Multinomial regression and Non-normality

[Gartland, Myles]

How robust are multinomial logistic regression estimates when the continuous variables are non-normal?

 

I could not find if the estimates are sensitive to this.