Measuring Policy Complexity: Evidence from EU Rules

• In Event: New Methods for Studying European Politics

Abstract

European Union scholars often assume that the complexity of rules correlates with response variables such as delegation, lobbying, and transposition delay. Obviously, measuring the complexity of public policy is challenging, which is why many researchers fall back on a narrow proxy by counting the number of recitals. As it is unlikely that a single proxy sufficiently captures a latent trait, I aim to arrive at a more reliable measure of policy complexity with the help of humans' ability to understand text. Specifically, I ask crowd-workers to encode complexity by conducting pairwise comparisons of policies by simply indicating which text is easier. To create suitable micro-tasks, crowd-workers read not the entire text of a policy but individual articles or summaries of legislation provided by the EUR-Lex database. Next, a random utility model produces posterior estimates for the policies’ positions on the complexity scale as well as the workers’ reliability. The final output is a score for each individual article/case on a continuous metric. This complexity score is brought together with additional information about the text of the policy (e.g. rarity of words) to understand why humans consider certain policies to be more complicated than others. In addition, I replicate previous findings in the literature using my new measure.

Author

• Roman Senninger, Aarhus University