Mapping discourses through Hierarchical Clustering on Principal Components

Discourses are essential elements in the human organization as they ascribe action and establish the content of relationships between actors. Discourses take different perspectives, and only in-depth analytical approaches can grasp the essential content of such discourses. This article introduces one of such approaches, the hierarchical clustering on principal component (HCPC), a straightforward technique to map typologies of discourses considering surveys’ data. HCPC presents at least three important benefits in discourses’ mapping. First, it allows significantly different responses to being isolated, which can be employed to identify the peculiarities of each discourse. Second, it systematically anchors typologies of discourses to survey participants which ultimately allows for exploring the perspective of each participant who represents different stances of a specific issue. Third, it is particularly flexible towards subgroups of questions that refer to the same dimension construct and data type as it can be employed considering either nominal, ordinal or continuous dataset. HCPC technique is presented here considering an empirical example based on a deliberative mini public experiment on climate change policy, an assembly of randomly selected participants which employ a talk-centric mode of deliberation and decision-making. The example shows HCPC potential in defining discourses and tracking their evolution during deliberative processes.