College of Medicine and Health Sciences

School of Public Health

ANOVA (ANalysis Of VAriance)

By Shambel.W (wollo university) 1

Outlines
♣Definition and basic concept of ANOVA

♣How ANOVA will be computed/ computing F statistics/

♣Post hock test

“In God we trust. All others must have data.”

W. Edwards Deming

4/28/2025 By Shambel.W (wollo university) 2

 ANOVA (ANalysis Of VAriance)

♦ Analysis of variance (ANOVA) is a statistical procedure concerned with comparing

means of several samples.

♦ The purpose is to test for significant differences between class means, and this is
done by analysis the variances.

♦ The original ideas of analysis of variance were developed by the English statistician
Sir Ronald A. Fisher (1890-1962) in his book “Statistical Methods for Research
Workers” (1925).

♦ Much of the early work in this area dealt with agricultural experiments.

4/28/2025 By Shambel.W (wollo university) 3

 ANOVA (ANalysis Of VAriance)

♣A t-distribution can be used for testing hypotheses about differences of means

for two independent samples. However, the usual two-sample t-test cannot
be applied when more complex sets of data comprising more than two
groups are considered.

♣In this case, analysis of variance (ANOVA) is used to compare the means of
several groups.

4/28/2025 By Shambel.W (wollo university) 4

Study design
Study design
 Observational study design
 Experimental design

Sample size
 Applicable in small sample size

4/28/2025 By Shambel.W (wollo university) 5

ANOVA
♣Dependent variable : A single continues outcome variable.

♣Independent variable: Categorical variable that divide the outcome variable in to

more than two groups. Also known as the grouping variable, factor or
intervention.
♣This variable divides cases into two or more mutually exclusive levels, or groups
Example :
 Socio economic status (of the mother) (LSE, MSE, HSE) effect on birth weight.
 Fertilizer(Chemical, biological, no fertilizer) effect on yield.
 Smoking (no smoking, >10 cig per day and <10 cig per day ) on mean distance covered
in km.

4/28/2025 By Shambel.W (wollo university) 6

Types of ANOVA
Based on the Independent/categorical variable
A. One way ANOVA : It is used when you have one IV (categorical) with three
or more levels (groups) and one DV (continuous).
B. Two way ANOVA: When there are two factors classifying the observations we
need two way analysis of variance, and so on.
C. K way ANOVA: Having more than two factors.
D. Repeated measures of ANOVA : a comparing outcome variation across
measurements that measured at different time period.

4/28/2025 By Shambel.W (wollo university) 7

One way ANOVA…
 A one-way analysis of variance is a variation test in group means that used
when the data are divided into groups according to only one factor.

 One-way analysis of variance is based on assessing how much of the overall

variation in the data is attributable to differences between the group means, and
comparing this with the amount attributable to differences between individuals
in the same group. Hence, the name analysis of variance is used.

4/28/2025 By Shambel.W (wollo university) 8

Conceptualization of One-way analysis of variance

Population: mean difference

 How many samples you have taken: n = (n1+n2+n3 with overall/grand/ mean = µ
 Do you think that individuals in a group have identical value? What is possible cause?
 Do you think that means across group
4/28/2025
have equal value ? What is possible cause?
By Shambel.W (wollo university) 9
Example

4/28/2025 By Shambel.W (wollo university) 10

One-Way ANOVA : Partitions Total Variation

4/28/2025 By Shambel.W (wollo university) 11

ANOVA…
♣SST measures the total variation in the data set, the variation of all individual
data points from the grand mean.

♣SSB measures the explained variation, the variation of individual sample means
from the grand mean. It is that part of the variation that is possibly expected, or
explained, because the data points are drawn from different populations. It’s the
variation between groups of data points.

♣SSE measures unexplained variation, the variation within each group that cannot
be explained by possible differences between the groups.

4/28/2025 By Shambel.W (wollo university) 12

ANOVA…
Sample variance involves the division of the sum of squared deviations from the
sample mean by the number of degrees of freedom. This principle is applied as
well to find the mean squared deviations within the analysis of variance.

𝑺𝑺𝑩
♠ Mean square between (MSB):
𝑲−𝟏
♠ Mean square error (MSE): 𝑺𝑺𝑾/(𝒏 − 𝒌)
♠ Mean square total (MST): 𝑺𝑺𝑻/(𝒏 − 𝟏)

4/28/2025 By Shambel.W (wollo university) 13

Summary of ANOVA

4/28/2025 By Shambel.W (wollo university) 14

Fisher/F /-test hypotheses

 The null and alternative hypotheses of one-way ANOVA can be expressed as:

 Ho : µ1 = µ2 = µ3 = ... = µk ("all k population means are equal")

 Ha: At least one µ different ("at least one of the k population means is not
equal to the others")

 µi is the population mean of the i group (i = 1, 2, ...,

4/28/2025 By Shambel.W (wollo university) 15

Decision and interpretation
P>0.05 which non significant
- Accept null hypothesis ( there is no difference across the group means )
- The observed differences between the means of the groups are simply due to chance
variation.
- The variation between these group means will be about the same as the variation within
individuals of the same type.
- The ratio (MSB/MSW) will tend to be near to 1
- Consider trend test for continuous outcome variable (Reading assignment)
P<0.05 which significant, then reject HO and accept HA.
- There is real difference or the intervention have brought a significant difference in mean
of the groups.
- The between groups variation larger than with in .
- The ratio (MSB/MSW) will tend to be greater than 1 .
4/28/2025 By Shambel.W (wollo university) 16
ANOVA...
♣ The One-Way ANOVA is considered an omnibus (Latin for “all”) test because
the F-test indicates whether the model is significant overall i.e., whether or not
there are any significant differences in the means between any of the groups.
(Stated another way, this says that at least one of the means is different from the
others.)

♣ However, it does not indicate which mean is different. Determining which

specie pairs of means are significantly different requires post hoc (Latin for
“after this”) tests (also called “a posteriori,” “unplanned,” or “unfocused”).

4/28/2025 By Shambel.W (wollo university) 17

Pair-wise comparisons of group means
 This post hock test includes:
 Bonferroni
 Tukey test
 Scheffe..
 Sidak

What did you conclude ?

4/28/2025 By Shambel.W (wollo university) 18

 Assumptions

1. We assume independent sample or observation from each of the k populations which means
error term are independent, mean zero and normally distributed. (no subject in either group can
influence subjects in the other group or cannot be in two or more group at the same time).

2. We assume that the k populations under study:

o The outcome variable is normally distributed in each group (KS &Shapiro-Wilk test of normality,
Skewness and Kurtosis, Histograms or box W and Normal Q-Q Plots.). If violated consider KW test.

o Equal variances(Homogeneity of variances), si2.

4/28/2025 By Shambel.W (wollo university) 19

Homogeneity of variances
 Checking for homogeneity of variances in SPSS using Levene's test of
homogeneity of variances. (Ho: Homogeneity of variance and HA: not homo)

• If the f statistics p-value not significant (P>0.05), then the assumption not
violated. Do Tukey post-hoc test for pair wise comparison.

• If you data does violate this assumption(p<0.05), you should carry out an
alternative test statistics w/c is assumption free, Welch ANOVA or Browne-
Forsythe , as well as alternative post-hoc tests (i.e., a Games-Howell test
instead of a Tukey post-hoc test). Which adjust for heterogeneity or done by
considering the heterogeneity.

4/28/2025 By Shambel.W (wollo university) 20

Exercise using SPSS and Stata
Command for SPSS:
Analyze....Compare Means....One-way ANOVA....Dependent lists: effect
treat.... Factor: enter group....click OK

4/28/2025 By Shambel.W (wollo university) 21

Exercise using SPSS and Stata

4/28/2025 By Shambel.W (wollo university) 22

Result

What is your conclusion based on the above table ?

4/28/2025 By Shambel.W (wollo university) 23

Post hock test : Multiple comparisons

4/28/2025 By Shambel.W (wollo university) 24

Stata commands for one way Anova
 One way ANOVA Command: One way outcome variable factor, tabulate
Based on our example: one way weight SES, tabulate

 Paired wise comparison command: One way outcome variable factor,

tabulate specific paired wise test like Bonferroni, Tukeys test
Based on our example: one way weight SES, tabulate bonferroni

 Assumption on normality will be check via: swilk outcome variable

swilk weight

4/28/2025 By Shambel.W (wollo university) 25

What are the limitation of ANOVA
 Fail to handle confounders
 Doesn’t show effect size
 Fail to accommodate more than outcome variable
 Fail to handle intra-correlation in repeated measurement

4/28/2025 By Shambel.W (wollo university) 26

No Parametric test
Parametric test No Parametric test
1. Paired t –test 1. Wilcoxon Signed Rank Test
HO: there is no mean difference (µd =
0) HO: there is no median difference in ranks.
HA: Not HO HA: Not HO
2. Unpaired t- test 2. Mann-whitney test
HO: there is no variation b/n two group HO: there is no variation b/n two group
means (µd = 0)
medians
HA: Not Ho
HA: Not Ho
3. ANOVA test
HO: there is no variation b/n group of 3. Kruskal-Wallis Test
means HO: there is no variation b/n medians across
HA: Not Ho group
Decision: P-value and CI HA: Not Ho
Decision: P-value and CI
4/28/2025 By Shambel.W (wollo university) 27
Thank you…!!!