9-Fundamentals of Designsf
9-Fundamentals of Designsf
=
, is called a contrast of s ' y
i
.
Fundamentals of Design of Experiments
II-129
Let y
1
, y
2
, ,y
n
be independent random variables with a common mean and variance
2
.
The expected value of the random variable C is zero and its variance is . l
n
1 i
2
i
2
In what
follows we shall not distinguish between a contrast and its corresponding random variable.
Sum of squares (s.s.) of contrasts. The sum of squares due to the contrast C is defined as
) C ( Var / C
2 2
=
=
n
1 i
2
i
2
l / C . Here
2
is unknown and is replaced by its unbiased
estimate, i.e. mean square error. It is known that this square has a
2 2
distribution with
one degree of freedom when the s ' y
i
are normally distributed. Thus the sum of squares
due to two or more contrasts has also a
2 2
distribution if the contrasts are independent.
Multiplication of any contrast by a constant does not change the contrast. The sum of
squares due to a contrast as defined above is not evidently changed by such multiplication.
Orthogonal contrasts. Two contrasts,
i
n
1 i
i 1
y l C
=
= and
i
n
1 i
i 2
y l C
=
= are said to be
orthogonal if and only if 0 m l
n
1 i
i i
=
=
. This condition ensures that the covariance between
1
C and
2
C
is zero.
When there are more than two contrasts, they are said to be mutually orthogonal if they are
orthogonal pair wise. For example, with four observations
4 3 2 1
y , y , y , y , we may write
the following three mutually orthogonal contrasts:
(i)
4 3 2 1
y y y y +
(ii)
4 3 2 1
y y y y +
(iii)
4 3 2 1
y y y y +
The sum of squares due to a set of mutually orthogonal contrasts has a
2 2
distribution
with as many degrees of freedom as the number of contrasts in the set.
Maximum number of orthogonal contrasts. Given a set of n values
n 2 1
y , , y , y L , the
maximum number of mutually orthogonal contrasts among them is n - 1. One way of
writing such contrasts is to progressively introduce the values as below:
(i)
2 1
y y
(ii)
3 2 1
y 2 y y +
: :
: :
(n) ( )
n 1 n 2 1
y 1 n y y y + + +
L .
Fundamentals of Design of Experiments
II-130
Another set of orthogonal contrasts for values of n is available in the Tables for Biological,
Agricultural and Medical Research prepared by Fisher and Yates (1963) under the name of
orthogonal polynomials.
To be specific about treatment effects let
i i
t l denote a treatment contrast, 0 l
i
i
=
.
The BLUE of
i i
t l is
i
i i
t
l ( Var
i
i
i
, where
i
t is the
parameter pertaining to the treatment effect i. The sum of squares due to contrast
i
i i
t l
is
i
i
i i
i
i
t l r a V t l
2
2
where
2
is the error variance estimated by the error mean
squares, MSE. The significance of the contrast can be tested using the statistic
i
i
i
i
i i
t l r a V
t l
t
which follows the Student's t-distribution with degrees of freedom same as that of error.
The null hypothesis is rejected at % level of significance if the tabulated value of
) edf , 2 / 1 (
t
is greater than computed t-value. Here edf represents the error degrees of
freedom. F-test can be used instead of t-test using the relationship that
1
1
n , 1
2
n
F t = .
Contrasts of the type
m i
t t in which experimenters are often interested are obtainable
from
i
i
i
t l
by putting 1 l , 1 l
m i
= = and zero for the other l's. These contrasts are called
as elementary contrasts and are useful for pairwise comparisons.
Besides hypothesis testing, the experimenter may also be interested in obtaining a
confidence interval. In the sequel, we shall give a formula for a confidence interval for an
individual contrast. If confidence intervals for more than one contrast are required, then
the multiple comparison methods should be used instead. A-100 (1 - )% confidence
interval for the contrast
i i
t l
is
i
i
i 2 / , edf i i i i i
i
i 2 / , edf i i
t
l r a V t t
l t l t
l r a V t t
l
.
We can write this more succinctly as
i i
t l
i
i
i 2 / , edf i i
t
l r a V t t
l
Fundamentals of Design of Experiments
II-131
where the symbol denotes that the upper limit of the interval is calculated using +and
the lower limit using - and edf is the number of degrees of freedom for error. The symbol
i i
t l mean that the interval includes the true value of contrast
i i
t l with 100(1 -
)% confidence.
The outcome of a hypothesis test can be deduced from the corresponding confidence
interval in the following way. The null hypothesis
=
i
i i 0
h t l : H will be rejected at
significance level in favor of the two-sided alternative hypothesis
i
i i 1
h t l : H if the
corresponding confidence interval for
i
i i
t l fails to contain h.
So far we have discussed experimental situations where one is interested in a single
treatment contrast. However, there may be situations when one is interested in a group of
treatment contrasts L t, where L is a p v matrix such that 0 1 L = , Rank (L) = p, and t
=( ) , , ,
2 1
v
t t t L is a v 1 vector of treatment effects. The sum of squares due to a set of
treatment contrasts L t is (L t
( )
L C L L t
, the
best linear unbiased estimator of L t, is D(L t
)=
2
( ) L C L
and C is the coefficient
matrix of reduced normal equations for estimating the linear functions of treatment effects.
The null hypothesis of interest say is L :
0
H t = 0 against L :
1
H t 0. The null
hypothesis H
0
is tested using the statistic F=
MSE
Contrasts) of SS(set
with p and edf (error
degrees of freedom) degrees of freedom. If L comprises of a complete set of (v-1)
linearly independent parametric functions, i.e., p = v-1, then we can get the treatment sum
of squares as we get in the ANOVA table. For more details on contrast analysis, a
reference may be made to Dean and Voss (1999).
In multi-factor experiments, the treatments are combinations of levels of several factors. In
these experimental situations, the treatment sum of squares is partitioned into sum of
squares due to main effects and interactions. These sums of squares can also be obtained
through contrast analysis. The procedure of obtaining sum of squares due to main effects
and interactions is discussed in the sequel.
2.2 Main Effects and Interactions
In general, let there be n-factors, say
n
F F F ..., , ,
2 1
and
th
i factor has
i
s levels, n i ,..., 1 = .
The ) (
1
=
=
n
i
i
s v treatment combinations in the lexico-graphic order are given by
n
a a a ...
2 1
where denotes the symbolic direct product and
( ) n i s
i i
,..., 2 , 1 ; ,..., 1 , 0
1
= =
a . Renumber the treatment combinations from 1 to v and
analyze the data as per procedure of general block designs for single factor experiments.
The treatment sum of squares obtained from the ANOVA is now to be partitioned into
Fundamentals of Design of Experiments
II-132
main effects and interactions. This can easily be done through contrast analysis. One has to
define the set of contrasts for each of the main effects and interactions. Before describing
the procedure of defining contrasts for main effects and interactions, we give some
preliminaries. The total number of factorial effects (main effects and interactions) are
1 2
n
. The set of main effects and interactions have a one-one correspondence with ,
the set of all n-component non-null binary vectors. For example a typical p-factor
interaction.
( ) n p n g g g F F F
p g g g
p
1 , ... 1 ,..., ,
2 1
2 1
corresponds to the element
( )
n
x x x ,...,
1
= of such that 1 ...
2 1
= = = =
p
g g g
x x x and 0 =
u
x for
p
g g g u ,..., ,
2 1
.
The treatment contrasts belonging to different interactions ( ) =
n
x
x x x F ,..., ,
1
are
given by
t P
x
, where
n
x
n
x x
x
P P P P = ...
2 1
2 1
where
i
x
i
i
P P = if 1 =
i
x
i
s
' 1 = if 0 =
i
x
where
i
P is a ( )
i i
s s 1 matrix of complete set of linearly independent contrasts of order
i
s and
i
s
1 is a 1
i
s vector of ones. For example, if 4 =
i
s , then
=
3 1 1 1
0 2 1 1
0 0 1 1
i
P .
For sum of squares of these contrasts and testing of hypothesis, a reference may be made to
section 2.1.
In the sequel we describe some basic designs.
3. Completely Randomized Design
Designs are usually characterized by the nature of grouping of experimental units and the
procedure of random allocation of treatments to the experimental units. In a completely
randomized design the units are taken in a single group. As far as possible the units
forming the group are homogeneous. This is a design in which only randomization and
replication are used. There is no use of local control here.
Let there be v treatments in an experiment and n homogeneous experimental units. Let the
i
th
treatment be replicated
i
r times (i = 1,2,, v) such that n r
v
1 i
i
=
=
. The treatments are
allotted at random to the units.
Normally the number of replications for different treatments should be equal as it ensures
equal precision of estimates of the treatment effects. The actual number of replications is,
however, determined by the availability of experimental resources and the requirement of
Fundamentals of Design of Experiments
II-133
precision and sensitivity of comparisons. If the experimental material for some treatments
is available in limited quantities, the numbers of their replication are reduced. If the
estimates of certain treatment effects are required with more precision, the numbers of their
replication are increased.
Randomization
There are several methods of random allocation of treatments to the experimental units.
The v treatments are first numbered in any order from 1 to v. The n experimental units are
also numbered suitably. One of the methods uses the random number tables. Any page of
a random number table is taken. If v is a one-digit number, then the table is consulted digit
by digit. If v is a two-digit number, then two-digit random numbers are consulted. All
numbers greater than v including zero are ignored.
Let the first number chosen be
1
n ; then the treatment numbered
1
n is allotted to the first
unit. If the second number is
2
n which may or may not be equal to n
1
then the treatment
numbered
2
n is allotted to the second unit. This procedure is continued. When the i
th
treatment number has occurred
i
r times, ( ) v ,..., 2 , 1 i = this treatment is ignored
subsequently. This process terminates when all the units are exhausted.
One drawback of the above procedure is that sometimes a very large number of random
numbers may have to be ignored because they are greater than v. It may even happen that
the random number table is exhausted before the allocation is complete. To avoid this
difficulty the following procedure is adopted. We have described the procedure by taking
v to be a two-digit number.
Let P be the highest two-digit number divisible by v. Then all numbers greater than P and
zero are ignored. If a selected random number is less than v, then it is used as such. If it is
greater than or equal to v, then it is divided by v and the remainder is taken to the random
number. When a number is completely divisible by v, then the random number is v. If v is
an n-digit number, then P is taken to be the highest n-digit number divisible by v. The rest
of the procedure is the same as above.
Alternative methods of random allocation
If random number tables are not available, treatments can be allotted by drawing lots as
below. Let the number of the i
th
treatment be written on
i
r pieces of papers ( ). v ,..., 2 , 1 i =
The
=
=
v
1 i
i
n r pieces of papers are then folded individually so that the numbers written on
them are not visible. These papers are then drawn one by one at random. The treatment
that is drawn in the t
th
draw is allotted to the t
th
plot ( ). n ,..., 2 , 1 t =
Random allocation is also possible by using a fair coin. Let there be five treatments each
to be replicated four times. There are, therefore, 20 plots. Let these plots be numbered
from 1 to 20 conveniently.
Fundamentals of Design of Experiments
II-134
When a coin is tossed, there are two events that is, either the head comes up, or the tail.
We denote the "head" by H and the "tail" by T. When the coin is tossed twice, there are
four events, that is, both times head HH; first head next tail HT: first tail next head TH and
both times tail TT. Similarly, when the coin is thrown three times, there are the following
eight possible events:
HHH, HHT, HTH, HTT, THH, THT, TTH, TTT.
Similar events can be written easily for four or more number of throws of the coin.
The five treatments are now labeled not by serial numbers as earlier but by any five of the
above eight events obtainable by tossing three coins. Let us use the first five events and
omit THT, TTH and TTT.
A coin is now thrown three times and the event happened noted. If the event is any of the
first five events described above, the treatment labeled by it is allotted to the first
experimental unit. If the event happened is any of the last three, it is ignored. The coin is
again tossed three times and this event is used to select a treatment for the second
experimental unit. If the same event occurs more than once, we are not to reject it until the
number of times it has occurred equals the number of replications of the treatment it
represents. This process is continued till all the experimental units are exhausted.
Analysis
This design provides a one-way classified data according to levels of a single factor. For
its analysis the following model is taken:
, r 1, j v; , 1, i , e t y
i ij i ij
L L = = + + =
where
ij
y is the random variable corresponding to the observation
ij
y obtained from the j
th
replicate of the i
th
treatment, is the general mean,
i
t is the fixed effect of the i
th
treatment
and
ij
e is the error component which is a random variable assumed to be normally and
independently distributed with zero means and a constant variance
2
.
Let ( ) v ,..., 2 , 1 i T y
i
j
ij
= =
=
Total sum of squares = . F . C y
v
1 i
r
1 j
2
ij
i
= =
Fundamentals of Design of Experiments
II-135
ANALYSIS OF VARIANCE
Sources of
variation
Degrees of
freedom (D.F.)
Sum of squares
(S.S.)
Mean squares
(M.S.)
F
Treatments v 1 SST
. F . C
r
T
v
1 i
i
2
i
=
=
MST = SST / (v - 1)
MST/MSE
Error n v SSE = by
subtraction
MSE =
SSE / (n - v)
Total n 1
. F . C y
ij
2
ij
The hypothesis that the treatments have equal effects is tested by F-test where F is the ratio
MST / MSE with (v - 1) and (n - v) degrees of freedom. We may then be interested to
either compare the treatments in pairs or evaluate special contrasts depending upon the
objectives of the experiment. This is done as follows:
For a completely randomized design, the BLUE of the treatment contrast
i i
t l is
i
i i
t
l =
i
i i
y l , where
i i i
r / T y = ,
=
i
i
2
i 2
i
i
i
r
l
) t
l ( Var , where
2
is the error
variance estimated by the error mean squares, MSE. The sum of squares due to contrast
i
i i
t
l is
i
i
2
i
2
i
i
i
r
l
/ Y l
The significance of the contrast can be tested by t test, where
=
i
i
2
i
i
i i
r
l
MSE
y l
t
where
) v n ( , 2 / 1
t
is the value of Student's t at the level of significance and degree of
freedom (n - v). Contrasts of the type
m i
t t in which experimenters are often interested
are obtainable from
i
i
i
t l
by putting 1 l , 1 l
m i
= = and zero for the other l's. Such
comparisons are known as pairwise comparisons.
Sometimes the levels of the treatment factors divide naturally into two or more groups, and
the experimenter is interested in the difference of averages contrast that compares the
average effect of one group with the average effect of the other group(s). For example,
consider an experiment that is concerned with the effect of different colors of exam paper
(the treatments) on students exam performance (the response). Suppose that treatments 1
Fundamentals of Design of Experiments
II-136
and 2 represent the pale colors, white and yellow, whereas treatments 3, 4 and 5 represent
the darker colors, blue, green and pink. The experimenter may wish to compare the effects
of light and dark colors on exam performance. One way of measuring this is to estimate
the contrast ( ) ( ), t t t
3
1
t t
2
1
5 4 3 2 1
+ + + , which is the difference of the average effects of
the light and dark colors. The corresponding contrast coefficients are
3
1
- ,
3
1
- ,
3
1
- ,
2
1
,
2
1
The BLUE of the above contrast would be
5 4 3 2 1
y
3
1
y
3
1
y
3
1
y
2
1
y
2
1
+ with
estimated standard error as . )
9r
1
9r
1
9r
1
4r
1
4r
1
( MSE
5 4 3 2 1
+ + + +
A )% 1 ( 100 confidence interval for the contrast
i
i i
t l is
+
i
2
i
2 / , v n i i i i
i
2
i
2 / , v n i i
r
l
MSE t y l t l
r
l
MSE t y l
.
4. Randomized Complete Block Design
It has been seen that when the experimental units are homogeneous then a CRD should be
adopted. In any experiment, however, besides treatments the experimental material is a
major source of variability in the data. When experiments require a large number of
experimental units, the experimental units may not be homogeneous, and in such situations
CRD can not be recommended. When the experimental units are heterogeneous, a part of
the variability can be accounted for by grouping the experimental units in such a way that
experimental units within each group are as homogeneous as possible. The treatments are
then allotted randomly to the experimental units within each group (or blocks). The
principle of first forming homogeneous groups of the experimental units and then allotting
at random each treatment once in each group is known as local control. This results in an
increase in precision of estimates of the treatment contrasts, due to the fact that error
variance that is a function of comparisons within blocks, is smaller because of
homogeneous blocks. This type of allocation makes it possible to eliminate from error
variance a portion of variation attributable to block differences. If, however, variation
between the blocks is not significantly large, this type of grouping of the units does not
lead to any advantage; rather some degrees of freedom of the error variance is lost without
any consequent decrease in the error variance. In such situations it is not desirable to adopt
randomized complete block designs in preference to completely randomized designs.
If the number of experimental units within each group is same as the number of treatments
and if every treatment appears precisely once in each group then such an arrangement is
called a randomized complete block design.
Fundamentals of Design of Experiments
II-137
Suppose the experimenter wants to study v treatments. Each of the treatments is replicated
r times (the number of blocks) in the design. The total number of experimental units is,
therefore, vr. These units are arranged into r groups of size v each. The error control
measure in this design consists of making the units in each of these groups homogeneous.
The number of blocks in the design is the same as the number of replications. The v
treatments are allotted at random to the v plots in each block. This type of homogeneous
grouping of the experimental units and the random allocation of the treatments separately
in each block are the two main characteristic features of randomized block designs. The
availability of resources and considerations of cost and precision determine actual number
of replications in the design.
Analysis
The data collected from experiments with randomized block designs form a two-way
classification, that is, classified according to the levels of two factors, viz., blocks and
treatments. There are vr cells in the two-way table with one observation in each cell. The
data are orthogonal and therefore the design is called an orthogonal design. We take the
following model:
,
r ,..., 2 , 1 j
; v ,..., 2 , 1 i
, e b t y
ij j i ij
=
=
+ + + =
where
ij
y denotes the observation fromi
th
treatment in j
th
block. The fixed effects
j i
b , t ,
denote respectively the general mean, effect of the i
th
treatment and effect of the j
th
block.
The random variable
ij
e is the error component associated with
ij
y . These are assumed to
be normally and independently distributed with zero means and a constant variance
2
.
Following the method of analysis of variance for finding sums of squares due to blocks,
treatments and error for the two-way classification, the different sums of squares are
obtained as follows: Let ( ) v ,..., 2 , 1 i T y
i
j
ij
= =
,
Sum of squares due to blocks . F . C
v
B
j
2
j
=
Fundamentals of Design of Experiments
II-138
ANALYSIS OF VARIANCE
Sources of
variation
Degrees of
freedom (D.F.)
Sum of squares
(S.S.)
Mean squares
(M.S.)
F
Blocks r - 1
SSB . F . C
v
B
j
2
j
=
MSB = SSB / (r - 1)
MSB/MSE
Treatments v - 1
SST . F . C
r
T
i
2
i
=
MST = SST / (v - 1)
MST/MSE
Error (r - 1)(v - 1) SSE = by subtraction MSE =
SSE / (v - 1)(r - 1)
Total vr - 1
. F . C y
ij
2
ij
The hypothesis that the treatments have equal effects is tested by F-test, where F is the
ratio MST / MSE with (v - 1) and (v - 1)(r - 1) degrees of freedom. We may then be
interested to either compare the treatments in pairs or evaluate special contrasts depending
upon the objectives of the experiment. This is done as follows:
Let
i i
t l denote a treatment contrast, 0 l
i
i
=
. The BLUE of
i i
t l is
i
i i
t
l =
i
i i
y l ,
where r / T y
i i
= ,
=
i
2
i
2
i
i
i
l
r
) t
l ( Var
, where
2
is estimated by the error mean
squares, MSE. The sum of squares due to contrast
i
i i
t
l is
r / l / y l
i
2
i
2
i
i
i
. The
significance of the contrast can be tested as per procedure described in sections 2 and 3.
The )% - (1 100 confidence interval for this contrast is
+
r / l MSE t y l
t l r / l MSE t y l
2
i 2 / , v ) 1 r )( 1 v ( i i
i i
2
i 2 / ), 1 r )( 1 v ( i i
As we know that the outcome of a hypothesis test can be deduced from the corresponding
confidence interval in the following way. The null hypothesis
=
i
i i 0
0 t l : H will be
rejected at significance level in favor of the two-sided alternative hypothesis
i
i i 1
0 t l : H if the corresponding confidence interval for
i
i i
t l fails to contain 0. The
interval fails to contain 0 if the absolute value of
i i
y l is bigger than
Fundamentals of Design of Experiments
II-139
i
2
i 2 / ), 1 r )( 1 v (
r / l MSE t
. Therefore, all possible paired comparisons between
treatment effects one may use the critical differences.
The critical difference for testing the significance of the difference of two treatment
effects, say
j i
t t is r / MSE 2 t . D . C
2 / ), 1 r )( 1 v (
= , where
2 / ), 1 r )( 1 v (
t
is the
value of Student's t at the level of significance and degree of freedom (v - 1)(r - 1). If the
difference of any two-treatment means is greater than the C.D. value, the corresponding
treatment effects are significantly different.
Example 4.1: An experiment was conducted to evaluate the efficacy of Londax 60 DF in
transplanted rice as pre-emergent application as stand alone and as tank mix with grass
partner against different weed flora. The weed counts were recorded. The details of the
experiment are given below:
The weed Count in Rice
Replications Treatment Dose
(gai/ha)
1 2 3
Londax 60 DF 30 72 60 59
Londax 60 DF 45 81 56 71
Londax 60 DF 60 66 49 56
Londax+Butachlor 30+938 8 9 4
Londax +Butachlor 45+938 10 17 6
Londax+Butachlor 60+938 4 8 3
Butachlor 50 EC 938 22 10 11
Pretilachlor 50 EC 625 4 8 10
Pyrazo.Eth.10 WP 100 g/acre 20 46 33
Untreated Control - 79 68 84
Analyze the data and draw your conclusions.
Procedure and Calculations
We compute the following totals:
Treatments totals (
. i
y ) Treatment means ( b / y y
. i . i
= )
=
. 1
y 72 +60 +59 =191 =
. 1
y 191/3 =63.6667
=
. 2
y 81 +56 +71 =208 =
. 2
y 208/3 =69.3333
=
. 3
y 66 +49 +56 =171 =
. 3
y 171/3 =57.0000
=
. 4
y 8 + 9+ 4 =21 =
. 4
y 21/3 =7.0000
=
. 5
y 10 +17 +6 =33 =
. 5
y 33/3 =11.0000
=
. 6
y 4 +8 +3 =15 =
. 6
y 15/3 =5.0000
=
. 7
y 22 +10 +11 =43 =
. 7
y 43/3 =14.3333
=
. 8
y 4 +8 +10 =22 =
. 8
y 22/3 =7.3333
=
. 9
y 20 +46 +33 =99 =
. 9
y 99/3 =33.0000
=
. 10
y 79 +68 +84 =231 =
. 10
y 231/3 =77.0000
Fundamentals of Design of Experiments
II-140
Replication (or Blocks) Totals (
j .
y ) Replication Means ( v / y y
j . j .
= )
=
1 .
y 72 + +79 =366 =
1 .
y 366/10 =36.6
=
2 .
y 60 + +68 =331 =
2 .
y 331/10 =33.1
=
3 .
y 59 + +84 =337 =
3 .
y 337/10 =33.7
Grand Total (of all the observations) = 0000 . 1034
. . ..
= = = =
j
j
i
i
i j
ij
y y y y .
Correction Factor =( ) vb / y
2
..
=(1034)
2
/30 = 35638.5333
Sum of Squares due to Trees = . F . C b / y
i
2
. i
=( ) 8 . 23106 3 / 231 191
2 2
= + + 35638.5333 L
Sum of Squares due to Replications = . F . C v / y
j
2
j .
( ) 0667 . 70 5333 . 35638 10 / 337 331 366
2 2 2
= + + = .
Total Sum of Squares = . F . C y
i j
2
ij
4667 . 24343 . . 84 81 72
2 2 2
= + + + = F C L .
Error Sum of Squares =Total Sum of Squares - Sum of Squares due to Trees - Sum of
Squares due to Replications =24343.4667 70.0667 23106.8000 = 1166.6000.
We now form the following Analysis of Variance Table:
ANOVA
Source D.F. S.S. M.S. F Pr > F
Due to Trees 9 23106.8000 2567.422 39.61 0.000
Due to Replications 2 70.0667 35.03335 0.54 0.592
Error 18 1166.6000 64.81111
Total 29 24343.4667
Critical Difference between any two tree means = b / MSE 2 x t
. f . d error ,
= ( ) 3 / 81111 . 64 2 101 . 2 = 13.810
On the basis of the critical difference we prepare the following table giving the
significance of the difference between two trees effects:
Fundamentals of Design of Experiments
II-141
Mean Treatment No.
A 77.0000 10
B A 69.3330 2
B A 63.6670 1
B 57.0000 3
C 33.0000 9
D 14.3330 7
D 11.0000 5
D 7.3330 8
D 7.0000 4
D 5.0000 6
Suppose now that treatment numbers 1, 2, 3 and treatment numbers 4, 5, 6 form two
groups as the treatments in group 1 are with Londax only where as group2 comprises of
treatments in which Butachlor is added along with Londax. Our interest is in comparing
the two groups. We shall have the following contrast to be estimated and tested:
1.
6 5 4 3 2 1
t t t t t t + + .
Similarly, suppose the other contrasts to be estimated and tested are:
7 6 5 4
10 9 8 7 6 5 4 3 2 1
3 . 3
9 . 2
t t t t
t t t t t t t t t t
+ +
+ + + + + + + +
We have the following table:
Sl. No. D.F. Contrast S.S. M.S. F Pr > F
1 1 13944.5000 13944.5000 215.16 0.0001
2 1 6030.2815 6030.2815 93.04 0.0001
3 1 100.0000 100.0000 1.54 0.2301
Suppose now that the interest of the experimenter is to test certain hypothesis concerning
the three treatments in the Group 1. The sum of squares for testing the equality of tree
effects can be obtained by defining four mutually orthogonal contrasts as ; t t
2 1
.
3 2 1
2t - t t +
Using these sets of contrasts we get the following:
Sl. No. D.F. S.S. M.S. F Pr > F
1 2 228.6667 114.3333 1.76 0.1997
Example 4.2: An initial varietal trial (Late Sown, irrigated) was conducted to study the
performance of 20 new strains of mustard vis-a-vis four checks (Swarna J yoti: ZC;
Vardan: NC; Varuna: NC; and Kranti: NC) using a Randomized complete Block Design
(RCB) design at Bhatinda with 3 replications. The seed yield in kg/ha was recorded. The
details of the experiment are given below:
Fundamentals of Design of Experiments
II-142
Yield in kg/ha
Replications Strain Code
1 2 3
RK-04-3 MCN-04-110 1539.69 1412.35 1319.73
RK-04-4 MCN-04-111 1261.85 1065.05 1111.36
RGN-124 MCN-04-112 1389.19 1516.54 1203.97
HYT-27 MCN-04-113 1192.39 1215.55 1157.66
PBR-275 MCN-04-114 1250.27 1203.97 1366.04
HUJ M-03-03 MCN-04-115 1296.58 1273.43 1308.16
RGN-123 MCN-04-116 1227.12 1018.74 937.71
BIO-13-01 MCN-04-117 1273.43 1157.66 1088.20
RH-0115 MCN-04-118 1180.82 1203.97 1041.90
RH-0213 MCN-04-119 1296.58 1458.65 1250.27
NRCDR-05 MCN-04-120 1122.93 1065.05 1018.74
NRC-323-1 MCN-04-121 1250.27 926.13 1030.32
RRN-596 MCN-04-122 1180.82 1053.47 717.75
RRN-597 MCN-04-123 1146.09 1180.82 856.67
CS-234-2 MCN-04-124 1574.42 1412.35 1597.57
RM-109 MCN-04-125 914.55 972.44 659.87
BAUSM-2000 MCN-04-126 891.40 937.71 798.79
NPJ -99 MCN-04-127 1227.12 1203.97 1389.19
SWAN JYOTI (ZC) MCN-04-128 1389.19 1180.82 1273.43
VARDAN (NC) MCN-04-129 1331.31 1157.66 1180.82
PR-2003-27 MCN-04-130 1250.27 1250.27 1296.58
VARUNA (NC) MCN-04-131 717.75 740.90 578.83
PR-2003-30 MCN-04-132 1169.24 1157.66 1111.36
KRANTI-(NC) MCN-04-133 1203.97 1296.58 1250.27
Analyze the data and draw your conclusions.
Procedure and Calculations: We compute the following totals:
Treatment
Total ( )
. i
y
Treatment Mean
( 3 /
. . i i
y y = )
Treatment Total
( )
. i
y
Treatment Mean
( 3 /
. . i i
y y = )
=
. 1
y 4271.77 =
. 1
y 1423.92 =
. 13
y 2952.04 =
. 13
y 984.01
=
. 2
y 3438.26 =
. 2
y 1146.09 =
. 14
y 3183.57 =
. 14
y 1061.19
=
. 3
y 4109.70 =
. 3
y 1369.90 =
. 15
y 4584.34 =
. 15
y 1528.11
=
. 4
y 3565.60 =
. 4
y 1188.53 =
. 16
y 2546.86 =
. 16
y 848.95
=
. 5
y 3820.28 =
. 5
y 1273.43 =
. 17
y 2627.89 =
. 17
y 875.96
=
. 6
y 3878.17 =
. 6
y 1292.72 =
. 18
y 3820.28 =
. 18
y 1273.43
=
. 7
y 3183.57 =
. 7
y 1061.19 =
. 19
y 3843.44 =
. 19
y 1281.15
=
. 8
y 3519.29 =
. 8
y 1173.10 =
. 20
y 3669.79 =
. 20
y 1223.26
=
. 9
y 3426.68 =
. 9
y 1142.23 =
. 21
y 3797.13 =
. 21
y 1265.71
=
. 10
y 4005.51 =
. 10
y 1335.17 =
. 22
y 2037.49 =
. 22
y 679.16
=
. 11
y 3206.72 =
. 11
y 1068.91 =
. 23
y 3438.26 =
. 23
y 1146.09
=
. 12
y 3206.72 =
. 12
y 1068.91 =
. 24
y 3750.82 =
. 24
y 1250.27
Fundamentals of Design of Experiments
II-143
Replication (or Blocks) Totals (
j
y
.
) Replication Means ( v y y
j j
/
. .
= )
=
1 .
y 29277.27 =
1 .
y 29277.27/24=1219.89
=
2 .
y 28061.73 =
2 .
y 28061.73/24=1169.24
=
3 .
y 26545.19 =
3 .
y 26545.19/24=1106.05
Grand Total (of all the observations) = 19 . 83884
. . ..
= = = =
j
j
i
i
i j
ij
y y y y .
Correction Factor =( ) vb y /
2
..
=(83884.19)
2
/72 = 97730396.53
Sum of Squares due to treatments = . . /
2
.
F C b y
i
i
=( ) 05 . 2514143 53 . 97730396 3 / 82 . 3750 77 . 4271
2 2
= + +L
Sum of Squares due to Replications = . . /
2
.
F C v y
j
j
( )
3283 . 156139
53 . 97730396 24 / 19 . 26545 73 . 28061 27 . 29277
2 2 2
=
+ + =
.
Total Sum of Squares = . .
2
F C y
i j
ij
13 . 3133406 . . 27 . 1250 69 . 1539
2 2
= + + = F C L .
Error Sum of Squares =Total Sum of Squares - Sum of Squares due to treatments - Sum of
Squares due to Replications =3133406.13-2514143.05-156139.33=463123.75.
We now form the following Analysis of Variance Table:
ANOVA (Yield: Bhatinda)
Source D.F. S.S. M.S. F Pr > F
Due to Treatments 23 2514143.05 109310.57 10.86 <0.0001
Due to Replications 2 156139.33 78069.66 7.75 0.0013
Error 46 463123.75 10067.91
Total 71 3133406.13
R-Square CV Root MSE Mean Yield
0.852198 8.612337 100.3390 1165.06
2. Critical Difference between any two tree means = b / MSE 2 x t
. f . d error ,
= ( ) 3 / 91 . 10067 2 10290 . 2 = 164.91
On the basis of the critical difference we prepare the following table giving the
significance of the difference between two treatment effects:
Fundamentals of Design of Experiments
II-144
Mean Treatment No. Mean Treatment No.
1528.11 15 A 1173.10 8 D E F
1423.93 1 A B 1146.10 23 E F G
1369.90 3 A B C 1146.10 2 E F G
1335.18 10 B C D 1142.22 9 E F G
1292.73 6 B C D E 1068.90 12 F G
1281.14 19 B C D E 1068.90 11 F G
1273.43 18 B C D E 1061.19 7 F G
1273.43 5 B C D E 1061.19 14 F G
1265.72 21 B C D E 984.02 13 G H
1250.27 24 C D E 875.97 17 H
1223.27 20 C D E F 848.96 16 H
1188.55 4 D E F 679.15 22 I
Suppose now that treatment numbers 19, 20, 22 and 24 are the checks and rest of the
treatments are test entries. It is clear from the above Table that treatment 15 is
significantly different from highest performing check. The above Table gives Our interest
is in comparing the checks with new entries. We shall have the following contrast to be
estimated and tested:
1.
24 22 20 19 23 21 18 2 1
20 20 20 20 4 4 4 4 4 t t t t t t t t t + + + + + L .
We have the following table:
Sl. No. D.F. Contrast S.S. M.S. F Pr > F
Checks vs
Entries
1 46128.89 46128.89 4.58 0.0376
Suppose the experimenter can test any other hypothesis of interest.
5. Latin Square Design
Latin square designs are normally used in experiments where it is required to remove the
heterogeneity of experimental material in two directions. These designs require that the
number of replications equal the number of treatments or varieties.
Definition 1. A Latin square arrangement is an arrangement of v symbols in v
2
cells
arranged in v rows and v columns, such that every symbol occurs precisely once in each
row and precisely once in each column. The term v is known as the order of the Latin
square.
If the symbols are taken as A, B, C, D, a Latin square arrangement of order 4 is as follows:
A B C D
B C D A
C D A B
D A B C
A Latin square is said to be in the standard form if the symbols in the first row and first
column are in natural order, and it is said to be in the semi-standard form if the symbols of
the first row are in natural order. Some authors denote both of these concepts by the term
standard form. However, there is a need to distinguish between these two concepts. The
Fundamentals of Design of Experiments
II-145
standard form is used for randomizing the Latin-square designs, and the semistandard form
is needed for studying the properties of the orthogonal Latin squares.
Definition 2. If in two Latin squares of the same order, when superimposed on one
another, every ordered pair of symbols occurs exactly once, the two Latin squares are said
to be orthogonal. If the symbols of one Latin square are denoted by Latin letters and the
symbols of the other are denoted by Greek letters, the pair of orthogonal Latin squares is
also called a graeco-latin square.
Definition 3. If in a set of Latin squares every pair is orthogonal, the set is called a set of
mutually orthogonal latin squares (MOLS). It is also called a hypergraeco latin
square.
The following is an example of graeco latin square:
A B C D
B A D C
C D A B
D C B A
A B C D
B A D C
C D A B
D C B A
We can verify that in the above arrangement every pair of ordered Latin and Greek
symbols occurs exactly once, and hence the two latin squares under consideration
constitute a graecolatin square.
It is well known that the maximum number of MOLS possible of order v is v - 1. A set of
v - 1 MOLS is known as a complete set of MOLS. Complete sets of MOLS of order v
exist when v is a prime or prime power.
Randomization
According to the definition of a Latin square design, treatments can be allocated to the v
2
experimental units (may be animal or plots) in a number of ways. There are, therefore, a
number of Latin squares of a given order. The purpose of randomization is to select one of
these squares at random. The following is one of the methods of random selection of Latin
squares.
Let a v v Latin square arrangement be first written by denoting treatments by Latin letters
A, B, C, etc. or by numbers 1, 2, 3, etc. Such arrangements are readily available in the
Tables for Statisticians and Biometricians (Fisher and Yates, 1974). One of these
squares of any order can be written systematically as shown below for a 55 Latin square:
D C B A E
C B A E D
B A E D C
A E D C B
E D C B A
Fundamentals of Design of Experiments
II-146
For the purpose of randomization rows and columns of the Latin square are rearranged
randomly. There is no randomization possible within the rows and/or columns. For
example, the following is a row randomized square of the above 55 Latin square;
B A E D C
C B A E D
D C B A E
A E D C B
E D C B A
Next, the columns of the above row randomized square have been rearranged randomly to
give the following random square:
A C E D B
B D A E C
C E B A D
E B D C A
D A C B E
As a result of row and column randomization, but not the randomization of the individual
units, the whole arrangement remains a Latin square.
Analysis of Latin Square Designs
In Latin square designs there are three factors. These are the factors P, Q, and treatments.
The data collected from this design are, therefore, analyzed as a three-way classified data.
Actually, there should have been
3
v
observations as there are three factors each at v levels.
But because of the particular allocation of treatments to the cells, there is only one
observation per cell instead of v in the usual three way classified orthogonal data. As a
result we can obtain only the sums of squares due to each of the three factors and error sum
of squares. None of the interaction sums of squares of the factors can be obtained.
Accordingly, we take the model
ijs s j i ijs
e t c r Y + + + + =
where
ijs
y denotes the observation in the i
th
row, j
th
column and under the s
th
treatment;
( ) v ,..., 2 , 1 s , j , i t , c , r ,
s j i
= are fixed effects denoting in order the general mean, the row,
the column and the treatment effects. The
ijs
e is the error component, assumed to be
independently and normally distributed with zero mean and a constant variance,
2
.
The analysis is conducted by following a similar procedure as described for the analysis of
two-way classified data. The different sums of squares are obtained as below: Let the data
be arranged first in a row column table such that
ij
y denotes the observation of (i, j)th
cell of table.
Fundamentals of Design of Experiments
II-147
Let ( ), v 1,2,..., i total row i y R
j
th
ij i
= = =
=
s
T sum of those observations which come from s
th
treatment (s= 1,2,,v),
. total grand R G
i
i
= =
Analysis of Variance of v v Latin Square Design
Sources of Variation D.F. S.S. M.S. F
Rows v -1
. F . C
v
R
i
2
i
Columns v - 1
. F . C
v
C
j
2
j
Treatments v - 1
. F . C
v
T
s
2
s
2
t
s
2
e
2
t
s / s
Error (v - 1)(v - 2) By subtraction 2
e
s
Total v
2
-1
. F . C y
ij
2
ij
The hypothesis of equal treatment effects is tested by F-test, where F is the ratio of
treatment mean squares to error mean squares. If F is not significant, treatment effects do
not differ significantly among themselves. If F is significant, further studies to test the
significance of any treatment contrast can be made in exactly the same way as discussed
for randomized block designs.
6. Illustrations for Combined Analysis of Data
Example 6.1: An initial varietal trial (Late Sown, irrigated) was conducted to study the
performance of 20 new strains of mustard vis-a-vis four checks (Swarna J yoti: ZC;
Vardan: NC; Varuna: NC; and Kranti: NC) using a Randomized complete Block Design
(RCB) design at four locations (Sriganganagar, Navgaon, Bhatinda and Hissar) with 2
replications at Sriganganagar and with 3 replications each at other 3 locations. The seed
yield in kg/ha was recorded. The data pertaining to Bhatinda is given in Example 1. The
data from the rest of 3 locations is given as below:
Fundamentals of Design of Experiments
II-148
Yield in kg/ha
Sriganganagar Navgaon Hissar
Replications Replications Replications
Strain
No.
1 2 1 2 3 1 2 3
1 778.00 667.00 533.28 488.84 799.92 945.68 1040.25 1040.25
2 556.00 444.00 444.40 488.84 466.62 567.41 945.68 803.83
3 556.00 444.00 977.68 888.80 799.92 1134.82 1182.10 1040.25
4 778.00 778.00 888.80 799.92 799.92 969.33 1229.39 1134.82
5 556.00 556.00 666.60 666.60 444.40 898.40 851.11 969.33
6 444.00 444.00 799.92 533.28 577.72 851.11 756.55 969.33
7 556.00 333.00 1066.56 1022.12 933.24 1134.82 1323.96 1040.25
8 556.00 444.00 1111.00 1066.56 1066.56 1229.39 1134.82 1134.82
9 444.00 556.00 666.60 888.80 844.36 1087.54 898.40 992.97
10 778.00 556.00 533.28 622.16 844.36 851.11 1134.82 945.68
11 667.00 778.00 1022.12 666.60 755.48 1040.25 1276.67 1229.39
12 444.00 444.00 799.92 666.60 622.16 803.83 945.68 992.97
13 333.00 556.00 799.92 666.60 688.82 992.97 1182.10 1323.96
14 444.00 333.00 888.80 933.24 666.60 1040.25 1134.82 1276.67
15 556.00 333.00 844.36 688.82 577.72 1182.10 1418.52 1229.39
16 333.00 333.00 711.04 622.16 622.16 1087.54 945.68 1040.25
17 556.00 333.00 799.92 577.72 533.28 969.33 1040.25 1040.25
18 333.00 333.00 1066.56 1111.00 999.90 969.33 1087.54 1040.25
19 444.00 444.00 933.24 711.04 711.04 1418.52 1040.25 945.68
20 444.00 444.00 755.48 799.92 733.26 1182.10 1134.82 1087.54
21 333.00 444.00 844.36 755.48 666.60 1087.54 1323.96 1040.25
22 444.00 333.00 666.60 533.28 488.84 992.97 803.83 992.97
23 556.00 333.00 755.48 799.92 1022.12 1134.82 992.97 1229.39
24
333.00 333.00 488.84 577.72 666.60 1040.25 992.97 1182.10
The data from each of the centers were analyzed separately using PROC GLM of SAS.
The results for Bhatinda center are given in Example 1. The results of the other 3 locations
are given in the sequel.
ANOVA (Yield: Hissar)
Source D.F. S.S. M.S. F Pr > F
Due to Treatments 23 1007589.069 43808.220 3.06 0.0006
Due to Replications 2 37465.039 18732.519 1.31 0.2795
Error 46 657493.58 14293.33
Total 71 1702547.68
R-Square CV Root MSE Mean Yield
0.613818 11.30376 119.5548 1057.65
Treatments are significantly different at 5% level of significance, where as replications are
not significantly different. None of the entries gave significantly higher yield than best
performing check (Swarna J yoti).
Fundamentals of Design of Experiments
II-149
ANOVA (Yield: Navgaon)
Source D.F. S.S. M.S. F Pr > F
Due to Treatments 23 1685581.90 73286.19 6.51 <0.0001
Due to Replications 2 73332.38 36666.19 3.26 0.0476
Error 46 518154.24 11264.23
Total 71 2277068.52
R-Square CV Root MSE Mean Yield
0.772447 14.15831 106.1330 749.6164
Both treatments and replications are significantly different at 5% level of significance.
New entry at serial number 8 gave significantly higher yield than best performing check
(Swarna J yoti).
ANOVA (Yield: Sriganganagar)
Source D.F. S.S. M.S. F Pr > F
Due to Treatments 23 699720.92 30422.65 4.03 0.0007
Due to Replications 1 31314.08 31314.08 4.15 0.0533
Error 23 173540.92 7545.26
Total 47 904575.92
R-Square CV Root MSE Mean Yield
0.808152 17.95781 86.86344 483.7083
Both treatments and replications are significantly different at 5% level of significance,
New entry at serial number 8 gave significantly higher yield than best performing check
(Swarna J yoti). Error mean squares and error degrees of freedom of the 4 locations are:
Bhatinda Hissar Navgaon Sriganganagar
Error degrees of freedom 46 46 46 23
Error Mean Square 10067.91 14293.33 11264.23 7545.26
In order to perform the combined analysis of the data for 4 locations (group of
experiments), the mean square errors for the 4 locations were tested for the homogeneity of
error variances using Bartletts
2
-test. The test is described in the sequel:
Let the experiment is conducted in k environments. The estimate of error variance for the
i
th
environment is
2
i
s (MSE for i
th
environment) with
i
f degrees of freedom (error degrees
of freedom).
Fundamentals of Design of Experiments
II-150
We are interested to test the null hypothesis
2 2
2
2
1 0
... :
k
H = = = against the alternative
hypothesis :
1
H at least two of the s
i
'
2
are not equal, where
2
i
is the error variance for
treatment i. (
2
i
is the error variance for the i
th
environment).
The procedure involves computing a statistic whose sampling distribution is closely
approximated by the
2
distribution with k - 1 degrees of freedom. The test statistic is
c
q
3026 . 2
2
0
=
and null hypothesis is rejected when
2
1 ,
2
0
>
k
, where
2
1 , k
is the upper
percentage point of
2
distribution with k - 1 degrees of freedom.
To compute
2
0
, follow the steps:
Step 1: Compute mean and variance of all v-samples.
Step 2: Obtain pooled variance
=
=
=
k
i
i
k
i
i i
p
f
s f
S
1
1
2
2
Step 3: Compute
= =
=
k
i
i i p
k
i
i
S f S f q
1
2
10
2
10
1
log log
Step 4: Compute
( )
+ =
=
k
i
k
i
i i
f f
k
c
1
1
1
1
1 3
1
1
Step 5: Compute
2
0
.
For this example, the computed
2
0
was found to be 3.28. The tabulated value of
81 . 7
2
05 . 0 , 3
= . Therefore, the null hypothesis is not rejected. Therefore, the error variances
were found to be homogeneous. Now the combined analysis of data can be carried out
using the following statements of SAS.
Data comb;
Input loc $ rep var yield;
Cards;
.
.
.
;
Fundamentals of Design of Experiments
II-151
proc glm;
class loc rep trt;
model yield =loc rep(loc) trt trt*loc;
random loc rep(loc) trt*loc/test;
run;
The results obtained are:
Combined Analysis of Data Over 4 Locations of Rapeseed-Mustard Initial Varietal
Trial
Source DF SS Mean Square F Value Pr>F
loc 3 16794186.86 5598062.29 497.31 <.0001
Replications(loc) 7 298250.83 42607.26 3.79 0.0008
Treatments 23 2153545.49 93632.41 8.32 <.0001
loc*Treatment 69 3495630.98 50661.32 4.50 <.0001
Error 161 1812312.49 11256.60
Total 263 24811785.12
R-square C.V. Root MSE Mean
0.93 11.81 106.10 898.59
Source Type III Expected Mean Square
loc Var(Error) +2.7273 Var(loc*treatment) +24 Var(rep(loc) +65.455
Var(loc)
rep(loc) Var(Error) +24 Var(rep(loc)
treatment Var(Error) +2.6667 Var(loc*treatment) +Q(treatment)
loc*treatment Var(Error) +2.7273 Var(loc*treatment)
Tests of Hypotheses for Mixed Model Analysis of Variance
Source DF SS MS F-Value Pr>F
loc 3 16794187 5598062 68.26 <.0001
Error: MS(rep(loc)) + MS(loc*treatment) - MS(Error)
rep(loc) 7 298251 42607 3.79 0.0008
loc*treatment 69 3495631 50661 4.50 <.0001
Error: MS(Error)
treatment 23 2153545 93632 1.88 0.0232
Error: 0.9778*MS(loc*treatment)+0.0222*MS(Error)
Fundamentals of Design of Experiments
II-152
Example 6.2: An experimenter was interested in comparing 49 treatments. The
experiment was laid out in a lattice design with four replications. There were seven blocks
per replication and seven treatments were allotted within each block. Observations were
recorded on several characters but for illustration purposes only one data set (one
character) is analyzed. The same design was repeated over two years. The layout of the
design is given below:
Blocks Replication - I
1. 1 2 3 4 5 6 7
2. 8 9 10 11 12 13 14
3. 15 16 17 18 19 20 21
4. 22 23 24 25 26 27 28
5. 29 30 31 32 33 34 35
6. 36 37 38 39 40 41 42
7. 43 44 45 46 47 48 49
Blocks Replication - II
1. 1 8 15 22 29 36 43
2. 2 9 16 23 30 37 44
3. 3 10 17 24 31 38 45
4. 4 11 18 25 32 39 46
5. 5 12 19 26 33 40 47
6. 6 13 20 27 34 41 48
7. 7 14 21 28 35 42 49
Blocks Replication - III
1. 1 9 17 25 33 41 49
2. 43 2 10 18 26 34 42
3. 36 44 3 11 19 27 35
4. 29 37 45 4 12 20 28
5. 22 30 38 46 5 13 21
6. 15 23 31 39 47 6 14
7. 8 16 24 32 40 48 7
Blocks Replication - IV
1. 1 37 24 11 47 34 21
2. 15 2 38 25 12 48 35
3. 29 16 3 39 26 13 49
4. 43 30 17 4 40 27 14
5. 8 44 31 18 5 41 28
6. 22 9 45 32 19 6 42
7. 36 23 10 46 33 20 7
The analysis was carried out using PROC GLM of SAS and using the option of contrast
for carrying out the contrast analysis. The results of the analysis of data for the first year
are as given below:
Fundamentals of Design of Experiments
II-153
RESULTS 1 (LATTICE DESIGN: FIRST YEAR)
Source DF SS Mean Square F Value Pr>F
Replications 3 186.04 62.01 7.53 0.0001
Block(replication) 24 358.94 14.95 1.82 0.0192
Treatments 48 3442.14 71.71 8.70 0.0001
Error 120 988.70 8.23
Total 195 6025.75
R-square C.V. Root MSE Mean
0.84 3.37 2.87 85.18
It may be noted that all sum of squares reported in the table are adjusted sums of squares
and that the adjustments have been made for all the other remaining effects. The CV is
very small and, therefore, the design adopted is appropriate. The interesting feature of the
design is that the blocks within replication sum of squares are significant and, therefore,
formation of blocks within replications has been fruitful. Thus, the formation of
incomplete blocks within replications has been very effective and the error mean square is
quite small. The treatment effects are also highly significant. The 49 treatments tried in
the experiment were formed into four groups on the basis of the nature of the treatments.
The groups are - Group 1: Treatments 1 - 15; Group 2: Treatments 16 - 30; Group 3:
Treatments 31 - 46; Group 4: Treatments 47 - 49. Contrast analysis was carried out to
study the equality of the treatment effects within groups and desired between group
comparisons. The results are as follows:
Contrast DF Contrast SS Mean Square F Value Pr > F
gr1 14 985.53 70.39 8.54 0.0001
gr2 14 1004.60 71.75 8.71 0.0001
gr3 15 1373.17 91.54 11.11 0.0001
gr4 2 60.27 30.13 3.66 0.0287
gr1 vs gr4 1 47.29 47.29 5.74 0.0181
gr2 vs gr4 1 92.69 92.69 11.25 0.0011
gr3 vs gr4 1 41.74 41.74 5.07 0.0262
gr1 vs gr2 1 18.86 18.86 2.29 0.1329
It may be seen that the group 1 vs group 2 comparisons are not significantly different
whereas all other comparisons are significantly different.
RESULT 2 (LATTICE DESIGN SECOND YEAR)
Source DF SS Mean Square F Value Pr>F
Replications 3 176.404 58.79 11.81 0.0001
Block(replication) 24 556.49 23.18 4.66 0.0001
Treatments 48 3353.21 69.85 14.03 0.0001
Error 120 597.30 4.97
Total 195 5413.92
R-square C.V. Root MSE Mean
0.89 2.50 2.23 89.31
Fundamentals of Design of Experiments
II-154
It may be noted again that all sum of squares reported in the table are adjusted sums of
squares and that the adjustments have been made for all other remaining effects. The CV
is very small and therefore the design adopted is appropriate. The interesting feature of the
design is that the blocks within replication sum of squares are highly significant and,
therefore, formation of blocks within replications has been fruitful. Thus, the formation of
incomplete blocks within replications has been very effective and the error mean square is
quite small. The treatment effects are also highly significant.
In order to perform the combined analysis of the data for two years (group of experiments),
the mean square errors for the two years were tested for the homogeneity of error
variances. The value of F statistic was obtained as F = MSE1 / MSE2 = 8.23 / 4.97 =
1.6559 (significant at 5 % level of significance). Therefore, for performing the combined
analysis weighted least squares was done, the weight being the reciprocals of the root mean
square error. The weighted least squares analysis is carried out by defining a new variable
newres = res/root mean square error. The analysis of variance is then performed on the
new variable. The following analysis is usually carried out for these situations.
RESULT 3 (LATTICE DESIGN COMBINED ANALYSIS FOR YEARS 1 & 2)
Source DF SS Mean Square F Value Pr>F
Year 1 4911.42 4911.42 3672.45 0.0001
Replications 3 19.27 6.42 4.80 0.0028
Block(replication) 24 93.56 3.90 2.91 0.0001
Treatments 48 1142.94 23.81 17.80 0.0001
Year*Treatment 48 137.75 2.87 2.15 0.0001
Error 267 357.08 1.34
Total 391 6780.42
R-square C.V. Root MSE Mean
0.95 3.34 1.16 34.66
The year*treatment interaction is highly significant. Therefore, treatment is tested against
the year*treatment interaction mean square. The results obtained are given as:
Source DF Type III SS Mean Square F Value Pr > F
treatment 48 1142.94 23.81 8.30 <.0001
In the above analysis, the degrees of freedom for the replications and blocks (replications)
are 3 and 24 respectively and are same as that of individual year analyses. Therefore, no
distinction is made in the replications and blocks (replications) of the two years. Hence,
this procedure is inappropriate.
The appropriate procedure, therefore, is to view the groups of experiments as a nested
design with several factors nested within one another. The locations are treated as big
blocks, with the experiments nested within these. The combined analysis of data,
therefore, can be done as that of a nested design. An advantage of this analysis is that there
is a further reduction in the error sum of squares because one more source of variability is
Fundamentals of Design of Experiments
II-155
taken out from the experimental error thus reducing the experimental error. This may also
lead to the reduction in the value of CV. If we take the data for two years together there
will be 56 blocks and hence the blocks will account for 55 degrees of freedom. The
analysis of variance just described accounts for 28 degrees of freedom. The remaining 27
degrees of freedom go into the error. However, if we analyze the data as a nested design,
we get 55 degrees of freedom for the blocks that can be split into various components. In
the sequel, we present the appropriate analysis of groups of experiments. This enables us
to further reduce the experimental error thus reducing the CV. The results obtained are
reproduced below:
RESULT 4. (LATTICE DESIGN COMBINED ANALYSIS CONSIDERING
NESTED CLASSIFICATIONS - REPLICATIONS NESTED WITHIN YEARS AND
BLOCKS NESTED WITHIN REPLICATIONS AND YEARS ON THE
TRANSFORMED DATA)
Source DF SS Mean Square F Value Pr>F
Year 1 4911.42 4911.42 4344.23 <.0001
Replications(Year) 6 58.83 9.80 8.67 <.0001
Blocks(Year*replication) 48 139.74 2.91 2.58 <.0001
Treatments 48 968.42 20.18 17.85 <.0001
Year*Treatment 48 130.86 2.73 2.41 <.0001
Error 240 271.33 1.13
Total 391 6780.42
R-square C.V. Root MSE Mean
0.96 3.07 1.06 34.66
It may be seen that the error sum of squares has a reduction of 27 degrees of freedom. The
CV has also reduced from 3.34 to 3.07. The sums of squares due to various components in
the model are highly significant. The advantage of analyzing the data as a nested design is
quite visible thus.
The year*treatment interaction is highly significant. Therefore, treatment is tested against
the year*treatment interaction mean square. The results obtained are given as:
Source DF Type III SS Mean Square F Value Pr > F
treatment 48 968.42 20.18 7.40 <.0001
In the above analysis, the proper error terms can also be identified using PROC GLM of
SAS along with random statement with TEST option. Using PROC GLM, the expected
mean squares for different effects in the model are given as
Fundamentals of Design of Experiments
II-156
Source Type III Expected Mean Square
Year Var(Error) +4 Var(year*treatment) +7 Var(block(year*rep)) +49
Var(rep(year)) +196 Var(year)
rep(year) Var(Error) +7 Var(block(year*rep)) +49 Var(rep(year))
block(year*rep) Var(Error) +5.25 Var(block(year*rep))
treatment Var(Error) +3.5 Var(year*treatment) +Q(treatment)
year*treatment Var(Error) +3.5 Var(year*treatment)
Tests of Hypotheses for Mixed Model Analysis of Variance
Source DF SS MS F-Value Pr>F
Year 1 4911.42 4911.42 422.37 <.0001
Error: MS(rep(year)) + 1.1429*MS(year*treatment) - 1.1429*MS(Error)
rep(year) 6 58.83 9.80 2.80 0.0232
Error: 1.3333*MS(block(year*rep)) - 0.3333*MS(Error)
block(year*rep) 48 139.74 2.91 2.58 <.0001
year*treatment 48 130.86 2.73 2.41 <.0001
Error: MS(Error)
treatment 48 968.42 20.18 7.40 <.0001
Error: MS(year*treatment)
It is observed that the year*treatment interaction is highly significant and the proper error
term for testing the equality of treatment effects is year*treatment interaction mean square.
The above discussions refer to the combined analysis of experiments conducted at different
locations or different times at the same location in general block designs with same
treatments in each of the environments. There may arise situations where all the
treatments are not common to the whole set. Only subsets of the treatments are common to
the whole set. This may happen due to some location specific treatments that cannot be
tried at all the locations. Different treatments at different locations do not give any problem
in the combined analysis of data so long as there are some common treatments over all the
locations.
7. Factorial Experiments
Example 7.1: An experiment was conducted at Ludhiana Centre of AICRP on Cropping
Systems using a balanced confounded design for factorial experiments with three factors,
viz., Nitrogen (40, 80 and 120 kg/ha), Phosphorous (0, 40 and 80 kg/ha) and Potassium (0
and 40 kg/ha). These 18 treatment combinations were arranged in 3 blocks of size 6 each.
The analysis of the data was performed using PROC GLM of SAS. The SAS commands
and the output is given in the sequel.
Fundamentals of Design of Experiments
II-157
Options linesize=72;
data ludh98k;
input rep block N P K trt yield;
cards;
1 1 40 0 0 1 7.79
1 1 120 80 0 17 10.30
1 1 40 80 40 6 10.08
1 1 120 40 40 16 11.66
1 1 80 0 40 8 9.13
1 1 80 40 0 9 10.56
1 2 40 0 40 2 6.12
1 2 120 0 0 13 8.44
1 2 120 80 40 18 11.44
1 2 80 40 40 10 9.13
1 2 80 80 0 11 9.40
1 2 40 40 0 3 6.85
1 3 80 0 0 7 6.25
1 3 120 0 40 14 7.78
1 3 40 40 40 4 6.66
1 3 80 80 40 12 9.42
1 3 40 80 0 5 6.50
1 3 120 40 0 15 11.82
2 1 120 0 0 13 7.86
2 1 120 40 40 16 10.15
2 1 40 80 40 6 7.50
2 1 80 0 40 8 7.89
2 1 80 80 0 11 8.00
2 1 40 40 0 3 6.40
2 2 120 0 40 14 8.50
2 2 80 80 40 12 9.86
2 2 40 40 40 4 7.70
2 2 120 80 0 17 10.79
2 2 80 40 0 9 7.87
2 2 40 0 0 1 6.30
2 3 80 0 0 7 7.00
2 3 40 80 0 5 8.00
2 3 120 80 40 18 10.90
2 3 40 0 40 2 6.62
2 3 80 40 40 10 9.62
2 3 120 40 0 15 9.50
3 1 80 80 0 11 10.00
3 1 120 80 40 18 10.86
3 1 40 40 40 4 7.58
3 1 80 0 40 8 6.35
3 1 120 40 0 15 9.40
3 1 40 0 0 1 5.94
Fundamentals of Design of Experiments
II-158
3 2 120 0 40 14 9.00
3 2 40 80 40 6 8.80
3 2 80 40 40 10 9.53
3 2 120 80 0 17 10.56
3 2 40 40 0 3 7.07
3 2 80 0 0 7 6.00
3 3 80 40 0 9 7.20
3 3 120 0 0 13 8.36
3 3 40 0 40 2 6.05
3 3 80 80 40 12 10.45
3 3 120 40 40 16 10.10
3 3 40 80 0 5 7.50
4 1 80 80 0 11 7.97
4 1 80 40 40 10 7.18
4 1 40 80 40 6 6.16
4 1 40 0 0 1 4.95
4 1 120 40 0 15 10.12
4 1 120 0 40 14 7.15
4 2 80 0 0 7 6.65
4 2 40 40 0 3 6.66
4 2 80 80 40 12 7.90
4 2 120 40 40 16 10.10
4 2 40 0 40 2 6.49
4 2 120 80 0 17 10.30
4 3 80 0 40 8 6.12
4 3 40 40 40 4 5.80
4 3 120 80 40 18 10.06
4 3 120 0 0 13 7.37
4 3 80 40 0 9 7.24
4 3 40 80 0 5 7.70
;
proc glm;
class rep block n p k;
model yield =rep block(rep) n p k n*p n*k p*k n*p*k;
run;
The output is given as:
Dependent Variable: yield
Source DF Sum of Squares Mean Square F-Vlaue Pr>F
Model 28 185.8448 6.6373 13.56 <0.0001
Error 43 21.0427 0.4894
Corrected Total 71 206.8876
R-Square Coeff Var Root MSE Yield Mean
0.8983 8.4444 0.699547 8.284
Fundamentals of Design of Experiments
II-159
Source DF Type III SS Mean Square F-Vlaue Pr>F
Rep 3 15.7187 5.2396 10.71 <.0001
Block(rep) 8 14.1946 1.7743 3.63 0.0027
N 2 89.1108 44.5554 91.05 <.0001
P 2 55.9270 27.9635 57.14 <.0001
K 1 3.2173 3.2173 6.57 0.0139
NP 4 4.2752 1.0688 2.18 0.0868
NK 2 0.7301 0.3650 0.75 0.4803
PK 2 0.1128 0.0564 0.12 0.8914
NPK 4 2.1958 0.5490 1.12 0.3588
From the above, it is clear that the blocks with in replication are significant indicating that
the incomplete blocks have help in reducing mean square error. All the three main effects
N, P and K are significant whereas none of the interaction is significant.
References
Kempthorne, O. (1977). Why randomize? Journal of Statistical Planning and Inference,
1, 1-25.
Dean, A. and Voss, D. (1999). Design and Analysis of Experiments. Springer Text in
Statistics, New York.
Fisher, R.A. and Yates, F. (1963). Statistical Tables For Biological, Agricultural and
Medical Research. Longman Group Ltd., England.
Fundamentals of Design of Experiments
II-160
Exercise 1.1: In order to select suitable tree species for Fuel, Fodder and Timber an
experiment was conducted in a randomized complete block design with ten different trees
and four replications. The plant height was recorded in cms. The details of the experiment
are given below:
Plant Height (Cms): Place Kanpur
Replications Name of Tree Spacing
1 2 3 4
A. Indica 4x4 144.44 145.11 104.00 105.44
D. Sisso 4x2 113.50 118.61 118.61 123.00
A. Procer 4x2 60.88 90.94 80.33 92.00
A. Nilotic 4x2 163.44 158.55 158.88 153.11
T. Arjuna 4x2 110.11 116.00 119.66 103.22
L. Loucoc 4x1 260.05 102.27 256.22 217.80
M. Alba 4x2 114.00 115.16 114.88 106.33
C. Siamia 4x2 91.94 58.16 76.83 79.50
E. Hybrid 4x1 156.11 177.97 148.22 183.17
A. Catech 4x2 80.2 108.05 45.18 79.55
Analyze the data and draw your conclusions.
Procedure and Calculations: We compute the following totals:
Treatments totals (
. i
y ) Treatment means ( b / y y
. i . i
= )
=
. 1
y 144.44 ++105.44 =498.99 =
. 1
y 498.99/4 =124.7475
=
. 2
y 112.50 + +123.00 =473.72 =
. 2
y 473.72/4 =118.4300
=
. 3
y 60.88 + +92.00 =324.15 =
. 3
y 324.15/4 =81.0375
=
. 4
y 163.44 + +153.11 =633.98 =
. 4
y 633.98/4 =158.4950
=
. 5
y 110.11 + +103.22 =448.99 =
. 5
y 448.99/4 =112.2475
=
. 6
y 260.05 + +217.8 =836.34 =
. 6
y 836.34/4 =209.0850
=
. 7
y 114.00 + +106.33 =450.37 =
. 7
y 450.37/4 =112.5925
=
. 8
y 91.94 + +79.50 =306.43 =
. 8
y 306.43/4 =76.6075
=
. 9
y 156.11 + +183.17 =665.47 =
. 9
y 665.47/4 =166.3675
=
. 10
y 80.20 + +79.55 =312.98 =
. 10
y 312.98/4 =78.2450
Replication (or Blocks) Totals (
j .
y ) Replication Means ( v / y y
j . j .
= )
=
1 .
y 144.44 + +80.20 =1294.67 =
1 .
y 1294.67/10 =129.4670
=
2 .
y 145.11 + +108.05 =1190.82 =
2 .
y 1190.82/10 =119.0820
=
3 .
y 104.00 + +45.18 =1222.81 =
3 .
y 1222.81/10 =122.2810
=
4 .
y 105.44 + +79.55 =1243.12 =
4 .
y 1243.12/10 =124.3120
Fundamentals of Design of Experiments
II-161
Grand Total (of all the observations) = 42 . 4951 y y y y
j
j .
i
. i ..
i j
ij
= = = =
.
Correction Factor =( ) vb / y
2
..
=(4951.42)
2
/40 = 612914.0004
Sum of Squares due to Trees = . F . C b / y
i
2
. i
=( ) 35 . 66836 00 . 612914 4 / 98 . 312 99 . 498
2 2
= + +L
Sum of Squares due to Replications = . F . C v / y
j
2
j .
( ) 43 . 569 00 . 6121914 10 / 12 . 1243 67 . 1294
2 2
= + + = L .
Total Sum of Squares = . F . C y
i j
2
ij
42 . 89101 . F . C 55 . 79 12 . 145 44 . 144
2 2 2
= + + + = L .
Error Sum of Squares =Total Sum of Squares - Sum of Squares due to Trees - Sum of
Squares due to Replications =89101.42 - 66836.35 - 569.43 = 21695.26.
We now form the following Analysis of Variance Table:
ANOVA
Source D.F. S.S. M.S. F Pr > F
Due to Trees 9 66836.35 7426.26 9.24 0.0001
Due to Replications 3 569.43 189.81 0.24 0.8703
Error 27 21695.26 803.53
Total 39 89101.04
Critical Difference between any two tree means = b / MSE 2 x t
. f . d error ,
= ( ) 4 / 53 . 803 x 2 x 05 . 2 = 41.09
On the basis of the critical difference we prepare the following table giving the
significance of the difference between two trees effects:
Mean Tree No.
A 209.085 6
B 166.368 9
C B 158.500 4
D C 124.748 1
E D C 118.430 2
Fundamentals of Design of Experiments
II-162
F E D 112.593 7
F E D 112.248 5
F E 81.038 3
F E 78.245 10
F 76.608 8
Suppose now that tree numbers 1, 2, 3, 4, 10 and trees numbers 5, 6, 7, 8, 9 form two
groups on the basis of some considerations. (The first group of trees is useful for fuel,
fodder and timber while the second group of trees is useful for fuel and fodder only). Our
interest is in comparing the two groups. We shall have the following contrast to be
estimated and tested:
1.
10 9 8 7 6 5 4 3 2 1
t t t t t t t t t t + + + + .
Similarly, suppose the other contrasts to be estimated and tested are:
10 9
10 8 7 6 5
9 8 7 6 5
10 4 3 2 1
9 4 3 2 1
10 9 8 7 6 5 4 3 2 1
t t . 7
t 4 t t t t . 6
t 4 t t t t . 5
t 4 t t t t . 4
t 4 t t t t . 3
t t t t t t t t t t 9 . 2
+ + +
+ + +
+ + +
+ + +
We have the following table:
Sl. No. D.F. Contrast S.S. M.S. F Pr > F
1 1 788.3285 788.3285 0.98 0.3307
2 1 4.1131 4.1131 0.01 0.9435
3 1 6680.2435 6680.2435 8.31 0.0076
4 1 5761.6546 5761.6546 7.17 0.0125
5 1 4801.1258 4801.1258 5.98 0.0213
6 1 7805.3981 7805.3981 9.71 0.0043
7 1 15531.1500 15531.1500 19.33 0.0002
Suppose now that the interest of the experimenter is to test certain hypothesis concerning
the five trees in the Group 1 (comprising of Trees Numbers 1, 2, 3, 4, and 10). The sum of
squares for testing the equality of tree effects can be obtained by defining four mutually
orthogonal contrasts as ; t t
2 1
; 2t - t t
3 2 1
+ ; 3t - t t t
3 3 2 1
+ +
10 4 3 2 1
4t - t t t t + + + .
Using these sets of contrasts we get the following:
Sl. No. D.F. S.S. M.S. F Pr > F
1 4 17854.0908 4463.5227 5.55 0.0021