Regression on Survey Data

#1
Good morning.

I'm trying to use different variables to see what factors influence how someone has answered: "Increase", "Decrease" or "Stay the Same" for a question. I've been using ordinal logistic regression, but I'm wondering if I could (and if there's a benefit or drawback), recoding the dependant variable from factoral to numerical -- could I do this as a linear regression instead?

Thanks!
 

obh

Active Member
#2
Hi Kiki,

Do you mean instead of DV :"Increase", "Decrease" or "Stay the Same", to use a continuous value of change? like (-1.12, 2,-1.1, 4.34, 0 ,6?
In this case why not linear regression?