Week Ten - MANOVA Flashcards
What is a MANOVA?
When we have more than one DV (can be on different scales)
what does a MANOVA promote?
Lower type I error
When should we use a MANOVA?
If the goal of the design is to discover whether behaviour, as reflected by the DVs, is changed by manipulation of the IV
What does MANOVA test for and what does this result in?
Differences on a linear combination of the DVs.
Results in potentially being more powerful/sensitive in identifying differences between groups - because it is examining differences along a combination of variables rather than just examining differences on a single variable
For MANOVA, the DVs must be? (2)
Conceptually related (theoretical reasons)
Moderately correlated (use ANOVA if not) - cannot be too highly correlated either (results in redundancy of power)
Too highly correlated DVs may be due to what? (2)
Multicollinearity: Very high correlations between variables
Singularity: One variable is a combination of a number of others in the analysis
A correlation score of .85 or above means what?
That the DVs are just measuring the same thing and we should just look at one or the other based on the theory
What are the assumptions of MANOVA?
INDEPENDENCE OF OBSERVATIONS: random allocation
MULTIVARIATE NORMAL DISTRIBUTION
HOMO OF VARIANCE (OF THE VARIANCE-COVARIANCE MATRICES)
What is HOV(VC) in MANOVA?
We need HOV for each group AND the correlation between DVs needs to be the same in each group (assumes that the data in each cell comes from a single population)
How do we check for HOV(CV)?
Levenes’s test
Box’s test: sig means violated
When should we pay attention to Box’s test?
When we have unequal sample sizes as it is very sensitive to violations of multivariate normality
MANOVA is reasonably robust to violations of HOV(CV) if?
There are equal numbers of participants per cell
What multivariate test of significance does MANOVA produce?
Wilks Lambda
Pillai’s Trace
When do we use Wilks Lambda?
Always unless there are some possible assumption violations
When do we use Pillai’s Trace?
When the design is flawed/violated - eg small sample size, unequal numbers