Initially I was going to use a binary logistic regression. I cannot find a good reason to make the beneish M-score into a binary variable. I did this because I saw this widely used in literature. However the suggestion of multivariate regression seems more appropriate. I have the beneish M-Score, a red flag questionaire and a perception questionaire. The questionaire are Likert Scale. Suggestions on literature?
If your DV is an ordinal variable then you want ordinal logistic regression. Attached is an excellent set of notes including examples. Best, David Booth
For a ratio variable, I would consider multiple linear regression. See:
https://www.amazon.com/s?k=Kutner+Applied+linear+statistical+models&i=stripbooks&ref=nb_sb_noss_2
for details. Best, D. Booth
how about multivariate regression or is this the same as multiple linear regression ? different terms are used in literature so confusing
Yes, very confusing. To a statistician multiple linear regression means multiple IVs. Multivariate means multiple DVs (i.e. response variables). To get full details see the above reference. I assumed the ratio variable is THE ONLY response. If so try multiple linear regression. If you have more than 1 DV that is ratio google multivariate linear regression for those details. If this is still unclear, please ask me again. Best wishes, David Booth
If you have single dependent variable I suggest ordinal logit models.For multiple dependent variables case, the multivariate models. If you are dealing with constructs, sem models. If your objective is prediction, PLSsem and if your objective is explanation, then CBsem.
Just in case I still misunderstand you see the notes below. For multivariate linear models just Google that term. Best again, David Booth