Non-linearity in OLS-Models, Multinominal Logit-Regression

#1
Hey there,

I have a question connected to the Ols-Model in case of nonlinearity between parameters. What should be done if the assummption of Linearity in OLS-Model ist not fulfilled and there is an non-linear relationship between the used parameters?

And my second question is if I can use a multinominal logit-regression in case of non-linearity between the parameters?

Thank you in advance,
L.
 
#3
Mr. Buis is surely correct. On top of that, MLR is appropriate when your DV is categorical (has nothing to do with linearity of the parameters).