Staying Tuned or Tuning Out? A Longitudinal Analysis of News-Avoiders on the Micro and Macro-Level

Abstract

European media landscapes have changed into high-choice environments. Additionally, research suggests an increasing number of people are staying away from the news (news-avoiders). In two studies, we approach the topic on the macro and micro-level. In Study 1, we analyze on the macro-level how news-avoiders in 18 EU countries have developed since the late 1980s and what role country variables play in this context. We use multilevel spline models to distinguish between two phases of technological advancement. Findings suggest an increase in news-avoiders in the phase when the Internet was available. The market share of public-service broadcasters is no discerning factor for the number of news-avoiders in different countries. Using German panel-data, Study 2 takes a closer look at the news-avoiders in the Internet phase. We also show the increase of news-avoiders on the micro-level, but do not identify an increasing impact of political interest on news avoidance over time.

Keywords

news use political interest longitudinal micro and macro-level multilevel spline models

In 2019, over 4,600 TV channels were available in the EU (Ene, 2020), 90% of EU households had Internet access (Eurostat, 2021a), and in 2018, 75% of EU citizens used a smartphone (Eurostat, 2021b). In this high-choice media environment, citizens have every chance to individualize their media use behavior. So-called news-avoiders use these opportunities to stay away from political information without having to forgo media consumption (e.g., Strömbäck et al., 2013). Several studies address the issue of individual motivation interfering with an evenly informed public, raising fears of an increasing participation gap (Prior, 2005), with the disengaged leaning toward populism (Spruyt et al., 2016), and societies disintegrating (Blekesaune et al., 2012). While such individual-level characteristics like political motivation play an important part in shaping citizens’ media diet (Lecheler & de Vreese, 2017), media systems can make news avoidance easy or guide users toward (occasional) news consumption (Iyengar et al., 2010; see also Toff & Kalogeropoulos, 2020). If and to what extent news avoidance has become an increasing issue across Europe is still debated (e.g., Blekesaune et al., 2012; Karlsen et al., 2020; Wonneberger et al., 2012). Especially, the question of whether technological advancements like the Internet have been driving forces needs to be resolved to strengthen the field’s theoretical understanding of changes and related processes in the media environment.

This study helps understand the extent of news avoidance as well as mechanisms that shape and have shaped news avoidance over time. This is achieved by analyzing news-avoiders from different comparative angles; this study (a) combines a macro and micro view on news-avoiders, (b) includes several EU countries, (c) views news-avoiders from a longitudinal perspective that distinguishes between phases of technological advancements, and (d) compares two forms of news-avoiders common in literature, but different in their implications for democracy. The findings of this study can help identify interventions and solutions that limit the impacts of news avoidance on the democratic process.

Opportunity Structures in Changing Media Environments

Environmental factors, like technological developments or the media system, restrict our behavior. Such external factors, shaping the space in which we can act, are defined as opportunity structures (e.g., McAdam, 1996). In the context of media use, opportunity refers to the availability of media choice (Luskin, 1990; Prior, 2007). In recent history, two technological advancements impacting opportunity structures stand out, namely, the implementation of private TV in the 1980s and 1990s in most of Europe and the widespread Internet access starting in the early 1990s but leading far into the new century, with the later addition of mobile Internet. In this paper, we will examine these two phases closer.

Private television had been introduced shortly before the first phase for most countries and gradually brought more electronic media options. Aalberg et al. (2013) offer first insight that the presence of more (or fewer) channels impact the ratio between time spent watching TV and watching the news. Given the expansion of TV channels throughout the phase, people were less likely exposed to the same (news) programs as they had been previously. Nonetheless, “by-product learning” (Prior, 2007, p. 4) was still relevant, especially for generalist channels, which are said to be the most watched (Esser et al., 2012). There, the news is a crucial part of the programming; thus, audiences would stumble upon political information unintentionally. Unintentional news exposure is not exclusive to public channels. Introducing private television can have a positive effect on the available news supply. But whether a private channel contributes to a dense information environment is dependent on its programming strategy (Esser et al., 2012).

In the second phase, the Internet was available to a majority in the country. A further increase in options, interactivity, and mobility mark the Internet phase. News is now available at any time, place, and of any length, loosening the more rigid news-programming schedules people were bound to before, Hermida (2010) labeled this type of environment as “ambient news” (para. 9). Scholars argue that users can come across news unintentionally on the Internet, especially on social media (Fletcher & Nielsen, 2018). People can also be specifically targeted (e.g., political micro-targeting, Jamieson, 2013). In contrast, algorithms personalizing content recommendations can interfere with unintended news viewing (Thorson et al., 2021). Some more modern entertainment forms, like streaming platforms, usually do not incorporate news at all in their portfolio. In the same sense, the increasing number of specialized TV channels usually do not incorporate news into their thematically focused program (Esser et al., 2012)¹.

These changes in opportunity structures, that is supply-side changes, made more options available for all citizens at around the time they were introduced or reached a certain threshold of dissemination. This does not entail citizens’ adaption to these changes at the same time because technological developments and social adaption move at their own speed (Livingstone, 1999). In this sense, ongoing changes throughout a phase are expected as people become familiar with new media forms or, as Pan and McLeod (1991) put it: “[. . .] the processes of inducing individuals to adopt innovations may result in social change” (p. 151). In such a situation, effects are only expected when a certain threshold is reached (e.g., Park, 2017). We, thus, differentiate the phases not by the mere introduction of a medium. The second phase begins when more than half of households have Internet access, in other words they have the opportunity to use the medium. We base this threshold on the diffusion of innovations by Rogers (2003), which explains the process of the dissemination of innovations in society. In the beginning, the innovation is only disseminated between a few people in society and is slowly gaining ground in other early segments. When the 50% boundary is reached, the dissemination starts to spread to the wider public. This discussion leads us to the need to include Internet use (as opposed to Internet access) as a control variable in the macro analysis.

Selection Processes and Their Consequences

In a high-choice environment, selection processes must take place, or as Tsfati and Cappella (2003) put it, “we must select, because we are not physically capable of paying continuous attention to information, political or otherwise” (p. 508). Most people favor content that is in line with their prior attitudes (Goldman & Mutz, 2011; but see Brenes Peralta et al., 2017 for the suggestion of a mixed media diet) or fits their preferences (Prior, 2005). Recipients are anticipated to choose between news and entertainment, with those having higher political interest being more inclined to select political information than those with lower political interest (Strömbäck et al., 2013), which can result in overall avoidance of news. Skovsgaard and Andersen (2020) define news avoidance as “low news consumption over a continuous period of time caused either by a dislike for news (intentional) or a higher preference for other content (unintentional)” (p. 463; italics in original). Consequently, news-avoiders are people with infrequent news usage, who can be further divided by their motivation for the avoidance. Several studies have shown the increase in news-avoiders over time (Blekesaune et al., 2012; Elvestad et al., 2014; Karlsen et al., 2020; Strömbäck et al., 2013). In contrast, Wonneberger et al. (2012) show that the share of avoiders fluctuates around a baseline in the Netherlands. These differences may have originated from different definitions of what low news exposure is. For some, news-avoiders must never have consumed news in a specific time period (Blekesaune et al., 2012; see Toff & Nielsen, 2018 for a similar approach), others include citizens with news usage up to several days a week (Papathanassopoulos et al., 2013) or relative to other news user groups (Strömbäck, 2017). While these differences appear to be mainly operational, they have consequences on the theoretical level as well. Starting from a normative view on democracy, the number of active and informed citizens is key, as they need to monitor politicians and make informed choices (e.g., Aalberg & Curran, 2012). The extent to which citizens inform themselves about current events impacts their level of political knowledge and participation (e.g., de Vreese & Boomgaarden, 2006). The consequences of complete news avoidance for democracy can, therefore, be more severe than infrequent news use.

In this paper, we want to illustrate the differences between common thresholds of news exposure in the news avoidance literature as a basis for further theoretical discussions. Therefore, we will run our models for two common operationalizations of news-avoiders. These operationalizations are (a) news-avoiders defined by an absolute cut-off, which includes people who never or very seldom consume news, and (b) news-avoiders defined relative to other citizens, specifically people consuming relatively less news than any other news user group (e.g., Blekesaune et al., 2012; Skovsgaard & Andersen, 2020; Strömbäck, 2017). In the following, we will refer to these two measurements as total and relative news-avoiders. The total news-avoiders represent the potentially more problematic group for democracy mentioned above. Relative news-avoiders make up a larger segment of the public and could—depending on the average frequency of news usage in a country—even use substantial amounts of news in a given time period or, put differently, they might be well informed about current affairs.

Both phases show means that can lead to unintentional news viewing (Schoenbach, 2008). However, especially mechanisms in the second phase, particularly personalized recommendations, specialized channels, and streaming services, make it easier for media users to avoid political information. Therefore, we expect the following:

H1: There has been no increase in (a) total and (b) relative news-avoiders per country in the first phase.

H2: There has been an increase in (a) total and (b) relative news-avoiders per country in the second phase.

In Study 2, we focus on the second phase and the individual-level in a one-country panel study. We are interested in whether we can identify the same dynamics on the micro-level.

H3: There is an increase in (a) total and (b) relative news-avoiders on the individual-level.

Media System Characteristics

Blekesaune et al. (2012) found that the number of citizens avoiding the news varies noticeably in European countries, identifying a strong difference between Nordic and Southern countries. This indicates that the incentives for news use vary with the media system or as Aalberg and Curran (2012) argue “the organization and structure of the media influences what information is available to national publics, and that national differences of media systems give rise to different national information diets.” (p. 6).

One main difference between media systems is the role public service broadcasting (PSB) plays (Hallin & Mancini, 2004). Research has identified positive effects of PSB on the individual-level, for example concerning cross-cutting exposure (Castro-Herrero et al., 2018), political interest (reciprocal relationship; Strömbäck & Shehata, 2010), or political knowledge (Fraile & Iyengar, 2014; Park & Gil de Zúñiga, 2020). Also, trust in PSB outlets is higher than in private channels (although we see differences between countries; Matsa, 2018). These positive relationships are (among others) related to several characteristics of public service-oriented media systems. First, in less commercialized or public service-oriented systems, the news supply, and the frequency with which news is broadcasted during prime-time, are higher than in more commercialized ones (Aalberg et al., 2010; Curren et al., 2009; Iyengar et al., 2010). Second, PSB lays a stronger emphasis on hard news content than commercial providers or systems (Curran et al., 2009; Iyengar et al., 2010; Reinemann et al., 2017). Based on the same arguments that legitimize publicly funded broadcasting, PSB systems can impose legal requirements for private broadcasters, or (as role models) set standards for reporting. Reinemann et al. (2017) indicate that the standards of a strong PSB tend to be mimicked by commercial channels. This is underpinned by Aalberg et al. (2010) who show that the availability of news for both public-service and private broadcasters is higher in PSB systems compared to more market-oriented ones (see also Esser et al., 2012).

Overall, countries with strong PSB make it easier for uninterested citizens to become inadvertent news audiences by providing a dense information environment and rendering motivation less important (Castro-Herrero et al., 2018; Iyengar et al., 2010). That means the news supply in a system influences the size of news user groups (Elvestad et al., 2014). But as Newton (2016) stresses, PSB systems vary between countries, as do their effects on the public (e.g., de Vreese & Boomgaarden, 2006), especially when broadcasters hold large audience shares and the pressure to adhere to commercial standards decreases.

H4: Having a strong PSB has a negative effect on the percentage of (a) total and (b) relative news-avoiders per country.

We control for other media system-related influences in Study 1 that focus on the social and political inclusiveness of the media system (see Measures) to ensure that we do not overestimate possible influences of PSB strength.

Motivation in a Changing Media Environment

Following the opportunity, ability, and motivation (OAM) framework, externally set opportunities and individual-level factors are entangled. Ability and motivation shape behavior in the context of political information acquisition on the individual-level. Ability describes the skills of understanding offered content. Motivation represents the inner drive to engage with certain media (Delli Carpini & Keeter, 1996; Luskin, 1990; Strömbäck et al., 2013). Prior (2005) finds that “motivation, not ability, is the main obstacle that stands between an abundance of political information and a well- and evenly informed public” (p. 589; but see Karlsen et al., 2020 for important insights on the increasing relevance of education for news avoidance). Thus, on the individual-level (Study 2), we focus on motivation while controlling for ability. On the macro-level (Study 1) we control for both. Luskin (1990) defines motivation in the context of information acquisition as political interest. Prior research has established a positive link between political interest and political knowledge with news use (Lecheler & de Vreese, 2017; Strömbäck & Shehata, 2010).

As suggested by Prior (2007) for the US-American context, the impact of motivation, here political interest, on news use should increase over time. While some studies find evidence for the increased importance of political interest on news use in European countries (Hopmann et al., 2016; Strömbäck et al., 2013; Wonneberger et al., 2012), other studies show opposing results (Aalberg et al., 2013; Elvestad et al., 2014). One explanation can be the different media systems. As argued before, there is variance in the ease with which a person can access news, depending on, for instance the frequency of news programming (e.g., Aalberg et al., 2010). When political information is always present all the time, avoiding it can take up more effort than consuming it, particularly for frequent media users. Schoenbach (2008) indicates that programming strategies can lead media users to consume news if they are not sufficiently motivated to change channels. This finding is in line with research about lead-in and lead-out effects that contribute to unintended news consumption (e.g., Wonneberger et al., 2012). Citizens might also have the “news-finds-me perception” (Gil de Zúñiga et al., 2017, p. 105). Here, they assume that relevant information will reach them through social media or informal contacts without additional effort. These aspects might counter an increasing relevance of political interest on news consumption. In light of these considerations, we ask:

RQ1: Has the impact of political interest on news avoidance increased over time on the micro-level?

Lastly, prior research has suggested that news-avoiders tend to be younger, female, less educated, unemployed, living without a partner, less interested in politics, and show lower political efficacy (Blekesaune et al., 2012). We will control for these aspects in the micro-level analysis (Study 2) to avoid overestimating the impact of political interest.

Study 1: Method

Data

The analysis of Study 1 is based on the data of the Voter study, a national representative survey included in the European Election Study (EES²). This survey is conducted in EU member states subsequent to the European Parliament (EP) elections. Therefore, it takes place in all EU member states, starting from the year of the first EP election after their accession. With a few exceptions, such as Luxembourg, around 1,000 participants were questioned per country per wave. Our sample comprises six waves in total, wherein the first wave took place in 1989 and the last in 2014.³ We considered all countries, which fulfilled two criteria: (a) the country took part in at least one wave before and one wave after the Internet was available to a majority in said country, and (b) the measures for news use were comparable. Therefore, the analysis includes the following 18 countries: Austria, Belgium, Cyprus, Czech Republic, Denmark, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Luxembourg, Poland, Portugal, Spain, the Netherlands, and the United Kingdom (for more information, see chapter Analysis and Table 1). From a macro perspective, the data is hierarchically structured, which means that repeated measures are nested within countries. Thus, in Study 1 we focus on the macro-level.

Table 1.

Percentage of Total and Relative News-Avoiders per Country per Year.

Country	1989	1994	1999	2004	2009	2014
Austria			1.2/19.6	1.1/21.3	1.4/22.2	4.0/7.2
Belgium	2.3/6.8	2.4/14.8	0.8/31.6	1.2/6.0	5.3/15.0	4.2/8.2
Cyprus				4.8/11.2	1.7/7.1	9.2/15.7
Czech Republic				1.2/7.2	3.6/10.0	9.1/14.1
Denmark	0.5/2.4	0.3/7.7	2.7/6.8	0.5/21.6	0.1/21.5	1.2/10.0
Finland			0.4/4.6	0.0/6.4	0.5/11.6	0.9/8.6
France	2.4/4.9	1.5/16.9	2.1/9.1	1.7/9.7	5.2/10.9	4.9/9.7
Germany	1.6/5.5	0.7/8.2	0.8/20.1	2.0/20.0	1.1/18.3	3.3/20.4
Greece	3.8/6.4	2.1/15.7	2.0/7.8	3.0/9.2	4.5/15.4	7.3/15.5
Hungary				1.8/22.8	2.1/9.1	10.4/15.2
Ireland	0.7/3.1	2.0/12.6	0.6/24.3	0.5/28.2	1.6/25	5.8/10.2
Italy	2.6/6.4	0.9/9.0	4.5/9.9	0.1/18.4	1.2/16.1	10.2/14.3
Luxembourg	1.3/2.3	0.8/7.0	0.3/10.6	4.9/8.1	1.6/19.4	3.0/6.7
Poland				2.5/11.9	4.2/13.2	6.8/14.6
Portugal	5.9/4.0	4.8/8.5	2.0/21.6	1.5/14.1	2.2/22.9	5.7/10.1
Spain	3.4/2.4	4.1/6.1	2.8/8.4	3.8/9.4	2.1/23.7	9.0/12.4
The Netherlands	0.3/2.2	1.1/6.6	1.1/13.3	0.6/18.1	0.5/14	1.4/11.4
United Kingdom	0.6/2.3	1.4/7.9	2.3/7.4	2.2/22.1	1.4/23.5	7.0/12.9

Note. Only countries that were included in the analysis. Values are percentages for total news-avoiders/relative news-avoiders.

Measures

News use

Although the question wordings and answer alternatives vary slightly throughout the years, they always ask how often people consume the news in the available outlets (see Supplemental Appendix A for question wording). News use on different platforms was condensed into one variable (overall news use) by assigning it the highest usage of any platform reported by an individual. Therefore, one variable holds the information about a person’s news use in the specific wave. Next, the news use scales of all waves were recoded into a 4-point scale to make it consistent over the six measuring points: 1 indicates once a month or less often, 2 indicates once or twice a week, 3 indicates several times a week, and 4 indicates (almost) every day. For the analysis, we condensed this measure into one value per year and country. The percentage of those falling into the category “once a month or less often” were coded as total news-avoiders. We classified relative news-avoiders by calculating one standard deviation below the mean for each country and wave. People falling on or below this value were identified as relative news-avoiders and aggregated by percentage (see Table 1).

Independent variables

Table 2 lists the independent variables (including control variables) used in Study 1. Due to different scales across predictors, we rescaled all independent variables ranging from 1 to 5. Political bias was taken from an expert survey, the European media systems survey (see Table 2 for sources), and is susceptible to variations in measurement, therefore, it was not possible to include it as time-variant. We used a mean of the two available waves. For Luxembourg, we assigned the overall political bias mean as Luxembourg was not part of the survey. When data was not available in a specific year for newspaper circulation, it was supplemented with the closest available data point.

Table 2.

List of Independent Variables (Including Control Variables).

Indicator	Measure	Original scale	Mean	SD	Source
Political interest	Mean of political interest per country	1–4	3.41	0.86	EES
Education	Leaving school before the age of 15	%	2.29	0.85	EES
PSB strength	PSB market share	%	2.86	0.94	EAO (1995, 2001, 2006, 2014)
Newspaper circulation	Newspaper circulation per 1,000 inhabitants		2.63	0.94	Hendriks (1999), Kleinsteuber et al. (1995), The Bertelsmann Stiftung (n.d.), UNESCO Institute for Statistics (2016), WAN-IFRA (2010, 2015).
Political bias	Country mean of “Medium X coverage is influenced by a political party”	0–10	2.90	1.41	Popescu et al. (2010, 2013)
Internet use	Internet use in last 3 months	%	2.72	1.38	The World Bank Group (2018)

Note. Mean and SD are based on the rescaled variable values of 1–5. EAO = European Audiovisual Observatory; WAN-IFRA = The World Association of News Publishers.

Study 1: Analysis

To investigate the development of news-avoiders over time, we rely on spline (or piecewise) regression analysis using multilevel modeling (e.g., Marsh & Cormier, 2002; Raudenbush & Bryk, 2002; Tilling et al., 2014). These models show several advantages when applied to longitudinal data. First, multilevel models consider the nested structure of the macro data. Thus, we can span a country-specific growth trajectory over repeated measures of the number of news-avoiders. The growth rate of avoiders over time is represented by a slope coefficient. Additionally, we can estimate differences between countries in their growth rates of the number of news-avoiders over time. Second, the spline regression approach allows to split the slope into two distinct phases (called splines), which makes it an ideal approach “to study the effect of an event on [a country’s] trajectory” (Grimm & Marcoulides, 2016, p. 90). Therefore, we can differentiate hypotheses for (spline 1) the time during which people relied on TV as well as newspapers and radio news, and (spline 2) the time during which the Internet was used as an additional mass medium.

The growth trajectory for the number of news-avoiders can be split into two sequences by setting a knot point. In this study, the knot point is the transition from the period with newspapers, radio, and TV only to the period with additional Internet access (see Table 3). However, the pre-knot and post-knot slopes do not represent the same time frames (e.g., the time interval from 2004 to 2014) for each country because the appearance of the Internet as an additional mass medium—described as 50% + 1—varies between each country, depending on when exactly the 50% + 1 boundary is achieved.⁴ The two slopes represent the phase without the Internet as an additional mass medium (spline 1) and the following phase including the Internet (spline 2), respectively. When the 50% + 1 boundary was not achieved in the specific year in which data was collected, we used the previous measurement occasion as the knot point to ensure that the post-knot slope merely captures the time with the Internet as a mass medium. Because we used unbalanced data and placed the knot points at different locations for each country, the growth trajectories vary in length. The following equation describes the full multilevel spline model, including all covariates:

\begin{array}{l} y_{t i} = (γ_{00} + u_{0 i}) + \sum_{k = 1}^{c + 1} ((γ_{10_{k}} + u_{1 i_{k}}) (t i m e_{i})) + γ_{20} (p o l . i n t_{i}) + γ_{30} (p r o p . l o w e d u c ._{i}) \\ + γ_{40} (I n t e r n e t u s e_{i}) + γ_{50} (m . s h a r e T V_{i}) + γ_{60} (n e w s p . c i r c ._{i}) + γ_{01} (b i a s_{i}) + r_{t i}, \end{array}

Table 3.

Included and Excluded Countries and their Knot-Points.

Country	Year	1989	1994	1999	2004	2009	2014
Included
Austria	2006			O	O	O	O
Belgium	2005	O	O	O	O	O	O
Cyprus	2009				O	O	O
Czech Republic	2009				O	O	O
Denmark	2002	O	O	O	O	O	O
Finland	2005			O	O	O	O
France	2007	O	O	O	O	O	O
Germany	2003	O	O	O	O	O	O
Greece	2011	O	O	O	O	O	O
Hungary	2009				O	O	O
Ireland	2006	O	O	O	O	O	O
Italy	2009	O	O	O	O	O	O
Luxembourg	2001	O	O	O	O	O	O
The Netherlands	2001	O	O	O	O	O	O
Poland	2009				O	O	O
Portugal	2010	O	O	O	O	O	O
Spain	2009	O	O	O	O	O	O
United Kingdom	2004	O	O	O	O	O	O
Excluded
Bulgaria	2012					O	O
Croatia	2009						O
Estonia	2007				O	O	O
Latvia	2007				O	O	O
Lithuania	2008				O	O	O
Malta	2007					O	O
Romania	2012					O	O
Slovakia	2008				O	O	O
Slovenia	2006				O	O	O
Sweden	2001			O	O	O	O

Note. O = participated; Highlighted = knot point in analysis; Year = knot point (empirical).

where $y_{t i}$ represents the number of news-avoiders for country i at time t. Further, $γ_{00}$ and $u_{0 i}$ represent the random intercept and the corresponding deviation, respectively. The Sigma sign indicates a minimum spline number of 1 and a maximum spline number of c + 1. Here, the number of knots equals one (c = 1), which leads to two splines in our model. $γ_{10_{k}}$ and $u_{1 i_{k}}$ represent the random slopes for the time variable and the corresponding deviation for spline k, respectively. Fixed effects of time-variant covariates (level-1 variables) are represented by $γ_{20}$ to $γ_{60}$ while the fixed effect of the time-invariant covariate (level-2 variable) is represented by $γ_{01}$ . Residual $r_{t i}$ is a time-specific random disturbance for outcome y.

We calculated the models separately for each outcome variable, that is, the proportion of relative and total news-avoiders. Further, we compared models with random slopes to models with fixed slopes and modeled the random slopes only if the model fit improved. This was only the case for the second slope (spline 2) in the total news-avoiders model, χ²(2) = 21.69, p < .001, indicating differences between countries in their growth rates of total news-avoiders during the second phase, while the relative news-avoiders model was built with a random intercept and fixed slopes for both splines. All models were calculated with the R package lme4 (Bates et al., 2015); we used lmertest (Kuznetsova et al., 2017) to determine p-values and MuMIn (Barton, 2018) to determine R². Variables at level 1 were group-mean centered and those at level 2 were grand-mean centered.⁵

Study 1: Results and Discussion

As shown in Table 4, we found an increase in the number of total news-avoiders during the phase with the Internet as an additional mass medium represented by spline 2 (coefficient $γ_{10_{2}}$ ), supporting H2a. In contrast, we did not find a development in the number of total news-avoiders for the phase in which the Internet was not used by more than 50% + 1 people (coefficient $γ_{10_{1}}$ ), supporting H1a. Further, we found a negative effect of aggregated political interest on the number of total news-avoiders indicating that higher levels of aggregated political interest are related to a lower number of total news-avoiders (coefficient $γ_{20}$ ). Market share of PSB did not affect the number of total news-avoiders (coefficient $γ_{50}$ ), thus, we reject H4a. In contrast, we found a negative effect on the number of total news-avoiders from newspaper circulation (coefficient $γ_{60}$ ) indicating that higher circulation led to lower-than-usual levels in the number of total news-avoiders and a positive effect from political bias (coefficient $γ_{01}$ ), which means that higher political bias in news coverage led to a higher number in total news-avoiders. For the model with relative news-avoiders as the dependent variable, we did not find a statistically significant development in the number of news-avoiders, nor did we find an effect of any of the predictors (see Appendix A). Thus, we accept H1b but reject H2b and H4b.

Table 4.

Fixed Effects on Number of Total News-Avoiders.

Predictor	Coef.	Proportion of total news-avoiders
Predictor	Coef.	Est.	SE	t	p
Intercept	$γ_{00}$	1.90	0.62	3.06	.004
Spline 1	$γ_{10_{1}}$	0.14	0.22	0.62	.539
Spline 2	$γ_{10_{2}}$	2.16	0.69	3.13	.004
Political interest	$γ_{20}$	−0.92	0.35	−2.60	.012
Prop. low. education	$γ_{30}$	−0.13	0.47	−0.28	.784
Internet use	$γ_{40}$	0.14	0.29	0.49	.629
Market share TV	$γ_{50}$	0.60	0.36	1.66	.102
Newspaper circulation	$γ_{60}$	−1.73	0.51	−3.40	.001
Political bias	$γ_{01}$	0.42	0.13	3.21	.005
R²		.43

Note. N = 18, 90 observations. Dependent variable describes the percentage of news-avoiders in each country. Marginal R² describes variance explained by fixed factors.

These results show that, together with the Internet, news avoidance has become of increasing relevance. In the first phase, the mechanisms causing unintended news exposure might have been strong enough to counter an increase in news-avoiders. However, with people getting accustomed to the ever-increasing number of outlets, more digital natives coming of age, and news and entertainment becoming increasingly mobile, citizens make use of their chances to individualize media exposure, helping them navigate the abundance of choices. The missing effect of the PSB market share could indicate that private channels assume the role of PSB in terms of information dissemination in countries with weaker PSB and/or that they complement public-service channels rather than having negative impacts on the availability of news, as has been indicated by Reinemann et al. (2017).

Study 2: Method

Data

Building upon the findings from Study 1, in Study 2 we take a closer look at the micro-level developments in phase two. We rely on individual panel data collected during the German federal elections in 2009, 2013, and 2017 (Rattinger et al., 2015, 2016; Roßteutscher et al., 2019). The German media system is marked by high newspaper circulation and low political bias. As the control variables in Study 1 indicate (see coefficients $γ_{60}$ and $γ_{01}$ in Table 4), this should lead to fewer news-avoiders among the public. Thus, by focusing on Germany, we choose a cautious approach, decreasing the chance of overestimating results. Additionally, in Study 1, Germany shows the typical increase in total news avoidance that we see over most of the countries between 2009 and 2014.

Overall, 592 individuals participated in all 3 years. In each year, the datasets include several repeated measures for each individual, capturing their news use during the weeks before the election. Since individuals’ news use changes over the course of German election campaigns (Gorski & Thomas, 2019) and our main interest is to show developments over longer periods, we averaged news use and corresponding predictors to get single values for each year representing individuals’ news use during German campaigns.

Measures

Within each year, news use was measured on five waves before the election, with questions asking on how many days during the last week individuals read or watched the news on several newspaper and TV outlets and a single question for the Internet. For each medium, we built a mean score and then built a combined score including newspaper, TV, and Internet news use for each wave. After that, we averaged the five combined scores within each year leading to three repeated measures for news use ranging from 2009 to 2017. Based on these three repeated measures (each representing overall news use during one election campaign), we built dummy variables for both relative and total news-avoiders for each year. Individuals with news use below one standard deviation from the mean were grouped as relative avoiders (=1). For the second dummy variable, we grouped all individuals that did not consume news or used it very rarely (values up to 0.02 which represent—except the value 0—the lowest value on the scale ranging from 0 to 7) as total avoiders (=1).

General Internet use, referring to non-campaign phases, was measured in a single wave within each year with an 8-point scale (0–7) and a single item asking on how many days individuals usually use the Internet for general usage (M = 6.59, SD = 1.03).⁶ In 2009, the item slightly differs, explicitly asking for private general Internet use.

Political interest was measured with a single item in five waves within each year for 2013 as well as 2017 and a single item in three waves within 2009 asking about individuals’ general interest in politics. These repeated measures were averaged within each year and were measured on a 5-point scale (1–5) (M = 3.34, SD = 0.94).

For internal political efficacy, we used two items that were asked in two waves within each year, measured on a 5-point scale (1–5). Again, we averaged the political efficacy scores across waves within each year (M = 3.27, SD = 0.91). Additionally, we used items asking for individuals’ employment status (4.4% of all occasions indicated no employment), whether they live in a relationship (30% of all occasions indicated no partner), and their age (M = 43.16, SD = 12.09), education (42.6% highly educated), as well as gender (50.2% female) reported in 2009.

Study 2: Analysis

In order to analyze predictors and the development of news avoidance on an individual-level over time, we rely—similar to Study 1—on multilevel modeling. In contrast to Study 1, however, we (a) use the advantages of individual-level panel data with repeated measures nested in individuals, and (b) use multilevel logistic models (e.g., Sommet & Morselli, 2017) with a dichotomous dependent variable (news-avoiders = 1) to determine the predictors of being a news-avoider. Further, we do not rely on spline regression anymore but model the development for the dichotomous news-avoiders variable over the whole period representing the later Internet phase. In addition, we assess an extended model testing an interaction between political interest and time in order to examine whether the effect of political interest increased over time. We modeled the random effect for the growth parameter only if it significantly contributed to the model. Predictors at level 1 were group-mean centered; predictors at level 2 were grand-mean centered. Again, all models were modeled with the R packages lme4 (using the glmer command for binomial outcomes), lmertest, and MuMin.

Study 2: Results and Discussion

Descriptive results in Table 5 show that both relative and total news-avoiders increased over time, supporting H3a and H3b. We also see that they did not constantly stay news-avoiders. Since the number of total news-avoiders was very small and the model for this group did not converge, we ran the analysis only for relative news-avoiders. The same was true for a growth model with a random slope. Therefore, we ran the model with a random intercept and a fixed slope. The results presented in Table 6 strengthen H3b: The likelihood of being a relative news-avoider increased over time (coefficient $γ_{10}$ ). Moreover, the likelihood of being a relative news-avoider was lower if individuals showed higher levels in political interest (coefficient $γ_{20}$ ). Besides that, we found negative effects of age (coefficient $γ_{01}$ ) as well as education (coefficient $γ_{03}$ ) and a positive effect for females (coefficient $γ_{02}$ ). Furthermore, the interaction term between time and political interest was not significant (coefficient $γ_{70}$ ) indicating that the effect of political interest on being a news-avoider did not increase over time, answering RQ1.⁷

Table 5.

Total and Relative Number of Total and Relative News-Avoiders per Year.

	2009 (%)	2013 (%)	2017 (%)	2009–2017 (%)
Relative news-avoiders	88 (15)	95 (16)	95 (17.6)	41 (6.9)
Total news-avoiders	15 (2.6)	21 (3.5)	31 (5.7)	5 (0.8)
Missing cases	5	0	52	0

Note. N = 592. The 2009–2017 column represents individuals that indicated on all occasions they participated in, that they are news-avoiders.

Table 6.

Fixed Effects on Relative News-Avoiders.

Predictor	Coef.	Relative news-avoiders (=1)
Predictor	Coef.	Est.	SE	z	p
Model 1
Intercept	$γ_{00}$	−7.57	0.92	−8.26	<.001
Time	$γ_{10}$	0.76	0.20	3.77	<.001
Political interest	$γ_{20}$	−2.71	0.48	−5.67	<.001
Political efficacy	$γ_{30}$	−0.58	0.40	−1.46	.145
No partner (=1)	$γ_{40}$	−0.40	0.66	−0.61	.544
Not employed (=1)	$γ_{50}$	0.58	1.09	0.53	.594
Internet use	$γ_{60}$	−0.18	0.24	−0.75	.453
Age	$γ_{01}$	−0.11	0.03	4.04	<.001
Female (=1)	$γ_{02}$	1.70	0.58	2.95	.003
Highly educated (=1)	$γ_{03}$	−1.35	0.60	−2.27	.023
R²		.09
Model 2
Interaction (main eff.)
Time X political interest	$γ_{70}$	−0.65	0.62	−1.05	.294
R²		.10

Note. N = 592, 1,485 observations. Model 2 shows only the main effects of the interaction term; other variables are not depicted but were included in the model. Marginal R² describes variance explained by fixed factors.

The results are in line with most prior research that sees news avoidance as a growing phenomenon (Blekesaune et al., 2012; Elvestad et al., 2014; Strömbäck et al., 2013). Albeit increasing in number, the results stress that the group of total news-avoiders is relatively small, whereas relative news-avoiders make up a substantial portion of the public. Although the relationship is not increasing over time, political interest drives differences in news exposure. As prior research has indicated that political interest is a rather stable characteristic (Prior, 2010), solutions to decrease news avoidance might need to be long-term and focused on younger generations. This is further strengthened as Table 5 shows a core of over-time news-avoiders who are (probably) not susceptible to short-time changes.

General Discussion

At the core of this study lies the question of whether news avoidance has become an increasing issue in the EU over the past decades and what shaped the developments. By focusing on several comparative angles, we provide important insights to contextualize news avoidance with research in the field of media use. There are three key findings: (a) news avoidance has become an increasing phenomenon in the Internet age, (b) the strength of PSB does not explain country differences in the number of avoiders, and (c) political interest is an important predictor for news avoidance but does not increase its impact over time.

First, we found that news avoidance has not increased, when a gradual expanse of television channels, mainly commercially-oriented, marked the main change in people’s media environment. Private broadcasters brought a wider selection of channels, but the modus operandi stayed largely the same: TV viewers subjected themselves to a set programming schedule, which often interjected news between entertainment programs (see generalist channels, as discussed above). Even when viewing alternatives are frequent, people choose to restrain their channel repertoire, which can help navigate high-choice media environments (e.g., Yuan & Webster, 2006). In contrast, together with the expansion of opportunities based on wide-ranging Internet access, the number of news-avoiders increased. This is true for our macro and micro-level study, and thus for the earlier and later developments of the Internet phase. While studies have shown that, for example, through social media people can be incidentally exposed to news (Bergström & Belfrage, 2018), news avoidance still increases. Fletcher and Nielsen (2018) find that incidental exposure to news on social media can have a positive effect on the number of online sources used, especially for the less politically interested. However, as our study shows, it is not a sufficient counter-mechanism for news avoidance. One reason might lie in the repelling effect unwanted news exposure can have on news-avoiders. As Marcinkowski and Došenović (2021) underline, we must differentiate between “unexpected and generally unwanted exposure to political online communication” (p. 457). This is related to the call to differentiate between intentional and unintentional news-avoiders (Skovsgaard & Andersen, 2020). We infer that an additional temporal dimension is appropriate to differentiate news-avoiders, as our data shows that only some individuals stay away from the news over a longer time period.

Second, while we find that the dissemination of a new medium coincides with behavioral changes, the question remains which mechanisms are the leading cause for country differences. Several studies have pointed out how public-service systems offer a media environment in which the cost of acquiring political information is low even for the politically uninterested (e.g., Aalberg et al., 2010). However, Toff & Kalogeropoulos (2020) suggests that behind positive effects of a strong PSB might be underlying assumptions about the actual character of such institutions, specifically aspects of press freedom. Hence, what the missing effect of PSB market share in our study shows, is that providing an environment in which people can easily encounter news is not enough to engage the unwilling. Moreover, knowledge about specific features that lead to incidental news exposure on different media could lead individuals to expect to encounter political information without having to seek news out (Gil de Zúñiga et al., 2017). As prior research has shown, if people expect news to find them without additional effort on their side, news usage on traditional and online media decreases. This relationship is strengthened the more frequently people encounter news unintentionally (Park & Kaye, 2020). As social media usage becomes more frequent and political interest remains an important indicator for news use, a further increase in people not actively seeking out news seems probable. The question then is, does incidental exposure on social media provide enough information to prevent a widening of the knowledge and participation gap? Recent research suggests that this is not the case (Heiss & Matthes, 2019). As such, interventions should identify what characteristics of news repel the politically unengaged and how to make it more interesting for the wider public.

Third, the impact of political interest on news avoidance does not increase over time. Mixed results concerning this question might be rooted in the focus on either exposure or avoidance in prior research. An increasing relevance of political interest for searching out political information (e.g., Hopmann et al., 2016), does not necessarily entail the same for the avoidance of news (see also selective exposure and avoidance, e.g., Schmuck et al., 2020). Further, our paper highlights the relevance of a common theoretical understanding of news-avoiders and comparable measures in the field, by demonstrating that changes in operationalization can lead to dissimilar results. While different measures are appropriate for different data and research questions, as Skovsgaard and Andersen (2020) discuss, it ultimately leads to constrained comparability of findings. We, therefore, should be careful with overgeneralizations of findings and consider the complex nature of news avoidance.

Some important limitations concerning the data analyzed need to be discussed. Prior (2009) finds that self-report measures of news use can strongly overestimate actual news use. This might be a reason why both samples include only a small number of news-avoiders (but, see, Elvestad et al., 2014; Strömbäck et al., 2013 for similar numbers). This could mask the effects for casual news users who would otherwise fall into the news-avoiders category. Moreover, data collection took place close to elections, which are highly salient events and might intensify any upwards bias or even actual news use compared to non-election times.

We must also discuss the construction of the news use measure. The explicit naming of the Internet in the question wording of Study 1 only took place in 2014. While the self-report measure might have underestimated the number of news-avoiders per country, this could have the inverse effect, possibly overestimating the increase in news avoidance between 2009 and 2014 instead of a constant smaller increase over time or even no clear trend. This issue cannot be bypassed, as there is a lack of secondary data, but there are some indications that the trend is, nonetheless, in the upward direction. First, the last measuring point of Study 1 (2014) shows an increase in avoiders, while including the Internet in the question wording. Second, in Study 2, we see the increase of news-avoiders over time as well, although just for one country. Third, this is in line with the findings of prior research (Blekesaune et al., 2012; Elvestad et al., 2014; Strömbäck et al., 2013). This hints at an overall increase in news avoidance in recent history, while possibly underestimating the increase of news avoidance earlier in the second phase.

While not spared of limitations, the current study has enhanced our understanding of the developments of news avoidance in the light of important technological advancements. It is part of a growing body of literature that sees news avoidance as an increasing phenomenon as opportunity structures in the media environment change, thereby emphasizing that it is important and valuable to look more closely at this group.

Supplemental Material

sj-pdf-1-crx-10.1177_00936502211025907 – Supplemental material for Staying Tuned or Tuning Out? A Longitudinal Analysis of News-Avoiders on the Micro and Macro-Level

Supplemental material, sj-pdf-1-crx-10.1177_00936502211025907 for Staying Tuned or Tuning Out? A Longitudinal Analysis of News-Avoiders on the Micro and Macro-Level by Lea C. Gorski and Fabian Thomas in Communication Research

Footnotes

Appendix A

Fixed Effects on Number of Relative News-Avoiders.

Predictor	Coef.	Proportion of relative news-avoiders
Predictor	Coef.	Est.	SE	t	p
Intercept	$γ_{00}$	12.37	2.71	4.57	<.001
Spline 1	$γ_{10_{1}}$	0.31	0.94	0.33	.743
Spline 2	$γ_{10_{2}}$	−1.04	1.70	−0.61	.543
Political interest	$γ_{20}$	0.12	1.64	0.07	.942
Prop. low. education	$γ_{30}$	−4.43	2.28	−1.94	.056
Internet use	$γ_{40}$	1.40	1.35	1.04	.301
Market share TV	$γ_{50}$	−1.08	1.67	−0.65	.519
Newspaper circulation	$γ_{60}$	2.73	2.37	1.15	.254
Political bias	$γ_{01}$	−0.26	0.56	−0.47	.644
R²		.28

Note. N = 18, 90 observations. Dependent variable describes the percentage of news-avoiders in each country. Marginal R² describes variance explained by fixed factors.

Acknowledgements

We thank Michaela Maier, Adam Shehata, and David Nicolas Hopmann for their helpful comments and time invested in discussing this paper. We would also like to thank the editor and the anonymous reviewers for their very helpful input as well as Nigel Cleave for proofreading this paper.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Lea C. Gorski

Supplemental Material

Supplemental material for this article is available online.

Notes

Author Biographies

Lea C. Gorski, MA (University of Koblenz-Landau), is a research assistant and PhD student at the Institute for Communication Psychology and Media Education at the University of Koblenz-Landau. Her research focuses on (political) media use and avoidance.

Fabian Thomas, MA (University of Koblenz-Landau), is a research assistant and PhD student at the Institute for Communication Psychology and Media Education at the University of Koblenz-Landau. He is interested in dynamic processes in communication research (e.g., the reciprocal relationship between media use and attitudes or behaviors) and the statistical modeling of these processes.

References

Aalberg

Blekesaune

Elvestad

(2013). Media choice and informed democracy: Toward increasing news consumption gaps in Europe? The International Journal of Press/Politics, 18(3), 281–303. https://doi.org/10.1177%2F1940161213485990

Aalberg

Curran

(2012). How media inform democracy: Central debates. In Aalberg

Curran

(Eds.), How media inform democracy: A comparative approach (pp. 3–14). Routledge. https://doi.org/10.4324/9780203803448

Aalberg

Van Aelst

Curran

(2010). Media systems and the political information environment: A cross-national comparison. International Journal of Press/Politics, 15(3), 255–271. https://doi.org/10.1177%2F1940161210367422

Andersen

Ohme

Bjarnøe

Bordacconi

M.J.

Albæk

de Vreese

(2021). Generational gaps in political media use and civic engagement. From Baby Boomers to Generation Z. Routledge.

Barton

(2018). MuMIn: Multi-model inference. Tools for performing model selection and model averaging (R package version 1.42.1) [Computer software]. R package. https://CRAN.R-project.org/package=MuMIn

Bates

Mächler

Bolker

Walker

(2015). Fitting linear mixed-effects models using lme4. Journal of Statistical Software, 67(1), 1–48. https://doi.org/10.18637/jss.v067.i01

Bergström

Belfrage

M. J.

(2018). News in social media. Incidental consumption and the role of opinion leaders. Digital Journalism, 6(5), 583–598. https://doi.org/10.1080/21670811.2018.1423625

Blekesaune

Elvestad

Aalberg

(2012). Tuning out the world of news and current affairs—An empirical study of Europe’s disconnected citizens. European Sociological Review, 28(1), 110–126. https://doi.org/10.1093/esr/jcq051

Brenes Peralta

Wojcieszak

Lelkes

de Vreese

(2017). Selective exposure to balanced content and evidence type: The case of issue and non-issue publics about climate change and health care. Journalism & Mass Communication Quarterly, 94(3), 833–861. https://doi.org/10.1177/1077699016654681

10.

Castro-Herrero

Nir

Skovsgaard

(2018). Bridging gaps in cross-cutting media exposure: The role of public service broadcasting. Political Communication, 35(4), 542–565. https://doi.org/10.1080/10584609.2018.1476424

11.

Curran

Iyengar

Lund

A. B.

Salovaara-Moring

(2009). Media system, public knowledge and democracy: A comparative study. European Journal of Communication, 24(1), 5–26. https://doi.org/10.1177%2F0267323108098943

12.

Delli Carpini

M. X.

Keeter

(1996). What Americans know about politics and why it matters. Yale University Press.

13.

Depp

C. A.

Schkade

D. A.

Thompson

W. K.

Jeste

D. V.

(2010). Age, affective experience, and television use. American Journal of Preventive Medicine, 39(2), 173–178. https://doi.org/10.1016/j.amepre.2010.03.020

14.

de Vreese

Boomgaarden

(2006). News, political knowledge and participation: The differential effects of news media exposure on political knowledge and participation. Acta Politica, 41, 317–341. https://doi.org/10.1057/palgrave.ap.5500164

15.

Elvestad

Blekesaune

Aalberg

(2014, September 28–31). The polarized news audience? A longitudinal study of news-seekers and news-avoiders [Paper presentation]. American Political Science Association Annual Meeting 2014, Washington, DC, United States.

16.

Ene

(2020). Supply of audiovisual media services in Europe. MAVISE insights 2019. European Audiovisual Oberservatory. https://rm.coe.int/supply-of-audiovisual-media-services-in-europe-mavise-insights-2019/16809c7874

17.

Esser

Claes

Strömbäck

Aelst

Aalberg

Stanyer

Lengauer

Berganza

Legnante

Papathanassopoulos

Salgado

Sheafer

(2012). Political information opportunities in Europe: A longitudinal and comparative study of thirteen television systems. The International Journal of Press/Politics, 17(3), 247–274. https://doi.org/10.1177/1940161212442956

18.

European Audiovisual Observatory (EAO) [Publisher]. (1995). Statistisches Jahrbuch 96. Filmindustrie, Fernsehen, Video und Neue Medien in Europa [Statistical yearbook 96. Film industry, television, video and new media in Europe].

19.

European Audiovisual Observatory (EAO) [Publisher]. (2001). Statistical yearbook 2001. Film, television, video and new media in Europe.

20.

European Audiovisual Observatory (EAO) [Publisher]. (2006). Yearbook 2006. Television in 36 European states (Vol. 1).

21.

European Audiovisual Observatory (EAO) [Publisher]. (2014). Yearbook 2014. Television, cinema, video and on-demand audiovisual services – the pan-European picture.

22.

Eurostat. (2021a). Households - level of internet access [Data set]. https://ec.europa.eu/eurostat/web/products-datasets/-/isoc_ci_in_h

23.

Eurostat. (2021b). Trust, security and privacy - smartphones (2018) [Data set]. https://ec.europa.eu/eurostat/web/products-datasets/-/isoc_cisci_sp

24.

Fletcher

Nielsen

R. K.

(2018). Are people incidentally exposed to news on social media? A comparative analysis. New Media & Society, 20(7), 2450–2468. https://doi.org/10.1177/1461444817724170

25.

Fraile

Iyengar

(2014). Not all news sources are equally informative: A cross-national analysis of political knowledge in Europe. The International Journal of Press/Politics, 19(3), 275–294. https://doi.org/10.1177/1940161214528993

26.

Gil de Zúñiga

Weeks

Ardèvol-Abreu

(2017). Effects of the news-finds-me perception in communication: Social media use implications for news seeking and learning about politics. Journal of Computer-Mediated Communication, 22(3), 105–123. https://doi.org/10.1111/jcc4.12185

27.

Goldman

S. K.

Mutz

D. C.

(2011). The friendly media phenomenon: A cross-national analysis of cross-cutting exposure. Political Communication, 28(1), 42–66. https://doi.org/10.1080/10584609.2010.544280

28.

Gorski

L. C.

Thomas

(2019, September 12). Dynamics of citizens' news consumption during pre-campaign and campaign phases in three German federal elections [Paper presentation]. ECREA Political Communication Section Interim Conference 2019, Poznań, Poland.

29.

Grimm

Marcoulides

(2016). Individual change and the timing and onset of important life events: Methods, models, and assumptions. International Journal of Behavioral Development, 40(1), 87–96. https://doi.org/10.1177%2F0165025415580806

30.

Hallin

D. C.

Mancini

(2004). Comparing media systems. Three models of media and politics. Cambridge University Press.

31.

Heiss

Matthes

(2019). Does incidental exposure on social media equalize or reinforce participatory gaps? Evidence from a panel study. New Media & Society, 21(11–12), 2463–2482. https://doi.org/10.1177/1461444819850755

32.

Hendriks

(1999). Newspapers: A lost cause? Strategic management of newspaper firms in the United States and the Netherlands. Springer.

33.

Hermida

(2010). From TV to Twitter: How ambient news became ambient journalism. Media and Culture Journal, 13(2), n.p. http://www.journal.media-culture.org.au/index.php/mcjournal/article/view/220

34.

Hopmann

D. N.

Wonneberger

Shehata

Höijer

(2016). Selective media exposure and increasing knowledge gaps in Swiss referendum campaigns. International Journal of Public Opinion Research, 28(1), 73–95. https://doi.org/10.1093/ijpor/edv002

35.

Iyengar

Curran

Lund

A. B.

Salovaara-Moring

Hahn

K. S.

Coen

(2010). Cross-national versus individual-level differences in political information: A media systems perspective. Journal of Elections, Public Opinion and Parties, 20(3), 291–309. https://doi.org/10.1080/17457289.2010.490707

36.

Jamieson

(2013). Messages, micro-targeting, and new media technologies. The Forum, 11(3), 429–435. https://doi.org/10.1515/for-2013-0052

37.

Karlsen

Beyer

Steen-Johnsen

(2020). Do high-choice media environments facilitate news avoidance? A longitudinal study 1997–2016. Journal of Broadcasting & Electronic Media, 64(5), 794–814. https://doi.org/10.1080/08838151.2020.1835428

38.

Kleinsteuber

H. J.

Rossmann

Wiesner

(1995). Essay 3: The mass media. In Shelley

Winck

(Eds.), Aspects of European cultural diversity (pp. 95–152). Routledge. https://doi.org/10.4324/9780203992999

39.

Kuznetsova

Brockhoff

P. B.

Christensen

R. H. B.

(2017). lmerTest package: Tests in linear mixed effects models. Journal of Statistical Software, 82(13), 1–26. https://doi.org/10.18637/jss.v082.i13

40.

Lecheler

de Vreese

C. H.

(2017). News media knowledge, and political interest: Evidence of a dual role from a field experiment. Journal of Communication, 67(4), 545–564. https://doi.org/10.1111/jcom.12314

41.

Livingstone

(1999). New media, new audiences? New Media & Society, 1(1), 59–66. https://doi.org/10.1177/1461444899001001010

42.

Luskin

R. C.

(1990). Explaining political sophistication. Political Behaviour, 12, 331–361. https://doi.org/10.1007/BF00992793

43.

Marcinkowski

Došenović

(2021). From incidental exposure to intentional avoidance: Psychological reactance to political communication during the 2017 German national election campaign. New Media & Society, 23(3), 457–478. https://doi.org/10.1177/1461444820902104

44.

Marsh

L. C.

Cormier

D. R.

(2002). Spline regression models. Sage.

45.

Marsh

Mikhaylov

(2008). EES trend file [Data set]. https://www.tcd.ie/Political_Science/people/michael_marsh/ees_trend_file.php

46.

Matsa

K. E.

(2018). Across Western Europe, public news media are widely used and trusted as sources of news. https://www.pewresearch.org/fact-tank/2018/06/08/western-europe-public-news-media-widely-used-and-trusted/

47.

McAdam

(1996). Conceptual origins, current problems, future directions. In McAdam

McCarthy

J. D.

Zald

M. N.

(Eds.), Comparative perspectives on social movements: Political opportunities, mobilizing structures, and cultural framings (pp. 23–40). Cambridge University Press.

48.

Newton

(2016). Public service and commercial broadcasting: Impacts on politics and society. The Political Quarterly, 87(1), 31–38. https://doi.org/10.1111/1467-923X.12214

49.

Organisation for Economic Cooperation and Development. (2005). OECD Factbook 2005. Economic, environmental and social statistics. OECD Publications. https://doi.org/10.1787/factbook-2005-en

50.

Pan

McLeod

J. M.

(1991). Multilevel analysis in mass communication research. Communication Research, 18(2), 140–173. https://doi.org/10.1177%2F009365091018002002

51.

Papathanassopoulos

Coen

Curran

Aalberg

Rowe

Jones

Rojas

Tiffen

(2013). Online threat, but television is still dominant. Journalism Practice, 7(6), 690–704. http://doi.org/10.1080/17512786.2012.761324

52.

Park

C. P.

Kaye

B. K.

(2020). What’s this? Incidental exposure to news on social media, news-finds-me perception, news efficacy, and news consumption. Mass Communication and Society, 23(2), 157–180. https://doi.org/10.1080/15205436.2019.1702216

53.

Park

C. S.

Gil de Zúñiga

(2020). Learning about politics from mass media and social media: Moderating roles of press freedom and public service broadcasting in 11 countries. International Journal of Public Opinion Research, edaa021, 1–21. https://doi.org/10.1093/ijpor/edaa021

54.

Park

S. S.

(2017). Gendered representation and critical mass: Women’s legislative representation and social spending in 22 OECD countries. Sociological Perspectives, 60(6), 1097–1114. https://doi.org/10.1177/0731121417710458

55.

Popescu

Buzarnescu

Czikora

Gosselin

Marincea

Santana Pereira

Toka

(2013). European media systems survey 2013 [Data set]. Median Research Centre. www.mediasystemsineurope.org

56.

Popescu

Gosselin

Santana Pereira

(2010). European media systems survey 2010 [Data set]. Department of Government, University of Essex. www.mediasystemsineurope.org

57.

Prior

(2005). News vs. entertainment: How increasing media choice widens gaps in political knowledge and turnout. American Journal of Political Science, 49(3), 577–592. https://doi.org/10.1111/j.1540-5907.2005.00143.x

58.

Prior

(2007). Post-broadcast democracy. How media choice increases inequality in political involvement and polarizes elections. Cambridge University Press.

59.

Prior

(2009). The immensely inflated news audience: Assessing bias in self-reported news exposure. Public Opinion Quarterly, 73(1), 130–143. https://doi.org/10.1093/poq/nfp002

60.

Prior

(2010). You’ve either got it or you don’t? The stability of political interest over the life cycle. The Journal of Politics, 72(3), 747–766. https://doi.org/10.1017/S0022381610000149

61.

Rattinger

Roßteutscher

Schmitt-Beck

Weßels

(2015). Short-term campaign panel (GLES 2009) (ZA5305 data file version 5.0.0) [Data set]. GESIS Data Archive, Cologne. https://doi.org/10.4232/1.12198

62.

Rattinger

Roßteutscher

Schmitt-Beck

Weßels

Wolf

(2016). Short-term campaign panel 2013 (GLES) (ZA5704 data file version 3.2.0) [Data set]. GESIS Data Archive, Cologne. https://doi.org/10.4232/1.12561

63.

Raudenbush

S. W.

Bryk

A. S.

(2002). Hierarchical linear models. Sage.

64.

Reinemann

Stanyer

Scherr

(2017). Hard and soft news. In de Vreese

C. H.

Esser

Hopmann

D. N.

(Eds.), Comparing political journalism (pp. 131–149). Routledge.

65.

Rogers

E. M.

(2003). Diffusion of innovations (5th ed.). Free Press.

66.

Roßteutscher

Schmitt-Beck

Schoen

Weßels

Wolf

(2019). Short-term campaign panel (GLES 2017) (ZA6804 data file version 7.0.0) [Data set]. GESIS Data Archive, Cologne. https://doi.org/10.4232/1.13323

67.

Schmitt

Hobolt

S. B.

Popa

S. A.

Teperoglou

, & European Parliament Directorate-General for Communication Public Monitoring Unit. (2016). European Parliament election study 2014, voter study, first post-election survey [Data set]. https://doi.org/10.4232/1.12628

68.

Schmuck

Tribastone

Matthes

Marquart

Bergel

E. M.

(2020). Avoiding the other Side? An eye-tracking study of selective exposure and selective avoidance effects in response to political advertising. Journal of Media Psychology, 32(2), 158–164. https://doi.org/10.1027/1864-1105/a000265

69.

Schoenbach

(2008). Trap effect. In Donsbach

(Ed.), The international encyclopedia of communication (Vol. XI, pp. 5176–5178). Wiley-Blackwell.

70.

Skovsgaard

Andersen

(2020) Conceptualizing news avoidance: Towards a shared understanding of different causes and potential solutions. Journalism Studies, 21(4), 459–476. https://doi.org/10.1080/1461670X.2019.1686410

71.

Sommet

Morselli

(2017). Keep calm and learn multilevel logistic modeling: A simplified three-step procedure using Stata, R, Mplus, and SPSS. International Review of Social Psychology, 30(1), 203–218. http://doi.org/10.5334/irsp.90

72.

Spruyt

Keppens

Van Droogenbroeck

(2016). Who supports populism and what attracts people to it? Political Research Quarterly, 69(2), 335–346. https://doi.org/10.1177/1065912916639138

73.

Strömbäck

(2017). News seekers, news avoiders, and the mobilizing effect of election campaigns: Comparing election campaigns for national and the European parliament. International Journal of Communication, 11, 237–258. https://ijoc.org/index.php/ijoc/article/view/5919

74.

Strömbäck

Djerf-Pierre

Shehata

(2013). The dynamics of political interest and news media consumption: A longitudinal perspective. International Journal of Public Opinion Research, 25(4), 414–435. https://doi.org/10.1093/ijpor/eds018

75.

Strömbäck

Shehata

(2010). Media malaise or a virtuous circle? Exploring the causal relationships between news media exposure, political news attention and political interest. European Journal of Political Research, 49(5), 575–597. https://doi.org/10.1111/j.1475-6765.2009.01913.x

76.

The Bertelsmann Stiftung. (n.d.). Per 1,000 adults, how many people buy a newspaper daily? Sustainable Governance Indicators. https://www.sgi-network.org/2014/Governance/Executive_Accountability/Media/Newspaper_Circulation

77.

The World Association of News Publishers (WAN-IFRA) [Publisher]. (2010). World press trends 2010 (Vol. 21).

78.

The World Association of News Publishers (WAN-IFRA) [Publisher]. (2015). World press trends 2015. The definitive guide to the global newspaper industry, in numbers, trends and changes.

79.

The World Bank Group. (2018). Individuals using the Internet (% of population) [Data set]. https://data.worldbank.org/indicator/it.net.user.zs

80.

Thorson

Cotter

Medeiros

Pak

(2021). Algorithmic inference, political interest, and exposure to news and politics on Facebook. Information, Communication & Society, 24(1), 183–200. https://doi.org/10.1080/1369118X.2019.1642934

81.

Tilling

Macdonald-Wallis

Lawlor

D. A.

Hughes

R. A

Howe

L. D.

(2014). Modelling childhood growth using fractional polynomials and linear splines. Annals of Nutrition & Metabolism, 65(2–3), 129–138. https://doi.org/10.1159/000362695

82.

Toff

Kalogeropoulos

(2020). All the news that’s fit to ignore: How the information environment does and does not shape news avoidance. Public Opinion Quarterly, 84(S1), 366–390. https://doi.org/10.1093/poq/nfaa016

83.

Toff

Nielsen

R. K.

(2018). “I just google it”: Folk theories of distributed discovery. Journal of Communication, 68(3), 636–657. https://doi.org/10.1093/joc/jqy009

84.

Tsfati

Cappella

J. N.

(2003). Do people watch what they do not trust? Exploring the association between news media skepticism and exposure. Communication Research, 30(5), 504–529. https://doi.org/10.1177%2F0093650203253371

85.

UNESCO Institute for Statistics. (2016). Daily newspapers: Total average circulation per 1,000 inhabitants [Data set]. http://data.un.org/Data.aspx?d=UNESCO&f=series%3AC_N_CDN#UNESCO

86.

Van Egmond

Van der Brug

Hobolt

Franklin

Sapir

E. V.

(2017). European Parliament election study 2009, voter study [Data set]. https://doi.org/10.4232/1.12732

87.

Wonneberger

Schoenbach

Van Meurs

(2012). Staying tuned: TV news audiences in the Netherlands 1988-2010. Journal of Broadcasting & Electronic Media, 56(1), 55–74. https://doi.org/10.1080/08838151.2011.648684

88.

Yuan

E. J.

Webster

J. G.

(2006). Channel repertoires: Using peoplemeter data in Beijing. Journal of Broadcasting & Electronic Media, 50(3), 524–536. https://doi.org/10.1207/s15506878jobem5003_10