The literature on democratization uses measures of either ethnic fractionalization or polarization in empirical analyses on the causes of democratic regress; some authors have argued that either of the two complicates democratization. This article detects a conceptual puzzle in this use of the two concepts: when we shift the attention from fractionalization to polarization we are not simply moving along a continuum but rather making an epistemic leap from facts to normative problems. But to treat the relation between a descriptive account of a state of affairs and a normative status as a continuum is a fallacy that remains unaddressed in this literature. This article exposes the limits of analyses that remove normative considerations from the big picture of dynamics of democratization and that narrow their focus to case histories of democratic development. It pleas for a return to normative insight and interdisciplinary dialogue.