I wanted to do an F-test to compare model L2 with L3 assuming heteroskedasticity. The variable KONK2 only has three levels and are for that reason replaced by dummyvariables KONKD1 and KONKD2 in L3. Hence making L4 equivalent to L2. L2<-lm(PRMRES ~ factor(NYPR) + KONK2 + OMS + factor(NYPR)*OMS ) L3<-lm(PRMRES ~ factor(NYPR)+ KONKD1+ KONKD2 + OMS + factor(NYPR)*OMS) L4<-lm(PRMRES ~ factor(NYPR)+ I(2*KONKD2+KONKD1) + OMS + factor(NYPR)*OMS) I tried Waldtest in lmtest, but no result due to non-nested models. I don't know whether linearHypothesis for car can be of any use restricting coefficients of KONKD1 and KONKD2 to be equal. So Im a little low on ideas Also being a novice I'm not quite sure of the setup. -- View this message in context: http://r.789695.n4.nabble.com/comparing-linear-models-tp4568825p4568825.html Sent from the R help mailing list archive at Nabble.com.