Assumptions for one way-ANOVA :

⑧
The population from which the samples are drawn are capproximately) normally distributed
· The population from which the samples are drawn have the same variances
·
The samples drawn from different populations are random and independent

i -

One-way ANOVA :
Completely Randomized Design (CRD)
① Hypothesis
Ho : Mi M2 = = M3 =... = Mr
Hi : At least one
pair of means are differ

② Test statistic (Fcal)

Fcal = MSA
MSE

- Construct ANOVA table :

NOTES
557 = EX 2 -

S(Ex)2
N
>
- This is
You
correction factor (C
can calculate first the value then
.
F)
[Tic
2
SSA = (Ex)
substitute into the other formula
-

D N .

(x)2
SSE =
CF =

SST- SSA

source of Degree of sum of mean of F

variation freedom square square

Treatments k - 1 (vi) SSA

#SA
k -
1
Fal
=
Error k(Vz)
M
N -

SSE

Total N -
1
SST

③ Critical value
Fx ,
V, Va

# Decision Rule
Reject Ho if Fcal > Fx ,
V, V

not
&
Decision :
compare based on value reject or do reject Ho
in 2 and G

⑤
Conclusion
D = Q DF
C
At % significance level ,
there is enough/not enough evidence ....
 Example CRD :

This question only have

one factor treatment
factor
(Drug)

T3 5 TB =
48 2 Tc= 27 4
TA
.
= .
.

nB = 6 nc = 5
ANOVA Table : NA = T

Step 1 : Ta = 73 5 .

,
Na = 7
; TB = 48 .
2 , B = 6 ; Tc = 27 4 .

, Ac = 5
2x = 149 .
1 Ex = 1350 :
01 N = 18
, ,

Step 2 : SST = EX* -

(EX) Note :(F = (EX12
N N
= 1350 .
01-(149 . 1)2 =
(149 .
1)2
Ig 18
= 114 965 .
= 1235 045 .

CEx
[TA T]
SSA =
+
To + -

RB N

+ 48 23 +
27.43] 1)2
2
(149
(73
=
.
5 . -
·

7 6 18
= 74 0637 .

SSE
= SST-SSA
= 114 965 .
-
74 0637 .

=
40 9013 .

Step 3 :

source of Degree of sum of mean of F

variation freedom square square

Drug 3 -1 =
2(V) 74 . 063774 .
0637 = 37 031937 0319
.
.
=
13 . 5807
2 2 .
7268

Error 9013 7268

40 901340
2
15 (82)
=
18 3
.
=
- .

13

Total 18 -1 17
=
114 965
.
 ① Ho : Ma = MB = Mc
Hi : At least one
pair of means are differ (C)(D)
or
cequal)
Ho : There is no difference in the mean blood glucose level in three different
type of
drug
c not equal)
H : There difference in the
is a mean blood glucose level in three different
type of drug (C) (D)
② Fcal = 13 .
5807
③ C V : Fo 05
.
.

,
2
,
15 = Fo 05
.

,
2
, 14 = 3 74
.

# D R Reject Ho if Fcal Fx
< Fo
. :
,
V, Va

Since Fcal = 13 .
5807 . 05
,
2
,
15 = 3 74
.

reject Ho
,

⑤ At 5 % significance level
,
there is enough evidence to conclude
that there is a difference in the mean blood glucose level in three different

type of
drug
.

MINITAB OUTPUT :

k 1 -
SSA MSA

N-K MSE Fal = MSA p-value

SSE -

MSE
N 1
SST
-

Note If the
:
question already give minitab output No NEED
, you a

to construct . Just take the value from there

ANOVA table .

Using p-value :
① Ho : Ma MB = = Mc
Hi · At least one
pair of means are differ (C)(D)
② p-value : 0 000 S < 0 05 .

<
= .

③ D R : Reject Ho if p-value &

.

Since p-value 0 000 <C 0 05 reject Ho = .

= .

# At 5 % significance level , there is enough evidence to conclude

that there is a difference in the mean blood glucose level in three different
type of
drug
.
ii -

One-way ANOVA : Randomized Complete Block Design (RCBD)

① Hypothesis

3
Ho : There is nodifference/no significant mean effect on
mean

dependent variable due to treatment factor T

Hi : There is a mean difference/a significant mean effect on
dependent variable due to treatment factor

3
Ho : There is nodifference/no significant mean effect on
mean

dependent variable due to blocking factor B

Hi : There is a mean difference/a significant mean effect on
dependent variable due to blocking factor

② Test statistic (Fcal)

ms
Fa = Treatment

FB
3
mSB
=
Block
MSE

- Construct ANOVA table :

SST = EX2-(x)2
N

SSA = [Tic -
(x)2
M

Di N
:
SSB = ETJ ~
(x)2
Dj N
SSE = SST-SSA-SSB

source of Degree of sum of meanof F

variation freedom square square

37
Treatment k 1 (V, ) SSA FA = MSA
SSA
-

k -
1 MSE

mSBT
Block b 1(V) SSB FB = B
SSB
-

b - 1

Error SSE
(k -
1) (b 1) (Vz)
-

SSE (k - 1) (b -
1)

Total N -
1 SST
③ Critical value VI Va

Fx V, Va : FC, r - 1
,
(k -
1) (b -
1) >
-
Treatment
,

Fa , b - 1
,
(k -
1)(b -
1) >
- Block

# Decision Rule
Reject Ho if Fcal Ex ,
V, Va

compare based
&
Decision : on value not Ho
reject or do reject
in 2 and G

⑤
Conclusion
D = Q DF
C
At % significance level ,
there is enough/not enough evidence .
 Example RCBD :

Factor : Drug
Block : Exercise

Ti = 22 . 3, n1 =3

T2 =
21 . 2, na = 3

73 24 9 My = 3
-
,

Ta = 27 7 . TB 26 = Tcz14 7 .

na = 3 nB = 3 nc = 3

A NOVA table :
Step 1 : Ta = 27 7 ·
sna =
3
; TB = 26
S nB =
3jTc = 14 7
.

, nc = 3
7 = 22 . 3 , n =
3 ; Tz = 21 2 .

-N2 = 3
; Tz
=
24 9 .

,
13 = 3
2X 567 76 N
= 68 4 .

-[x = .

9
= 9

Step 2 : SST = EX
=
-
(Ex) Note :(F = (EX12
N N
2
567 76 168 4 S
= . -
168 4)2 .
= .

9 9
=
47 92 .
= 519 84 .

(Ex
[TA T2]
SSA =
+
TB t -

RB N

J
(68 4)2
2
27
07 1407
= t 26 + -
.

3 3 3 9
= 33 2867 .

(EX)2
[ J
2 2
SSB = 7, t
T2 t T3 -

n. n2 M3 N

3)
32 22 + 24
92 168 4)2
=
22
t 21
. .
. .
-

3 3 9
= 2 .
4067
SSE = SST-SSA-SSB
= 47 9233 28672 4067
. . .

= 12 . 2266
 Step 3 :

of sum of of
source
Degree of mean F

variation freedom square square

2867 643416 6434
Drug 3 -
1 = 2(Vi)33 . 286733 .
= 16 .
.
= 5 .
4449
2 3 . 0567
Exercise .
2 4067 1 2034 1 2034 0 3937
(vi) 4067
= =
2
.

3-1
.
2
.
= .

2 3 . 0567
Error 12 2266 3 0567
4(V2) 2266
. = .

2 x2 =
12 .

Total 9 1 S
- =
47 92 .

&1 :
& Ho : There is no difference in the mean blood glucose level in three different
type of
drug (D)
H : There is a difference in the mean blood glucose level in three different
type of cas
drug
② Fca 1 = 5 . 4449

③ C V Fo 05 92 4
. : .
,
= 6 94
.

⑪ D R Reject Ho if
. : Fcal > Fx ,
V, Va
Since Fcal = 5 .
4449F0 . 05 , 2 , 4 = 6 94.

S
do not reject Ho

⑤ At 5 % significance level
,
there is not enough evidence to conclude
that there is a difference in the mean blood glucose level in three different

type of
drug
.

Q2 :
① Ho : There is no difference in the mean blood glucose level in three different
type of exercise (D)
H : There is a difference in the mean blood glucose level in three different
type of exercise (a)
② Fcal =
0 .
3937

③ C V Fo 05 92 4
. : .
,
=
6 94 .

⑪ D R Reject Ho if
. : Fcal > Fx
,
V, Va
Since Fcal = 0 .
3937 < Fo . 05 , 2 , 4 = 6 94
.

S
do not reject Ho

⑤ At 5 % significance level
,
there is not enough evidence to conclude
that there is a difference in the mean blood glucose level in three different

type of exercise.
 MINITAB OUTPUT :

k 1 >Treatment

A
-
SSA MSA -

b - 1 SSB mSB >

- Block

(k 1)(b
- -
1) SSE MSE mSE

N - 1 SS7

Using p-value :
&1 :

① Ho : There is no difference in the mean blood glucose level in three different

type of
drug (D)
H : There is a difference in the mean blood glucose level in three different
type of drug cas
② p-value : 0 072 S < 0 05
.

&
= .

③ D R : Reject Ho if p-value &

.

Since p-value 0 072 7 C 0 05 do not reject

= .
= .

,
Ho

# At 5 % significance level , there is not enough evidence to conclude

that there is a difference in the mean blood glucose level in three different
type of
drug
.

Q2 :
& Ho : There is no difference in the mean blood glucose level in three different
type of exercise (D)
H : There is a difference in the mean blood glucose level in three different
type of exercise (a)
② p-value : 0 698
S
.

&
<
0 05 = .

③ D R : Reject Ho if p-value &

.

Since p-value 0 698 >C 0 05

= do not reject
.

= .
Ho
,

O
. At 5 %
4
significance level , there is not enough evidence to conclude
that there is a difference in the mean blood glucose level in three different
type of exercise.
 iii-Two-way ANOVA with interaction / Factorial Design
ANOVA
Assumptions for
two
way
-

·
The population from which the samples are drawn are capproximately) normally distributed
· The population from which the samples are drawn have the same variances
·
The samples drawn from different populations are random and independent
sample size
· The groups must be equal in

① Hypothesis

3
Ho : There is no mean difference/no significant mean effect on

A
dependent variable due to treatment A

Hi : There is a mean difference/a significant mean effect on

dependent variable due to treatment A

3
Ho : There is no mean difference/no significant mean effect on

dependent variable due to treatment B B

Hi : There is a mean difference/a significant mean effect on

dependent variable due to treatment B

3
AB
Ho : There is no effect between TA and
interaction TB on dependent variable
H1 : There is an interaction effect between TA and TB on dependent variable

② Test statistic (Fcal)

3
FA - mSA Treatment A

MSE

FB
3
= MSB Treatment B
MSE

FAXB Interaction AB
3
= MS AXB

MSE

- Construct ANOVA table :

SST = EX2-(Ex)2
N
SSA = EAi2 -
(Ex)2
. of treatment Bx replicate = br
No N
:
SSB = EBJ _
(x)
2

. of treatment A xreplicate
No = ar N
SSAXB =
[ABij-SSA-sSB - (2x)2
replicate = r N

SSE =
SST-SSA-SSB-SSAXB
 source of Degree of sum of meanof F

variation freedom square square

msA
SSA
A a - 1 (v) SSA - FA =

9 -
1

MSB 3
SSB FB
B b -
1 (V) SSB =

b - 1

msax]AB
(a 1) (b 1) (v) SSAXB Fax
SSAXB
- -

AXB
=

(a -
1)(b -
1) MSE

Error ab(r 1) SSE

(Vz)
-

SSE
ab(r -
1)

Total abr -
1 SS7

③ Critical value VI Va

Fx FC, Treatment A
,
V, Va : a - 1
ab(r-1) >
-

F2 , b-1 ab(r -1) >

-
Treatment B
,

Fa , (a-1) (b- 1) ab(r 1) --

- Interaction AB
,

# Decision Rule
Reject Ho if Fcal , Fx ,
V, Va

not
compare based
]
Decision : on value reject or do
reject Ho
in 2 and G

⑤
Conclusion
D = Q DF
C
At % significance level ,
there is enough/not enough evidence .
 Example Factorial Design :

① e O
3
TAI= TBI= Tl = Ti = 49 .
2
22.
7 17 7 8 S .

47.
.

⑪A2=TB2 = ⑥Tcz =

18 I
.
17 9 . II .
S n2= 6
G
TTA3 &TB3 g
97
-

477
=
= =
73 = .

15 2 14- 2
18 3
.
.

nz =

TA = TB 8
5034Ta= 8
5
. .
I

nA nB 26 ni = 6

1 : Ta
Step 59 1s Na 6
; TB 50 8 6 j Tc 34 8 Mc = 6
=
= . = .

, B = = ·

7) Tz 6
49 2 71 6 47 8 ; Tz 47 7 D3 6
=
;
= = = =
z =
.
, .
,
,
.

Tal = 22 . 7 Tal = 2 ; TAz = 18 1 ·

RAz = 2 ; TA3 = 18 3 .

; RA3
= 2 ; TB1 = 17 7 ; RB1 = 2
·

, ,

TB2 = 17 9 , rB2
·
= 2 ; TB3 = 15 . 2
, rB3 = 2
; Ta = 88 .

,
c = 2
; Tcz = 11 . 8, raz = 2

Tc= 14 ·
2 rc3 =; Ex 144 7 = . Ex = 1256 .
99 N = 18
, , 9

EX
=
(Ex) Note :(F (EX12
Step 2 : SsT =
= -

N N

= 1256 99 .
-
(144 7j2 : = (144 . 7)2
IS 18

=
93 7628 . = 1163 . 2272

((x)2
[Taz Tc]
2
SSA =
t
TB + -

No .
of Exercise , b = 3 O br br br N

J
59 12 82 82 7)
2

2144
34

=
r= 2 50
t t
. . ·
.
-

3x2 6 -6 6
6
=

= 50 .
8544
(EX)2
[ I
2
SSB =
t
Tz2 t T3 -

No
. of Drug a = 3 ar ar N
,

J
49 22 t 47
82 + 47 72 (144 7) 2
[
r= 2 =
. . .
-
·

16
3x2 = 6 - 6 6 18
F 0 .
2344

[ + TAz & 7A3" +

TB Tre TB3 T T T
SSAXB = TAP + + + + +

r r r
2
Note : make sure followSSA-SB-EX)
S C

treatment combination - ① ② Ng 3 ⑪ O
S ⑥
(axb)3x3 9 =

23)
2
2
82 +
[22
2
17 9 15 2 t 8 82
2
12 t
-

=
.
7 t 18 .
18 3 :
t 17 7 .
2
+ . t . .
11 .
+ 14 .

No
· of replicate --2 2 2 2 2 2 222

-
30 .
8544 -
0 .
2344 - C14 4 .
732
= 16 109 .
18
 SSE = SST-SSA-SSB-SSAXB
= 93 7628 .
-
30 .
8544 -
0 .
2344 -
16 109
.

= 26 365 .

Step 3 :

source of Degree of sum of mean of F

variation freedom square square

Drug 3 -
1 = 2
(V,)50 .
854450
.
8544 = 25 . 427225 4272 .
= 8 6144 .

2 2 9517
.

Exercise 0 2344 1172 1172 0397

(V,) 0 0
.

2344
=
2 0
3 1
.

0
.
- = .
.
=

2 2 9517
.

Drug X 16 109 16 109 =

4 02734 0273 1 3644
4(V)
. .

2x2 = .
.
= .

Exercise 4 2 9517 .

Error . 365
26 2 9517
3x3(1) =
9(V2)26 .
565
= .

Total
3x3x2 -
1 =
1793 7628 .

&1 :
& Ho : There is no interaction effect between type ofdr and ty pe of exercise (DS
Hi : There
ug
interaction effect between type of and type of exercise (d)
is an
drug
② Fcal = 1 .
3644
③ C V Fo 05 . : .
, 4 , 9 = 3 . 63

# D R Reject . : Ho if Fcal Y Fx ,
V, Va
Since Fcal = 1 .
3644 < Fo . 05 · 459 = 3 . 63
,
do not reject Ho

⑤ At 5 % significance level
S
there is not
enough evidence to conclude
interaction effect between
that there is an
type of drug and type of exercise .

&: should we test for the main effects ?

A : Yes since there is no interaction effect between type of drug and type of exercise.
,

Q2 -

g1 Ho : There is no difference in the mean blood glucose level in three different

type of
drug
H : There is a difference in the mean blood glucose level in three different
type of
drug (a)(D)
② Fcal = 8 6144 :

③ C V Fo 05
. : .
,
2
, 9 = 4 26
.

# D R Reject . : Ho if Fcal > Fx ,

V, Va
Since Fcal = 8 6144 .

7 Fo . 05 , 29 = 4 26.

reject Ho
,

⑤ At 5 % significance level
,
there is enough evidence to concludethat
There is
-

a difference in the mean blood glucose level in three different type

of
drug
.
 Q3 :

& Ho : There is no difference in the mean blood glucose level in three different
type of exercise (D)
H : There is a difference in the mean blood glucose level in three different
type of exercise (a)
② Fcal = 0 .
0397

③ C V Fo 05. : .
,
2
,
9 = 4 26 .

# D R Reject . : Ho if Fcal > Fx ,

V, Va
Since Fcal = 0 0397
.
< Fo . 05 , 2 ,
9 = 4 26
.

S
do not reject Ho

⑤ At 5 % significance level
,
there is not enough evidence to conclude
that there is a difference in the mean blood glucose level in three different

type of exercise.

MINITAB OUTPUT :

9 - 1
- A
b - 1
- B
CInteractions (a 1)(b -
1)
-

>
- AB
ab(r 1) -

abr -1

Using p-value :
&1 :
& Ho : There is interaction effect between type of dr
no and type of exercise (D)
Hi : There
ug
is an interaction e ffect between type of and type of exercise (d)
drug
② p-value : S
0 320.

&
< 0 05 = .

③ D R : Reject Ho if p-value &

.

0 05 do not reject
Since p-value 0 320 Ho
- C = .
= .

# At 5 % significance level S there is not enough

G ·

evidence to conclude
interaction effect between
that there is an
type of drug and type of exercise .

02 :

& Ho : There is no difference in the mean blood glucose level in three different
type of
drug
H : There is a difference in the mean blood glucose level in three different
type of
drug (a)(D)
② p-value : 0 0 08 S < 0 05 .

&
= .

③ D R : Reject Ho if p-value &

.

Since p-value 0 008 < C 0 05

reject Ho
= .
= .

① At 5 % significance level , there is enough evidence to concludethat

There is
-

a difference in the mean blood glucose level in three different type

of
drug
.
 Q3 :

& Ho : There is no difference in the mean blood glucose level in three different
type of exercise (D)
H : There is a difference in the mean blood glucose level in three different
type of exercise (a)
② p-value : 0 961 S < 0 05
.

&
= .

③ D R : Reject Ho if p-value &

.

Since p-value 0 961 0 05 do

not reject Ho
7 C = .

= .

O
4 At 5 % significance level , there is not enough evidence to conclude
that there is a difference in the mean blood glucose level in three different
type of exercise.