Quantifying Political Relationships

Abstract

In this article, I introduce a method that uses large-scale event data and latent factor network models to provide a new comparative measure of cooperation and conflict in public relationships among politicians, nonpartisan political actors, and societal actors. The approach has a number of advantages over existing techniques: It captures public relationships in a multitude of venues on a continuous basis, incorporates both partisan and nonpartisan actors, allows quantifying the relationship between any pair of actors, reflects that communication is not unidirectional but rather a back and forth, and can be applied to a large number of countries over time. I apply the method to 13 Western European countries from 2001 to 2014 and demonstrate that party relationships are determined by coalition status as well as policy differences. The measure is publicly available and can be incorporated into standard research designs.

Elite Interactions and Voters’ Perceptions of Parties’ Policy Positions

“…To measure interparty cooperation and conflict as reported in the media, we apply the approach introduced by Weschle (2018), which uses a combination of large‐scale, machine‐coded event data and latent factor network models to produce a dyadic variable that quantifies the public relationships between political parties (also see Weschle 2019). Here, we provide a nontechnical explanation of this variable's construction.…”

“…Weschle (2018) uses a latent factor network approach to transform these data into a measure that quantifies the public relationships among political parties as well as other societal actors. The central idea is to think about the events reported by the media as emerging from a network of relationships among political and societal actors.…”

“…While we believe our measure represents the most comprehensive media‐based estimate of interparty cooperation and conflict devised to date, we are under no illusions that it perfectly captures either the true nature of parties’ underlying interactions or citizens’ perceptions of these interactions as filtered through the media. As regards coverage, Weschle (2018) notes that the national media reports that are coded in the ICEWS data may focus disproportionately on national party leaders’ behavior, so coverage of local politicians and backbench members of parliament may be more limited. In other words, the event data filter the total set of interactions and report only a fraction of it.…”

“…Empirically, we analyze the degree of cooperation and conflict in public relationships among political parties from 13 Western European democracies between 2001 and 2014. Our measure is based on latent factor network models of machine‐coded news stories that report tens of thousands of interactions between elites from different political parties, along with politicians’ interactions with nonpartisan actors (see Weschle 2018). We show that the degree of interparty cooperation and conflict varies sharply across different pairs of parties, and that, although governing coalition partners on average have more cooperative relationships than other party pairs, there is surprising variation in the tenor of coalition partners’ relations.…”

PLOVER and POLECAT: A New Political Event Ontology and Dataset

“…Text-derived event data, or structured records of who-did-what-to-whom, are an important source of data for scholars of international relations and comparative politics (e.g., Goldstein and Freeman, 1990;Reuveny and Kang, 1996;Colaresi, 2004;Chiba and Gleditsch, 2017;Weschle, 2018;Blair and Sambanis, 2020;Kibris, 2021). However, existing machinecoded event datasets such as the Integrated Crisis Early Warning System (ICEWS; O'Brien, 2010; Boschee et al, 2015) and Phoenix Historical Dataset (Althaus et al, 2019) have several major limitations: they use a Conflict and Mediation Event Observations (CAMEO) coding ontology (Schrodt, Gerner and Yilmaz, 2009) that is overly complex and rigid in how it represents events, they rely on coding software that is opaque and difficult to update, they use dictionaries that have limited coverage and that go out of date, and they often lack sufficient validation against other event datasets.…”