Digital Public-Administration Research Drawing from Bayesian Inspiration: Latent Trait Scaling and Topic Modeling Examination of Budgetary Legislation in Thirteen Countries

Abstract

This article extends the methodological and empirical scope of public administration research and applies two Bayesian-inspired computational research methods - unsupervised latent trait scaling and topic modeling. The article uses these methods to examine government budgeting in thirteen Western countries, utilizing budgetary legislation as the research material. The empirical research question is: How are legal system traditions present in the words of texts of legislation on government budgeting? According to the results, at one end of the latent trait scale we find overseas inheritors of Britain’s common law legal system, Canada, Australia, and New Zealand, and at the other end two representatives of civil law of the Napoleonic subtype, Italy and Spain. The other countries situate themselves in intermediate positions between the extremes. The topic modeling indicates three reasonably homogeneous groups of countries: the three overseas inheritors of common law andmore weakly the United Kingdom, three countries representing the Napoleonic heritage, and, more weakly German-speaking and Nordic countries. In general, the article and its results emphasize the opportunities to extend Bayesian-inspired research in this research field.