Categories
Technology

[Q] Regression model comparison for ordinal data

Hey everyone! I’m currently working on my thesis and want to analyse whether certain health markers have additional value in predicting symptoms to just using an already existing marker . Basically: does variable A explain variance in addition to variable B of variable C?

I analyse two different markers in two different analyses. For one marker, all variables A, B, and C are 5-point Likert, for another marker all variables are on a 0-10 scale.

My supervisor told me I could use multiple linear regression analysis with B as a covariate if my variables are approximately normally distributed so I could simply see if A has additional value in the output. Unfortunately they are not, and even transforming didn’t really help, so now I’m considering ordinal regression analyses.

Now I’m struggling with a few problems:
1. Should I just go for linear regression anyway ?
2. If not, how can I find out whether or not A explains additional variance? I haven’t been able to figure out how to check for it in my R output, but I’ve stumbled upon the likelihood ratio test, could that be an option?

Sorry this is a long read and probably basic stuff, but I’m not super competent in statistics and especially R. I’d really appreciate any sort of feedback!