Statistical issues in multisite effectiveness trials: the case of brief strategic family therapy for adolescent drug abuse treatment