Voting in large elections appears to be both ethically motivated and influenced by strategic considerations. The predominant approach to capture this interplay models voters' reasoning according to a rule-utilitarian calculus that abstracts away from heterogeneity in the intensity of support (Feddersen and Sandroni 2006, Coate and Conlin 2004). I show that, when such heterogeneity is considered, the approach delivers unsatisfactory results, since it implies that voters' preferences are irrelevant for participation behavior, contrary to the empirical evidence. A model of Kantian optimization à la Roemer (2015, 2019), based on the maximization of individual utility under a universalization principle, recovers the role of idiosyncratic policy preferences and can thus provide an important link between ethical participation and the spatial theory of voting.