Professional Documents
Culture Documents
Analysis of Variance and Design of Experiments
Analysis of Variance and Design of Experiments
Analysis of Variance and Design of Experiments
Experiments
Experimental Designs and Their Analysis
:::
Lecture 18
Completely Randomized Design
Shalabh
Department of Mathematics and Statistics
Indian Institute of Technology Kanpur
2
Completely randomized design (CRD)
• The number of replications for different treatments need not
be equal and may vary from treatment to treatment
depending on the knowledge (if any) on the variability of the
observations on individual treatments as well as on the
accuracy required for the estimate of individual treatment
effect.
3
Completely randomized design (CRD)
Example:
Suppose there are 4 treatments and 20 experimental units, then
the treatment 1 is replicated, say 3 times and is given to 3
experimental units,
the treatment 2 is replicated, say 5 times and is given to 5
experimental units,
the treatment 3 is replicated, say 6 times and is given to 6
experimental units and
finally, the treatment 4 is replicated [20 ‐ (6 + 5 + 3) =]6 times
and is given to the remaining 6 experimental units.
4
Completely randomized design (CRD)
• All the variability among the experimental units goes into
experimented error.
• CRD is well suited for the small number of treatments and for
the homogeneous experimental material.
5
Completely randomized design (CRD): Layout of CRD
Following steps are needed to design a CRD:
6
Completely randomized design (CRD) : Procedure
Let the v treatments are numbered from 1,2,...,v and
ni be the number of replications required for ith treatment such
v
that n n.
i 1
i
8
Completely randomized design (CRD): Analysis
There is only one factor which is affecting the outcome –
treatment effect. So the set‐up of one‐way analysis of variance is
to be used.
yij : Individual measurement of jth experimental units for ith
treatment i = 1,2,...,v , j = 1,2,...,ni.
yij : Independently distributed following N ( i , 2 )
v
with ni i 0.
i 1
: overall mean
i : ith treatment effect
9
Completely randomized design (CRD): Analysis
H 0 : 1 2 ... v 0 Treatments
1 2 ... v
H1 : All s are not equal.
i
'
_____________
The data set is arranged as follows: y11 y21 ... yv1
y12 y22 ... yv 2
y1n1 y2n2 ... yvnv
_____________
T1 T2 ... Tv
ni
where Ti yij is the treatment total due to ith effect,
j 1
v v ni
G Ti yij is the grand total of all the observations.
i 1 i 1 j 1
10
Completely randomized design (CRD): Analysis
In order to derive the test for H0, we can use either the likelihood
ratio test or the principle of least squares.
Since the likelihood ratio test has already been derived earlier, so
we choose to demonstrate the use of the least‐squares principle.
S v ni
0 2 (y
i 1 j 1
ij i ) 0
v v ni
or n ni i yij
i 1 i 1 j 1
S ni
0 2 ( yij i ) 0, i 1, 2,..., v
i j 1
ni
or ni ni i yij
j 1
12
Completely randomized design (CRD): Analysis
v
Solving them using n
i 1
i i 0 , we get
ˆ yoo
ˆi yio yoo
where
1 ni
yio
ni
y
j 1
ij is the mean of observation receiving the ith treatment
and
1 v ni
yoo yij is the mean of all the observations.
n i 1 j 1
13
Completely randomized design (CRD): Analysis
The fitted model is obtained after substituting the estimate ̂ and
ˆi in the linear model, we get
Squaring both sides and summing over all the observation, we have
v ni v v ni
ij oo i io oo ij io
( y y
i 1 j 1
) n ( 2
y y )
i 1
( y y ) 2
2
i 1 j 1
Sum of squares
Total sum Sum of squares
or of squares = due to treatment + due to error
effects
or TSS SSTr SSE 14
Completely randomized design (CRD): Analysis
v ni
Since ( y
i 1 j 1
ij yoo ) 0, so TSS is based on the sum of (n ‐1)
squared quantities. Thus TSS carries only (n ‐ 1) degrees of freedom.
v
Since ni ( yio yoo ) 0, so SSTr is based only on the sum of (v ‐1)
i 1
squared quantities. Thus SSTr carries only (v ‐ 1) degrees of
freedom.
ni
Since n (y
i 1
i ij yio ) 0 for all i = 1,2,...,v, so SSE is based on the
(y
j 1
ij yio ) 0, so SSE carries (n – v) degrees of freedom.
15
Completely randomized design (CRD): Analysis
• Using the Fisher‐Cochran theorem,
TSS = SSTr + SSE with degrees of freedom partitioned as
(n – 1) = (v ‐ 1) + (n – v).
To ensure that the equality holds exactly, we find one of the sums
of squares through subtraction.
v 2 ni
G
yij2
i 1 j 1 n
v ni
G2
G yij : Grand total, : correction factor
i 1 j 1 n
ni
SSTr ni ( yio yoo ) 2
j 1
Ti 2 G 2
v
i 1 ni n
ni
Ti yij Treatment total.
j 1
17
Completely randomized design (CRD): Analysis
Now under H 0 : 1 2 ... v 0 , the model become
Yij ij ,
v ni
and minimizing S ij
2
i 1 j 1
S G
with respect to gives 0 ˆ yoo .
n
v ni
The SSE under H0 becomes SSE ( yij yoo )2
i 1 j 1
and thus TSS = SSE. This TSS under H0 contains the variation only
due to the random error whereas the earlier TSS = SSTr + SSE
contains the variation due to treatments and errors both.
18
Completely randomized design (CRD): Analysis
This TSS under TSS = SSE contains the variation only due to the
random error whereas the earlier TSS = SSTr + SSE contains the
variation due to treatments and errors both.
19
Completely randomized design (CRD): Distributions and
decision rules:
Using the normal distribution property of ij ' s, we find that
yij’s are also normal as they are the linear combination of ij ' s.
Under H0 SSTr ~ 2 (v 1)
2
SSE
~ 2 (n v).
2
SSTr and SSE are independently distributed.
MStr
Under H0 , ~ F (v 1, n v)
MSE
Reject H0 at * level of significance if F F *, 1,n .
[Note: We denote the level of significance here by * because
has been used for denoting the factor] 20
Completely randomized design (CRD): ANOVA table
21
Completely randomized design (CRD): Expectations
v ni
E ( SSE ) E ( yij yio ) 2
i 1 j 1
v ni
E ( ij io ) 2
i 1 j 1
v ni v
E ( ) i io. )
i 1 j 1
n E ( 2 2
ij
i 1
v
2
n ni 2
i 1 ni
(n v) 2
SSE
E ( MSE ) E 2
nv
22
Completely randomized design (CRD): Expectations
v
E ( SSTr ) ni E ( yio yoo ) 2
i 1
v
ni E ( i io oo ) 2
i 1
v
v 2
ni ni E ( io ) nE ( oo )
i
2 2
i 1 i 1
v
v
2
2
ni i ni
2
n
i 1 i 1 ni n
v
ni i2 (v 1) 2
i 1
SStr 1 v
E ( MSTr ) E
v 1 v 1 i 1
n
i i
2
2
.
In general E MSTr
2