For un linear mutivariate model as :

Y ~ OM + SOCA + STIR + Uncov

built on 120 observation

when I compute

predict(Mod,data.frame(OM=15, SOCA =10, STIR=130, Uncov =3), interval =”prediction” )
I get the prediction for Y and its prediction interval for the values of each predictiv variable, but when I compute
Predict(ModArticle, DataGLM, interval =”prediction”)
I get 120 values of Y and their prediciton interval. My question is : are they calculated for the observed levels of OM, SOCA, STIR and Uncov, of for randomly choosen levels ? In this last case, how to know these levels?