Comparing Apples and Oranges?

Stream: Panel 79 - Comparative Politics: Theorising and Measuring Democracy 
Date: Wednesday, 28 September 2016
Time: 11.00 am – 12.30 pm


One of the core validity problems of data based on expert surveys is their cross-national comparability. This is particularly relevant as expert surveys are increasingly used at a worldwide scale, requiring concepts and data that can ‘travel’ (Sartori 1970). Particularly when complex concepts are involved, such as democracy, corruption and election integrity, expert evaluations may be contextual and contingent on the particular historical and cultural context. Yet, paradoxically, the capacity of experts to provide contextualized, evaluative judgments is also one of the key reasons for using them as a source of data (Schedler 2012). In this paper we explore this tension and evaluate empirically to what extent experts make context-dependent evaluations of election integrity. When experts judge elections to be free and fair, what aspects of the electoral process do they take into account? Do certain aspects weigh more strongly in experts’ overall judgments than others? Does this depend on the context in which evaluation takes place or is it rather shaped by expert characteristics? In this paper, we test empirically to what extent contextual factors such as closeness of the elections, post-conflict elections, and levels of democracy and development affect how experts weight specific aspects of election integrity in their overall judgment of election integrity, controlling for individual expert characteristics such as ideological position and cultural background. We use data from the Perceptions of Election Integrity (PEI) dataset –version 4.0- on over 160 elections in 140 countries from 2013 to 2015, evaluated by 2000 experts.


Ferran Martinez i Coma (Presenter), University of Sydney

Carolien van Ham (Presenter), UNSW