In an extensive series of articles, Rabinowitz, Macdonald, and Listhaug have launched a new, directional theory of electoral choice. According to their claims, the new theory is superior to the classical Downsian proximity model. Such a conclusion, however, is not tenable. First, it fails to take proper account of the lower falsifiability of the directional model. Second, it rests on empirical analyses that do not test the two theories on the predictions they actually make. Both are theories of individual choice, which implies intrapersonal comparisons of utility. Prior tests, however, have predominantly taken the form of interpersonal comparisons. A reanalysis based on the proper type of comparison yields an outcome diametrically opposed to that previously obtained. It also reveals an eccentricity effect that largely accounts for the apparent success of the directional model.