MANOVA Flashcards
(11 cards)
What is a MANOVA
A one way (between subjects) ANOVA, but with more DVs
What will MANOVA tell you
There is an overall difference between the groups across all the DVs you are looking at
MANOVA post-hoc test workflow
Is the MANOVA test significant?
Yes –> Are the individual ANOVAs significant?
Yes –> Does the IV have more than 2 levels?
Yes –> Post-hoc test
Why would you do a MANOVA instead of individual ANOVAs
- Situations where part of your research question is about whether your IV has an overall effect on all of the DVs
e.g. does pet ownership effect mental health? - usually only happens when individual DVs are a smaller part of a larger construct - Controlling error rates by avoiding multiple testing
but don’t correct for multiple comparisons in MANVOVA
Pillai’s trace
A type of multivariate test
conservative - lower power, but robust against violations of assumptions
Assumptions of MANOVA
DV should be at scale level
Data should be normally distributed
Equal variances - levenes test has to be done for all the DVs
- assumption is violated if any of the tests come back as significant
Equality of covariance matrices
Equality of covariance matrices
Correlation between the difference DVs should be the same across the groups of the between subjects IV
Box’s tests
Used to test equality of covariance matrices
But its very sensitive to potential violations
Because of this, people often do the significance test using an lower alpha level of 0.01 instead of 0.05
Use a lower p value to reject the null hypothesis and say the assumption is violated
not significant = assumption is met - there is equality of covariance matrices which is good
When is effect size reported
Not for the whole MANOVA
Is reported for the individual ANOVAs
How is MANOVA written up
There was a statistically significant overall effect of … on the dependent variables (Pillai’s Trace = , F(df) = , p=)
How are the individual ANOVAs written up
f (df) = , p = , np2=
post hoc
m = , sd= , p=
