On the validity of the effective field theory approach to SM precision tests