This is a request I've had quite a few times in private conversations, and something I've had to do more than once in my own projects, so I thought it'd be good to "share the wisdom". <hansl> function list autodrop(list X, scalar criterion[1.0e-12]) list drop = null matrix R qrdecomp({X}, &R) scalar ndrop = 0 loop i=1..nelem(X) --quiet if criterion > abs(R[i,i]) drop += X[i] ndrop++ endif endloop if ndrop > 0 printf "%d collinear variables dropped (%s)\n", \ ndrop, varname(drop) return X - drop else return X endif end function </hansl> I'm just about to add the above trick to the "Cheat Sheet" chapter of the User's guide. ------------------------------------------------------- Riccardo (Jack) Lucchetti Dipartimento di Scienze Economiche e Sociali (DiSES) Università Politecnica delle Marche (formerly known as Università di Ancona) r.lucchetti(a)univpm.it http://www2.econ.univpm.it/servizi/hpp/lucchetti -------------------------------------------------------