The 3 IVs would be measured on a Likert Scale and the 2 DVs would also be measured on a Likert Scale. So they could either be continuous or categorical variables.
The hypothesis is that an increase in the levels of 3 IVs and their interaction would lead to higher levels of the 2 DVs.
I also want to compare the difference in the levels of IV and DV between 7 groups(ethnicities).
Would a 3-way factorial MANOVA, F-tests and then planned comparisons be all I need? Do I need to treat the demographics of each group as nonparametric data?
Multivariate ordinal logistic regression because your two DVs are likert and thus ordinal. Two DVs mean multivariate. Google this for details if you aren't familiar with these methods. Best wishes, David Booth
If your data set is large, say at least 300, it might be very informative to run a canonical analysis between the IVs and the DVs. While the data points themselves are ordinal since you used Likert scales, it would not meet the assumptions from a strict standpoint, but due to the law of large numbers, it would provide you with some excellent information such if there is one or two dimensions to the relationship and the nature of the relationship(s). For example, which IVs are most closely related to which DVs. If these results were cross-validated using the method suggested by David Booth, I think you it would add much understanding the the nature of your relationships.
Good luck,
Jim McLean
- Badges
- Science topic
- Similar topics
- Papers