I'm conducting a study on mental health terminology on social media for a group project. We have three interval DVs measured through likert scales and three IVs. The IVs are social media usage (categorical ordinal), age (interval) and gender (categorical nominal).
Currently, I am trying to run a power analysis but can't without first deciding the type of statistical test to run. I believe the appropriate test would be a multivariate multiple linear regression. Yet, others in my group believe a MANOVA is the right test for these variables.
As it is our first time conducting a study I was hoping someone more experienced might be able to offer their opinion and steer us in the right direction.
I'd recommend that you think about things like ordinal logistic regression if you have any ordinal DVs. BTW I MEAN multivariate ordinal. If no ordinal variables look at multinomial logistic regression. Best, David Booth
You could run a Manova or a repeated measures ANOVA, wich as Independent variables: MediaUsage, AgeCategories (max. about 6 categories) and Gender.and the 3 Likert scales as dependent variables. If you are interested in the differences between these scales, then run a repeated measures ANOVA, otherwise run a MANOVA. The advantage of a repeated measures ANOVA is that you can correct for Sphericity (homogeneity of covariance). If Mauchly's test for Sphericity is significant, you may want to 'go back' to a MANOVA.
Hello Abbie,
Without knowing your specific research questions, it's difficult to answer your query. There are many ways to look at how three IVs do or don't relate to scores on three DVs. What I can say is, if your DVs are three individual questions each having a Likert-type response scale, then you'd be best served looking at some method that treats the DVs as ordinal values (as Professor Booth suggests).
Good luck with your work.
- Badges
- Science topic
- Similar topics
- Biological Science
- Kinesiology
- Running