Polling Aggregation and the Coverage of Opinion Polls

Opinion polls are an integral part of media coverage of election campaigns, but the way in which media cover these polls has been criticized by both academics and pollsters themselves. Studies show that opinion poll reporting, while increasing significantly, does in many cases not confirm to minimal disclosure standards and focuses on non-significant changes (Bhatti & Pedersen, 2016; Oleskog Tryggvason & Strömbäck, 2017; Pétry & Bastien, 2013; Szwed, 2011). One way in which observers try to remedy this is by reporting on ‘poll of polls’ rather than on separate surveys. Potentially, this could change both the quantity of polling news as well as the way in which polls are covered, i.e. focusing more on trends and less on (noisy) outliers. The proposed paper analysis the quality and quantity of opinion poll reporting in the Netherlands between 2010 and 2017. During this period, the use of a polling aggregation model (Peilingwijzer, or polling indicator) has increased from being non-existent to a widely used instrument. Therefore, this presents a good case to analyze the impact of these models on how opinion polls are presented in news media. We analyze opinion poll reporting in (both quality and tabloid) newspapers and on news websites during 5 weeks before the election, examining both the extent of the coverage and its quality. Our study shows that the quality of Dutch coverage of the polls is not very impressive, which is in line with previous studies on poll reporting. Poll reports on Peilingwijzer mention the margin of error more often than media reports of other polls.