Analysis of Variance

ANOVA for One-Way Classification

• To test the equality of several means,

researchers utilize a procedure known as
the analysis of variance. One-factor
analysis of variance is a procedure that
uses a set of calculations on several
variances to test the hypothesis that
several populations have the same mean.
ANOVA for One-Way Classification

• The application of ANOVA requires three basic assumptions:

1. The various groups of measurements are assumed to be
obtained from a population with normal distribution.
2. Each group is randomly selected and independent of all
other groups.
3. The variables from each group come from distributions
with approximately the same standard deviation.
ANOVA for One-Way Classification

The solution and formulas for the one-way analysis

of variance are as follows:
Step 1: Determine the Null and Alternative
Hypotheses
Step 2: Find (Sum of squares total)
Step 3: Find (Sum of squares between)
Step 4: Find (Sum of squares within)
Step 5: Find Variance Estimates (Mean Squares)
ANOVA for One-Way Classification

Degrees of freedom:
1. and
2. K is the number of groups and N is the
total sample size
ANOVA for One-Way Classification

Step 6: Find the F-ratio and complete the

ANOVA test.
 Example

• Three brands of milk were tried and

compared on a sample of three groups of 9
children to find out if they increase the
weight of the subjects. The data is
reflected in the following table in terms of
weight gain in pounds.
 Weight Gains Among Children Using Three Milk Brands – A (in Pounds)

Respondent Brand A Brand B Brand C

No.
1 4.4 3.1 2.9 Determine if
2 4.0 2.9 2.7 there is no
significant
3 3.5 3.7 3.1
difference in
4 5.2 3.8 3.5 the average
5 4.7 4.1 3.4 weight gain (in
6 2.6 3.0 3.4 lbs) among the
three groups of
7 4.2 3.9 2.8
children using
8 3.7 3.2 3.5 the 3 brands of
9 3.5 3.0 3.3 milk at 0.05
 Solution

Step 1: There is no significant difference in the

average weight gains (in lbs) among the
three
groups of children using the 3 brands of
milk.
There is significant difference in the mean
weight gains among the three groups of
subjects.
Respondent
No.
A B C
Computation:
1 4.4 Comparative
3.1 2.9 Weight
19.36 Gains
9.61 8.41
Among Children Using Three Milk Brands
2 4.0 2.9 2.7 19.0 8.41 7.29
3 3.5 3.7 3.1 12.25 13.69 9.61
4 5.2 3.8 3.5 27.04 14.44 12.25
5 4.7 4.1 3.4 22.09 16.81 11.56
6 2.6 3.0 3.4 6.76 9.0 11.56
7 4.2 3.9 2.8 17.64 15.21 7.84
8 3.7 3.2 3.5 13.69 10.24 12.25
9 3.5 3.0 3.3 12.25 9.0 10.89
Cont’n…
Computation of the different sums of
squares

Step 2:

Step 3:
Computation of the different sums of
squares

Step 4:

Step 5:
 CRITICAL VALUE: At 1% level of
significance, with 2 and 24 df, the
critical value is 5.61.
Summary table for the One-Factor ANOVA of the Mean-
Weight Gain of the Three Sample Groups
Source Sum of Degre Mean Compu Tabula Decisi Interpr
s Square es of Square ted F rF on etation
Variati s Freedo s (0.01)
on m
Betwee 3.04 2 1.52
n Not
Column 5.10 5.61 Accept signific
Within 7.15 24 0.298 ant
Column
 Conclusion

Since the computed F-value of 5.10 is

less than the tabular value of 5.61, the
null hypothesis is accepted. This implies
that the average increase in the weights
of the subjects using the three brands of
weight-enhancing vitamins is just the
same at 1 percent level of significance.
 ANOVA for Two-Way Classification

• The two-way analysis of variance incorporates two

independent variable in the same analysis, thus allowing
a more powerful test of differences among means.
• In this two-independent variable research design, one of
the independent variables is set up in columns and the
other in rows to form a block of cells.
• A cell is a combination of treatment conditions that is
unique to one group of subjects, or where a row
intersects a column.
 Steps in solving for the two-way ANOVA

Step 1: Determine the Null and Alternative

hypotheses.
Step 2: Find the total sum of squares
Step 3: Find the columns sum of squares
Step 4: Find the rows sum of squares
Step 5: Find the interaction sum of squares
 Steps in solving for the two-way ANOVA

Step 6: Find the within sum of squares

Step 7: Compute for the degrees of freedom
Degrees of Freedom
Rows:
Columns:
Interaction: (
Within:
Total:
 Steps in solving for the two-way ANOVA

Step 8: Find the Variance estimate (mean

squares)
Step 9: Find the F-ratio and complete the
two-way ANOVA test
 Example
A researcher wants to determine the effect of a new teaching methodology for
elementary mathematics. She believes that the proficiency of the grade four pupils under
the new methodology may be related to the pupils’ prior achievement in conventional
elementary mathematics and their tendency to persevere in past study patterns. A
research consultant designs a one-grading period investigation in which the dependent
variable is the achievement in elementary mathematics, as measured by a specially
designed achievement test.
The grade three mathematics level of achievement of the sample subjects is
designated as the first independent variable. Based on this prior level of achievement,
grade four pupils who will be using the new methodology are classified as high achievers,
average achievers and low achievers.
The second independent variable is designated as pupils’ tendency to persevere in
prior routines, as measured by a “rigidity” test given to all student subjects. According to
their test scores, the subjects were divided into three groups designated “high rigidity” ,
“average rigidity”, and “low rigidity”.
Using a two-way ANOVA, test if significant differences exist between the given
 Elementary Science 6 Achievement Test Score of 27
Pupils Classified According to Prior Elementary
Science Achievement and Performance on a Test of
Rigidity
Prior Elementary Science Achievement
Rigidity Low Achievers Average Achievers High Achievers Total
Tendency

High 40 1600 53 2809 61 3721

Tendency