A test group of 78 students in advanced journalism courses at Ohio University were assigned the role of "editor" in an attempt to determine whether newspaper editors can predict the preferences of their readers. Seventy-eight other students from classes in mass communication, introductory psychology, and journalism were assigned the role of newspaper reader and were requested to rank their level of interest in a set of news stories taken from late December 1971 and January 1972 issues of the "New York Times,""Los Angeles Times,""Detroit News,""Washington Post,""Columbus Citizen-Journal," and "Athens Messenger." On the basis of these data, editors were asked to predict how the same audience would rank a second set of similar news stories. The newspaper readers also took a Rokeach personality test. From the resulting information about readers' values,the editors predicted the preferences of individual readers. The accuracy of the editors' predictions, along with the degree to which editors projected their own preferences onto the audience, was examined. Resulting data supported the hypothesis that making judgments and comparisons from one sample of stories to another aided prediction of readers' levels of interest in news stories. Surprisingly, informing editors about audience values and preferences did not reduce the projection of views onto audiences. (EE)