Political Frames and How to Find Them

• In Event: Preprocessing, Annotation, and Interpretation in Computational Social Science

Abstract

Framing is to highlight a certain aspect of an issue over the others, nudging readers or listeners towards a particular problem definition on the issue even without making a biased argument. Despite their theoretical and practical significance, scholars have struggled to measure frames empirically, largely due to the complexity of the ways in which they can be expressed by political actors. Traditionally the approach used to be a multiply-coded manual analysis using an apriori codebook. Today with the great abundance of digital communication data, the development of automated approaches is an important goal in the area of frame analysis. Topic modeling, supervised classification, and word embeddings are three general approaches that scholars used to measure framing from text. However, there is no standard method that is easy to use and not disputed in its capabilities to which extent the construct generated by these computational methods are actually frames, in the sense used by the traditional manual frame analysis. This is especially the case with unsupervised methods such as topic modeling.

The paper pursues two goals. First, it compares three different computational approaches currently used to measure frames (i.e. topic modeling, supervised classification, and word embeddings), and highlights their advantages and disadvantages. Second, it defines and provides a framework for future research on frame detection based on a combination of methods to automatically identify and measure relevant frames in different settings, reliably reducing the number of problems associated with current methods. In particular, the paper aims to resolve the argument that computational methods are not capable of grasping the complexity of frames as they often reduce them to a single indicator or topic at the moment.

Authors

• Meysam Alizadeh, University of Zurich

• Maël Dominique Kubli, University of Zurich

• Fabrizio Gilardi, University of Zurich