Hello R-Users, again me! I have a data.frame with 291 rows, 82 columns. Tha variables in the columns are factors, numerics and ordered factors. The response variable is a factor with two levels. I would like to find the best model by trying every possible variable combination using a logistic regression and comparing the AIC (still have to find out how to do this fast and easy). I am aware of the step function but would like to do it also without step. But unfortunatley I have allready a problem with the first model using all variables: df.glm<-glm(RespVar~.,family="binomial",data=df) I get back this: Fehler in `contrasts<-`(`*tmp*`, value = "contr.treatment") : Kontraste k?nnen nur auf Faktoren mit 2 oder mehr Stufen angewendet werden Error in `contrasts<-`(`*tmp*`, value = "contr.treatment") : contrasts can only be used for factors with two or more levels There are no variables with less than two levels in my data.frame but here are missing values (NA). I could imagine that some observations are deleted due to missingness and perhaps after that, some of the variables contain only one level. I would be happy if somebody could tell me if my speculation could be true and perhaps also how I could find out which variables I have to delete to avoid this problem. Many thanks in advance. B. ----- The art of living is more like wrestling than dancing. (Marcus Aurelius) -- View this message in context: http://www.nabble.com/glm-error%3A-factor-levels-tp18730487p18730487.html Sent from the R help mailing list archive at Nabble.com.