On estimating the top quark mass from global analyses ofe + e − collision data