Agencies on the parliamentary radar: Exploring the relations between media attention and parliamentary attention for public agencies using machine learning methods

Abstract The news media frame political debate about public agencies, and enable legislators with incomplete information to monitor and act upon agency (mal)performance. While studies show that the news media matters for parliamentary attention, the contingent nature of this relation has been understudied. Building on agenda‐setting theory, this study theorizes that the effect of newspaper coverage is contingent on the sentiment of coverage, the majority vs. opposition role of legislators, and the locus (committee vs. plenaries) of parliamentary questions. Supervised machine learning methods allow to code sentiment towards agencies in newspapers and parliament, after which a balanced panel relates these data to the questioning behavior of legislators in parliament over time. Results show that media attention for public agencies precedes parliamentary attention. Sentiment matters, as positive media attention, was related to (positive) parliamentary attention in the same month. Negative media attention had broader and more enduring influences on parliamentary questioning behavior.