ABSTRACT
In this chapter we test two techniques of textual analysis for measuring populism across large numbers of party systems. We first use holistic grading, applying the technique to electoral manifestos from 144 parties in 27 countries from Western Europe and the Americas; as a validity check, for about half the sample we also code campaign speeches by candidates or party leaders. Our results show that manifestos perform relatively well alone, and that populism is stronger in Latin America than in Europe; indeed, the level of populism among some European parties typically considered as examples of radical right populism might be overstated. We then contrast the results with automated textual analysis, applying supervised learning methods to generate a model that can correctly classify documents as populist or not. In this sample, the results are not fully satisfactory, performing well only if we restrict populist classification to a binary choice, but they indicate that with more data automated techniques could be a viable option.
