A further comment on the Winkler, Kanouse, and Ware method of controlling for acquiescent response bias