2134 1A 7359

Module 2: Design Of Expeiments

a) Define "De siqn of Experiment" Write about the impotant

terms and definitions in Experimental Designs )
sol Desiqn of Experiments: T+ is defined as a branch of
applied statistics that deals with planning, conducting
test of t evaluate
analysing & interpreting controlled
value Ot a Parame ter Cor)
the factors that control the
gioup ofOf parame ters .
a
Purpose of DOE The purpose of DOE is to find
ation ot input variables that wil) resutt in an
combin
value.
Output variable with the desived
Importance of DOE:
DOE s used in the development Optini'satton of
I- Manufactuving process products, such as enqines, Semi
conductors.
2Analysis of dtuq
drug compounds in the pharma ceutleal
industries or in chemio research.
3. Angly tical methods or instruments to dete rrmine theit
best (or) most Sensitive pertormance,
4. Agricultural crop yields by Studying varbus factore
effecting the outcomes Such as sol types, fertilizers, plant
Spacing, surrounding tempetatures etc..
* Design of Experiments follows Completely Randomiged
Design (CRD), Randomised block Design (eeD).
(RBD), Latin
Square Design (1SD)
 Completely randomized Design (eRD)
CRD is one where the treatrnents are assigned
Completely at Yandom SO that each experimentaf unit
has the Same chance of receiving any one treatment
în CRD, total variation is diuided into & componentc
ie, treatment and error
Randomized Block Design (RBD):
în this design the total area is de s'gned into
no- of bocks. A the treatnments are allocateo in to each
block without teplication with The toto woinq conditionc
No of treatments = No. of plots in each block
2ND of times a treatment is tepeated -n0:of blocks .
Lat tice Square Design (Lso):
area 's diuided into ome
In this desian the total
no of tows 8 columns are Same no. of treatments
alloca te d in the plots at a andom mannet.
are
But, oa restricti'on is
no of rou)s no Ot colunnS= nO Of reatments oe

wetl as each treatment is represen ted only once in

each R each column
 b) Explain atout the Principles of on Experimental Design)
ol:
Principles of Design of Fxperlments
. Randomization : In onder to mininize bias 2 ensure
that the results obtained are repreSen tative Of the
Population, it is impot tant 06 o design randomls
experiment al units t0 ttterent treatme nts/ Conditone.
D. Replication: Replication inuo lves repeating the experinent
multiple times under ident'cal onditions to encure
|that the results obtained are consistent &reliable.
3 Blocking: ît involves grouping sinilar xperimental units
together to reduce the etfects of extranious vanableg
that mas effect the tesults.
4 Contyol: A Control group is use d to prou'de a baselin -
aganst which the effects of the treatment can 6e
conpared .
5 factorial design :A factorial cesign ivoves manipulatina
muttiple independent variables Cor) fac tots Simuttaneousk.
to deterroine tb
thein individual 2 Combined effects on the
response variable.
2 Carry out analysis of varian ce for the
Designs with 4
folloing pevimerld
treatments f, B, C, D at y. L0:s
A A
(2) (4) (2) (s)
B B A A
4) (3) (5) (8)
A D
(5) (1) (6) (1)
sol'
D Mean
48
4 3 35
53
55

24

Null ttypothesi's (Ho): There is no significoance diffe rence

blw the 4 treatments.
Alternative ttypothes's (H):There is significance
blw the tre atments. difference
leuel of Significance (a);

V, k-1
= 1)-4
feri (3, 8) at =759
-3

Calcul ation:
I) Grand total (a) - 22 Xij = 2t4t5t8 t5+ 4t3+2t7t7+6+5
= 58

i) Coerection factor (cr) = %- 58/ a80-3

38 759
significance difte tence betuseen the 4 trea thnept
Hi) Rou sum of squares (Rss) Ex;;- a't 45
94+5tt6 +5
RSS- 322
iv) Total sum of Squares (1ss) RSS- CE
-322- 2803 417
2

v) Sum of squaves of columns (ssc)- 1,3

+
t 280"3
5 2 3 2 ’ 523
V)Error Sum Of Squares CEs s) - 1S-SsC
= 4:7- 523 3647
vii) Mean sum of squares of colunns (Mssc) SSC
= 523
’74
vii) Error Mean sum t 3

Squares (EMss) = ESS 3614) =45

V 8
Anova One way Table:
Source Of Sum of Mean sur)
DOf f-test
Variation squares O7 squarer

Between the 3 MSSC =l:74 t 74

SsC =5 23
Column s 45

=038
nithin ESS - 36.47 EMSS - 45
Errors

Total
Conclusion: fcal fcri
038 < 59 Accept to
*.There is no significance diFlerence between the 4 treatnents.
 DeSce nding order Mean oifferences
A
4:8
55
5:3 O2
D 5'3
55 05
B 48 o5
3:5 3-5 (.2o:3

3 Define RBD. write about the advantage s and the

disadvantaqes of RBD.
solk Randomised Block Design (RBD):
Tn this design the total aLrea is desiqned into no of block.
Al| the treatments are allocated into each 6lock uithout
replication with the tolloving a Condition ss
block,
· Noof treatments = No of plots in each
2-Noof times treatment is repeated - no of 6lock.
Advantages 0f R8D
(. The precision is more in RBO.
2 The amount Of in formation 0btained in
RBO i's more as
Compared to CRO.
3 RBO is more tlexible Stat'sttal
easy
analysis i's simple 2
h. Even if sorne values are
be done by using missing, stll the
missing plot echnique . analys's can
 4)
Disadvantages of RBO:
.when the no of treatments is ino ea se d, the block size
wt incvease.
2Te the block size is large mointaining homogeneity is
difficutt hence where no of treatments is present this
design may not be suitable.

Carry out analysis of vaniance for the folloning experinenla

|designs with y heatnents AB,CD at 5/ t0-s

Block- 1 Block-2 Block-3

A
() (6) (1) 20

C D

(8) (2) (5) 15

B C
() (3) ()
A
(5) (4) (8)
29 (5 27

Null ttypothesis(Ho): here is no signiftcante difference betuaer

the 4 treatments
Alternative ttypothesis (H): There is siqnifican ce difference betoeer
the 4 teatments.
level of Siqni f'cance (a) : fcol cri (26) at 5/ a
Freol eri = ((s-), (3-a)) (3, 6) ’
 frawci - (C4-), (3)l4-)) ’ (3, G) ) froi (36) at 5%.

Calculations
Grand 7otal (G) - 22X/ = 9)
2 Cortection factor (cr) : = 42008
12
3- Rouw Surn of Squares (Rss)
1t8t4t+
4.70tal Sum of Squares (TSS) RSS -CF 50.92
5: Surm oE squares of columns (ssc) - 2q, 5
4 -420·08
= 2867
2
6" Sum of squ ares Of rows (sSR) 20 19
3 3 3 42008
4.92
7: Error Sum of squares (ESS) TSS - SSC - SSR
|733
Mean Surn ot squares of columns (MSSC) Sse
28'67
(4:335
q Mean Surn Of squares Of YoUOs (MSSR) = SSR

= l64
3
0. Error Yean Sum Of
Squ ares (EMSs) ESS
I7:33
2X3
 Anova Two wu lable

Meon
Source Of variat'on Sum o DO:f
squares Sum o f-test
Squares
Columns 867 3-l:2 I4335

Rovos 492 4-l:3 (64

Error 733 288 rcal =l75

Total S092

Conclusion -

Fecal < fc chi Frouw cal <

t497 < 54 (75 4:76

Aceept Ho
.There is no significance difference be tweern the 4
treatments.

8 Carry out analssis of vaiance torDthe following Gaperimenat

4 treatments A Br C, at 5y. boc
Desgns with

A D A C
tto) (2) () (5) (n)
A B
(30) (30) (16) (i5) (e)
A 8
(5) (10) (1) (18) 6)
sol
A

0 t5 (2

14 30 15 Mean
20
18 20 I514
8 l6 t633
15 20:75

17 4.66
49 83 44

Null ttypothesis There i's no Significance difference

betyeen the 4 treatments.
Alternative thypothesis (H): There i's significance difference
between the 4 treatments.
level of Significance (a):

F(3,() at 5/ d= g59
- 15-4
3 =||

Caleulations:
Grand Total (G)= 22X(j - Tt 2t3t Ty 977+49+83+49
= 253
2
2 Coreeetion factor CCe) G7 253 = 4267: 26
I5
3
3: Rouw Sum of Squares (Rss) =o+
o 5ntt+16 +r5
= 4553
4 Surm of squares of Columns (ssc) 77'+ 49
5 3
t4 t
3
SSC 8645 - 4261: 26
85 Total sum of Squaves (1ss)
=RsS-CF
= 45 53- 4267. 26 )
285-74
6 Error Surn of Squares (Ess) 1Ss - Ssc
= 285-74- 86.45 19929
7 Mean Sum of squarr s of columns (MSSc) SSC
= 8645

8 Er ror Mean surm of Squares (E mss)

3 28-81
ESS 199:29

Anova One-was Table

Source Of Sum dk D'f Mean sur
f-test
variation Squares o7 squaes

Columns 86-45 3 2881

Ercors 8"|
|99:29

Total 28574 t4

Conclusion'

Fcal < fen

:59 < 3:59

Acept Ho
There is no signi ficance differe nce between the
4 theatments.
Descending order Mean Differences
2075
4 42
B l6:33 349
O93 3-3
A 154 ot9
D 4:66 O74.
 5Carry out analysis of variance for the follooing
Experimental Designs with 4 treatments A, B, C,O at 7-(O.6

D B towtota
291 184 29:4 57 826

A 8
t64 21:2 G69
EA= 30:8
A D B
EB= 864
59 38-8 24-0 37·0 105 2
EC = 124.5
B A
24:9 q5 28-9 I05

column
75'8 t091 84:4 90-] 359.7
total
Solk Null typothesis (Ho) :There is Sinificance
no difference
between the rows, Columns treatments .
Alternative typothesis (H):There is significance diffetence
between the roKS, columns 2 trea trnents.

level of Significance a):

feol o (ce-), t-)e) (u-), (u-d(4-2))
(36)

(3,6)
frow ci (3,6) at Y. d = 4.78
fcol ct (3r6) at l . d= 9:78
 Calculations:
I) Gtand total (a)= 2xj = 359.1
) Correetion factor Cce) - G : 359:7 2
= 80865
3) Row Sum of
of Squares (Rss) - (2q-)4 (1g.4....
t. +(a5)+(38
- t0033·43
)Total Sum of Squares (TSS)- RSS- CF )
t0033·43-80865
= 1946.93
5) Surn of Squares of colurmns (SsC) = 2

15:8109.84-an.7
+4
4
2

- 80865
SSC = |504375

6) Sum of squares of towS ( SSR) =82.6 2

66.9 105-2
4

SSR= D6t"1025 8086.5

1)Surm of squares of treatnents (sst) A
2
R sl
n
2
SST - 30:8 -CP
86:4 +124:52 18 2

4 4 - 80865
4
13729625
8) Error Sum of Squares (ess) = 7ss -ssc- SSR- SST

=
1946:93-t50.4375 -26{·1025 -1372.961 5
ESS - 624275
4) Mean Sum of squares of colurnns (MSSc) = SSc
(50:4375 C-|
 SSR
0) Mean Sum of Squares of tows (MsSR -
261025
3

MSSR : 87.03

) Mean Sum of Squaec of tea trnents (MssT) SsT

t-1
= 1372 9625
)457:65
3
2) Error Mean Sum of Squares (E MSS) - ESS
Cc-) (r-)(t-)
162 4275
3X3x3

ANOVA Table

dof Sum Oß Mean Sum

Soutce Of f-ratio
vaiatiorn squarrs c squae!

SSC=150-4375 50-(4 colcal -8-39

colurnns c-l- 3

rows r-l-3 261 1025 87:03 fr cal 4:48

Treatment t-l=3 1372.9625 457:65

Error (t-)et-2) t6 24275

=6
Total 85 (946.93

Conclusion:
Fcol cal < col cri Frow cal 7 frow ot
8-34 q:78

9-78
 Reject Ho, Accept H
.There is Siqnificance difference between the
vows,
so
colunns & theatments .

CHical difference (C.D): [Link] 2(MESS)

t

te 2 (6-o1)
('tE:d:f tat o-o05) ’ 3:707
4
= 3·707

G Consider the results qven in the folloroing table for an

experinnent involuing six treatment sin four randonize d
blocks. he tieatments are indicated by numbers within
parantheSeS -

Blocks Vield for a tandomized block expenment tiratnent

and yield
Col
(3) (2) (4) (5) (6) total
()
27: 2016 tb2 lb-2 249 1303
(2) (4) (6) (5)
(3)
227 288 27:3 15·0 22:5 17o 133·3

(4) () (3) (a) (5)

(4)
3 263 38-5 36-8 3915 15-4 116.
(5) (2) () () (3) (6)
17.) 31·0 28-5 34:9 226 128
YOuw total 94 t07 (t4:9 82-1 I13-1 19:9
 Test whether the teatments differ significantly. Also detrmhio
the citical diffelence between the means of any thuo
treatment s )

sol': Null ttypothesis (Ho):"here is no sgnifilcance difference

between the 4 bocks
Al4ernative Hypothesi's (H):There is sianifitance di fference
between the 4 blocks.
level of Significance («); O-05 (5/ a)
Frouw ci = ( Cr-), cc-)(r-)
=( C6-), (u-) (6-)) (5,15) - 2.90
Fcol cvi ((c-), (e-) r-d) ((u-0, (*-) (6-0) (3,15)- 3-29
Caleulations
)Grand total (e) : 2Xi; 588-5
) Correc tion factor Ccr) - s'o - 5885
= (44305|
24
3) Roo Surm of squares (RSs) =
S2xij (24-1)+ (22.7)4....
RSS = 580-77
+ ('5-9) (32-6)
91otal Sum of Squares (1ss) Rss -ce
=
5)Suro of 15780:17- 4430·5| ) 1350-26
squares of colurnns (SsC) =(30-3,
+ (33+3 T%.tu9.
(76.1 2
148-8
6 6 4
6 6
- 144305)
SSC = 2lq42
 2.
6) Surn of Squares of vDws (sSR) o7 (43 89 13.

+ 79.
-|4430:5}
SSR= 30509
) Error Sum of Squares (ESS) = 1SS-
SSc-SSR
= 350"26 -
219:42-305'09
825"74
8) Mean Sum of squarer Ok colunnns (mssc) -
$SC219.42
C-) 3
= 134
) Mean Sum of
Squares Of rows (MSsR) SSR
305-09
5

0) Error Mean Sum of squares (EMSS) ESS

Cc-0tr-)
825:74
3x5 ) 55-04

Anova Two way Table

Source of Sum of Meansum F-test

variation Squares o squares

Columns 219:42 4-[=3 13"4 fe cal = t32

Rows 305:09 G-l=5 Gl·01 fr cal =10
Error 82574 I5 55-04
T0tal 350.26 23
Conclusion:

Fecal < fe cri fr cri

32 2:90

Accept Ho
There is no sonificance difference be tween the y 6locke

Critical difference
2(ME SS)
c-0=dfJ t

= tis 2x55 04
’ 2-131 2X55o4
’9.12
(' tedf - tis at o.o25 &)

Efficiency of R8O relative to its layout as CR0 is giren

by E (r-)(MSSR) +r(c-)(MESS)
r(c-) (MESS)
= (6-) (6l'o)+ G(4-)(55-04)
·3079
G(4-D(55o4)
A Define cRO: Write about the advantages 2 disadvantages
Of CRD?
sol Complete Randomise d Block (cRO): CRO is based on the
principle of randomization and replicaton. In this design
total area is diuided into number f tows number of
columns treatrment allo catt d into he parts at random
manner.

Aduantages Disadvantaqes
: It is easy to ayout ( Tn certain circumstanes
the design the design suffers from the
disadvantage of being inherent
-ly tess informative other
more sophistica td layouts.
[Link] desiqn allows for 2 Since the tandomisation
the maximum nurnber Of is not resthicted in any
deqre es of freedom in the direction to ensure that the
error Sum 0t Squares. units receiuing one treatment
are similar tD those
teceiuing the other treatme

3 A CRD has the simplust 3. The whole variationg

|analysis of all experimental among -the experimental
design o rt Statistical analysig. units is include d in
Unequalno of repe tat'ons for the residual variance
the van'ous treatments may be
ineludcd without complicating
the analysi's in most cages.