‘Empirical Type’: Towards a Multi-Method Approach of Classification and Concept Building

Conventional classification approaches in the social sciences, typically rooted either in qualitative typologies or quantitative taxonomies, have struggled to generate concepts and categories that are simultaneously empirically grounded and theoretically meaningful. This article proposes a novel approach, termed “empirical type,” which integrates quantitative clustering techniques and qualitative case studies. Leveraging the complementary strengths of both traditions, this method adopts a sequential and iterative procedure encompassing scope determination, data analysis, and typological conceptualization. Using an illustrative example focused on transitional economies, this approach yields results suitable for constructing categorical variables or sets, identifying ontological or causal dimensions, and developing a typological system of concepts. The paper also introduces cutting-edge machine learning tools for time-series clustering and dimensional reduction, which outperform conventional techniques in classification tasks. Consequently, the “empirical type” approach contributes to the expanding array of multimethod strategies that enable conventional typological tools to effectively address the complexities and opportunities presented by the era of big data and machine learning.