The causal approach to measurement error in panel analysis : some further contingencies. [TR 48]