The current approach to model standardization - taking the std. deviation of all input data - does not necesarilly work in the case of factor interactions. For example, if a factor is being interacted with a numeric value (i.e., Years), you can get a different result as a product of the interaction (:) operator in the pre- and post-standardization cases. The root cause of this is currently not clear.
The current approach to model standardization - taking the std. deviation of all input data - does not necesarilly work in the case of factor interactions. For example, if a factor is being interacted with a numeric value (i.e., Years), you can get a different result as a product of the interaction (:) operator in the pre- and post-standardization cases. The root cause of this is currently not clear.