MANOVA Flashcards
When is MANOVA used?
When there are two or more DVs
What 3 questions can MANOVA help answer?
(1) Do changes in the independent variables (IVs) have significant effects on the DVs?
(2) What relationships exist among DVs?
(3) What relationships exists between IVs?
How does MANOVA answer these three questions?
MANOVA uses the covariance between DVs in testing significance of mean differences
Running a single MANOVA avoids inflated _____ __ error rates that would occur from running multiple ANOVAs
type 1
MANOVA tests the difference between underlying ____ _____ while ANOVA only tests differences among ____ ___ __ _______ ______
latent variables
groups on an observed variable
MANOVA is more useful to social scientists because most topics research latent constructs that are not directly observable
Warne, 2014
When running a MANOVA, the null hypothesis is that…
all independent groups are equal on all dependent variables
4 statistical tests that should be run in order to reject null hypothesis
Pilai’s Trace,
Wilk’s Lambda,
Hotelling’s Trace,
Roy’s Largest Root
Despite differences in their formulae, all 4 test the same null hypothesis and can be converted into…
an F-statistic which can then be used to calculate a p-value (Olson, 1976)
If there are a large amount of DVs, the 4 tests may be contradictory due to differences in their ____ ____
statistical power
Haase & Ellis, 1987
Roys largest root has the most power when…
DVs are highly correlated and represent a single construct.
Pilai’s Trace, Wilk’s Lambda and Hotelling’s Trace are more powerful for….
disparate outcome measures (Huberty & Oljenik, 2006)
If the data violates any assumptions of the MANOVA, ____ _____ should be used as it is highly robust
Pillai’s trace
Olson, 1976
If these tests are significant and null hypothesis is rejected, the researcher will still not know which particular DV is influenced most by a particular IV….
what two follow ups?
Separate ANOVAs and Discriminant Function analysis
Who says that following a sig. MANOVA, subsequent ANOVAs are protected against type 1 errors?
Bock (1975)