We have several independent variables (some are continuous with more than 5 levels, some binary and some quasi-interval (5 levels - categorical). We also have 5 dependent variables that share a common construct. Is it useful to conduct MANOVA with all the continues/quasi-interval as covariates, and the binary as factor variables - or preform a 5 separate multiple regression analysis?
Thank you
In general, it is inadvisable to perform multiple (univariate) analyses when you can replace them with one (multivariate). Neglecting this can result in Type I errors when a suitable correction is not applied. It might also lead to overlooking an effect arising from interaction between variables analysed separately.
With this in mind, you would probably be better off performing a single MANOVA. Covariates, however, should be determined by your experimental design rather than by the type of a variable (nominal, categorical, and continuous).