News for you! News consumption in a world of news sites, algorithms and social media

This PhD project is conducted by Susan Vermeer, MSc. It is a eerste geldstroom-project hosted at and financed by the Department of Communication Science, UvA. Supervisors are Sanne Kruikemeier, Damian Trilling, and Claes de Vreese.

Being informed about relevant public issues is a prerequisite for a functioning democracy. News media are often the main source of information about public issues (Eveland & Schmitt, 2014). There exists a rich amount of literature showing that news consumption affects various forms of political engagement, such as knowledge (e.g., De Vreese & Boomgaarden, 2006), interest (e.g., Strömbäck & Shehata, 2016), and political participation (e.g., Moeller & De Vreese, 2013). Today, most research on how citizens consume news and how it impacts their levels of political engagement, and ultimately democracy, focuses, on news consumption in the digital age (see e.g., Dimitrova et al., 2014). Yet, theoretical and empirical assumptions about online news consumption and its implications often ignore the combined influence of three important components, which are of decisive influence online. The influence of content (which information is consumed online, such as the topic of news) consumer related factors (individual characteristics, such as political interest), and especially context (e.g., news delivered by news recommendation algorithms and/or online networks as opposed to a “one-size-fits-all” context like a traditional website or a newspaper). Regarding the context, in an online media environment, social networks may, in part, decide which news is consumed, as information acquisition happens in a network setting. Citizens come across information that is send, shared, or liked by their social ties, such as friends and family. Similarly, in the case of a news recommender system, citizens might be increasingly exposed and consume news that algorithms ‘choose’ for them.

Interestingly, these developments can have several plausible but opposite consequences. On the one hand, news users might effectively consume a highly individualized and personalized news diet (e.g., Pariser, 2011; Stroud, 2011). Due to the algorithmic nature of news recommendations, citizens are only exposed to information that algorithms ‘think’ you like (resulting in filter bubbles). In addition, when online networks are homogeneous, we might only be exposed to information that is recommended by like-minded friends (resulting in ego chambers). Equally detrimental, in an online media environment, it is easier to simply avoid political news: Since we are more empowered and have more control over our information flows online, we can therefore decide not to consume any news at all. News avoiding behaviour might be more prevalent among those who are less interested in politics (Bos, Kruikemeier, & de Vreese, 2016; Prior, 2007).

This could have serious and harmful consequences for society at large. When we are not exposed to (a diverse set of) news, citizens might not become interested in (a diverse set of) political and societal issues and “[t]o be politically interested is one of the most important norms from a democratic perspective” (Strömbäck & Shehata, 2010, p. 575). Traditionally, political interest has been shown to have a strong positive relationship with important other political variables, such as knowledge, participation, or voting. Nowadays, also other political- related activities, with a lower threshold, such as sharing of and engagement with political news, can be a consequence of political interest (see Van Deth, 2014). Having more interest in diverse public issues, arguably, will lead to more respect to different positions, ideas and values that are present in society (although some contest such arguments, see, e.g., Mutz, 2002). In this sense, political interest plays a dual role in this project. It might be an important predictor as it ties in with the context variables (citizens who are more interested in politics are more likely to be directed to diverse political news by their ties or algorithms), but also a dependent variable as increased news exposure to public issues also advances interest. Such ideas are strongly rooted in the theoretical notions of virtuous circles (Norris, 2000) and reinforcing spirals (Slater, 2007).

However, it should be acknowledged, that not all scholars are convinced that the consequences of the increasing impact of a changing context are largely negative. For instance, it can be objected that the online news market is still dominated by very few platforms (e.g., Hindman, 2009) and on social media like Facebook, topics that are already popular tend to attract even more shares (e.g., Trilling, Tolochko, & Burscher, 2016). Also, some studies show that social networks users regularly link to news articles from different perspectives (Barbera, Jost, Nagler, Tucker, & Bonneau, 2015; Morgan, Shafiq, & Lampe, 2013), and algorithmic recommendations of news articles can be designed to include diverse topics (Ge, Delgado-Battenfeld, & Jannach, 2010; Sridharan, 2014).

Summing up, previous research shows conflicting results regarding news consumption and its effect on interest. Mainly due to a lack of focus on the most important factors of online news consumption (i.e., the contextual factors), it is largely unknown how such different and counteracting forces play together in shaping someone’s news diet and affecting interest in (different) public issues.

Research question

To address this gap, this project assesses the change of the media ecosystem from the user’s perspective, focusing on two overarching research questions (1) to what extent do the consumer, content, and context features of online media interact in affecting how citizens consume news and (2), in turn, affect interest in politics and engagement with political news.


The first factor that can help explain whether someone is exposed to news are content characteristics. With regard to news avoidance, people might be more likely to be exposed to ‘entertainment’ or ‘soft’ types of news (e.g., local news, sports), while others might be more exposed to ‘hard news’ (political, economic, and foreign news). Furthermore, when people decide to consume news, they might be more often exposed to news that fits with their pre-existing interest and ideas, because they are exposed to information that recommender-systems show or friends decide to share. Thus in this way, topics can directly be linked to the readers’ interests, but they also help explaining news diffusion patterns between online media (Buhl, Günther, & Quandt, 2016). Two studies by Trilling et al. (2016) and García-Perdomo et al. (2017) show that topics influence how news is shared on social media. However, it is still not well understood how the role of such content features varies depending on consumer features and context features in an online environment. It is expected that some citizens (likely those who are more interested) are also more likely to be exposed to political content, while others are not.


Not all citizens are equally likely to follow the news: Characteristics like socio-demographics or political interest are influencing news media use, and many argue that the influence of user characteristics becomes increasingly important in the online media environment (see e.g., Prior, 2007). Research on selective exposure argues that (extremism of) political leaning, at least in the US, exerts an influence on the selection of and interest in news items. While these effects have been demonstrated, their role in complex media-environments in multi-party systems is not fully understood yet, as consistent exposure to a fully-coherent partisan media diet hardly occurs (Zuiderveen Borgesius et al., 2016). Nevertheless, it is reasonable to assume that such political attitudes matter in shaping someone’s news consumption and how news consumption affects interest (in diverse topics).

Two consumer characteristics are of special interest to us (both as driver of news consumption, and consequence of news consumption: political interest and participation). On a general level, political interest refers to someone’s general interest in politics. Traditionally, political participation manifests itself in actions like voting, taking part in demonstrations or political meetings, or becoming member of a party or being active in a political movement. Next to these, a number of ‘low threshold’-forms of participation have evolved, like engaging with political news content by means of commenting, sharing, or ‘liking’ it. It is expected that interested citizens are more likely to be exposed to political news compared to less interested citizens, and that due to context in which news consumption happens, those effects are accelerated. Consequently, it can be expected that it are also those who are already interested that will show increased levels of political participation due to their media consumption.


While content features and consumer features have an established place in social-scientific research into media use and media effects, the context in which news is distributed received less attention and is of crucial importance online. This is also important, as our media system is becoming increasingly hybrid (see e.g., Chadwick, 2013). In this project, we focus on two specific contexts, which differ from “traditional” one-size-fits all contexts (like reading a newspaper or visiting a static news site): (1) the context of social networks (in which news exposure is shaped by the consumer features, most importantly their strong ties, such as close friends, and weak ties, such as acquaintances), and (2) the context of algorithmic news recommenders (in which news exposure is based on use features, including past behaviour, socio-demographics and inferred similarity to other users). It is expected that those who have an online network with ties who are more interested in politics, are more likely to be exposed to political news, while those with ties who are less interested in politics are less exposed to political news. Even more specifically, as selective exposure theory suggests, citizens in a more homogenous network are likely to be mainly exposed to a subset of the news that represents topics and ideological stances that their ties prefer as well. The more homogenous a network is, the stronger these effects are expected to be. In a more heterogeneous network, in general, people are more likely to be exposed to political content, as at least some of their ties that will be interested in politics – a typical example of the benefits of social capital. Furthermore, similar feedback loops might occur with respect to recommender systems. If a recommender system (the algorithms that decide which news is shown to the user) ‘thinks’ a citizen likes political news, (s)he is more likely to be exposed to political news, while if recommender systems believe that a citizen does not like political news, (s)he is not exposed to political news. Such processes effectively fuel a spiralling process and thus can have serious implications for citizens’ opinions, their interest in politics, and their participation.

Combined role of CCC

Taken together, it is of crucial importance to understand which news is consumed online by whom, and how (in which contexts). The influence of our online social network (e.g., when consuming news via Facebook) may result in echo chambers and the influence of technological features (e.g., algorithms) may result in less exposure to diverse content (leading to filter bubbles). As explained above, we argue that it is precisely these two key features of today’s online media environment that interact with the content and consumer characteristics in shaping both exposure to news, which ultimately affects different levels of political interest and participation among citizens. In this project, we integrate content features, consumer features, and context features into one framework in order to better understand what shapes news use in the current media ecosystem and its effects on different forms of political interest and participation.


  • Susan Vermeer (UvA)
  • Sanne Kruikemeier (UvA)
  • Damian Trilling (UvA)
  • Claes de Vreese (UvA)