This study analyzes issues used repeatedly by the ANES in studying presidential races, 1972-88. Three conclusions stand out in the findings:(1) on issues offering concentrated benefits and diffuse or deferred costs, there is a correlation between issue position and issue salience; (2) those issues tend to be most salient to strong proponents, and of almost no salience to strong opponents; (3) consequently, the optimal candidate location on such issues is not at the mean preference of the electorate, but between that preference and the preference of the strong proponents of each issue. An important conclusion of much of the work on spatial modeling is that &dquo;in mass elections ... centrist or other outcomes reasonably reflective of voter opinion are to be expected&dquo; (Enelow and Hinich 1984b: 222). The dominance of centrist outcomes was first mentioned in the work of Hotelling (1929). That theme was continued in the pioneering work of Downs (1957), andDavis, Hinich, andOrdeshook (1970), where the emphasis was on the need for candidates to appeal to the &dquo;median voter.&dquo; With the development of more complex models dealing with multidimensional space and probabilistic voting, the dominance of centrist outcomes continues to be a major theme (Enelow and Hinich