Do you mean each variable or each observation?
If it were variable ,then the estimated coefficient before this variable is what you want.
If it were observations, I remember there is a somehow distinance ( I have forgotten,need to find in documention) which you are needed.
I think the original question is extremely unclear. What does the original poster mean by "contribution". To me, this could mean many different things, and in the case of multiple correlated predictor variables, the idea of "contribution" is meaningless in my opinion.