The European Union’s use of scientific knowledge in SDG policymaking: a topic model approach

Abstract

The European Union (EU) is committed to grounding its policies for achieving the sustainable development goals (SDGs) in evidence-based policymaking. While there is abundant research on the science–policy interface and the use of scientific knowledge at the scale of research projects, little is known about more aggregate patterns of the use of scientific knowledge in policy. This study proposes a novel measurement method combining Latent Dirichlet Allocation topic models with beta regression analysis to study these aggregate patterns. We demonstrate the method by analysing the alignment between topics in (EU-funded) scientific publications and policy documents produced in different parts of the EU’s policy cycle. Our dataset encompasses publication abstracts from 2010 to 2022 and EU legal documents related to the SDGs from 2015 to 2022. We find that the number of publications on academic topics is positively related to the subsequent prevalence of these topics in policy documents. This relation is strongest in the agenda-setting stage and robust for different time lags. For topics in EU-funded publications, the pattern deviates. Statistically, there is no overall topical alignment, but only with topics found in policy documents related to later policy cycle stages. By confirming established insights as well as generating new ones, our paper contributes to the quest for better methods that enable the monitoring and enhancing of scientific knowledge uptake in policymaking. We conclude by discussing potential applications of the method in research and practice geared towards enhancing science–policy alignment.