Choice of structural model via parsimony: A rationale based on precision.
It is shown that, in large samples, the more parsimonious of two competing nested models yields an estimator of the common parameters that has smaller sampling variance. The use...