I spent a whole hour looking through my stat textbook (Field, 2018 and Pallant, 2020), google scholar and the search energy of my library trying a combination of terms until I stumble upon this video...thank you for this!
Hello, I'm trying to understand the best way to include an ordinal predictor in a multinomial logistic regression, do the approaches you discuss in this video apply to logistic regression? My variable is Education and has 5 levels (less than high school, completed high school, completed college/trade school, completed university, graduate school or higher).
I think the simple approaches (e.g. binary coding scheme) could be used for multinomial logistic regression, too. However, I am not an expert in logistic regression, so I am not sure.
Unfortunately not, I have done the staircase coding manually in Excel. The regression itself was a simple multiple regression in R: library(jtools) reg.fit
I spent a whole hour looking through my stat textbook (Field, 2018 and Pallant, 2020), google scholar and the search energy of my library trying a combination of terms until I stumble upon this video...thank you for this!
Hi! What are the drawbacks of using an ordinal IV as a continuous variable? Do you know any resources on this? Thank you.
For that you could look into the paper by Williams (2020), which you can find via:
www.regorz-statistik.de/en/regression_ordinal_predictor.html
Hi! I am currently working on a problem related to this. Did you find any info relevant to your question on this?
Hello, I'm trying to understand the best way to include an ordinal predictor in a multinomial logistic regression, do the approaches you discuss in this video apply to logistic regression? My variable is Education and has 5 levels (less than high school, completed high school, completed college/trade school, completed university, graduate school or higher).
I think the simple approaches (e.g. binary coding scheme) could be used for multinomial logistic regression, too. However, I am not an expert in logistic regression, so I am not sure.
@@RegorzStatistik ok thank you for your reply!
Hi, could you share the script for the OLS model specification used in the staircase coding approach please?
Unfortunately not, I have done the staircase coding manually in Excel. The regression itself was a simple multiple regression in R:
library(jtools)
reg.fit
Thnks, just wanted to be sure lm() was ok to use and no other terms had been included in the formula besides the covariates.
How would you interpret the Y intercept in these cases?
in general: Intercept = predicted value of the criterion variable if all predictors have a value of zero