... a project on Affective Polarisation among Political Elites by three Research Master's Students at the University of Amsterdam, as part of an internship under supervision of Assoc.-Prof. Dr. Gijs Schumacher. This project is concerned with the following question: How does affective polarisation manifest in political text, namely parliamentary speeches?.
(scroll down for extensive description)
Time Frame: September 2019 - January 2019
Final Outputs:
- Measurement tools for affective polarization at the elite level
- Application of measurement across time for one case
- Application for relevant research questions
- Translation of framework to different contexts
see Issues for details
- dictionary of addressees, such as party names, member names, constituencies (see here)
- construction of data frame of outgroup-references within parliamentary speeches (here)
- sentiment analysis
- construction of network of references (here)
Other than the classical concept of ideological polarization, affective polarization describes a polarization not in terms of policy positions, but as based on social identity and related to in- and out-group attitudes (Iyengar et al. 2019: 130). Despite the current trend of talking about ‘divided electorates’ and ‘polarized elites’ among pundits, the research on this concept - especially in Europe - is still in its infancy. Hence, we believe it to be a fruitful starting point for further research (for an overview: Iyengar et al. 2019).
We are interested in the study of affective polarization on the elite level. This in itself is already novel, as current measurements of affective polarization focus on the electorate rather than elites and use survey-based measures such as self-reported bias and implicit association tests. In this project, we will try to develop a measure of affective polarization using methods of automated content analysis. This will have a twofold output: on the one hand, we aim to develop a scalable tool to identify affective references to outgroups in political texts. On the other hand, we will apply this tool to a dataset of Irish parliamentary speeches, which can then be rated according to their affective polarization. If time permits, we will make use of the expertise in our team to develop similar tools for German and Portuguese texts, and, if possible, generate data on affective polarization in different contexts.
Automated text analysis is exceptionally well suited to tackle affective polarization. The very definition of this type of polarization as affective and group related makes it prone to be a property of texts. Second, the concept has been studied on the voter level, but not on the elite or media level, which already makes our approach novel. Automated content analysis was specifically developed for the large-scale study of text and the development of scalable tools.
This internship is relevant both in terms of the output it will generate and the learning process it will engender. The contribution to the literature, as already detailed above, can be significant: affective polarisation is a relatively new topic in mainstream political science, with most of existing research focusing on survey data at the electorate level. This also tends to be done in a bi-partisan context (i.e. the U.S.), with many assumptions likely to change when in a multi-party context. By attempting to develop a measure of affective polarisation amongst political elites through text, we believe this project taps into an important and yet underdeveloped field of the literature.
Iyengar, Shanto et al. 2019. “The Origins and Consequences of Affective Polarization in the United States.” Annual Review of Political Science 22(1): 129–46.