MANOVA Flashcards
What is a multivariate test?
A test involving multiple DVs.
In which 3 scenarios might be MANOVA be most useful?
If there are multiple variables for one construct, in clinical research (e.g. an intervention) and to help us understand a complex pattern of results (e.g. one DV affected but not other).
What are the two requirements of DVs?
Chosen on an empirical or theoretical basis and related but not linearly dependent.
What is the main purpose of MANOVA?
To examine mean differences on linear combinations of quantitative variables and explain this variability.
What are the assumptions of MANOVA?
DVs are continuous, IV is categorical, normality, independence of observations, no multicollinearity, homogeneity of error variance and homogeneity of variance-covariance.
What is multicollinearity?
When the DVs correlate highly and should really just be combined and used in univariate analysis.
When can you ignore homogeneity of variance-covariance?
When group sizes are equal and the sample of large.
What is the most commonly reported MANOVA result?
Wilks’ Lambda.
When might you report Pillai’s Trace?
When the sample is small and/or Box’s test is signficant.
When might the F tests be identical for all multivariate tests?
When the IV only has 2 levels.
How do we correct for multiple comparisons in the between-subjects effects?
Bonferroni correction: 0.05 / No. of comparisons.
What are the two main purposes on discriminant function analysis?
To predict group membership from a set of variables and to help us understand how the DVs were combined to give us a significant MANOVA.
What is a discriminant function?
A new variable based on a combination of the DVs that is weighted to discriminate best between the levels of the IV.
What does DFA do with the discriminant function?
Classifies participants based on their scores on it and then compares this to their real group membership.
What is the Eigenvalue?
The amount of variance accounted for by the discriminant function.
