Politicians’ claims of representation and its relation to feeling represented

In recent decades, scholars of the so-called constructivist turn in representation have argued that we should pay attention to the claims of representation that politicians make about groups in society. Nevertheless, despite recent advances, empirical research on politicians’ claims of representation and how citizens react to those claims remains relatively scant. This article aims to make two main contributions to the literature. First, we focus on whom politicians claim to represent or not. While major advancements have been made in operationalizing and coding claims of representation for quantitative analysis, the available studies make use of labor-intensive manual coding. Thereby, most studies are rather narrow in scope, focusing on a short period in time, often limited to specific types of claims in very specific domains, such as gender quota. However, since the effects of claims of representation on citizens’ identity and their feeling of being represented are likely to become apparent mostly after years of exposure to claims, a large-scale longitudinal analysis of the various claims made by politicians, regardless of the issue, seems called for. By training state-of-the-art machine learning models to automatically classify claims of representation in political speech, this study offers a renewing analysis of Flemish politicians’ claim-making, on large scale, and in longitudinal fashion. Specifically, we trained models (based on manually coded data) to code all claims made by Belgian (Flemish) politicians on their personal Facebook pages and in the federal parliament (during the weekly question hour) from 2010 to 2022. Using a detailed coding scheme, we take into account not only which politician or party (the subject) is claimed (not) to represent which group (the object), but also the kind of representation that is being performed through claims-making (substantive vs. symbolic vs. descriptive), and whether some groups are explicitly defined as a non-deserving ‘out-group’ (e.g., ‘my opponent represents the interests of immigrants, while I do not!’). Second, we turn our attention to how these claims relate to citizens’ attitudes, and specifically, how well citizens of (non-)targeted groups feel represented. Drawing on cross-sectional survey data from 2021 among Flemish citizens, which included both relevant socio-demographic information that allow us to classify respondents into groups and information on the extent to which respondents feel represented, we try to disentangle the link between how often and in what way societal groups are claimed to be represented, and how well citizens belonging to these groups feel represented by their representatives.