Accountability and the Democratic Mandate: Analysing Pledges, Party Competition, Media Coverage, and Public Opinion

Abstract

This dissertation examines electoral accountability and the democratic mandate by reassessing the concept of election pledges, by analysing when parties emphasise the present and past rather than the future, by classifying media coverage of political promises, and revealing the circumstances under which parties lose public support during the electoral cycle. In the first paper, I revisit the scholarly conceptualisation of election pledges. Comparing the most extensive expert reliability exercise to the coding of crowd workers reveals that the experts have a much narrower perception of election promises. Besides raising awareness for differences in the understanding of pledges, the paper exemplifies how crowd coding can be used to reassess established concepts. The second paper introduces the difference between prospective and retrospective campaign communication. Using quantitative text analysis, I uncover the circumstances under which parties emphasise the past and present, and when parties instead focus on the future. I show that large parts of manifestos do not address the future at all. As hypothesised, governments employ more positive language only in statements about the past and present. Incumbents and opposition parties are almost equally positive when it comes to describing the future. The third paper investigates how the media report on campaign promises. The analysis of almost 500,000 statements about promises across 33 electoral cycles in four countries shows that newspaper coverage of promises peaks before elections. The analysis also reveals a substantial negativity bias. These findings help us to understand why voters often struggle to recall the fulfilment of salient pledges and why most citizens do not believe that parties keep their promises. The fourth paper investigates how government support changes throughout the legislative cycle. Based on over 25,000 opinion polls from 171 electoral cycles, I show that government parties, on average, lose most support in the first half of the cycle. Second, the previously assumed curvilinear effect is more likely to occur under single-party government and in countries where the prime minister can dissolve the parliament. Third, since the 2000s, government parties rarely recover from early losses.