Professional Documents
Culture Documents
Recovered PowerPoint 12
Recovered PowerPoint 12
Harmaya University
March, 2019
Multivariate Factorial Design
Number of DV’s = q
Number of factors (IV’s) = 2, plus their
interaction: A, B, AxB
T=B+W
Bfactor 1 Binteraction
Bfactor 2
Three different hypotheses to test.
Significance Tests
Use Wilk’s to test each main factor and
the interaction:
W
h
Bh W
h denotes the particular null hypothesis
being tested.
If effects are significant?
Follow-ups should be used to try and
understand the source of differences
Especially if there are interactions!
Look at the DVs individually, their
correlations, and use stepdown tests
Consider discriminant analysis if the
interaction is significant (may also be
helpful if only main effects sig).
Example (Tatsuoka, 1988)
Speed and accuracy scores from 60 subjects learning
typing
Factor A: method (J=2)
Traditional versus experimental method
Factor B: condition (K=3)
Three levels, distributed towards massed practice (1,2,3)
Statistical Approach
Data cleaning, checking distributions, correlations and
descriptive statistics (r=.749 for the two DVs)
Plots of means
HOC
Omnibus and Follow-ups
Stepdown and Discriminant analysis
Speed (words per minute)
45
40
35
30
25
traditional
experim ental
20
15
10
0
1 2 3
condition
Accuracy
30
25
20
traditional
15
experim ental
10
0
1 2 3
condition
Research Questions
Main effect for Method:
Averaging across conditions, is there a difference in
the two methods (on the composite score created for
the DV’s?)
Main effect for Condition:
Averaging across the two methods, is there a
difference in the three conditions (on the composite
scores created for the DV’s?)
Interaction effect:
Does the effect of condition differ as a function of
method (and vice versa)?
HOC?
Results of a Box-M test indicated that
homogeneity of covariance for the 2x3=6
cells (groups) was safely assumed, Box-
M=23.41, F(15, 15949)=1.413, p=.131.
Omnibus tests?
Manova brief summary
Effect p partial 2
MxC .63609 .000 .202
Cond. .13415 .000 .634
Method .90771 .077 .092
Findings…
There is a significant interaction between method
of learning and practice time, Wilk’s =.63609,
F(4, 106)=6.73, p=.000, partial 2 = .202.
Since the interaction is significant, we go on to
try and fully interpret the interaction. This could
be through univariate tests focusing on
interaction; stepdown tests for the interaction; and
interaction contrasts.
In many situations with factorial interaction,
better to proceed to discriminant analysis.
If there were no interaction…
If there were no interaction, we would report:
Averaging across methods, there is evidence for a differential
learning effect across the three conditions, Wilks =.13415,
F(4,106)=45.85, p=.000, partial 2 = .634.
Averaging across conditions, there is no statistical evidence for
a difference between experimental versus traditional methods,
Wilks =.13415, F(2,53)=2.69, p=.077, partial 2 = .092.
Note: These interpretations make no sense for our
problem because of the significant interaction!
For follow-ups (if no interaction), we would consider the
pattern of means, the univariate and step-down tests, and
perhaps the contrast results.
Stepdown Approach
Since the DV’s are correlated, we’ll use the Roy-Bargman stepdown
F-tests to try to understand the interaction effect. We’ll assume
speed is more important (given automatic spell checks). The order of
importance is the order specified on the MANOVA statement.
First analysis is from a univariate analysis of variance just on speed.
Notice result is the same as in the univariate section just above it.
There are statistically significant interaction effects between method
and condition on speed, F(2,54)=6.24, p=.004, 2=.188.
Second analysis is from a univariate analysis of covariance, with
DV=accuracy, COV=speed.
The effect of speed is removed from the test on interaction differences
in accuracy across combinations of method and condition.
After adjusting for the effects of speed, a statistically significant
interaction effect on accuracy remains across the combinations of
method and practice condition, stepdown F(2,53)=7.34, p=.002,
2=.270.
Effect size in the Stepdown tests…
For first variable, use 2 from the univariate
analysis.
2 = SSB/SST (where SST=SSB+SSW)
For second variable (effect adjusted for the first
variable:
Compute SSB=Hypoth MS * Hypoth DF
Compute SSW=Error MS * Error DF
Compute SST = SSB + SSW
2 = SSB/SST
Contrasts…
Orthogonal Contrasts included for this
example, but could use any kind.
See written handout for calculation and
interpretation.