Professional Documents
Culture Documents
Pol Com 2017
Pol Com 2017
net/publication/320823097
The Mix of Media Use Matters: Investigating the Effects of Individual News
Repertoires on Offline and Online Political Participation
CITATIONS READS
40 542
3 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Jesper Strömbäck on 11 November 2019.
To cite this article: Jesper Strömbäck, Kajsa Falasca & Sanne Kruikemeier (2017): The Mix of
Media Use Matters: Investigating the Effects of Individual News Repertoires on Offline and Online
Political Participation, Political Communication, DOI: 10.1080/10584609.2017.1385549
Download by: [Gothenburg University Library] Date: 03 November 2017, At: 01:02
Political Communication, 00:1–20, 2017
Copyright © Taylor & Francis Group, LLC
ISSN: 1058-4609 print / 1091-7675 online
DOI: https://doi.org/10.1080/10584609.2017.1385549
For democracy to function well, a prerequisite is that people are willing to engage. Not
only does political participation help ensure that those in political office are responsive, it
also helps to foster pro-democratic norms and to strengthen political efficacy (Pateman,
1970). Dahl (1998) consequently describes effective participation as a key criterion for
democratic processes. Moreover, it is essential that inequalities in political participation are
not too great, as political equality in a broader sense might otherwise be undermined
(Dahl, 2006; Verba, Schlozman, & Brady, 1995).
In this context, media use is crucial. As shown by numerous studies, there is a
positive relationship between media use and political participation (Eveland & Scheufele,
2000; Liu, Shen, Eveland, & Dylko, 2013; McLeod, Scheufele, & Moy, 1999). Although
research shows that there are many nuances, the overall pattern is that news media use
contribute to political participation. There are also studies showing the same pattern for
digital and social media use (Boulianne, 2009; Gil De Zúñiga, Jung, & Valenzuela, 2012;
Holt, Shehata, Strömbäck, & Ljungberg, 2013), although a meta-analysis “raise[s] ques-
tions about whether the effects are causal and transformative” (Boulianne, 2015, p. 11).
1
2 Jesper Strömbäck et al.
Despite this, there are several gaps in our knowledge about the effects of media use on
political participation. First, most studies focus on the effects of using traditional news media
or digital media on political participation, and there are only a few studies that simulta-
neously investigate the effects of using traditional offline, online, and social media. Second,
there are only a few studies that examine how people combine the use of offline and online
media and how their “political information repertoires” or “news diets” influence political
participation (Edgerly, 2015; Wolfsfeld, Yarchi, & Samuel-Azran, 2016). This is problematic
insofar as the transition from low- to high-choice media environments means that people
increasingly mix the use of various media into personal news repertoires (Hasebrink &
Domeyer, 2012; Ksiazek, Malthouse, & Webster, 2010; Strömbäck, Djerf-Pierre, & Shehata,
2013; Taneja, Webster, Malthouse, & Ksiazek, 2012; Wolfsfeld et al., 2016). Third, most
studies are based on cross-sectional data, and these are inherently limited when the goal is to
Downloaded by [Gothenburg University Library] at 01:02 03 November 2017
studies classify those who are tuning out from news about politics and current affairs as
“news-avoiders” (Ksiazek et al., 2010; Strömbäck et al., 2013; Van Den Bulck, 2006),
others use the term “disconnected citizens” (Blekesaune et al., 2012). For these people,
increasing media choice has offered them the opportunity to take part of various media
without having to expose themselves to news and current affairs.
A key problem with most studies, however, is that they do not take into account how
people mix their use of various forms of media and media content. In contemporary, high-
choice, hybrid and fragmented media environments (Chadwick, 2013), people increasingly
form their personal “information repertoires” or “news diets” by using a mix of traditional
offline TV news, radio news and newspapers, online versions of traditional TV news,
Web-only media, and social media (Edgerly, 2015; Hasebrink & Domeyer, 2012; Helles,
Ørmen, Radil, & Jensen, 2015; Lee & Yang, 2014; Taneja et al., 2012). In the end, what
Downloaded by [Gothenburg University Library] at 01:02 03 November 2017
probably matters most for how media use influences political participation or other out-
come variables is not people’s use of particular media, but their combinatory use of various
media in their online or offline formats. Ksiazek and colleagues (2010, pp. 552–553) have
thus argued that it might “be deceiving” to use “medium-by-medium” data, where the
impact of using specific media (such as television or print newspapers) is investigated. To
understand the impact of preferences on media use, or the effects of media use on outcome
variables such as political participation, it is therefore necessary to investigate how people
mix the use of various media into personal news repertoires.
While most studies do not investigate the combinatory use of various media, there are
exceptions. One approach is to construct additive indices based on the use of a range of
various media. Ksiazek and colleagues (2010) and Strömbäck and colleagues (2013), for
example, follow this approach in their studies on news-seekers and news-avoiders. The
downside with such an approach is that it only captures how much people use various
media. It does not capture how individuals mix the use of various media.
Another approach is to investigate people’s “news diets,” “political information
repertoires,” or “news repertoires.” Let us give a few examples. Bos, Kruikemeier, and
De Vreese (2016), for example, used latent profile analysis (LCA with continuous vari-
ables) to identify different user profiles based on their use of a range of media spanning
from the main commercial and public service broadcast news to the most used online news
sources and a range of newspapers. Their study showed that four user profiles can be
distinguished: minimalists, public news consumers, popular news consumers, and omni-
vores. Somewhat similarly, Edgerly (2015) measured people’s use of a range of offline and
online media, including partisan blogs. Using principal component analysis and cluster
analysis, she identified six “news repertoires”: news avoiders, online only, TV+print,
liberal+online, conservative only, and news omnivores. With the goal of investigating
how people mix the use of various media, but following another analytical approach,
Wolfsfeld and colleagues (2016) also measured the use of a range of traditional and social
media. They then constructed a “political exposure through traditional media” and a
“political exposure through social media” scale (p. 2105). Based on individuals’ scores
on these scales, they identified four ideal types representing distinct “political information
repertoires”: avoiders, traditionalists, socials, and eclectics. As a final example, Mosca and
Quaranta (2016) measured the use of different media types (newspapers, television, social
media, etc.) instead of individual media, and differentiated people’s media use by cross-
classifying the use of traditional and digital sources of information. According to their
analysis, four “news diets” can be identified: occasional or intermittent, traditional uni-
vores, digital univores, and omnivores (see also Hasebrink & Domeyer, 2012; Helles et al.,
2015; Lee & Yang, 2014; Taneja et al., 2012; Yuan, 2011).
4 Jesper Strömbäck et al.
Based on the research just reviewed, we expect that it will be possible to identify
different news repertoires also in the case under study here: Sweden. Since the number of
news repertoires identified in previous studies differs, it is, however, less clear how many
news repertoires we can expect to find. Against this background, instead of posing a
hypothesis we ask the following research question (RQ):
RQ: What news repertoires, in terms of media user profiles, can be identified?
and communication research. One important conclusion from extant research is that there
are no universal effects across all people or across all forms of media use. The effect on
political engagement rather seems to depend on what type of media people use, why they
use it, and the level of use (Choi, 2016; Gil De Zúñiga, Molyneux, & Zheng, 2014;
Livingstone & Markham, 2008). Political engagement can be measured in different ways,
but here we focus on citizens’ political participation. Political participation is commonly
defined as a behavioral part of political engagement, and it involves activities with the
intention to influence political policy either directly or indirectly (Verba & Nie, 1972).
With respect to media effects, several studies have shown a positive relationship between
the use of different media for getting the news and political participation (Norris, 2000; Shah,
Cho, Eveland, & Kwak, 2005). Nevertheless, evidence showing citizens’ participating more
after having been exposed to various media remains scarce. As a result, it is not clear to what
extent positive relationships between media use and political participation reflect a selection
effect—where those who participate more are more avid news (media) consumers—or a media
effect—where those who consume particular (news) media or are more heavy consumers than
others become more likely to participate politically.
With respect to the use of digital and social media, and with the reservation that
questions of causality remain, previous research suggests positive effects on political
engagement (for an overview, see Boulianne, 2009). On theoretical grounds, it is often
argued that the unique social networking mechanisms might explain why citizens who use
digital and, in particular, social media for political information are more likely to engage in
political life (Boulianne, 2015). First, on social media it is expected that citizens, through
their social network ties, are incidentally exposed to politically and societally relevant
news (Kruikemeier & Shehata, 2016; Mitchell, Kiley, Gottfried, & Guskin, 2013). More
precisely, in an online environment, news is coming from trusted sources such as friends
and acquaintances, and the information might be encountered unexpectedly in the sense
that people are not looking for it but might still be exposed to political information when
friends or acquaintances share political news (Boulianne, 2015). Second, because the
information is coming from people’s own networks, such information might mobilize
citizens to become politically active to a larger extent than information coming directly
from news organizations. Third, online information might be contagious, in the sense that
seeing others reading news or participating in politics can be engaging by itself
(Boulianne, 2015). Fourth, in a digital environment, networks are often larger. This
increases the likelihood that people are being confronted with content from weak ties,
which facilitates information flows and makes it more likely that people are exposed to
politically relevant news (Boulianne, 2015). Fifth, online information is more immediate,
The Mix of Media Use Matters 5
as information can be consumed at any time and in an interactive way; for example, by
finding more information via hyperlinks.
In line with these arguments, used to explain why using social media might contribute
to political participation, Gil De Zúñiga and colleagues (2014) found that social news
consumption has a direct positive effect on offline political participation and an indirect
positive effect on online participation. Using cross-sectional data, Bakker and De Vreese
(2011) also found that online news use is positively related to online and offline participa-
tion. In a similar fashion, Wolfsfeld and colleagues (2016) found that exposure to political
content on social media positively predicts offline and digital participation. Furthermore,
Mosca and Quaranta (2016) showed that media diets that include digital news consump-
tion positively affect offline participation such as boycotts, demonstrations, and occupa-
tions. Differentiating between types of digital media and using panel data, Dimitrova,
Downloaded by [Gothenburg University Library] at 01:02 03 November 2017
Shehata, Strömbäck, and Nord (2014) in addition found a positive participation effect from
using social media but not from using online news, suggesting that the effects might differ
between different types of digital media.
Some studies are, however, less optimistic about the impact of online news use on
political participation. The main rationale for the more pessimistic arguments relates to the
literature on selective exposure. Online, people are simply more likely to self-select
political information (Bennett & Iyengar, 2008; Kruikemeier & Shehata, 2016; Prior,
2007). In an online environment, people might also receive more information that is in
line with their previous dispositions and engagement levels, due both to personal choices
and to algorithm-based filtering and network effects (Pariser, 2011). This suggests that in
an online media environment, politically mobilizing information mainly comes to those
who are already engaged politically, simultaneously making it less likely that non-engaged
citizens become mobilized to participate. Such arguments are, to some extent, also
supported by research (Dimitrova et al., 2014; Edgerly, 2015).
The effects of using different kinds of media notwithstanding, theoretically it can be
expected that the combinatory use of various media (i.e., news repertoires) differs from the
use of specific media. First, in real life, individuals mix the use of various media and
media format, so accounting for the effects of individual news repertoires gets closer to
how people actually use news media. Second, if the effects of using one type of media or
media format differ from the effects of using another, either by being stronger or weaker or
by being negative as opposed to positive, then the total media effect is better captured by
investigating the effects of news repertoires than the effects of using specific media.
Thus far, there is, however, no study investigating the effects of news repertoires
using panel data. While Edgerly (2015), Mosca and Quaranta (2016), and Wolfsfeld and
colleagues (2016) link news repertoires to political participation, they rely on cross-
sectional data, which are limited in terms of identifying the chain of causality. As a result,
our knowledge of the effects of different news repertoires on political participation is
highly limited.
With the reservation that the question of causality largely remains open, overall
research nevertheless suggests that the use of traditional, digital, as well as social
media has positive effects on political participation, and that the effects of using digital
and social media are greater than the effects of using traditional media. Making a
distinction between news websites and social media, research also suggests that the
effects on political participation are stronger for social media than for other digital
media such as news websites (Dimitrova et al., 2014). Against this background, we
expect that different news repertoires will have different effects on offline and online
political participation.
6 Jesper Strömbäck et al.
As we do not yet have the answer to our RQ—asking what types of news repertoires
can be identified—it is difficult to be precise in terms of the effects of different news
repertoires on offline and online political participation. Nevertheless, given previous
research we expect that
news repertoires with lower levels of news consumption will have a negative
effect on both online (H1) and offline (H2) participation compared with news
repertoires with higher levels of news consumption.
Measures
To measure media exposure, which we subsequently use to investigate what news repertoires
can be identified, we used 21 items, including all the most important national news media in
both their traditional and online formats as well as social media. More specifically, we asked
respondents how often they listen to news on public service radio (via radio or via Internet or
cell phone), watch the public service TV news program Aktuellt or Rapport (via television or
via Internet or cell phone), watch the commercial TV news program TV4 Nyheterna (via
The Mix of Media Use Matters 7
television or via Internet or cell phone), and read a local newspaper, a tabloid (Aftonbladet or
Expressen) or a broadsheet (Dagens Nyheter and SvenskaDagbladet) newspaper (either in print
or via Internet or cell phone). All these variables were measured on a scale ranging from 1
(daily) to 6 (never), which were then reverse-coded. Reading news via social media was
measured by asking respondents how often they use social media, such as Facebook and
Twitter, to get updates about the community, share news about politics and society, and get tips
on political news and events. These variables were measured on a five-point scale ranging from
1 (daily) to 5 (never), which were reverse-coded. All these variables were used for the latent
class analyses. Summary statistics can be provided upon request. We also measured several
control variables: sociodemographics (age, gender, and educational level) and general political
interest (measured on a scale ranging from 1 = not at all interested to 4 = very interested).
Online political participation was measured with the question “During the last month,
Downloaded by [Gothenburg University Library] at 01:02 03 November 2017
how often have you done any of the following on the Internet?” The question included
eight items (see Appendix). The response scale ranged from 1 (daily) to 5 (never). Again
these items were reversed. A mean score was calculated (range 1–5) and used to measure
online political participation (Wave 1: eigenvalue = 3.62, explained variance = 45.3%,
Cronbach’s α = 0.81, M = 1.29, SD = 0.49; Wave 2: eigenvalue = 3.65, explained
variance = 45.6%, Cronbach’s α = 0.82, M = 1.34, SD = 0.54).
Offline political participation was measured using the question “During the past
month, have you done the following with the intention to influence a societal issue?”
The question included nine items (see Appendix). The responses were 1 (several times), 2
(once), and 3 (not); this was recoded into dichotomous variables for the analysis (1 and 2
were recoded as 1; 3 was recoded as 0). After performing a Mokken scale analysis (Van
Schuur, 2003) to examine whether the nine dichotomous variables form one scale, a mean
score was calculated (range 0–1) and used to measure offline political participation (Wave
1: H-coefficient = .45, M = .13, SD = .18; Wave 2: H-coefficient = .46, M = .16, SD = .19).
Analytical Strategy
While most studies investigating news repertoires have used principal component analysis,
cluster analysis, or simpler statistical techniques (Edgerly, 2015; Hasebrink & Domeyer,
2012; Lee & Yang, 2014; Taneja et al., 2012; Wolfsfeld et al., 2016), in this study we have
used latent class analysis (LCA; Hagenaars & McCutcheon, 2002; sometimes referred to
as latent profile analyses [see Oberski, 2016]) to examine different types of media users
(for similar approaches, see Bos et al., 2016; Van De Pol, Holleman, Kamoen, Krouwel, &
De Vreese, 2014). A latent class analysis “recovers hidden groups from observed data”
(Oberski, 2016, p. 275), and differs from other classification techniques in that it “fits a
model to the data rather than providing an ad hoc classification of the given data” (Van De
Pol et al., 2014, p. 402). It is particularly useful when variables are measured on different
scales. In our study, we identified individuals who share similar news repertoires, and
those individuals with the same news repertoire are then grouped together (see also Bos
et al., 2016). Using Mplus software (version 6), we included the 21 media exposure
variables mentioned earlier and performed the LCA seven times to determine which
solution fits our data. To avoid local maxima (Van De Pol et al., 2014), we used different
starting values to estimate our models. To identify the best models, we consulted several
commonly used fit indices. These include the Bayesian information criterion (BIC), where
a lower value means better fit; the likelihood ratio test (LRT), which compares whether a
model with K classes is significantly better then K-1 classes using significance tests; and
the entropy measure that assesses, in our case, how well the media exposure variables
8 Jesper Strömbäck et al.
predict the membership of the groups (closer to 1 means better fit; Hagenaars &
McCutcheon, 2002; Van De Pol et al., 2014). After obtaining the best solution, we stored
the new variables: (a) variables for each group, indicating how likely it is that the
respondent belongs to this group (i.e., individual posterior probabilities; the number of
groups depends on the number of K classes) and (b) one variable indicating the best
solution (i.e., every individual respondent is classified in one specific group—this is a
nominal variable).
Second, we examined the extent to which different news repertoires affect offline and
online participation. To take full advantage of the panel data, we included the lagged value
of participation in our model. By doing this, we are able to examine the change in
participation over time. If we find that news use is positively associated with a change
in participation, this will give us a stronger indication of a causal relationship that runs
Downloaded by [Gothenburg University Library] at 01:02 03 November 2017
from media use to participation (Finkel, 1995). This approach has been widely used by
others in political communication science (see, for example, Dimitrova et al., 2014).
Last, because the distribution of some of the dependent variables deviates from
normal—the distribution of “individual probability of membership of a class” and “parti-
cipation” was in some cases positively skewed—we used robust methods to examine the
relationship between types of news consumption and participation (Field, 2013). More
specifically, we deployed bootstrap tests using 5,000 subsamples.
Results
Turning to the results, we begin by addressing which news repertoires, in terms of media
user profiles, can be identified (RQ). To investigate this, as a first step we examined the
number of latent classes in the data (see Table 1). We found that the five-class and six-
class solution offered the best fit. Although the log likelihood and the BIC were lower for
each class that was included, the LRT test showed that the six-class solution was preferred
as the seven-class solution was not significantly better than the former. However, when we
compared the entropy score of the six-class with the five-class solution, the entropy score
of the five-class solution was slightly higher. Moreover, the difference in significance
score for the six-class solution compared to the five-class solution using the LRT test was
small. We also believe that a five-class solution makes more sense, as it is more in line
with previous work (see, e.g., Bos et al., 2016; Edgerly, 2015; Lee & Yang, 2014; Mosca
& Quaranta, 2016; Taneja et al., 2012).4
Table 1
LCA results with different fit indices
Building on the five-class solution, the results show that five news repertoires (unique
groups of media consumers) can be distinguished (see Table 2). In Table 2, the estimated
mean scores (based on the LCA) of using the investigated media per news repertoire are
included. The scale of the offline and online media ranges from 1 (never) to 6 (daily) and
for social media from 1 (never) to 5 (daily). To accompany the analysis and based on a
regression analysis, in Table 3 we present the predictors of membership in each group in
terms of sociodemographic factors and political interest.
The first class of respondents is labeled minimalists. As Table 2 shows, minimalists
consume very little news, although they still consume some news (e.g., the online version
of the tabloid newspaper Aftonbladet). Minimalists are often younger, higher educated, and
less interested in politics (see Table 3). In addition, this is a large group (38.8%). The
second class of respondents is labeled public news consumers, a rather small group (4.5%),
Downloaded by [Gothenburg University Library] at 01:02 03 November 2017
that is likely to listen to the radio, watch public broadcasting news on television, and read
a quality newspaper. In addition, they are also likely to read a local morning newspaper.
As shown in Table 3, public news consumers are often older, have a higher education, and
are more likely to be interested in politics. The third class is labeled local news consumers.
Besides the fact that they are more likely to read a local newspaper in print (slightly more
than the public news consumer), they are also more likely to consume both public and
private news on television and listen to the radio. The main reason why the local news
consumer and the public news consumer differ is the fact that the local news consumer
does not often read a national quality newspaper. The group of local news consumers is
also fairly large (31.2%). People that are local news consumers are often older and have a
lower educational level. The fourth group, social media news consumers, is the third
largest group (18.1%). Social news consumers are more likely to consume news from
social media, such as Facebook. Social news consumers are more likely to be younger,
female, and interested in politics. Finally, popular online news consumers are omnivores.
These individuals consume news from public and commercial broadcasters (both online
and offline) and they are also more likely to read the online version of Aftonbladet. People
in this class are more often interested in politics. This class is quite small (7.4%).
To sum up this part of the analysis, in line with other studies from countries such as the
Netherlands (Bos et al., 2016), Germany (Hasebrink & Domeyer, 2012) and the United States
(Taneja et al., 2012), the results show that it is possible to identify distinct news repertoires
(i.e., groups of media user profiles). Although the number and kind of media included in the
analyses vary across studies, which affects the precise groups identified and the labels used,
there were also some similarities in terms of the number of groups identified and in the fact that
the largest group consists of minimalists(Bos et al., 2016; Lee & Yang, 2014).
Table 2
The five-class solution showing the estimated means of media use for each media repertoire
Minimal- Public News Local News Social Media News Popular Online News
ists Consumers Consumers Consumers Consumers
PB radio (Public radio news) 2.41 4.29 3.93 2.92 3.61
PB radio (Online) 1.46 1.82 1.66 1.96 3.70
Aktuellt on SVT (Public 2.20 3.94 4.82 2.95 4.57
broadcasting news)
Aktuellt on SVT (Online) 1.27 1.46 1.47 1.58 4.67
Rapport on SVT (Public 2.27 4.32 5.03 3.08 4.44
broadcasting news)
Rapport on SVT (Online) 1.26 1.56 1.47 1.59 4.38
Nyheterna on TV4 (Commercial 2.34 3.23 4.17 2.82 4.19
news)
Nyheterna on TV4 (Online) 1.25 1.37 1.36 1.49 3.59
Aftonbladet (Tabloid newspaper) 1.31 1.37 1.76 1.42 2.10
Aftonbladet (Online) 3.01 2.73 3.06 3.96 4.57
Expressen (Tabloid newspaper) 1.19 1.33 1.61 1.32 1.92
Expressen (Online) 1.87 2.21 2.22 2.63 3.41
Local morning paper 2.57 3.62 4.36 2.87 3.49
Local morning paper (Online) 2.16 1.95 2.40 3.07 3.36
Dagens Nyheter (Quality 1.34 2.14 1.74 1.61 2.13
newspaper)
Dagens Nyheter (Online) 1.70 2.18 1.69 2.73 3.01
Svenska Dagbladet (Quality 1.10 5.58 1.22 1.28 1.77
newspaper)
Svenska Dagbladet (Online) 1.40 2.89 1.47 2.27 2.63
SNSs updated 1.77 1.60 1.46 4.28 2.71
SNSs share news 1.13 1.32 1.11 2.59 1.98
SNSs receiving 1.29 1.47 1.18 3.88 2.24
Note. SNSs, Social Network Sites (Facebook and Twitter).
Downloaded by [Gothenburg University Library] at 01:02 03 November 2017
Table 3
Predicting membership of the classes using sociodemographics and political interest
Public News Local News Social Media News Popular Online News
Minimalists Consumers Consumers Consumers Consumers
B (bootstrap SE) B (bootstrap SE) B (bootstrap SE) B (bootstrap SE) B (bootstrap SE)
Constant 1.25 (.05)*** –.11 (.02)*** –.20 (.05)*** .07 (.04) –.00 (.04)
Female –.02 (.02) –.01 (.00) .01 (.01) .03 (.01)* –.01 (.01)
Age –.01 (.00)*** .00 (.00)*** .01 (.00)*** –.01 (.00)*** .00 (.00)
Medium educational level .06 (.03)* .03 (.01)* –.06 (.03)† –.04 (.02)† .01 (.02)
(ref. = low)
High educational level .08 (.03)** .07 (.01)*** –.10 (.03)** –.04 (.02) –.00 (.02)
Political interest –.18 (.01)*** .01 (.00)*** .01 (.01) .13 (.01)*** .03 (.01)***
N 2746 2746 2746 2746 2746
R2 .2290 .0433 .2290 .1388 .0088
*** p < .001. ** p < .01. * p < .05. †p < .10.
12 Jesper Strömbäck et al.
age, educational level, and political interest. More precisely, membership of a class means
that if a respondent belongs to one class (e.g., social media news consumers), she or he is
not able to also be part of another class (e.g., popular online news consumers). We
included the membership of each classification (coded as 0 – not a member of this
class, or 1 – member of this class) as a dummy variable in the models. In this way, we
can compare the impact of membership in one class to the impact of membership in
another class, using minimalists as the reference group.
As shown in Table 4, the results show that only social news consumers are more
likely to become politically active online (b = .13, p < .001) compared to the minimalists.
In contrast, the results also show that local news consumers are less likely to become more
politically active online (b = –.03, p = .039) compared to the minimalists. Turning to the
relationships between news repertoires and offline participation, the regression analyses
Downloaded by [Gothenburg University Library] at 01:02 03 November 2017
again show that only social news consumers are more likely to become active in politics
offline (b = .06, p < .000), compared to the minimalists. All in all, these results lend
support for H3 and H4, while at the same time not lending support for H1 and H2.
Robustness Checks
To further test hypothesis H1 and H2 and examine the robustness of the effects, we also
ran two regression analyses predicting online and offline political participation using the
individual probability of belonging to a specific class as the independent variable. This
Table 4
Predicting participation using membership classification (most likely class membership) as
predictor
analysis was different from the previous ones, because in the previous analyses we used
the membership classification as predictors. The results are presented in Tables 5 and 6.
Importantly, the results are very similar to the ones found in the previous analyses.
Specifically, we again find a robust effect of membership of the social news consumption
group, compared to membership of the other groups, on both forms of participation. We
also find, in some comparisons, a positive significant effect of membership in the popular
online news consumption group on online participation and a negative significant effect of
membership in the local news consumption group on online participation. With respect to
membership in the minimalist group, no support was found for H1 and H2.
As another robustness check, we also examined the results of the six-class solution
(results not shown here).5 Using the same regression analyses described in the latter
analyses, our analyses showed that the results remained largely the same. More specifi-
Downloaded by [Gothenburg University Library] at 01:02 03 November 2017
cally, we again found that membership in the social media news consumer group posi-
tively affects online and offline participation, compared to membership in the minimalist
group. We also found that members of the popular online news consumers group are more
likely to become active in politics online, compared to members of the minimalist group.
Yet, we did not find that local news consumers, compared to the minimalist group, become
less likely to participate online.
Finally, we included the 21 media exposure items as separate predictors in the models,
to examine their individual effect on both forms of political participation (results not
shown here). We again found that the social media items have, in particular, a positive
effect on both online and offline participation.
Table 5
Predicting online participation using probabilities of class membership
Online Participation Online Participation Online Participation Online Participation Online Participation
(Wave 2) (Wave 2) (Wave 2) (Wave 2) (Wave 2)
B (bootstrap SE) B (bootstrap SE) B (bootstrap SE) B (bootstrap SE) B (bootstrap SE)
†
Constant .13 (.05)** .16 (.06)** .10 (.05) .27 (.06)*** .20 (.06)**
Online participation (Wave 1) .86 (.03)*** .86 (.03)*** .86 (.03)*** .86 (.03)*** .86 (.03)***
Minimalists ref. category –.03 (.03) .03 (.02)† –.14 (.03)*** –.07 (.04)†
Public news consumers .03 (.03) ref. category .06 (.03)* –.11 (.04)** –.04 (.05)
Local news consumers –.03 (.02)† –.06 (.03)* ref. category –.17 (.03)*** –.10 (.04)*
Social media news consumers .14 (.03)*** .11 (.04)*** .17 (.03)*** ref. category .07 (.05)
Popular online news consumers .07 (.04)† .04 (.05) .10 (04)* –.07 (.05) ref. category
Female .02 (.01)† .02 (.01)† .02 (.01)† .02 (.01)† .02 (.01)†
Age –.00 (.00)*** –.00 (.00)*** –.00 (.00)*** –.00 (.00)*** –.00 (.00)***
Medium educational level .03 (.02)† .03 (.02)† .03 (.02)† .03 (.02)† .03 (.02)†
(ref. = low)
High educational level .02 (.02) .02 (.02) .02 (.02) .02 (.02) .02 (.02)
Political interest .04 (.01)*** .04 (.01)*** .04 (.01)*** .04 (.01)*** .04 (.01)***
N 2397 2397 2397 2397 2397
R2 .7239 .7239 .7239 .7239 .7239
*** p < .001. ** p < .01. * p < .05. †p < .10.
Downloaded by [Gothenburg University Library] at 01:02 03 November 2017
Table 6
Predicting offline participation using probabilities of class membership
Offline Participation Offline Participation Offline Participation Offline Participation Offline Participation
(Wave 2) (Wave 2) (Wave 2) (Wave 2) (Wave 2)
B (bootstrap SE) B (bootstrap SE) B (bootstrap SE) B (bootstrap SE) B (bootstrap SE)
Constant –.06 (.02)** –.07 (.02)** –.07 (.02)** .01 (.02) –.04 (.02)
Offline participation (Wave 1) .67 (.03)*** .67 (.03)*** .67 (.03)*** .67 (.03)*** .67 (.03)***
Minimalists ref. category .01 (.01) .01 (.01) –.07 (.01)*** –.02 (.02)
Public news consumers –.01 (.01) ref. category –.00 (.01) –.07 (.02)*** –.03 (.02)
Local news consumers –.01 (.01) .00 (.01) ref. category –.07 (.01)*** –.03 (.02)†
Social media news consumers .07 (.01)*** .07 (.02)*** .07 (.01)*** ref. category .04 (.02)*
Popular online news consumers .02 (.02) .03 (.02) .03 (.02)† –.04 (.02)* ref. category
Female .02 (.01)** .02 (.01)** .02 (.01)** .02 (.01)** .02 (.01)**
Age –.00 (.00) –.00 (.00) –.00 (.00) –.00 (.00) –.00 (.00)
Medium educational level .01 (.01) .01 (.01) .01 (.01) .01 (.01) .01 (.01)
(ref. = low)
High educational level .02 (.01)† .02 (.01)† .02 (.01)† .02 (.01)† .02 (.01)†
Political interest .03 (.00)*** .03 (.00)*** .03 (.00)*** .03 (.00)*** .03 (.00)***
N 2397 2397 2397 2397 2397
R2 .5276 .5276 .5276 .5276 .5276
*** p < .001. ** p < .01. * p < .05. †p < .10.
16 Jesper Strömbäck et al.
With respect to the other news repertoires, the results show, however, few significant
effects on political participation, and there was no support for the hypotheses predicting
that news repertoires with lower levels of news consumption would have a negative effect
on political participation offline and online. Instead we found a negative effect of the local
news repertoire on political participation, even when comparing with minimalists. An
explanation might, however, be that local news consumers are often older and less
educated, and therefore less susceptible to being mobilized by using media.
Nevertheless, the overall pattern is that while a social media news repertoire has a
positive effect on offline and online political participation, the other news repertoires do not.
Broadening the perspective, this study has several important implications. First, and
to reiterate a point already made, one important implication is that research on media use
and its effects needs to move beyond medium-by-medium comparisons. While this study
adds to a still quite limited number of studies investigating how people mix and combine
the use of various news media, more research is needed to fully understand the
antecedents and effects of different news repertoires. Second and related, the recurrent
finding that the group of news-avoiders or minimalists is the or among the largest groups
calls for more research on why some groups of people choose not to follow the news.
One key part of the answer is a lack of political interest (Prior, 2007; Strömbäck et al.,
2013), but it is clearly not the full answer. Third, there is a need for further research on
the stability of people’s news repertoires. To our knowledge, thus far there is no study on
how stable people’s news repertoires are, but the degree of stability of people’s news
repertoires has major implications for how we understand different news repertoires,
their antecedents, and effects.
Fourth, this study raises conflicting expectations as to whether offline and online
political participation is likely to increase or decrease in the future. On the one hand, the
migration from traditional news media to social media suggests that political participation
might increase. On the other hand, the fact that minimalists constitute the largest group
while several studies have shown an increase in the share of news-avoiders (Aalberg et al.,
2013; Strömbäck et al., 2013) suggests a limit to the degree to which political participation
will take place or increase. If “monitorial citizenship” (Schudson, 1998) is becoming more
common, the democratic potential of social media to increase political participation might
be offset by minimalists being the most common news repertoire.
Fifth, there is a great need to understand the effects of different news repertoires on
other, democratically relevant, outcome variables. This holds particularly true for the
effects on political knowledge, not least since some studies suggest that social media
have a positive effect on political participation but not on political knowledge, while
traditional media have a positive effect on political knowledge but not on political
The Mix of Media Use Matters 17
participation (Dimitrova et al., 2014). Thus, it might be the case that the same news
repertoire that has a positive effect on political participation has a negative effect on
political knowledge. In order to understand the democratic implications of contemporary
news repertoires, it is thus essential to explore their effects on political knowledge as well.
Notes
1. While it could be considered problematic using a Web panel when investigating—among other
things—Internet use, in Sweden 93% of the population has access to Internet in their homes, and
it is primarily among those older than 76 years that Internet penetration is lower
(Internetstiftelsen i Sverige, 2016). As our sample does not include those older than 75 years
old, this is less of a problem.
Downloaded by [Gothenburg University Library] at 01:02 03 November 2017
2. The waves were part of a larger study with multiple waves. For this study, we used the third and fourth
waves, because the data for these waves were collected around the Swedish national election.
However, in this article, for interpretation purposes, we refer to them as Wave 1 and Wave 2.
3. In 2014, the mean age in the population was 41.2 years, the gender distribution 50% male and
50% female, while 12% had low education, 46% medium education, and 39% higher education.
4. We also performed all the analyses using the six-class solution to examine similarities and
differences (see Robustness Checks section for detailed information and results).
5. The groups that were identified are the minimalists (34.55%), local news consumers (24.28%),
social media news consumer (16.60%), public news consumer – reading quality newspapers in
print (4.51%), the public news consumer – reading an online tabloid news website (14.89%),
and popular online news consumer (5.17%).
References
Aalberg, T., Blekesaune, A., & Elvestad, E. (2013). Media choice and informed democracy toward
increasing news consumption gaps in Europe? The International Journal of Press/Politics, 18,
281–303. doi:10.1177/1940161213485990
Arceneaux, K., & Johnson, M. (2013). Changing minds or changing channels? Partisan news in an
age of choice. Chicago, IL: University of Chicago Press.
Bakker, T. P., & De Vreese, C. H. (2011). Good news for the future? Young people, Internet use, and
political participation. Communication Research, 38, 451–470. doi:10.1177/
0093650210381738
Bennett, W. L., & Iyengar, S. (2008). A new era of minimal effects? The changing foundations of
political communication. Journal of Communication, 58, 707–731. doi:10.1111/j.1460-
2466.2008.00410.x
Blekesaune, A., Elvestad, E., & Aalberg, T. (2012). Tuning out the world of news and current affairs:
An empirical study of Europe’s disconnected citizens. European Sociological Review, 28, 110–
126. doi:10.1093/esr/jcq051
Bos, L., Kruikemeier, S., & De Vreese, C. (2016). Nation binding: How public service broadcasting
mitigates political selective exposure. PloS One, 11, e0155112. doi:10.1371/journal.
pone.0155112
Boulianne, S. (2009). Does Internet use affect engagement? A meta-analysis of research. Political
Communication, 26, 193–211. doi:10.1080/10584600902854363
Boulianne, S. (2015). Social media use and participation: A meta-analysis of current research.
Information, Communication & Society, 18, 524–538. doi:10.1080/1369118X.2015.1008542
Chadwick, A. (2013). The hybrid media system: Politics and power. Oxford, UK: Oxford University
Press.
Choi, J. (2016). Differential use, differential effects: Investigating the roles of different modes of
news use in promoting political participation. Journal of Computer-Mediated Communication,
21(6), 436–450. doi:10.1111/jcc4.12176
18 Jesper Strömbäck et al.
Livingstone, S., & Markham, T. (2008). The contribution of media consumption to civic participa-
tion. The British Journal of Sociology, 59, 351–371. doi:10.1111/j.1468-4446.2008.00197.x
McLeod, J. M., Scheufele, D. A., & Moy, P. (1999). Community, communication, and participation:
The role of mass media and interpersonal discussion in local political participation. Political
Communication, 16, 315–336. doi:10.1080/105846099198659
Mitchell, A., Kiley, J., Gottfried, J., & Guskin, E. (2013). The role of news on Facebook: Common
yet incidental. New York, NY: Pew Research Center.
Mosca, L., & Quaranta, M. (2016). News diets, social media use and non-institutional participation
in three communication ecologies: Comparing Germany, Italy and the UK. Information,
Communication & Society, 19, 325–345. doi:10.1080/1369118X.2015.1105276
Nir, L. (2012). Public space: How shared news landscapes close gaps in political engagement. Journal
of Broadcasting & Electronic Media, 56, 578–596. doi:10.1080/08838151.2012.732145
Norris, P. (2000). A virtuous circle: Political communications in postindustrial societies. Oxford,
Downloaded by [Gothenburg University Library] at 01:02 03 November 2017
Webster, J. G. (2014). The marketplace of attention: How audiences take shape in a digital age.
Cambridge, MA: MIT Press.
Wolfsfeld, G., Yarchi, M., & Samuel-Azran, T. (2016). Political information repertoires and political
participation. New Media & Society, 18, 1–20. doi:10.1177/1461444815580413
Wonneberger, A., Schoenbach, K., & Van Meurs, L. (2011). Interest in news and politics—Or
situational determinants? Why people watch the news. Journal of Broadcasting & Electronic
Media, 55, 325–343. doi:10.1080/08838151.2011.597466
Yuan, E. (2011). News consumption across multiple media platforms. Information, Communication
& Society, 14, 998–1016. doi:10.1080/1369118X.2010.549235
Appendix
Online political participation was measured with the question “During the past month, how
Downloaded by [Gothenburg University Library] at 01:02 03 November 2017
often have you done any of the following on the Internet?” The scale ranged from
1 = Daily to 5 = Never, but was reverse-coded in our analyses. The different forms of
online political participation asked about were as follow:
1. Visit a website of a political party/youth organization
2. Read a blog about politics and society
3. Write texts about social and political issues on your own blog
4. Comment on or discuss issues related to politics and society online
5. Follow any politician or political party via Twitter
6. Follow any politician or political party via Facebook
7. Follow any politician or political party via YouTube
8. Follow any politician or political party via Instagram
Offline political participation was measured using the question “During the past month,
have you done the following with the intention to influence a societal issue?” The response
alternatives were 1 = Yes, several times, 2 = Yes, once, and 3 = No. In our analyses,
responses were recoded into a dichotomous variable and reverse-coded. The different
forms of offline political participation asked about were as follow:
1. Sign a petition
2. Contact a politician
3. Write a letter or debate article for a newspaper
4. Write a debate article or blog post on the Internet
5. Argue for your views in a political discussion
6. Contact mass media
7. Attend a demonstration
8. Attend a political meeting
9. Try to convince others to vote for a particular party