Invited Commentary: Variable Selection versus Shrinkage in the Control of Multiple Confounders | Synapse