MSE
SSC c-1 MSC = MSE
Columns c −1
(or)
MSE
SSE FC = if MSE  MSC
Error SSE N-c MSE = MSC
N−c

Total SST N-1

Step 8: Find the F table value at  % level of significance with ( v1 , v2 ) degrees of freedom.
Step 9: Conclusion:
Calculated value < Table Value, the we accept Null Hypothesis H 0 (or)
Calculated value > Table Value, the we reject Null Hypothesis H 0
1. The following are the number of mistakes made in 5 successive days by 4 technicians
working for a photographic laboratory. Test whether the difference among the four
sample means can be attributed to chance. (Test at a level of significance  = 0.01 )

Technicians
I II III IV
6 14 10 9
14 9 12 12
10 12 7 8
8 10 15 10
11 14 11 11
Solution:
H0: There is no significant difference between the technicians
H1 : Significant difference between the technicians
X1 X2 X3 X4 X12 X22 X32 X42
6 14 10 9 36 196 100 81
14 9 12 12 196 81 144 144
Total 10 12 7 8 100 144 49 64
8 10 15 10 64 100 225 100
11 14 11 11 121 196 121 121
49 59 55 50 517 717 639 510

3
 Unit-II Design of Experiments

Step1: N= Total No of Observations = 20

Step 2: T=Grand Total = 49+59+55+50=213

(Grand total )2 T 2 (213) 2
Step 3: Correction Factor = = = = 2268.5
Total No of Observatio nsN 20
Step 4: SST =  X1 +  X 2 +  X 3 +  X 4 − C.F = 37 + 37 + 39 + 10 − 8.45 = 114.5
2 2 2 2

Step 5:

( X ) + ( X ) + ( X ) + ( X )
2 2 2 2

SSC =
1 2 3 4
-C.F
c1 c2 c3 c4
(−1)2 92 52
= + + + 0 − 8.45
5 5 5
SSC = 0.2 +16.2 + 5 − 8.45 = 12.95 Where ci = Number of elements in each column=5
Step 6: SSE=SST-SSC = 114.5 −12.95 = 101.5
Step 7: ANOVA TABLE
Source of Sum of Degree of
Mean Square F- Ratio
Variation Squares freedom

Between SSC
=4.317
SSC=12.95 C-1= 4-1=3 MSC = FC =
MSE
Columns C −1 MSC
6.35
SSE = =1.471
Error SSE=101.5 N-C=20-4=16 MSE = =6.35 4.317
N−C

Total SST=114.5 19

Cal FC = 1.471
Table value: FC (v2,v1)=FC (16,3)=26.83 at 1% level of significance.
Step 8 Conclusion: Cal F < Tab F0.01  There is no significance difference between the technicians
2. A completely randomized design experiment with 10 plots and 3 treatments gave the
following results.
Plot No : 1 2 3 4 5 6 7 8 9 10
Treatment : A B C A C C A B A B
Yield : 5 4 3 7 5 1 3 4 1 7
Analyse the results for treatment effects.
Solution:

4
 Unit-II Design of Experiments

A B C
5 4 3
7 4 5
3 7 1
1
Null Hypothesis H0: There is no significant difference in treatments
Alternate Hypothesis H1 : Significant difference in treatments
X1 X2 X3 X12 X22 X32
5 4 3 25 16 9
7 4 5 49 16 25
Total
3 7 1 9 49 1
1 1
16 15 9 84 81 35
N= Total No of Observations = 10, T=Grand Total = 40
(Grand total )2 T 2 402
Correction Factor = = = = 160
Total No of Observatio ns N 10
SST =  X12 +  X 22 +  X 32 − C.F = 84 + 81 + 35 − 160 = 40
( X ) + ( X ) + ( X )
2 2 2
(16)2 152 92
SSC = − C.F = + + − 160 == 64 + 75 + 27 − 160 = 6
1 2 3

c1 c2 c3 4 3 3
SSE=SST-SSC = 40 − 6 = 34
ANOVA TABLE
Source of Sum of Degree of
Mean Square F- Ratio
Variation Squares freedom

SSC
MSC =
Between C −1
SSC=6 C-1= 3-1=2
Columns 6
= =3 FC =
MSE
2 MSC
SSE 4.86
MSE = = = 1.62
N−C 3
Error SSE=34 N-C=10-3=7
34
= = 4.86
7

Total SST=40 9

Cal FC = 1.62
Table value : FC (7,2)=19.35 at 5% level of significance.

5
 Unit-II Design of Experiments

Conclusion : Cal F< Tab F0.05, We accept Null Hypothesis  There is no significance difference
in treatments
3. As head of the department of a consumers research organization you have the responsibility of
testing and comparing life times of 4 brands of electric bulbs. suppose you test the life time of 3
electric bulbs each of 4 brands, the data is given below, each entry representing the life time of an
electric bulb, measured in hundreds of hours. Perform an analysis of variance of the data test
the difference between the brands.
A B C D
20 25 24 23
19 23 20 20
21 21 22 20
Solution:
H0: There is no significant difference in four brands of electric bulbs.
H1: There is significant difference in four brands of electric bulbs.
X1 X2 X3 X4 X12 X22 X32 X42
20 25 24 23 400 625 576 529
19 23 20 20 361 529 400 400
21 21 22 20 441 441 484 400
TOTAL 60 69 66 63 1202 1595 1460 1329
N= Total No of Observations = 12, T=Grand Total = 258
(Grand total )2 T 2 2582
Correction Factor = = = = 5547
Total No of Observatio ns N 12
SST =  X12 + X22 + X32 + X42 -C.F=39
( X ) + ( X ) + ( X ) + ( X )
2 2 2 2

SSC = -C.F = 15
1 2 3 4

c1 c2 c3 c4
SSE=SST-SSC = 39 −15 = 24
ANOVA TABLE
Source of Sum of Degree of
Mean Square F- Ratio
Variation Squares freedom

Between SSC
=5
SSC=15 C-1= 4-1=3 MSC = MSC
Columns C −1 FC =
MSE
SSE 5
MSE = =
N−C 3
Error SSE=24 N-C=12-4=8
=3 = 1.667

Total SST=39 11

Cal FC = 1.667 & Table value of F (3,8)=4.07 at 5% level of significance.

Conclusion : Cal F < Tab F0.05 Hence we accepted H0

6
 Unit-II Design of Experiments

4. As part of the investigation of the collapse of the roof of a building, a testing laboratory is given
all the available bolts that connected all the steel structure at three different position on the roof.
The forces required to shear each of these bolts are as follows:
Position 90 82 79 98 83 91 -
1
Position 105 89 93 104 89 95 86
2
Position 83 89 80 94 - - -
3
Test the significant of the different position.
Solution:
H0: There is no significant difference in sample mean at the three positions.
H1: There is significant difference in sample mean at the three positions.
For simplifying calculation we take the average of given samples 90 and subtract 90 from
each given sample values as follows:

X1 X2 X3 X12 X22 X32

0 15 -7 0 225 49
-8 -1 -1 64 1 1
-11 3 -10 121 9 100
8 14 4 64 196 16
-7 -1 - 49 1 -
1 5 - 1 25 -
- -4 - - 16 -
TOTAL -17 31 -14 299 473 166
N= 17, T=0, Correction Factor =
T2
N
= 0, SST =  X + X + X -C.F = 938 − 0 = 938 ,
2
1
2
2
2
3

( X ) + ( X ) + ( X ) ( −17 ) ( 31) ( −14)

2 2 2 2 2 2

SSC = -C.F = + + − 0 = 234.452

1 2 3

c1 c2 c3 6 7 4
SSE=SST-SSC = 938 − 234.452 = 703.548
ANOVA TABLE
Source of Sum of Degree of
Mean Square F- Ratio
Variation Squares freedom

SSC
Between MSC =
SSC=234.452 C-1= 3-1=2 C −1 MSC
Columns FC =
=117.226 MSE
SSE 117.226
MSE = =50.253 = = 2.333
Error SSE=703.548 N-C=17-3=14 N−C 50.253

Total SST=938 16

Cal F = 2.333 & Table value of F (2, 14) =3.74 at 5% level of significance.

7
 Unit-II Design of Experiments

Conclusion : Cal F < Tab F0.05 Hence we accepted H0

There is no significant difference in sample mean at the three positions.
5. Four different manufacturing process were tried at three different stations and the average
measurements of quality characters of the product by these processes are given in the following
table. Perform an analysis of variance of the date and test for the difference between the
processes.
Process
A B C D
1 7 14 11 11
Station 2 15 16 14 10
3 8 15 10 12
Solution:
H0: There is no significant difference between the processes.
H1: There is a significant difference between the processes.
X1 X2 X3 X4 X12 X22 X32 X42
7 14 11 11 49 196 121 121
15 16 14 10 225 256 196 100
8 15 10 12 64 225 100 144
TOTAL 30 45 35 33 338 677 417 365
N= Total No of Observations = 12, T=Grand Total =143
(Grand total )2 T 2 1432
Correction Factor = = = = 1704.08
Total No of Observatio ns N 12
SST =  X12 + X22 + X32 + X42 -C.F = 1797 − 1704.08 = 92.92
( X ) + ( X ) + ( X ) + ( X ) (30) ( 45) (35) (33)
2 2 2 2 2 2 2 2

SSC = − C.F = + + + − 1704.08 = 42.25

1 2 3 4

c1 c2 c3 c4 3 3 3 3
SSE=SST-SSC = 92.92 − 42.25 = 50.67

ANOVA TABLE
Source of Sum of Degree of
Mean Square F- Ratio
Variation Squares freedom

Between SSC
=14.08
MSC
SSC=42.25 C-1= 4-1=3 MSC = FC =
Columns C −1 MSE
14.08
SSE =
MSE = =6.33 6.33
Error SSE=50.67 N-C=12-4=8 N−C
= 2.224

Total SST=92.92 11

Cal F = 2.224 & Table value of F (3,8) =4.07 at 5% level of significance.

Conclusion : Cal F < Tab F0.05 Hence we accepted H0. There is no significant difference the process

8
 Unit-II Design of Experiments

Two Way Classification (or) Randomized Block Design (R.B.D):

The entire experiment influences on only two factors rows and columns (say, treatments and
blocks) is two way Classification.

Working Procedure ( Two – Way classification )

Null Hypothesis H 0 : There is no significance difference between rows and between columns.

Alternate Hypothesis H1 : There is a significance difference between rows and between

columns.
Step 1: Find N= number of observations
Step 2: Find T = The total value of observations

T2
Step 3: Find the correction Factor = C.F=
N
Step 4: Calculate the total sum of squares = SST = (  X12 +  X 22 +  X 32 + ...) − C.F

 ( X )2 ( X )2 ( X )2 
Step 5: Find column sum of Square SSC =   1
+
 2
+
 3
+...  − C.F
 c1 c2 c3 
 
Where ci = Total number of observations in each column (i = 1, 2, 3,...)
 ( Y ) ( Y ) ( Y )2 
  
2 2

Step 6: Find Row sum of square = SSR =  1

+
2
+
3
+...  − C.F
 r1 r2 r3 
 
Where rj = Total number of observations in each Row ( j = 1, 2, 3,...)
Step 7: Find SSE=SST-SSC-SSR
Step 8: Prepare the ANOVA TABLE to calculate F-ratio .
Degree
Source of Sum of
of Mean Square F- Ratio
Variation Squares
freedom
MSC
If MSC  MSE , FC =
Between SSC MSE
SSC c-1 MSC =
Columns c −1 MSE
If MSE  MSC , FC =
MSC
MSR
If MSR  MSE , FR =
Between SSR MSE
SSR r-1 MSR =
Rows r −1 MSE
If MSE  MSR , FR =
MSR
SSE
Error SSE N-c-r+1 MSE =
N − c − r +1

9
 Unit-II Design of Experiments

Total SST rc-1

Step 9: Find the table value for both FC & FR (use F table) at  % level of significance with
( v1 , v2 ) degrees of freedom.
Step 10: Conclusion:
If Calculated value < Table Value, then we accept Null Hypothesis H 0 (or)
If Calculated value > Table Value, then we reject Null Hypothesis H 0 .

1. A Company appointments four salesmen A, B, C and D and observes their sales in 3

seasons: summer, winter and monsoon. The figures (in lakhs of Rs.) are given in the
following table:
Salesman
Season
A B C D
Summer 36 36 21 35
Winter 28 29 31 32
Monsoon 26 28 29 29
i) Do the salesmen significantly differ in performance?
ii) Is there significant difference between the seasons?
Solution:
Null Hypothesis H 0 : There is no significant difference between the sales in the 3 seasons and
also between the sales of the 4 salesmen.
Alternate Hypothesis H1 : There is a significant difference between the sales in the 3 seasons and
also between the sales of the 4 salesmen.

Test statistic:
To simplify calculations we deduct 30 from each value

Seasons A B C D Seasons
X12 X22 X32 X42
X1 X2 X3 X4 Total

Y1 Summer 6 6 -9 5 8 36 36 81 25

Y2 Winter -2 -1 1 2 0 4 1 1 4

Y3 Monson -4 -2 -1 -1 -8 16 4 1 1

Total 0 3 -9 6 0 56 41 83 30

Step1: N= Total No of Observations = 12

Step 2: T=Grand Total = 0

10
 Unit-II Design of Experiments

(Grand total )2 T 2 02
Step 3: Correction Factor = = = =0
Total No of Observatio ns N 12
Step 4: SST =  X1 + X 2 + X3 + X 4 -C.F = 56 + 41 + 83 + 30 − 0 = 210
2 2 2 2

Step 5:

( X ) + ( X ) + ( X ) + ( X )
2 2 2 2

SSC = − C.F
1 2 3 4

c1 c2 c3 c4
02 32 (−9)2 62
= + + + −0
3 3 3 3
SSC = 0 + 3 + 27 + 12 − 0 = 42
Where ci = Number of elements in each column

( Y ) + ( Y ) + ( Y )
2 2 2
82 02 (−8)2
Step 6: SSR = − C.F = + + − 0 = 16 + 0 + 16 − 0 = 32
1 2 3

r1 r2 r3 4 4 4
Where rj = Number of elements in each row

Step 7: SSE=SST-SSC-SSR = 210 − 42 − 32 = 136

Step 8: ANOVA TABLE:

Source of Sum of Degrees of Mean Sum of Variance F – ratio
Variation Squares Freedom Squares At 5% level
of
significance
Between SSC=42 c-1=4-1=3 SSC MSE
MSC = FC =
Columns c −1 MSC
(Salesmen) 22.67 FC (6,3) = 8.94
42 =
= = 14 14
3
= 1.619
Between SSR =32 r-1=3-1=2
MSR =
SSR
FR =
MSE FR (6, 2) = 19.3
rows r −1 MSR
(Seasons) 22.67
32 =
= = 16 16
2
= 1.417
Error SSE=136 N-c-r +1=6 SSE
MSE =
N − c − r +1
136
= = 22.67
6
Total 210 11
Conclusion:

11
 Unit-II Design of Experiments

1) Cal FC < Table FR,0.05 (6,3) , Hence we accept the H 0 and we conclude that there is no
significant difference between sales in the three seasons.
2) Cal FR < Table FR,0.05 (6, 2) , Hence we accept the H 0 and we conclude that there is no
significant difference between in the sales of 4 salesmen.
2. The following data represent the number of units of production per day turned out by
different workers using 4 different types of machines.

Machine type A B C D

Workers 1 44 38 47 36
2 46 40 52 43
3 34 36 44 32
4 43 38 46 33
5 38 42 49 39
(1) Test whether the five men differ with respect to mean productivity and
(2) Test whether the mean productivity is the same for the four different machine types.
Solution:
Null Hypothesis H0: There is no significant difference between the Machine types the Workers.
Alternate Hypothesis H1 : Significant difference between the Machine types between the Workers
Test statistic:
To simplify calculations we deduct 46 from each value

Machine type
workers
X12 X22 X32 X42
Total
worker A B C D
s X1 X2 X3 X4

Y1 -2 -8 1 -10 -19 4 64 1 100

Y2 0 -6 6 -3 -3 0 36 36 9

Y3 -12 -10 -2 -14 -38 144 100 4 196

Y4 -3 -8 0 -13 -24 9 64 0 169

Y5 -8 -4 3 -7 -16 64 16 9 49

Total -25 -36 8 -47 -100 221 280 50 523

Step1: N= Total No of Observations = 20

Step 2: T=Grand Total = -100

12
 Unit-II Design of Experiments

(Grand total )2 T 2 (−100) 2 10000

Step 3: Correction Factor = = = = = 500
Total No of Observatio nsN 20 20
Step 4: SST =  X1 +  X 2 +  X 3 +  X 4 − C.F = 221 + 280 + 50 + 523 − 500 = 574
2 2 2 2

Step 5:

( X ) + ( X ) + ( X ) + ( X )
2 2 2 2

SSC = − C.F
1 2 3 4

c1 c2 c3 c4
(−25)2 (−36) 2 82 (−47) 2
= + + + − 500 = 338.8
5 5 5 5

( Y ) + ( Y ) + ( Y ) + ( Y ) + ( Y )
2 2 2 2 2

Step 6: SSR = − C.F

1 2 3 4 5

r1 r2 r3 r4 r5
(−19)2 (−3) 2 (−38) 2 (−24) 2 (−16) 2
= + + + + − 500 = 161.5
4 4 4 4 4
Step 7: SSE=SST-SSC-SSR = 574 − 338.8 −161.5 = 73.7

Step 8: ANOVA TABLE

Source of Sum of Degrees of Mean Sum of Variance F – ratio
Variation Squares Freedom Squares At 5% level of
significance
Between SSC=338.8 c-1=4-1=3 SSC MSC
MSC= FC =
Columns c-1 MSE
(Salesmen) 338.8 112.9 FC (3,12) = 3.49
= =
3 6.14
= 112.9 = 18.39
Between SSR =161.8 r-1=5-1=4
MSR=
SSR
FR =
MSR FR (4,12) = 3.26
rows r-1 MSE
(Seasons) 161.5 40.4
= =
4 6.14
= 40.4 = 6.58
Error SSE=73.7 N-c-r +1 SSE
MSE=
=20-4-5+1 N-c-r+1
=12 73.7
=
12
= 6.14
Total SST=574.3 19
Conclusion:
1) For Columns: Calculated value > table value
Hence we reject the H 0 and we conclude that there is significant difference between the types of
machines

13
 Unit-II Design of Experiments

2) For Rows: Calculated value > table value

Hence we reject the H 0 and we conclude that there is significant difference between workers.
Hence there is significant difference between the Machine types and between the Workers
3. A laboratory technician measures the breaking strength of each of 5 kinds of linen threads
by using 4 different measuring instruments, and obtains the following results, in ounces.
I1 I2 I3 I4
Thread 1 20.9 20.4 19.9 21.9
Thread 2 25 26.2 27.0 24.8
Thread 3 25.5 23.1 21.5 24.4
Thread 4 24.8 21.2 23.5 25.7
Thread 5 19.6 21.2 22.1 22.1
Perform a 2 – way ANOVA using the 0.05 level of significance for both tests.
Solution:
Null Hypothesis H0: There is no significant difference between in breaking strength of various
threads and between measuring instruments.
Alternate Hypothesis H1: There is a significant difference between in breaking strength of
various threads and between measuring instruments.
To simplify calculations we deduct 22.1 from each value

Instruments
Thread I1 I2 I3 I4 Total X12 X22 X32 X42

X1 X2 X3 X4

Y1 -1.2 -1.7 -2.2 -0.2 -5.3 1.44 2.89 4.84 0.04

Y2 2.9 4.1 4.9 2.7 14.6 8.41 16.81 24.01 7.29

Y3 3.4 1 -0.6 2.3 6.1 11.56 1 0.36 5.29

Y4 2.7 -0.9 1.4 3.6 6.8 7.29 0.81 1.96 12.96

Y5 -2.5 -0.9 0 0 -3.4 6.25 0.81 0 0

Total 5.3 1.6 3.5 8.4 18.8 34.95 21.61 31.17 25.58

T 2 (18.8) 2
N= 20 , T= 18.8, Correction Factor = = = 17.672
N 20

14
 Unit-II Design of Experiments

SST =  X12 +  X 22 +  X 32 +  X 42 − C.F = 34.95 + 21.61 + 31.17 + 25.58 − 17.672 = 96.348

( X ) + ( X ) + ( X ) + ( X )
2 2 2 2

SSC = − C.F
1 2 3 4

c1 c2 c3 c4
(5.3) 2 (1.6)2 ( 3.5 ) (8.4)2
2

= + + + − 17.672 = 5.02
5 5 5 5
( Y ) + ( Y ) + ( Y ) + ( Y ) + ( Y )
2 2 2 2 2

SSR = − C.F
1 2 3 4 5

r1 r2 r3 r4 r5
(−5.3)2 (14.6)2 (6.1) 2 (6.8) 2 (−3.4) 2
= + + + + − 17.672 = 66.393
4 4 4 4 4
SSE=SST-SSC-SSR = 96.348 − 66.393 − 5.02 = 24.935

ANOVA TABLE
Source of Sum of Degrees of Mean Sum of Variance F – ratio
Variation Square Freedom Squares (F)
Between SSR R–1=4 MSR=16.598 MSC F0.05 (4,12) = 3.26
FR =
Rows =66.393 MSE
16.598
= = 7.987
2.078
Between SSC=5.02 C–1=3 MSC=1.673 MSE F0.05 (12,3) = 8.74
FC =
Columns MSC
2.078
= = 1.242
1.673
Error SSE (C-1) (R-1) MSE=2.078
=24.935 =12
Total SST N – 1 = 11
=96.348
Conclusion:

1) FR  3.26 . Hence, we reject H 0 and we conclude that there is significant difference between

threads.

2) FC  8.74 . Hence, we accept H 0 and we conclude that there is no significant

difference between instruments.

4. Four varieties A, B, C, D of a fertilizer are tested in a randomized block design with 4
replications. The plot yields in pounds are as follows:

15
 Unit-II Design of Experiments

Column / Row 1 2 3 4

1 A(12) D(20) C(16) B(10)

2 D(18) A(14) B(11) C(14)

3 B(12) C(15) D(19) A(13)

4 C(16) B(11) A(15) D(20)

Analyse the experimental yield.

Solution:
H0: There is no significant difference between the fertilizers and replication
H1 :Significant difference between the fertilizers and replication

Variety Block Total

varieties
1 2 3 4
X12 X22 X32 X42
(X1) (X2) (X3) (X4)

A (Y1 ) 12 14 15 13 54 144 196 225 169

B (Y2 ) 12 11 11 10 44 144 121 121 100

C (Y3 ) 16 15 16 14 61 256 225 256 196

D (Y4 ) 18 20 19 20 77 324 400 361 400

Total 58 60 61 57 236 868 942 963 865

Block

(Grand total)2 (236)2

N=16, T=Grand Total = 236, Correction Factor = = = 3481
Total No of Observations 16

SST =  X12 +  X 22 +  X 32 +  X 42 − C.F = 868 + 942 + 963 + 865 − 3481 = 157

( X ) + ( X ) + ( X ) + ( X )
2 2 2 2

SSC = − C.F = 841 + 900 + 930 + 812 − 3481 = 2

1 2 3 4

c1 c2 c3 c4

16
 Unit-II Design of Experiments

(Y ) + (Y ) + (Y ) + (Y )

2 2 2 2

SSR = − C.F
1 2 3 4

r1 r2 r3 r4
= 729 + 484 + 930.25 + 1482.25 − 3481
= 144.5

SSE = SST – SSC – SSR = 157-2-144.5=10.5

ANOVA Table

Source of Sum of Degree of Mean

F- Ratio FTab Ratio
Variation Squares freedom Square

Between MSR=
SSR=144.5 r - 1= 3 FR = 41.17 F5%(3,9) = 3.86
varieties 48.17

Between
SSC=2 c – 1=3 MSC = 0.67
blocks
FC = 1.75 F5%(9,3) = 8.81
(r – 1)( c – 1) MSE
Residual SSE = 10.5
=9 = 1.17

Total SST=157 15

Conclusion:

Cal FC<Tab FC and Cal FR> Tab FR  Therefore null hypothesis is rejected. Hence four varieties are
not similar. But the varieties are similar along block wise.

5. Consider the results given in the following table for an experiment involving 6 treatments in 4
randomized blocks. The treatments are indicated by numbers with in the parenthesis.

(1) (3) (2) (4) (5) (6)

I
24.7 27.7 20.6 16.2 16.2 24.9
(3) (2) (1) (4) (6) (5)
II
22.7 28.8 27.3 15 22.5 17
(6) (4) (1) (3) (2) (5)
III
26.3 19.6 38.5 36.8 39.5 15.4
(5) (2) (1) (4) (3) (6)
IV
17.7 31 28.5 34.9 34.9 22.9

Solution:
Null Hypothesis H0: There is no significant difference between the treatments and the blocks.
Alternative Hypothesis H1: Significant difference between the treatments and the blocks.

Test statistic:
To simplify calculations we deduct 25 from each value

17
 Unit-II Design of Experiments

Treatments
Total X12 X22 X32 X42 X52 X62

Blocks X1 X2 X3 X4 X5 X6

Y1 -0.3 -4.4 2.7 -8.8 -8.8 -0.1 -19.7 0.09 19.36 7.29 77.44 77.44 0.01

Y2 2.3 3.8 -2.3 -10 -8 -2.5 -16.7 5.29 14.44 5.29 100 64 6.25

Y3 13.5 14.5 11.8 -5.4 -9.6 1.3 26.1 182.25 210.25 139.24 29.16 92.16 1.69

Y4 3.5 6 9.9 9.9 -7.3 -2.1 19.9 12.25 36 98.01 98.01 53.29 4.41

Total 19 19.9 22.1 -14.3 -33.7 -3.4 9.6 199.88 280.05 249.83 304.61 286.89 12.36

N= Total No of Observations = 24, T=Grand Total = -11.5,

T 2 (9.6)2
Correction Factor = = = 3.84
N 24

SST =  X12 +  X 22 +  X 32 +  X 42 +  X 52 +  X 62 − C.F

= 199.88 + 280.05 + 249.83 + 304.61 + 286.89 + 12.36 − 3.84
= 1329.78
( X ) + ( X ) + ( X ) + ( X ) + ( X ) + ( X )
2 2 2 2 2 2

SSC = − C.F
1 2 3 4 5 6

c1 c2 c3 c4 c5 c6
(19)2 (19.9)2 (22.1)2 (−14.3)2 (−33.7)2 (−3.4) 2
= + + + + + − 3.84
4 4 4 4 4 4
= 645.44

( Y ) + ( Y ) + ( Y ) + ( Y )
2 2 2 2

SSR = − C.F
1 2 3 4

r1 r2 r3 r4
(−19.7) 2 (−16.7) 2 (26.1) 2 (19.9) 2
= + + + − 3.84
6 6 6 6
= 286.85

SSE = SST-SSC-SSR = 1329.78 − 645.44 − 286.85 = 397.49

18
 Unit-II Design of Experiments

ANOVA TABLE:
Source of Sum of Degrees Mean Sum of variance F – ratio
Variation Squares of Squares
Freedom
Between SSC=645.44 c-1=6-1=5 SSC MSC
MSC= FC =
Columns c-1 MSE
645.44 129.08 Fc (5,15) = 2.90
= =
5 26.49
= 129.08 = 4.87
Between SSR =286.85 r-1=4-1=3
MSR=
SSR
FR =
MSR FR (3,15) = 3.24
rows r-1 MSE
286.85 95.61
= =
3 26.49
= 95.61 = 3.60
Error SSE=397.49 N-c-r +1 SSE
MSE =
=24-6-4+1 N-c-r+1
=15 397.49
=
15
= 26.49
Total SST=1329.78 23
Conclusion:
1) Calculated value > table value
Hence, we reject H 0 and we conclude that there is significant difference between treatments
2) Calculated value > table value
Hence we reject H 0 and we conclude that there is significant difference between blocks.
Latin Square Design:
Latin Square design controls variation in two direction of the experimental materials as
rows and columns resulting in the reduction of experimental error. The analysis of the design
results in a three way classification of analysis of variance. Data from Latin Square experiments
form a three-way classification according the factors rows, columns, and treatments.
In Latin Square Design the treatments are grouped into replicates in two different ways,
such that each row and each column is a complete block, and the grouping for balanced
arrangement is performed by imposing the restriction that each of the treatment must appear
once and only once in each of the rows and only once in each of the column. The experimental
material should be arranged and the experiment conducted in such a way that the differences
among the rows and columns represent a major source of variation.
Hence a Latin Square Design is an arrangement of k treatments in a k x k squares, where
the treatments are grouped in blocks in two directions. It should be noted that in a Latin Square
Design the number of rows, the number of columns and the number of treatments must be equal.

19
 Unit-II Design of Experiments

In other words unlike Randomized Completely Block Design (RCBD) and Completely
Randomized Design (CRD)a Latin Square Design is a two restriction design, which provided the
facility of two blocking factors which are used to control the effect of two variable that influences
the response variable. Latin Square Design is called Latin Square because each Latin letter
represents the treatment that occurs once in a row and once in a column in such a way that in
respect of one criterion (restriction) rows are completely homogeneous blocks and in respect of
another criterion (second restriction) columns are completely homogeneous blocks.
The application of Latin Square Design is mostly in animal science, agriculture, and
industrial research, etc. A daily life example can be a simple game called Sudoku puzzle is also a
special case of Latin square design. The main assumption is that there is no contact between
treatments, rows and columns effect.
CRD RBD LSD
To influence one factor To influence two factor To influence more than
two factor

No restriction further No restriction on The number of

treatments treatment and replication of each
replications treatment is equal to
the number of
treatment
- Use only rectangular Use only Square filed
or Square field
The advantages of the Latin square design over other designs are:
(i) With a two way stratification or grouping, the Latin square controls more of the variation
than the CRD or the randomized completely block design. The two way elimination of variation
often results in small error mean square.
(ii) The analysis is simple.
(iii) Even with missing data the analysis remains relatively simple.
Working Procedure ( Three – Way classification )
we have seen data from a latin square experiment result in a three way classification
result in a three way classification say
(i) variety seeds
(ii) types of spacing(rows)
(iii) the letters for different manure treatment

20
 Unit-II Design of Experiments

H0: There is no difference between columns, between rows and between treatments
H1 : Not all are equal.

Step 1: Find N= number of observations

Step 2: Find T = The total value of observations

T2
Step 3: Find the correction Factor = C.F =
N
Step 4: Calculate the total sum of squares = SST = (  X12 +  X 22 +  X 32 + ...) − C.F

( X ) + ( X ) + ( X )
2 2 2

Step 5: Find column sum of Square SSC = + ... − C.F

1 2 3

c1 c2 c3
Where ci = Total number of observations in each column (i = 1, 2, 3,...)

( Y ) + ( Y ) + ( Y )
2 2 2

Step 6: Find Row sum of square = SSR = + ... − C.F

1 2 3

r1 r2 r3
Where rj = Total number of observations in each Row ( j = 1, 2, 3,...)
Step 7: Find SSK for letters and find SSE=SST-SSC-SSR-SSK
Step 8: ANOVA Table for three-way classification:

Source of Sum of Degree of

Mean Square F- Ratio
Variation square freedom

MSC
If MSC>MSE, FC =
Column SSC MSE
SSC n-1 MSC =
Treatment n −1 MSE
If MSC<MSE, FC =
MSC

MSR
If MSR>MSE, FR =
Row SSR MSE
SSR n-1 MSR =
Treatments n −1 MSE
If MSR<MSE, FR =
MSR

MSK
Between If MSK>MSE, FK =
SSK MSE
Treatments SSK n-1 MSK =
n −1 MSE
(Letters) If MSK<MSE, FK =
MSK

Error (or) SSE

SSE (n-1) (n-2) MSE =
Residual (n − 1)(n − 2)

21
 Unit-II Design of Experiments

Total SST n2-1

Step 9: Conclusion

1. Analyse the variance in the following latin square of yields (in kgs) of paddy where A, B,
C, D denote the different methods of cultivation.
D 122 A 121 C 123 B 122
B 124 C 123 A 122 D 125
A 120 B 119 D 120 C 121
C 122 D 123 B 121 A 122
Examine whether the different methods of cultivation have given significantly different
yields.
Solu.:
H0: There is no difference between columns, between rows and between treatments
H1 : Not all are equal.
We shift the origin Xij = xij – 120;

Variety X1 X2 X3 X4 TOTAL X1 2 X2 2 X3 2 X4 2
D2 A1 C3 B2 8
Y1 4 1 9 4
B4 C3 A2 D5 14
Y2 16 9 4 25
A0 B -1 D0 C1 0
Y3 0 1 0 1
C2 D3 B1 A2 8
Y4 4 9 1 4

TOTAL 8 6 6 10 30 24 20 14 34

n = 4, N = 16, T=Grand Total = 30 ;

(Grand total)2 (30)2

Correction Factor = = = 56.25
Total No of Observations 16

SST =  X12 +  X 22 +  X 32 − C.F = 92 − 56.25 = 35.75

( X ) + ( X ) + ( X ) + ( X )
2 2 2 2

SSC = − C.F = 81 − 56.25 = 24.75

1 2 3 4

c1 c2 c3 c4

22
 Unit-II Design of Experiments

( Y ) + ( Y ) + ( Y ) + ( Y )
2 2 2 2

SSR = − C.F = 59 − 56.25 = 2.75

1 2 3 4

r1 r2 r3 r4

Letters Total
A 1 2 0 2 5
B 2 4 -1 1 6
C 3 3 1 2 9
D 2 5 0 3 10
Total 30

SSK =
 (A) 2

+
 (B) 2

+
− C .F
 (C) 2

+
 (D) 2

4 4 4 4
25 36 81 100
= + + + − 56.25
4 4 4 4
= 60.5 − 56.25 = 4.25
SSE = SST – SSC – SSR-SSK = 35.75 – 24.75 – 2.75 – 4.25 = 4
ANOVA Table

Source of Sum of Degree of FTabRatio

Mean Square F- Ratio
Variation Squares freedom ( 5% level)

Between
SSR=24.75 n - 1= 3 MSR=8.25 FR= 12.31 FR(3, 6)=4.76
Rows

Between
SSC=2.75 n - 1= 3 MSC = 0.92 FC = 1.37 Fc(3, 6)=4 .76
Columns

Between
SSK = 4.25 n - 1= 3 MSK = 1.42 FK = 2.12 FK(3, 6)=4 .76
Letters
(n – 1)(n – 2)
Residual SSE= 4 MSE = 0.67
=6
Total 35.75

Conclusion :

Cal FC<Tab FC , Cal FK< Tab FK and Cal FR>Tab FR  There is significant difference between the rows , no
significant difference between the letters and no significant difference between the columns

2. The following is a Latin square of a design when 4 varieties of seeds are being tested. Set
up the analysis of variance table and state your conclusion. The following is a Latin square
of a design when 4 varieties of seeds are being tested. Set up the analysis of variance table
and state your conclusion. You may carry out suitable change of origin and scale.

23
 Unit-II Design of Experiments

A 105 B 95 C 125 D 115

C 115 D 125 A 105 B 105
D 115 C 95 B 105 A 115
B 95 A 135 D 95 C 115
(APRIL / MAY ‘17)
Solu.:

H0 : Four varieties are similar, H1 : Four varieties are not similar.

Let us take 100 as origin and divide by 5 for simplifying the calculation.

Variety X1 X2 X3 X4 TOTAL X1 2 X2 2 X3 2 X4 2

Y1 A1 B -1 C5 D3 8 1 1 25 9

Y2 C3 D5 A1 B1 10 9 25 1 1

Y3 D3 C -1 B1 A3 6 9 1 1 9

Y4 B -1 A7 D -1 C3 8 1 49 1 9
6 10 6 10 32 20 76 28 28
N=Total No of Observations = 16 , T=Grand Total = 32

(Grand total )2
Correction Factor = = 64
Total No of Observatio ns

SST =  X12 + X2 2 + X32 + X4 2 − C.F = 20 + 76 + 28 + 28 − 64 = 88

( X 1 ) 2 ( X 2 ) 2 ( X 3 ) 2 ( X 4 ) 2 (6)2 (10) 2 (6) 2 (10) 2

SSC = + + + − C.F = + + + − 64 = 4
c1 c2 c3 c4 4 4 4 4

( Y1 )2 ( Y2 ) 2 ( Y3 ) 2 ( Y4 ) 2 (8)2 (10)2 (6)2 (8)2

SSR = + + + − C.F = + + + − 64 = 2
r1 r2 r3 r4 4 4 4 4

To find SSK:
Treatment I II III IV Total
A 1 1 3 7 12
B -1 1 1 -1 0
C 5 3 -1 3 10
D 3 5 3 -1 10

24
 Unit-II Design of Experiments

( A)2 ( B) 2 ( C) 2 (  D) 2
SSK= + + + − C.F = 22
k1 k2 k3 k4

SSE= SST − SSC−SSR−SSK = 88-4-2-11=60

ANOVA Table
Degree
Source of Sum of of Mean FTabRatio
F- Ratio ( 5% level)
Variation Squares freedo Square
m
SSC FC(3, 6)=4.76
Between MSC = MSE
SSC=4 n-1=3 n −1 FC = =7.52
Column MSC
=1.33
SSR FR(3, 6)=4.76
Between MSR = MSE
SSR=2 n-1=3 n −1 FR = =14.9
Row MSR
=0.67
SSK FK (3, 6)=4.76
Between MSK = MSE
SSK=22 n-1=3 n −1 FK = =1.36
letters MSK
=7.33
MSE
Error (or) (n-1) (n- SSE
SSE=60 =
Residual 2)=6 (n − 1)(n − 2)
= 10

Conclusion:
(i) For columns: FC >FC(3, 6), There is a significant between columns.
(ii) For rows: FR >FR(3, 6), There is a significant between rows.
(iii) For letters: FK >FK(3, 6), There is no significant difference between treatments.
3. The following data resulted from an experiment to compare three burners B1, B2, and B3. A
Latin square design was used as the tests were made on 3 engines and were spread over 3
days.
Engine 1 Engine 2 Engine 3
Day 1 B1-16 B2-17 B3-20
Day 2 B2-16 B3-21 B1-15
Day 3 B3-15 B1-12 B2-13
Test the hypothesis that there is no difference between the burners. (Nov/Dec 2018)
Solution:

25
 Unit-II Design of Experiments

H0 : There is no significant difference between the burners.

H1 : There is significant difference between the burners.

Variety X1 X2 X3 TOTAL X1 2 X22 X3 2

B1-16 B2-17 B3-20
Y1 53 256 289 400
B2-16 B3-21 B1-15
Y2 52 256 441 225
B3-15 B1-12 B2-13
Y3 40 225 144 169

TOTAL 47 50 48 145 737 874 794

T 2 (145)
2

N=9, T = 145, correction Factor = C.F = = = 2336.11

N 9
SST =  X12 +  X 22 +  X 32 − C.F = 737 + 874 + 794 − 2336.11 = 68.89

( X ) + ( X ) + ( X ) ( 47 ) (50) ( 48)
2 2 2 2 2 2

SSC = − C.F = + + − 2336.11= 1.56

1 2 3

c1 c2 c3 3 3 3
( Y ) + ( Y ) + ( Y ) (53) (52) ( 40)
2 2 2 2 2 2

SSR = − C.F = + + − 2336.11 = 34.88

1 2 3

r1 r2 r3 3 3 3
To find SSK
Treatment I II III Total
B1 16 15 12 43
B2 17 16 13 46
B3 20 21 15 56

(  B1) + (  B 2) + (  B 3) ( 43) ( 46) (56)

2 2 22 2 2 2

SSK = − C.F = + + − 2336.11= 30.88

k1 k2 k3 3 3 3

SSE= SST − SSC−SSR−SSK = 68.86 - 1.55 - 34.88 - 30.88=1.55

ANOVA Table

Source of Sum of Degree of

Mean Square F- Ratio FTabRatio
Variation Squares freedom
( 5% level)

MSC FC(2,2)=19
Between SSC FC =
SSC=1.56 n-1=2 MSC = MSE
Column n −1
=1.006

26
 Unit-II Design of Experiments

=0.780

SSR FR(2,2)=19
Between MSR = FR =
MSR
SSR=34.88 n-1=2 n −1 MSE
Row
=17.44 =22.50

SSK MSK FK (2,2)=19

Between MSK = FK =
SSK=30.88 n-1=2 n −1 MSE
Letters
=15.44 =19.92

SSE
Error (or) (n-1) (n- MSE =
SSE=1.55 (n − 1)(n − 2)
Residual 2)=2
=0.775

Total SST=68.86 n2-1=8

Conclusion: Since FC <FC(2,2), FR >FR(2.2), FK >FK(2,2), Hence there is no significant between

columns and but there is a significant between rows and between burners (treatments).
4. The following is the Latin Square of a design when 4 varieties of seed are being tested. Set up the
analysis of variance table and state your conclusion. You can carry out the suitable change of origin
and scale. P 110 Q100 R 130 S 120
R 120 S 130 P 110 Q 110
S 120 R 100 Q 110 P 120
Q 100 P 140 S 100 R 120
Solution:
Null hypothesis H0 : Four varieties are similar
Alternative hypothesis H1 : Four varieties are not similar
Let us take 100 as origin and divide by 5 for simplifying the calculation

Variety X1 X2 X3 X4 TOTAL X12 X22 X32 X42

Y1 P2 Q0 R6 S4 12 4 0 36 16

Y2 R4 S6 P2 Q2 14 16 36 4 4

Y3 S4 R0 Q2 P4 10 16 0 4 16

Y4 Q0 P8 S0 R4 12 0 64 0 16
10 14 10 14 48 36 100 44 52

27
 Unit-II Design of Experiments

(Grand total )2 T2
N=16, T=48 , Correction Factor = = = 144
Total No of Observatio ns N

SST =  X12 + X2 2 + X32 + X4 2 − C.F = 36 + 100 + 44 + 52 − 144 = 88

( X ) + ( X ) + ( X ) + ( X )
2 2 2 2
(10)2 (14)2 (10) 2 (14) 2
SSC = − C.F = + + + − 144 = 4
1 2 3 4

c1 c2 c3 c4 4 4 4 4
( Y ) + ( Y ) + ( Y ) + ( Y )
2 2 2 2
(12)2 (14)2 (10)2 (12)2
SSR = − C.F = + + + − 144 = 2
1 2 3 4

r1 r2 r3 r4 4 4 4 4

To find SSK
Treatment 1 2 3 4 Total
P 2 2 4 8 16
Q 0 2 2 0 4
R 6 4 0 4 14
S 4 6 4 0 14

(  P ) + (  Q ) + (  R ) + (  S)
2 2 2 2
162 42 142 142
SSK = − C.F = + + + − 144 = 22
k1 k2 k3 k3 4 4 4 4

SSE= SST − SSC−SSR−SSK = 88-4-2-22=60

ANOVA Table
Source of Sum of Degree of
Mean Square F- Ratio
Variation Squares freedom
SSC
Between MSC = MSC
SSC=4 n-1=3 n −1 FC = =7.52
columns MSE
=1.33
SSR
Between MSR = MSR
SSR=2 n-1=3 n −1 FR = =14.9
rows MSE
=0.67
SSK
Between MSK = MSK
SSK=22 n-1=3 n −1 FK = =1.36
Letters MSE
= 7.33

28
 Unit-II Design of Experiments

MSE
Error (or) SSE
SSE=60 (n-1) (n-2)=6 =
Residual (n − 1)(n − 2)
= 10

Total SST=88 15

Table value F(3,6) degrees of freedom 8.94

1) Calculated value < table value
Hence, we accept H 0 and we conclude that there is significant difference between columns.
2) Calculated value > table value
Hence, we reject H 0 and we conclude that there is significant difference between rows.
3) Calculated value < table value
Hence, we accept H 0 and we conclude that there is significant difference between treatments.

5. A farmer wishes to test the effects of four different fertilizers A, B, C, D on the yield of wheat.
In order to eliminate sources of error due to variability in soil fertility he uses the fertilizers in
a Latin square arrangement as indicated below where the number indicate yields in Kilograms
per unit area. Perform an analysis of variance to determine if there is a significant difference
between the fertilizers at 0.01 level of significance.
A 18 C 21 D 25 B 11
D 22 B 12 A 15 C 19
B 15 A 20 C 23 D 24
C 22 D 21 B 10 A 17
(April/May 2019)
Solution: Subtract 15 from all the items

X1 X2 X3 X4 Total X21 X22 X23 X24

Y1 3 6 10 -4 15 9 36 100 16

Y2 7 -3 0 4 8 49 9 0 16

Y3 0 5 8 9 22 0 25 64 81

Y4 7 6 -5 2 10 49 36 25 4

Total 17 14 13 11 55 107 106 189 117

H0: There is no significant difference between fertilizers and yield.

H1 There is a significant difference between them.

29
 Unit-II Design of Experiments

T2
N = 16, T = 55, C.F =
N
= 189.06 , SST = X +X +X X
1
2
2
2
3
2 2
4 − C.F = 329.94

( X ) + ( X ) + ( X ) + ( X )
2 2 2 2
1 2 3 4
SSC = − C.F = 4.69
c1 c2 c3 c4

( Y ) + ( Y ) + ( Y ) + ( Y )
2 2 2 2
1 2 3 4
SSR = − C.F = 29.19
r1 r2 r3 r4

To find SSK

A B C D

3 -4 6 10

0 -3 4 7

5 0 8 9

2 -5 7 6

TOT 10 -12 25 32

(10) ( −12 ) ( 25) ( 32 )

2 2 2 2
T2
SSK = + + + − = 284.19
4 4 4 4 N

SSE = SST – SSC – SSR – SSK = 329.94 - 4.69 – 29.19 – 284.19 = 11.87
ANOVA TABLE

S.V DF SS MSS F cal F

tab

Between SSC MSE

3 SSC=4.69 MSC = = 1.56 FC = = 1.27 8.94
Column K −1 MSC

Between SSR MSR 4.76

3 SSR=29.19 MSR = = 9.73 FR = = 4.92
Row K −1 MSE

Between SSK MSK 4.76

3 SSK=284.19 MSK = = 94.73 FK = = 47.89
Treatment K −1 MSE

SSE
Error 6 SSE=11.87 MSE = = 1.978
(K − 1)(K − 2)

Tot 15 329.94

Conclusion:

(i). Cal Fc  Tab Fc , Accept H 0 (ii). Cal FR  Tab FR , Reject H 0 (iii). Cal FK  Tab FK ,Reject H 0

30
 Unit-II Design of Experiments

Factorial experiment:
A factorial experiment in which each of m factors at ‘S’ is called a symmetrical factorial
experiment and is often known as Sm factorial design. In factorial experiment, the effect of
several factors of variation are investigated simultaneously, the treatment being all the
combinations of different factors under study.
Definition:
2m - Factorial experiments means a symmetrical factorial experiment where each of the m-
factors is at two levels.

22 Factorial Design Experiment:

22 -a factorial experiment means a symmetrical experiment where each of the factors is at

two levels.

Note: 22 Factorial = 4 treatment (Let besay1,a,b,ab or 1,k,p,kp)

Procedure:
Step 1. Find N, T

T2
Step 2. Find Correction factor = C.F=
N
Step 3. We proceed two-way classification between treatment and blocks.

Step 4. For 2  2 or 22 factorial :

1 1 1
a+ab-b-(1)  , SSB =  b+ab-a-(1) , SSAB = ab+(1)-a-b 
2 2 2
Find SSA =
N N N
Step 5. Find SSE = SST-SSA-SSB-SSAB
Step 6. ANOVA Table:

Source of Sum of Degree of

Mean Square F- Ratio
Variation Square freedom

MSC
If MSC>MSE, FC =
Between SSC MSE
SSC n-1 MSC =
Column n −1 MSE
If MSC<MSE, FC =
MSC

MSR
If MSR>MSE, FR =
Between SSR MSE
SSR n-1 MSR =
Row n −1 MSE
If MSR<MSE, FR =
MSR

31
 Unit-II Design of Experiments

MSA
If MSA>MSE, Fa =
MSE
A SSA 1 MSA = SSA
MSE
If MSA<MSE, Fa =
MSA

MSB
If MSB>MSE, Fb =
MSE
B SSB 1 MSB = SSB
MSE
If MSB<MSE, Fb =
MSB

MSAB
If MSAB>MSE, Fab =
MSE
AB SSAB 1 MSAB = SSAB
MSE
If MSAB<MSE, Fab =
MSAB

Error (or) SSE

SSE N − c − r + 1 MSE =
Residual N − c − r +1
Step 7. Conclusion
Problem:
1. Given the following observation for the 2 factors A & B at two levels, Compute (i) the main effect (ii) make
an analysis of variance.

Treatment
Replication I Replication II Replication III
Combination

(1) 10 14 9

A 21 19 23

B 17 15 16

AB 20 24 25

Solution:
H0: There is no significant difference between treatment combinations.
H1: There is significant difference between treatment combinations.
We code the data by subtracting 20,

Variety X1 X2 X3 TOTAL X1 2 X2 2 X3 2
-10 -6 -11 -27 100 36 121
(1) Y 1

32
 Unit-II Design of Experiments

1 -1 3 3 1 1 9
(a) Y 2
-3 -5 -4 -12 9 25 16
(b) Y 3
0 4 5 9 0 16 25
(ab) Y 4

TOTAL -12 -8 -7 -27 110 78 171

N = 12, T= Grand total = -27,

(Grand total)2 (−27)2
correction Factor = = = 60.75
Total no of Observations 12

SST =  X12 + X2 2 + X32 − C.F = 110 + 78 + 171 − 60.75 = 298.25

( X ) + ( X ) + ( X ) ( −12) ( −8) ( −7 )
2 2 2 2 2 2

SSC = − C.F = + + − 60.75 = 3.5

1 2 3

c1 c2 c3 4 4 4
( Y ) + ( Y ) + ( Y ) + ( Y )
2 2 2 2

SSR = − C.F
1 2 3 4

r1 r2 r3 r4
( −27 ) ( 3) ( −12 ) (9)
2 2 2 2

= + + − 60.75 = 260.25 +
3 33 3
1 1
SSA =  a+ab-b-(1)  = = 3 + 9 − ( −12 ) − ( −27 )  = 216.75
2 2

N 12
1 1
SSB =  b+ab-a-(1) = ( −12 ) + 9 − 3 − ( −27 )  = 36.75
2 2

N 12
1 1
SSAB =  ab+(1)-a-b  = 9 + ( −27 ) − 3 − ( −12 )  = 6.75
2 2

N 12
SSE = SST – SSA–SSB–SSAB = 298.25 –216.75–36.75–6.75= 38

ANOVA Table

Degree FTab
Source of Sum of
of Mean Square F- Ratio
Variation Degrees
freedom

If MSA>MSE F5%(1, 6) = 5.99

MSA
A SSA=216.75 1 MSA=216.75 FA =
MSE
= 34.24

33
 Unit-II Design of Experiments

If MSB>MSE F5%(1, 6) = 5.99

MSB
B SSB=36.75 1 MSB=36.75 FB =
MSE
= 5.81

If MSAB>MSE F5%(1, 6) = 5.99

MSAB
AB SSAB=6.75 1 MSAB=6.75 FAB =
MS E
= 1.07

Error SSE=38 6 MSE=6.33

Conclusion:
Cal FA > Tab FA , Reject H0 , Cal FB < Tab FB , Accept H0 , Cal FAB < Tab FAB , Accept H0

2. Analyse 22 factorial experiment for the following table;

Block Treatment
(1) (k) (p) (kp)
I
64 25 30 6
(p) (kp) (1) (k)
II
50 33 75 14
(k) (p) (kp) (1)
III
12 41 17 76
(kp) (1) (k) (p)
IV
10 75 33 25

Solution:

Treatment
I II III IV

(l) 64 75 76 75
(k) 25 14 12 33
(p) 30 50 41 25
(kp) 6 33 17 10

H0: There is no significant difference between the treatments.

H1: There is significant difference between the treatments.

34
 Unit-II Design of Experiments

We shift the origin Xij = xij – 37;

Variety X1 X2 X3 X4 TOTAL X1 2 X2 2 X3 2 X4 2
27 38 39 38 1444
(1) Y 1 142 729 1444 1521
-12 -23 -25 -4 16
(k) Y 2 -64 144 529 625
7 13 4 -12 144
(p) Y 3 12 49 169 16
-31 -4 -20 -27 729
(kp) Y 4 -82 961 16 400

TOTAL -9 24 -2 -5 8 1883 2158 2562 2333

N = 16, T= Grand total = 8,

T 2 ( 8)
2

correction Factor = C.F = = =4

N 16
SST =  X 12 +  X 22 +  X 32 +  X 42 − C.F
= 1883 + 2158 + 2562 + 2333 − 4
= 8932

( X ) + ( X ) + ( X ) + ( X )
2 2 2 2

SSC = − C.F
1 2 3 4

c1 c2 c3 c4
( −9 ) ( 24 ) ( −2 ) ( −5)
2 2 2 2

= + + + − 4 = 167.5
4 4 4 4
(Y ) + (Y ) + (Y ) + (Y )
2 2 2 2

SSR = − C.F
1 2 3 4

r1 r2 r3 r4
(142 ) ( −64 ) (12 ) ( −82 )
2 2 2 2

= + + + − 4 = 7778
4 4 4 4
1 1
 k + kp − p − (1) = ( −64 ) + ( −82 ) − (12 ) − 142  = 5625
2 2
SSk=
N 16

1 1
 p+kp-k-(1) = (12 ) + ( −82 ) − ( −64 ) − 142 = 1369
2 2
SSp =
N 16

1 1
 kp+(1) - k - p  = ( −82 ) +(142) - ( −64 ) - (12 ) = 784
2 2
SSk p=
N 16

SSE = SST -SSk -SSp-SSkp = 8932 -5625 − 1369 − 784 = 1154

35
 Unit-II Design of Experiments

ANOVA Table
Degree FTab
Source of Sum of
of Mean Square F- Ratio
Variation Degrees
freedom

Between If MSC<MSE
SSC
SSC=167.5 3 MSC = = 55.83
Column n −1 FC =
MSE F0.05(9,3)=8.812
MSC
= 2.30

If MSR>MSE
SSR
Between MSR = MSR F0.05(3 ,9)=3.86
SSR=7778 3 n −1 FR =
Row = 2592.67 MSE
= 20.22

F0.05(1 ,9)=5.12
If MSk>MSE

MSk = SSk = 5625 MSk

k SSk=5625 1 Fk =
MSE
= 43.87

If MSp>MSE F0.05(1 ,9)=5.12

MSp = SSp = 1369 MSp

p SSp=1369 1 Fp =
MSE
= 10.68

If MSkp>MSE F0.05(1 ,9)=5.12

MSkp = SSkp MSkp
kp SSkp=784 1
= 784 Fkp =
MSE
= 6.114

SSE
Error (or) MSE =
SSE=1154 9 N − c − r +1
Residual
= 128.22

Conclusion : Cal Fk > Tab Fk , Cal Fp > Tab Fp and Cal Fkp> Tab Fkp There is significant difference
between the treatments.

36