Towards a real comparison of left-right indices

Abstract

In a recent article in Party Politics, Detlef Jahn proposed an alternative way of generating left–right scores using data from the Comparative Manifestos Project (CMP). Despite presenting innovative ideas, the article requires comment and correction. First, Jahn’s claim of a deductive approach is not convincing. He proceeds at least partly inductively. Furthermore, he ignores the theoretical ground of the competing approaches. Second, and more importantly, the central criterion by which to evaluate different approaches should be construct validity; however, Jahn does not conduct a full test of the construct validity. In this comment, I first explain the theoretical base of different approaches in generating left–right position scores using CMP. Second, I conduct a test of construct validity using not only expert data but also survey data. The results clearly suggest that approaches using a context-specific scheme for determining left–right issues are superior to the existing alternatives. This comment should be of interest to all scholars concerned with the question of how to create valid party position indicators.