Nudging News Engagement: Avenues Towards and Consequences of Normatively Motivated News Recommender Design

This dissertation examines avenues towards (RQ1) and consequences of (RQ2) nudging news engagement in digital news environments. It includes four sub-studies that feature a combination of theoretical, quantitative, and qualitative work. The studies examine the effects of changes to a) the ranking/filtering, b) the presentation, and c) the content of news. The implications of the nudging effects that emerge are assessed through the lens of different models of democracy. Chapter 2 lays the theoretical groundwork for this dissertation by synthesising insights from a literature review into a theoretical motivation of five different diversity nudges that are empirically tested in subsequent chapters. Chapter 3 reports a mixed-methods study that tests whether such nudges can facilitate the engagement with current affairs news over other genres and explores how nudges are perceived and interpreted more broadly. Chapter 4 uses a between-subjects experimental design to test the effects of deliberately increased exposure diversity on democratically relevant outcome variables. Finally, chapter 5 reports the results of a one-week field experiment that tested the effects of nudging environmental news on engagement and recall in highly externally valid setting. Overall, the results of this dissertation show that nudging news engagement is possible and can, at least in some cases, produce measurable media effects (such as better recall of nudged news). These effects tend to be relatively small (e.g. being 1.5 times as likely to click on a nudged article) and are often heterogeneous, but they may nonetheless compound over time. As such, this dissertation shows that nudging news engagement holds potential for democratic news recommender design but is by no means a silver bullet for addressing concerns about non-diverse and non-political news consumption. Moreover, finding comparably strong effects of how articles are filtered and ranked, the results also point towards the need to carefully reflect on the implications that editorial and increasingly also algorithmic decisions have for what news people see and engage with.