Analysis of Variance

ANOVA, also called as F-test

F-Test
a parametric test used to compare the means of two or more groups of independent samples

also known as the Analysis of Variance (ANOVA)

three (3) kinds of analysis of variance:

one way analysis of variance - only 1 variable involved

two way analysis of variance - 2 variables involved, the column and the row variables; used to know if there are significant differences
between and among columns and rows

three way analysis of variance - 3 variables involved

💡 2 way analysis is already long and complex, so usually 3 way is not really used

Why do we use ‘F-Test’?

→ to find significant difference between and among the means of two or more independent groups
→ but if its two only, z-test and t-test can still be used because it is easier, of course, f-test can still be used

When do we use ‘F-Test’?

→ if there is normal distribution and when the level of measurement is expressed in interval or ratio date (numeral/quantity like t-test and z-test)

How do we use ‘F-Test’?

→ to get f-computed value, use
compute the following to construct the ANOVA table
the formula
1. TSS - the total sum of squares minus CF [Correction Factor]
(GT )2
CF = ​
2. BSS - the between sum of squares minus the CF
N
3. WSS - within sum of squares or it is the difference between TSS minus the BSS
GT = grand total / sum of
everything
N = total number of
observations

ANOVA Table

Sources of Variation SS [sum of squares] df [degrees of freedom] MS [mean squared] F-Computed F-Tabular

Between the Groups BSS K-1 BSS / df F = MSB / MSW *see the table @

Within the Groups WSS (N-1)-(K-1) WSS / df desired level of significance with

Total TSS N-1 df between and within groups*

if F-Computed Value > F-Tabular Value → disconfirm null hypothesis in favor of the research hypothesis, accept alternative
→ means there is a significant difference between and among the means of the different groups

Analysis of Variance 1
 One-Way ANOVA
sample problem—

A sari-sari store is selling 4 brands of shampoo. The owner is interested if there is a significant difference in the average sales of the four brands of
shampoo for one week. The following data are recorded:

[perform the analysis of variance and test the hypothesis at .05 level of significance that the average sales of the four brands of shampoo are equal]

Brand A Brand B Brand C Brand 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

Solving by the Stepwise Method:

1. Problem

→ is there a significant difference in the average sales of the four brands of shampoo?

2. Hypotheses
→ H0: There is no significant difference in the average sales of the four
brands of shampoo
(Σx1 + Σx2 + Σx3 + Σx4)2
→ H1: There is significant difference in the average sales of the four brands CF =
n1 + n2 + n3 + n4
​

of shampoo
TSS = Σx12 + Σx22 + Σx32 + Σx42 − CF
CF = 869.14 (Σx1)2 (Σx2)2 (Σx3)2 (Σx4)2
BSS = + ​ + ​ + −C​ ​

TSS = 144.86 n1 n2 n3 n4
BSS = 72.29 W SS = TSS − BSS
WSS = 72.57

Sources of
df SS MS F-Computed F-Tabular
Variation

(K-1) 3 72.28 24.09 7.98 3.01

(N-1)-(K-1) 24 72.58 3.02

Total 27 144.86

3. Decision Rule
→ if the F-computed value is greater than the F-tabular value, disconfirm H0

4. Conclusion
→ since F-computed > F-tabular (7.98 > 3.01) at .05 level of significance and 24 degrees of freedom, the null hypothesis is disconfirmed. This
means that there is significant difference in the average sales of the four brands of shampoo

Two-Way ANOVA
→ involves two variables, the column and the row variables
→ used to find out if there is an interaction effects between two variables

Analysis of Variance 2
 sample problem—
Forty-five language students were randomly assigned to one of three instructors and to one of the methods of teaching then achievement was
measured on a test administered at the end of the
term. Use the two-way ANOVA with interaction effect at 0.05 level of significance to test the following hypotheses:
I.

H0: There is no significant difference in the performance of the three groups of students under three different instructors

H1: There is a significant difference in the performance of the three groups of students under three different instructors

II.

H0: There is no significant difference in the performance of the three groups of students under three different methods of teaching

H1: There is a significant difference in the performance of the three groups of students under three different methods of teaching

III.

H0: Interaction effects are not present

H1: Interaction effects are present

A [teacher factor] B [teacher factor] C [teacher factor]

40 50 40
41 50 41
Method of Teaching 1 40 48 40
39 48 38
38 45 38

40 45 50
41 42 46
Method of Teaching 2 39 42 43
38 41 43
38 40 42

40 40 40
43 45 41
Method of Teaching 3 41 44 41
39 44 39
38 43 38

Solving by the Stepwise Method:

1. Problem
→ Is there a significant difference in the performance of students under the three different teachers?
→ Is there a significant difference in the performance of students under the three different methods of teaching?
→ Is there an interaction effect between teacher and method of teaching factors?

2. Hypotheses
→ [given]

level of significance - a=0.05

df total = N-1 → 45-1 = 44

df within = k(n-1) → 9(5-1) = 36

df column = c-1 → 3-1 = 2
df row = r-1 → 3-1 = 2
df c(r) = (c-1)(r-1) → 2(2) = 4

Analysis of Variance 3
 A [teacher factor] B [teacher factor] C [teacher factor] Grand Total (row)

Method of Teaching 1
198 249 197 636
(TOTAL)

Method of Teaching 2
196 210 224 630
(TOTAL)

Method of Teaching 3
201 216 199 616
(TOTAL)

Grand Total (column) 595 667 620 1882

CF = 78709.42
SStotal = 508.42
(GT )2
SSwithin = 129.2
CF = ​

N total
SScolumn = 178.18
SS total = Σx12 + Σx22 + .... Σxn2 − CF
SSrow = 14.05 (Σx1)2 (Σx2)2 (Σ
SS within = Σx12 + Σx22 + .... Σxn2 − [ + + .... ​ ​

SScr = 187.15 n1 n2
(GT column1)2 + (GT column2)2 + .... (GT columnn)2
SS column = −C ​

F-Value Computed N
(GT row1)2 + (GT row2)2 + .... (GT rown)2
SSrow = − CF
Columns = MSC /MSW → 89.09/3.59 = 24.82
​

N
Rows = MSR /MSW → 7.02 / 3.59 = 1.95 SS cr = SS total − SS within − SS column − SS row
Interaction = MSI /MSW → 46.79 / 3.59 = 13.03
F-Value Tabular

Columns df = dfC / dfW → 2 / 36 = 3.26

Rows df = dfR / dfW → 2 / 36 = 3.26

Interaction df = dfCR / dfW → 4 / 36 = 2.63

Sources of Variation SS df MS F-Computed F-Tabular Interpretation

Between Columns 178.18 2 89.09 24.82 3.26 S

Between Rows 14.05 2 7.02 1.95 3.26 NS

Interaction 187.15 4 46.79 13.03 2.63 S

Within 129.20 36 3.59

Total 508.58 44

3. Decision Rule

→ if the F-computed value is greater than the F-tabular/critical value, disconfirm H0

4. Conclusion

→ since F-computed (column) > F-tabular [24.82 > 3.26] at .05 level of significance with 2 and 36 degrees of freedom, the null hypothesis is
disconfirmed. This means that there is significant difference in the performance of three groups of students under three different instructors

→ since F-computed (row) < F-tabular [1.95 < 3.26] at .05 level of significance with 2 and 36 degrees of freedom, the null hypothesis is
confirmed. This means that there is no significant difference in the performance of the students under three different methods of teaching

→ since F-computed (interaction) > F-tabular [13.03 > 2.63] at .05 level of significance with 4 and 36 degrees of freedom, the null hypothesis
is disconfirmed. This means that an interaction effect is present between the instructors and their methods of teaching

5. Interpretation

Analysis of Variance 4
 → students under instructor B have better performance under methods of teaching 1 and 3 while students under instructor C have better
performance under method 2.

Analysis of Variance 5