Transcript one-way anova

ANALYSIS OF VARIANCE
By
ADETORO Gbemisola Wuraola
ANOVA is sub-divided into
 One- way ANOVA
 Two- way ANOVA
 “n” – way ANOVA
ONE-WAY ANALYSIS
It is used to determine possible effect of a single
non-metric independent variable (factor) on a
metric dependent variable.
In one-way ANOVA we seek to determine if three
or more categories of an independent variable is
significantly different in terms of average values of
a continuous dependent variable.
ONE-WAY ANOVA
 It compares three or more means
What you need:
 --One categorical variable with three or more
categoriesindependent
 --One continuous variable---
STEPS
 --Analyse
 --Compare means
--One-way ANOVA
--Move dependent variable into Dependent list box
--Move independent variable into Factor box
--Option (click descriptive & means plot)
---and make sure a dot is the Exclude cases analysis
by analysis box
--Continue
--OK
Interpretation
When the ANOVA value has a p-value less than or
equal to 0.05, it is said that the categories are
significantly different; otherwise it is not.
POST-HOC TEST
 Post hoc test becomes important when p-value
indicates that the categories are significantly
different. This test enables us to identify which of
the group (groups) are significantly different.
 This is indicated with placement of star (*) in front
of the categories.
Two-way ANOVA
 Two-way ANOVA is used when the goal is to test
the effect of two categorical independent variables
on a continuous dependent variable—this is the
test for “main effect”.
 It also test for “interaction effect”. E.g. Education
with 4 categories and Sex (male, female), and
income as the dependent variable.
 Is educational categories significantly different
among males or females.
Steps
Analyse
 Select general linear model
 Select univariate
 click on dependent variable & move it to
the dependent variable box
 click on the independent variable & move it
into fixed factor box

 Click on options , select descriptive, estimate of
effect size, homogeneity test.
 Click continue
 Click post-hoc
 Click on independent variables with at least
three (3) categories & move into the post-hoc box.
 Select the test for it – turkey.
 Click continue
 Ok
INTERPRETATION
Descriptive statistics

Check that the statistics are ok
Levene's Test of Equality of Error Variance Table

This test an underlying assumption: The sig. value must be greater
than 0.05 (or 0.01).
Main output—Test of Between-Subject effects Table


Interaction effects(two indep. Variables separated by *): if Sig.
less than 0.05 it means there is interaction effect—there is
significant difference in first independent variable GIVEN the
second one.
Main Effects—check each of the indep. Variables and their Sig.
value; if less than 0.05, that variable is having a main effect (i.e.
the categories are significantly different in terms of the dep.
Variable). Otherwise not sig. different.
 Post Hoc
If your main effects or interaction effects is
established through the Sig. value(s) for the
independent variables is significant.


Then under the table Multiple Comparisons
check the significant variable and see where you have *.
Where this appears indicates the categories are the ones
significantly different.