Hello,
I am working on one database (around 870 observations). I saw one continuous variable highly correlated with 8 continuous variables. I would like to examine the effect of that one continuous variable (1 IV) on the 8 continuous variables (8 DVs). Since the IV is not categorical. I cannot use MANOVA. Also, I don't know if I should conduct 8 different regression models for each pair of IV and DV? Since the story that I want to build in my study is 1 IV results (e.g. impact) on 8 DVs, I cannot build the reversed order. Could you give me any suggestions in this situation? Thank you very much!
Hello Anh,
1. Yes, you can run a manova. Enter the continuous IV as a covariate, then the DVs as normal. The test of the covariate (as an IV) is the one of interest. Note that the overall F test for intercept will be uninformative (it only tells you whether the cetroid for your data set on the 8 DVs = 0).
However, you might prefer to:
2 Run an ordinary least squares regression using the 8 DVs as predictors and the lone IV as a DV. This would give you information as to how the "best" linear composite of DVs is identified to maximize the relationship with the IV. That sounds like the information you want.
Good luck with your work.
