In order to compute error-components 3SLS estimates of this model, which requires different instruments for different equations, we propose a generalization of the Fuller-Battese transformation to ...

In the case where the time component is improperly omitted from the two-way model, we show that the difference between the true and estimated coefficient variance is of order greater than N-1 in ...

We conclude that, generally in epidemiology, clear differention between classical and Berkson error components is essential in the assessment of error sources and for establishing an error model ...