Hello,
I am wondering if it is possible and/or accurate to get negative values for least-squares means? I have a glm model with both x and y variables that are categorical (as.factor), and when I run the lsmeans function in R with the GLM, the resulting lsmeans are negative. Does this seem correct? I have checked for interaction among my predictor variables with no significant p-value, and there are no negative values in my dataset. I am unable to post data unfortunately as I am not allowed to share with outside sources. The general model setup is GLM<-as.factor(y)~as.factor(x1)+as.factor(x2), family="binomial". My y variable only has two levels being a categorical; 1 or 2. Is this an appropriate setup for my GLM model? And if so, why am I receiving negative results for my lsmeans. Any help is much appreciated.