Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
OF
VARIANCE
(ANOVA)
Quantitative Techniquesin Decision Making
Fauni, Katrizia
Ferrer, Fatima
Lacap, Normand
Lusabia, John
Valeroso, Jasmin
ANALYSIS OF VARIANCE (ANOVA or F-TEST)
ANOVA was developed by R.A Fisher in 1923. The F-Test used in ANOVA is named after him. It is a
technique in statistics designed to test whether or not more than two samples are significantly different
from each other.
We learned in the previous topics that z-test and t-test are used to test non-significance of difference
between single pair of samples. The t-test is used to test non-significance of difference between samples
taking them one pair at a time so that the t-test is applied as many times as there are pairs among the
samples. While ANOVA takes all the samples all at the same time and the test is applied only once. It
minimizes time and effort expended when computing and testing more than two samples.
Suppose: samples A B C D have to be tested for non-significance of difference. If t-test is to be used, we
have to compute separately for the following pairs such as A-B, A-C, A-D, B-C, B-D, and C-D. There is a
possibility that none of the pairs are significantly different, so much time and effort would have been
spent uselessly.
Now, using ANOVA, all the four samples are taken without pairing. If, however, the conclusion arrived at is
that of significant difference, then t-test must be used to find which of the pairs are significantly different.
ANOVA data must be arranged into rows and columns where the row represent the items in a sample and
the columns represent the sample classification.
There are two main types of ANOVA:
1. One-Way ANOVA
2. Two-Way ANOVA
ONE-WAY ANOVA
One-way ANOVA is used when there is only onefactor being studied as the independent variable. Thus,
also called One-Factor ANOVA.
Factors are the independent variables in a study. Let’s say, the researcher wants to know the effect of
three different methods of teaching on the grades of the students. The independent variable here is the
“method of teaching”, and the dependent variables are the grades of students and since there are 3
methods of teaching being compared, the factor has 3 levels or groups.
The null hypothesis (H0) is the equality in all population means, while alternative hypothesis (H 1) will be
the difference in at least one mean.
One-way ANOVA is based on the following assumptions:
Normal distribution of the population from which the samples are drawn.
Measurement of the dependent variable is at interval or ratio level.
Two or more than two categorical independent groups in an independent variable.
Independence of samples
Homogeneity of the variance of the population.
Page 1 of 20
ONE-WAY ANOVA TABLE
Source of Sum of Squares Degrees of Mean Sum of F Value
Variation (SS) Freedom (df) Squares (MSS)
2
Within-column ∑ x Ai k–1 SS wit
(wit) - ¿¿
n df wit
Where:
x ij = value of individual observation in the i th column and j th row
N = total number of entries
k = no. of levels or groups
n = no. of data values within each group
TWO-WAY ANOVA
Two-Way ANOVAis used when there are two factors being studied as the independent variable. Thus, also
called Two-Factor ANOVA.
Two-way ANOVA examines the effect of the two factors on the continuous dependent variable. It also
studies the inter-relationship between independent variables influencing the values of the dependent
variable, if any.
For example, the researcher wants to know whether there is an interaction between age and nationality
on life satisfaction. Here, there are two factors being studied,age and nationality.
Assumptions of two-way ANOVA:
Normal distribution of the population from which the samples are drawn.
Measurement of dependent variable at continuous level.
Two or more than two categorical independent groups in two factors.
Categorical independent groups should have the same size.
Independence of observations
Homogeneity of the variance of the population
Page 2 of 20
TWO-WAY ANOVA TABLE
Source of Sum of Squares (SS) Degrees of Mean F Value
Variation Freedom Sum of
(df) Squares
(MSS)
2
Factor A
∑ x Ai SS A MSS A
(A) - ¿¿ a–1
bn df A MSS wit
2
Factor B
∑ x Bj - SS B MSS B
(B) ¿¿ b–1
an df B MSS wit
AB 2
∑ ( x AiBj ¿ )2 ¿- ∑ x Ai - ∑ x Bj
2 (a – 1) (b – 1) SS AB MSS AB
(Interaction) −¿ ¿
n bn an df AB MSS wit
(AB)
Where:
a = no. of levels or groups in factor A
b = no. of levels or groups in factor B
Page 3 of 20
STEPS IN THE ANALYSIS OF VARIANCE
Step 1. State the null hypothesis
Step 5. Locate the tabular value of F by getting the intersection of d 1 (df of greater MSS) and d 2 (df of
smaller MSS)
Step 7. Compare the computed F value with the tabular F value then state the conclusion arrived at.
If the F Value < Tabular F value, the H o is accepted.
If the F Value > Tabular F value, the H o is rejected.
APPLICATION
ONE-WAY ANOVA
Suppose a researcher wants to test whether the grades of the students differ between the teaching
methods being used in a certain school:
1. lecture method
2. interactive method
3. content-focused method
The researcher took a sample of six (6) students for each of method. The grades of the students are taken
at the end of the semester and enumerated according to groups.
1,560 136,484
Total Sum of Squares ( SStot ) orTotal Variation – is the aggregate dispersion of the individual data values
across the factor levels
SStot = ∑ x ij2 - ¿ ¿
∑ x ij= 1,560
∑ x ij2 = 136,484
= 136,484 - ¿ ¿
2,433,600
= 136,484 -
18
= 136,484 - 135,200
SStot = 1,284
Page 5 of 20
Between-Column Sum of Squares ( SSbet ) or Between-Sample Variation – is
thedispersion among the factor sample means
2
SSbet = ∑ x Ai - ¿¿
n
(285,156+216,255+314,721)
= - 135 , 200
6
= 136,022−135,200
SSbet = 822
Within-Column Sum of Squares ( SSbet ) or Within-Sample Variation – is thedispersion that existsamong the
data values factor sample means within a particular factor level
= 1,284 – 822
SSwit = 462
df tot = N – 1
= 18 – 1
df tot = 17
df bet = k – 1
k = no. of levels or groups
=3–1
df bet = 2
= 17 – 2
Page 6 of 20
df wit = 15
SS bet
MSS b =
df bet
822
=
2
MSS bet = 411
SS wit
MSS wit =
df wit
462
= 15
13.34
Within-column 462 15 30.8
Total 1,284 17
Step 5. Locate the tabular value of F by getting the intersection of d 1 (df of greater MSS) and d 2 (df of
smaller MSS)
d1 = 2
d 2 = 15
Tabular Value
@ 5% significance level = 3.68
Page 7 of 20
@ 1% significance level = 6.36
Step 6. Calculate the F value
MSS bet
F=
MSS wit
411
= 30.8
F = 13.34
Step 7. State the conclusion
Note:
Tabular values:
@ 5% significance level: 3.68
@ 1% significance level: 6.36
TWO-WAY ANOVA
A study examining the differences in life satisfaction between young adult, middle adult, and older adult
American, British, Mexican and German. Each individual who participated in the study completed a life
satisfaction questionnaire. A high score on the test indicates a higher level of life satisfaction. Test scores
are recorded below.
Young Middle Older Adult
Adult Adult
7 10 10
13 15 7
American 8 9 7
15 9 17
11 5 13
12 15 10
11 5 4
British 11 13 8
16 8 12
4 12 11
Mexican 12 12 6
13 9 7
16 6 9
Page 8 of 20
13 14 8
9 12 8
4 9 11
7 12 8
German 14 3 10
8 11 16
9 7 9
SS A =
∑ x Ai - ¿ ¿
bn
Page 9 of 20
( 213 )2+ ( 196 )2+ ( 191 )2 (600)2
= -
( 4 ) (5) (3)(4 )(5)
= 6,013.30 – 6,000
SS A = 13.30
SS =
∑ x Bi - - ¿ ¿
B
an
= 6,013.33 – 6,000
SSB = 13.33
2 2
( x ¿ )2
= ∑ AiBj ¿- ∑ Ai - ∑ Bj
x x
SSAB +¿ ¿
n bn an
SSAB = 83.77
SSwit = ∑ x 2- ∑ ( x AiBj ¿ )2 ¿
n
= ( 7 )2 + ( 13 )2 + ( 8 )2 + ( 15 )2 + ( 11 )2 + ( 12 )2 +. .. ( 16 )2 + ( 9 )2 – 6,110.40
= 6.672 - 6,110.40
SSwit = 561.60
SStot = 672
df A = 2
df B = b – 1
b = no. of levels or groups
= 4– 1
df B = 3
df AB = (a – 1) (b – 1)
= (2 - 1) (3 - 1)
= (1) (2)
df AB = 2
df wit = ab ¿ )
= 3*4(5-1)
df wit = 48
df tot = N – 1
= 60 – 1
df tot = 59
SS A
MSS A =
df A
13.30
= 2
MSS A = 6.65
Page 11 of 20
SS B
MSS B =
df B
13.33
= 3
MSS B = 4.44
SS AB
MSS AB =
df AB
83.77
=
2
MSS AB = 41.89
SS wit
MSS wit =
df wit
561.60
= 48
MSS wit = 11.70
AB 83.77 2 41.89
Total 672 55
Page 12 of 20
Step 5. Locate the tabular value of F by getting the intersection of n1 (df of greater MSS) and n2 (df of
smaller MSS)
MSS A
FA =
MSS wit
6.65
= 11.70
F A = 0.57
MSS B
FB =
MSS wit
4.44
= 11.70
F B=¿ 0.38
MSS AB
F AB=
MSS wit
41.89
=
11.70
F AB = 3.58
Page 13 of 20
If the F Value > Tabular F value, the H o is rejected.
Since F values for Factors A and B are less than the tabular values such that:
Answer:
Page 14 of 20
Page 15 of 20
Page 16 of 20
Page 17 of 20
Page 18 of 20