THE importance of cognition and affect in the consideration ofcommunication issues is widely acknowledged; the measure-ment of these sets of constructs has been, to say the least, problematic. One technique that has enjoyed a recent history of wide use is multidimensional scaling (MDS) , a tool with the potential to measure cognition and affect with a single measurement procedure, and to do so in a relatively unobtrusive manner, by not specifying dimensions for respondents to use in making comparisons among concepts. 2

To use spatial models to represent both beliefs and attitudes, one must represent affect within a spatial model that is typically seen as a structure of cognitions. One approach is suggested by Osgood, Suci, and Tannenbaum (1957), who regard evaluation as one dimension of a threedimensional space whose other dimensions are activity and potency. Woelfel and his associates, however, reject this approach on the grounds that the three dimensions are not the only dimensions used by respondents when making evaluations, that they are not uncorrelated (orthogonal), and that they may not share a common origin (see Woelfel & Fink, 1980, p. 78 H.).3 Instead, Woelfel and his associates propose that an appropriate self-referent concept be included in the space as a maximum preference location.