I am running Ordinal Logistic Regression on a data with Ordered Dependent Variables (1 to 5). In my multiple linear regression analysis, I have standardise my dataset with scale() before running lm command. Should I do so for Ordinal Logistic Regression or it is not relevant?


1 Answer 1


Standardization has no effect on the predicted values in multiple linear (logistic, ordinal) regression, because both the scale factors and the parameters are multiplicative constants that compensate each other.

There are however circumstances when standardization might make sense:

  1. Some (mostly older) text books on linear regression recommend variable standardization in order to compare effect sizes ("standardized" or "beta coefficients"). This is debatable, though, because it makes randomness assumptions about the predictors.
  2. In penalized regression (ridge, LASSO), variables must be standardized to avoid unfair shrinking. This is however automatically taken care of by software implementations of penalized regression like glmnet for R.

This site is temporarily in read-only mode and not accepting new answers.

Not the answer you're looking for? Browse other questions tagged .