Analysis of Variance (ANOVA)

What is the F-test?

The F-test is a parametric test used to analyze or to test the significance of differences among
means of three or more groups of independent samples. It is also known as the Analysis of
Variance (ANOVA).

Simple analysis of variance or One-way ANOVA

Used when there is only one variable involved. It is based on two sources of variations:
1. Actual differences of the means due to treatment. This is represented by the squares
between columns (SSb).
2. Chance or experimental error. This is represented by the sum of squares within
columns. (SSw).

How do we use the F-test?

The mathematical notations are:
1. Total Sum of Squares: TSS = Σx2 – ¿ ¿
2. Sum of Squares Between Columns:
1
SSb = Σ¿
no . of rows
3. Sum of Squares Within Columns:
SSw = TSS – SSb

DEGREES OF FREEDOM
1. Total Degree of Freedom: dft = N – 1
2. Between Column Degree of Freedom: dfb = k – 1 (k is the number of column)
3. Within Columns Degree of Freedom: dfw = dft – dfb

SUM OF SQUARES
1. Mean Sum of Squares Between Columns:
SS b
MSSb =
df b
2. Mean Sum of Squares Within Columns
SS w
MSSw =
df w

Computed F:
MSS b
F=
MSS w

ANOVA TABLE
The ANOVA table has five columns:
1. Sources of variations
2. Degrees of freedom
3. Sum of squares
4. Mean squares
5. F-value, both the computed and the tabular values

Sources of Variations
1. between the groups,
2. within the group itself and
3. the total variation

ANOVA TABLE
Sources of Degrees of Sum of F-Value
Variation Freedom Squares Mean Square Computed Tabular
S Sb See the table
Between K−1 S Sb at desired
Groups df MS Sb level of
F=
MS Sw significance
Within ( N−1 )( K −1 ) S Sw S Sw with df
Groups df between and
within groups
Total N−1 TSS

Interpretation of Results:
The F-computed value must be compared with the F-tabular value at a given level of
significance with the corresponding degrees of freedom of BSS and WSS

If the F-computed value is greater than the F-tabular value, disconfirm the null hypothesis and
the research hypothesis is confirmed which means that there is a significant difference between
and among the means of the different groups.

Example:
The following table indicates the number of bottles of four popular brands of soy sauce sold by
a supermarket on six randomly selected days. Test at 0.05 level of significance that there is no
significant difference in the average number of bottles sold of the four brands of soy sauce

A B C D
29 23 45 23
36 19 60 40
22 41 33 42
34 27 36 29
29 12 31 53
 45 35 40 32

2
816
TSS = 30450− = 2706
24

Sum of Squares Between Columns:

2
1 816
SSb = (1952 +157 2+245 2+219 2)− = 699.33
6 24

Sum of Squares Within Columns:

SSw = 2706 – 699.33 = 2006.67

Mean Sum of Squares Between Columns:

699.33
MSSb = = 233.11
3

Mean Sum of Squares Within Columns

2006.67
MSSw = = 100.33
20

F computed:
233.11
F= = 2.3234
100.33

ANOVA TABLE
Sources of Degrees of Sum of F-Value
Variation Freedom Squares Mean Square Computed Tabular
699.33
Between 3 699.33
=233.11
Groups 3 233.11 3.10
F= =2.32
100.33
Within 20 2006.67 2006.67
=100.33
Groups 20
Total 23 2706

The 5-step solution

1. HO: There is no significant difference in the average sales of the four brands of soy sauce
H1: There is significant difference in the average sales of the four brands of soy sauce.
2. α = 0.05; df = 3 and 20; Critical value = 3.10
3. Decision Rule: Reject Ho if
2.32│ ≥ │3.10│
4. 4. Decision: Do not reject Ho since 3.10 > 2.32
 5. Conclusion: Since the F-computed value of 2.32 is less than the F-critical value of 3.10 at
0.05 level of significance with 3 and 20 degrees of freedom, the null hypothesis is
accepted which means that there is no significant difference in the average sales of the
four brands of soy sauce.

A sari-sari store is selling four brands of detergent. The owner is interested if there is a
significant difference in the average sales of the four brands of detergent for one week. The
following data recorded:

BRAND
A B C D
7 9 2 4
3 8 3 5
5 8 4 7
6 7 5 8
9 6 6 3
4 9 4 4
3 10 2 5

2
156
TSS = 1014− = 144.86
28

Sum of Squares Between Columns:

2
1 156
SSb = (37 2+57 2+ 262 +362 )− = 72.28
6 28

Sum of Squares Within Columns:

SSw = 144.86 – 72.28 = 72.58

Mean Sum of Squares Between Columns:

72.28
MSSb = = 24.09
3

Mean Sum of Squares Within Columns

72.58
MSSw = = 3.02
24

24.09
F computed: F = = 7.98
3.02

ANOVA TABLE
Sources of Degrees of Sum of F-Value
Variation Freedom Squares Mean Square Computed Tabular
72.28
 Between 3 72.28
=24.09
Groups 3 24.09 3.01
F= =7.98
3.02
Within 24 72.58 72.58
=3.02
Groups 24
Total 27 2706

The 5-step solution

1. HO: There is no significant difference in the average sales of the four brands of detergent
H1: There is significant difference in the average sales of the four brands of detergent
2. α = 0.05; df = 3 and 24; Critical value = 3.01
3. Decision Rule: Reject Ho if
│7.98│ ≥ │3.01│
4. Decision: Reject Ho since 7.98 > 3.01
5. Conclusion: Since the F-computed value of 7.98 is greater than the F-critical value of
3.01 at 0.05 level of significance with 3 and 24 degrees of freedom, the null hypothesis is
rejected in favor of the alternative hypothesis which means that there is significant
difference in the average sales of the four brands of detergent.

If there is significant difference, we have to determine where the difference lie. The test to be
used is the Scheffe’s Test

2
' ( x1−x 2 )
F= 2
SW (n1 +n2 )
n 1 n2
Where:
F’ = Scheffe’s test
x1 = mean of group 1
x2 = mean of group 2
n1 = number of sample in group 1
n2 = number of sample in group 2
SW2 = within mean squares

A vs B
2
(5 ,28−8.14)
'
F= =9.51
3.02(7+ 7)
(7 )(7)
A vs C
2
(5 ,28−3.71)
'
F= =2.87
3.02(7+7)
(7)(7)
A vs D
 2
' (5 ,28−8.14)
F= =9.51
3.02(7+ 7)
(7 )(7)
B vs C
2
' (8.14−3.71)
F= =22.82
3.02(7+ 7)
( 7)(7)
B vs D
2
(8.14−5.14)
F '= =10.46
3.02(7 +7)
(7)(7)
C vs D
2
' (3.71−5.14)
F= =2.38
3.02(7+7 )
(7)(7)
Between ( F 0.05 ¿ (k−1)
Brand F’ (3.01)(3) Interpretation
A vs. B 9.51 9.03 Significant
A vs. C 2.87 9.03 Not Significant
A vs. D 0.03 9.03 Not Significant
B vs. C 22.82 9.03 Significant
B vs. D 10.46 9.03 Not Significant
C vs. D 2.38 9.03 Significant

Interpretation of Results
The above table shows that there is a significant difference in the sales between brand A and B,
brand B and C and also brand B and D. However, brands A and C, A and D and C and D do not
differ significantly in their average sales. This implies that brand B is more saleable than brands
A, C and D.

Problems encountered by teachers in teaching reading comprehension

Problems Problem with Problem with

with Parents Teachers Pupils Rank1 Rank2 Rank3
2.80 2.80 3.00 13.5 13.5 16.5
2.20 1.40 2.60 3 1 9.5
4.20 3.20 3.00 24 18.5 16.5
2.40 4.80 2.60 6 27.5 9.5
2.60 2.20 2.80 9.5 3 13.5
4.00 4.80 5.00 21 27.5 29.5
2.20 3.20 2.40 3 18.5 6
2.60 4.20 2.80 9.5 24 13.5
 2.40 4.00 4.20 6 21 24
4.60 5.00 4.00 26 29.5 21

Sum 121.5 184 159.5

H= ( 12
Σ
N (N +1) n
R2
)−3(N +1)

[ ( )]
2 2 2
12 121.5 184 159.5
H= + + −3(30+ 1)
30 (30+1) 10 10 10

H= ( )
12
930
( 1476.225+3385.6+ 2544.025 )−(93)
H=2.5594

Using the chi-square table with 2-degree of freedom at 0.05 level of significance, the critical
value is 5.99. Since the computed value is less than the tabular value, the null hypothesis is
accepted which means that there is no significant difference in the problems encountered
teachers in teaching reading comprehension.
The 5-step solution
1. HO: There is no significant difference in the problems encountered by teachers in
teaching reading comprehension
2. H1: There is no significant difference in the problems encountered by teachers in
teaching reading comprehension
α = 0.05; df = 2; Critical value = 5.991
3. Decision Rule: Reject Ho if
│2.559│ ≥ │5.991│
4. Decision: Do not reject Ho since 2.559 < 5.991
5. Conclusion: Since the F-computed value of 2.551 is less than the chi-square critical value
of 5.991 at 0.05 level of significance with 3 and 24 degrees of freedom, the null
hypothesis is accepted which means that there is no significant difference in the
problems encountered by teachers in teaching reading comprehension.