© The Author(s) 2017
Reprints and permissions:
Antecedents of internal
political efficacy incidental
news exposure online and
the mediating role of political
Universidad de La Laguna, España
University of Vienna, Austria
Homero Gil de Zúñiga
University of Vienna, Austria
Universidad Diego Portales, Chile
Internal political efficacy has long been associated with news use and political discussion. Yet,
as more people are inadvertently exposed to news and political discussion online, it remains
unclear whether incidental news exposure also has a discursive effect on political efficacy. In a
two-wave panel study, we applied the O-S-R-O-R model of communication effects to test these
relationships. We found that political discussion with weak ties, but not strong ties, is a mediator
between incidental news exposure and internal political efficacy.
incidental news exposure, internal political efficacy, network tie strength, news media, political
Received: 18th March 2016; Revised version received: 14th December 2016; Accepted: 18th December 2016
Alberto Ardèvol-Abreu, Departamento de Ciencias de la Comunicación y Trabajo Social, Facultad de
Ciencias Políticas, Sociales y de la Comunicación, Universidad de La Laguna, Av. César Manrique, Campus de
Guajara, S/C de Tenerife 38071, España.
693251POL0010.1177/0263395717693251PoliticsArdèvol-Abreu et al.
Scholars have long emphasized the important role political self-efficacy plays in stimulat-
ing political participation, civic engagement, and other behaviours related to democratic
norms (e.g. Gastil and Xenos, 2010; Rosenstone and Hansen, 1993; Verba and Nie, 1972).
The basic idea is that individuals are more likely to participate in the democratic process
if they feel that they understand their political environment (internal efficacy), and that
their actions can make a difference (external efficacy) (Bandura, 2002; Craig, 1979;
Morrell, 2003). In spite of evidence for a strong influence of political efficacy on a variety
of pro-democratic outcomes, only a few studies have explored its antecedents or predic-
tors (Kenski and Stroud, 2006; Marx and Nguyen, 2016; Moeller et al., 2014). This is
particularly the case for the internal dimension of political efficacy. In contrast, research
suggests that external efficacy and political participation have reciprocal predictive rela-
tions; in other words, participatory behaviours are not only the outcome but also the
antecedent of external efficacy (Finkel, 1985).
Previous research has called for further investigation into the behavioural antecedents
of internal efficacy that, when compared to other forms of political engagement, require
a higher ‘emotional or cognitive activation’ (Finkel, 1985: 907). The current study adds
to this strand of literature by exploring individual demographic and political orientations
that predict people’s perception of their ability to participate in political life. Next, the
study seeks to advance the role of media use and discussion network effects in boosting
internal political efficacy. In particular, the study proposes a model for how incidental and
intentional news exposure, along with interpersonal reasoning mechanisms (political dis-
cussion with strong and weak ties), creates opportunities to learn about, and discuss, poli-
tics. These are salient research issues, given the ubiquity of media content from social
media and online sources, and the numerous chances to simultaneously discuss politics
online and offline.
Drawing on the framework of an O-S-R-O-R (Orientation-Stimulus-Reasoning-
Orientation-Response) model of communication effects (Cho et al., 2009; Jung et al.,
2011), political discussion is tested as a mediator of the relationship between news expo-
sure and internal political self-efficacy. Using panel survey data, we find that a set of
social orientations – strength of partisanship, political knowledge, discussion network
size, and political interest – are all strong predictors of internal efficacy. This study also
tests how incidental exposure to news might influence political discussion with strong
and weak discussion network ties and, in turn, help explain a sense of political compe-
tence. Analyses reveal that both intentional news use and incidental exposure to news
have only indirect effects on internal efficacy, through political discussion with weak ties,
but not strong ties. These results not only extend our understanding of when and how
people generate a sense of political efficacy but also the mechanisms for how news and
discussion intertwine to foster internal efficacy.
Developing the O-S-R-O-R model of media effects
Research on media effects, particularly with regard to campaign media exposure and
participatory behaviours, has progressively abandoned the simple stimulus–response
(S-R) models in favour of more articulated and complex ones, also called ‘indirect para-
digms of media effects’ (Holbert and Stephenson, 2003). Rooted in behavioural psychol-
ogy (Markus and Zajonc, 1985), the O-S-O-R model proposes a three-step model of
Ardèvol-Abreu et al. 3
communication effects that flows: from previous orientations (O) to media stimuli (S),
from media stimuli to subsequent orientations (O), and from subsequent orientations to a
cognitive or behavioural response (R). This framework was introduced in political com-
munication research as a useful tool for explaining certain mediating mechanisms of
direct media effects. For example, media attention, interpersonal communication, and
cognitive elaboration have all been used in this framework to explain the relationship
between media exposure and behavioural or cognitive outcomes (Eveland, 2001; McLeod
et al., 2001).
In the O-S-O-R framework, the first set of orientations (O) corresponds with ‘struc-
tural, cultural, and motivational characteristics of the audience’ (McLeod et al., 1994:
146) that influences the selection and overall impact of a message received from the
media. The stimulus (S) usually refers to news consumption and exposure – though inter-
personal communication is sometimes tested as a stimulus – as a mediator of ‘the effects
of demographic, dispositional, and structural factors on cognitive and behavioral out-
comes’ (Cho et al., 2009: 71). A second set of orientations (O) derives from media expo-
sure, and represents ‘what is likely to happen between reception of the message and the
response of the audience’ (McLeod et al., 1994: 147). In turn, a final response (R) refers
to eventual outcome behaviours of media exposure, such as political participation or civic
engagement (e.g. Cho et al., 2009), although some studies have explored cognitive out-
comes as responses to media exposure (Eveland et al., 2003).
Authors concerned with understanding the complex relationships between mass media
exposure, interpersonal communication, and the underlying cognitive processes of media
effects have proposed an updated version of the O-S-O-R framework. This initial media-
tion model has recently been expanded in order to better account for various activities that
result from media exposure, like conversations, reflection, and cognitive elaboration.
These processes, some authors argue, are not truly ‘subsequent orientations’ stemming
from direct media effects, but are instead reasoning processes taking place between some
media stimuli (S) and subsequent outcome orientations (second O) (Cho et al., 2009; Jung
et al., 2011; Shah et al., 2007). For this reason, Cho et al. (2009) proposed an O-S-R-O-R
(Orientation-Stimulus-Reasoning-Orientation-Response) chain of causation. The addi-
tion of reasoning (R) to the model allows researchers to account for reflection on media
content, and includes various forms of reasoning: anticipation of conversation, composi-
tion of ideas for expression, and cognitive elaboration. Reasoning can also refer to inter-
personal forms of reasoning (e.g. political discussion).
In line with this literature, this study argues that demographics and political attributes
(Orientations) will predict news consumption patterns and incidental news exposure
(Stimulus). This media stimulus will have a further influence on social reasoning mecha-
nisms: political discussion with both strong and weak ties (Reasoning). These ties, in
turn, should influence people’s perceptions of how well equipped they are to partake in
the political process (internal political self-efficacy, Outcome). This theoretical and
empirical model also answers a call made by Cho et al. (2009) for further research regard-
ing outcome orientations. Thus, we advance the O-S-R-O-R model of communication
effects by introducing internal efficacy as an outcome orientation.
Demographic and social orientations (O)
Demographic variables have been found to explain most of the variance in political self-
efficacy (Kenski and Stroud, 2006). This is because political self-efficacy is determined,
at least in part, by an individual’s tendency to reflect on their experiences in order to
achieve goals, solve problems, or master a task (Bandura, 2002). In the political realm,
reflection on political experiences depends on one’s education, employment level, social
status, or past experience with politics (Marx and Nguyen, 2016; Verba and Nie, 1972).
This rationale is in line with a resource model of political participation, where certain
individuals are better equipped to connect their political needs to certain goals or oppor-
tunities (e.g. Brady et al., 1995). For example, Kenski and Stroud found that, among
demographic predictors of political efficacy, education and income showed a positive
influence on both internal and external efficacy, while age was a negative predictor. Social
and political orientations, strength of partisanship, political interest and political discus-
sion with strong ties were also positively related to internal efficacy in that study. Another
study suggests that unemployment also influences internal political efficacy (Marx and
Nguyen, 2016). However, these studies either (1) relied on cross-sectional data analysis or
(2) did not include several key social orientation variables, such as political knowledge,
trust in the media, or discussion network size. Therefore, it is necessary to further explore
the role of these social orientations, not only to assess their importance as predictors of
internal efficacy, but also to control for potential confounds in more complex models.
Thus, we pose the following research question:
RQ1. Which demographic and social orientation characteristics (O) (W1) predict
internal efficacy (W2)?
News media use and incidental news exposure (Stimuli)
In the O-S-R-O-R framework, media use is a stimulus for direct effects on both political
discussion and political self-efficacy. The following section first outlines the effects of
media use on political discussion attributes. Previous research has generally shown that
intentional media use for informational purposes predicts frequency of political discus-
sion (Jung et al., 2011; Kim et al., 1999; Xenos and Moy, 2007). Some studies have
explored in further detail the effects of news use on political discussion by distinguishing
between discussion with so-called ‘strong ties’ and ‘weak ties’. Close ties (people one
knows well, like family or friends) are characterized by ‘intimacy, trust, respect, access,
and mutual regard’ (Huckfeldt and Sprague, 1991: 125). Accordingly, strong ties are
thought to influence political behaviours, like voting or protest, but have less of an impact
on exposure to information diversity (Sinclair, 2012; Valenzuela et al., 2012). In contrast,
weak ties (people one does not know well, like co-workers and acquaintances) are often
composed of people from diverse social, cultural, and economic backgrounds. Exposure
to this diversity gives individuals the opportunity to acquire more diverse, non-redundant
information that could enhance learning and foster political participation (Lake and
Huckfeldt, 1998). Along these lines, Gil de Zúñiga and Valenzuela (2011) found that the
information people gather from the news media tends to spur conversations about public
affairs with both strong and weak ties.
Although the habit of intentional news seeking may influence political conversation,
the acquisition of information about politics and current events does not always occur
intentionally, particularly in the online arena (Tewksbury et al., 2001). In the contempo-
rary news environment, many individuals will be inadvertently exposed to news via
search engines, social media, and the Internet (Kim et al., 2013). People can stumble onto
news while they are doing other activities, despite not actively seeking for information
(Tewksbury et al., 2001). Literature concerning the effects of unintentional exposure to
Ardèvol-Abreu et al. 5
news online is scarce, and to the authors’ best knowledge, no previous studies have
explored whether the effects of this incidental form of consuming news is strong enough
to foster political discussion. Considering these previous findings, but also the gap in the
literature on the subject, we pose the following hypotheses and research questions:
H1. News media use (S) (W1) predicts political discussion (W2) with weak ties (H1a)
(R) and strong ties (H1b) (R).
RQ2. What is the impact of incidental news exposure (W1) on political discussion (W2)
with weak (RQ2a) (R) and strong ties (RQ2b) (R)?
This article also examines the direct relationship between news exposure (intentional
and incidental) and internal political efficacy. Research in this field offers mixed results,
but in general, informational uses of news media have been found to predict internal effi-
cacy (Moeller et al., 2014; Pinkleton et al., 1998). At the disaggregated level, newspapers,
political talk radio, television, and Internet use for news have all been associated, at vary-
ing degrees of magnitude, to internal efficacy (Jung et al., 2011; Kenski and Stroud, 2006;
McLeod et al., 1999; Tewksbury et al., 2008). Based on these previous findings, the fol-
lowing hypothesis is given:
H2. News media use (S) (W1) will be positively related to internal efficacy (W2).
Studies on the political outcomes of incidental exposure to news have focused mainly
on its impact on political knowledge and participation. Previous research partially sup-
ports the assumption that incidental exposure to news predicts political knowledge
(Tewksbury et al., 2001; Zukin and Snyder, 1984), narrows the knowledge gap between
entertainment-oriented and news-oriented Internet users, and enhances both offline and
online political participation (Kim et al., 2013; Kobayashi and Inamasu, 2014). If inci-
dental exposure to news has a positive effect on political knowledge, and helps generate
the informational resources for people to mobilize and participate in politics, it seems
likely that incidental exposure to news will also influence one’s self-perception of their
ability to understand political information, which is a key part of internal political effi-
cacy. On the other hand, occasional and unintended information exposure simply may not
be sufficient for people to change how they think about their ability to participate in poli-
tics. The following is posed as a research question:
RQ3. What is the effect of incidental exposure to news (S) (W1) on internal political
Political discussion: Weak and strong ties (R, reasoning)
Political discussion is a central, normative pillar of liberal democratic systems. This is
because discussion is a way for groups to make decisions based on interpersonal reason-
ing (Dryzek, 1994). Cho et al. (2009) explicitly mention political discussion as a form of
social reasoning in their O-S-R-O-R model. Similarly, Jung et al. (2011) emphasize inter-
personal political discussion and online political messaging as a form of reasoning (R) in
response to media exposure. The model suggested by Jung and colleagues explored the
effects of news media use (stimulus) on political participation, both online and offline
(response). The authors did not find any direct effect of news use on participation.
However, the influence of media consumption on participation was mediated via interper-
sonal discussion/political messaging (reasoning) and political knowledge/internal effi-
cacy (subsequent orientations).
Recent studies have also turned their attention to the effects of the strength of political
discussion ties. Given that political discussions can take place among individuals with
varying degrees of closeness, intimacy, or emotional intensity, it seems logical to question
whether discussions with a ‘close’ or ‘strong tie’ will have the same effects as discussion
with a ‘loose’ or ‘weak tie’. Thus, previous research has confirmed the important role of
the number of weak ties as a predictor of civic engagement (Son and Lin, 2008), and as a
mediator of the relationship between discussion network size and participation (Gil de
Zúñiga and Valenzuela, 2011). However, no previous studies have addressed the relation-
ship between discussion tie strength and internal political efficacy. Considering this gap
in the literature, we ask the fourth research question:
RQ4. How do political discussion network attributes of weak (RQ4a) and strong ties
(RQ4b) (W2) relate to internal efficacy (W2)?
Internal political efficacy (second O, outcome orientation)
Early research on political efficacy conceived political efficacy as a one-dimensional
construct, defined as ‘the feeling that individual political action does have, or can have,
an impact upon the political process, i.e., that it is worthwhile to perform one’s civic
duties’ (Campbell et al., 1954: 187). Thereafter, political efficacy was decomposed into
two different components, which scholars refer to as ‘internal’ and ‘external’ dimensions
of political efficacy (Craig, 1979). Internal political self-efficacy corresponds to internal
evaluations of personal characteristics and experiences that facilitate or impede an indi-
vidual’s response to their political environment (Bandura, 2002; Craig, 1979; Morrell,
2003). External efficacy refers to evaluations of the capacity, or incapacity, of the system
itself to provide an appropriate response (Coleman and Davis, 1976). Since the current
study explores personal characteristics and habits of news consumption and political dis-
cussion, internal efficacy is the outcome variable in all the models tested.
Implicit in the O-S-R-O-R framework of media effects, as proposed by the research
questions and hypotheses, is the indirect relationship between news media use and inter-
nal efficacy. News media use often leads to political discussions (e.g. Kim et al., 1999),
and political discussions foster internal efficacy (Kenski and Stroud, 2006). Thus, it
seems reasonable to expect a mediating role of political discussion on the relationship
between news use and internal efficacy. Individuals receive information about political
affairs and current events from the news media. Naturally, and depending on individual
orientations, this information should increase one’s confidence in their ability to under-
stand and participate in politics. Information received from the media helps motivate a
person to maintain discussions about political issues or recent events, which in turn rein-
forces their self-perception of political competence (indirect effect). Nevertheless, con-
sidering the amount of alternative paths involved in this model, at this stage, we cannot
predict which paths will be significant explanatory mechanisms of internal political effi-
cacy. Thus, we are forced to ask our last research question:
RQ5. Is there a mediation effect of political discussion (either with weak or strong ties)
(W2) in the relationship between news use (either intentional or incidental) (W1) and
internal efficacy (W2)?
Ardèvol-Abreu et al. 7
This study relies on a custom online survey administered in the United States between
December 2013 and March 2014. The media-polling group Nielsen recruited respondents
from approximately 200,000 registered people. Selection of respondents was conducted
according to a quota based on age, gender, education, and income, in order to match US
Census estimates as closely as possible. This method minimizes the drawbacks of sam-
ples based on Internet users (see, for example, Bode et al., 2013; Garret et al., 2012). The
survey was administered using the online survey tool Qualtrics, supplied by a university-
wide subscription account.
The first wave of the survey was distributed between 15 December 2013 and 5 January
2014 to 5000 individuals, of which 1813 provided valid information. The response rate,
computed using the American Association of Public Opinion Research (AAPOR) (2011)
response rate calculator, was 34.6%, an acceptable percentage for Internet-based surveys
(see Sax et al., 2003). The second wave was collected between 15 February and 5 March
2014. Valid information of 1024 cases was included, and the retention rate was 57%.
These figures have been considered acceptable to maintain representation (Watson and
Wooden, 2006). The sample is comparable to the US Census, and is equivalent to other
surveys employing similar sampling procedures (e.g. Pew Research Center for the People
and the Press, 2013).
This study used five key variables and an exhaustive set of control variables in order to
minimize potential confounds. All of these controls have been shown in previous litera-
ture to have an impact in some of our endogenous variables, as explained below. For
specific wording of the items, see Appendix 1.
Key variables. The operationalization of our main dependent variable, ‘internal efficacy’,
was based on universally tested items (Morrell, 2003, 2005) (2-item averaged scale; W2,
Spearman-Brown Coefficient = .87; M = 5.34, standard deviation (SD) = 2.56). Depend-
ing on the degree of closeness and intimacy of the relationship, we distinguished between
‘political discussion with weak ties’ (4-item averaged scale; W2, Cronbach’s α = .86; M =
2.19; SD = 1.57) and ‘political discussion with strong ties’ (5-item averaged scale; W2,
Cronbach’s α = .78; M = 4.00; SD = 2.05). Building on previous measures of the construct
(Gil de Zúñiga et al., 2016), the measure of ‘news media use’ was based on participants’
responses to nine questions about their frequency of intentional exposure to a variety of
news media (W1, Cronbach’s α =.68; M = 3.71; SD = 1.50). Conversely, ‘incidental news
exposure’ measured the amount of unintentional exposure to news, mainly in the online
arena (adapted from Kim et al., 2013; Tewksbury et al., 2001; 6-item averaged scale; W1,
Cronbach’s α = .84; M = 3.56; SD = 1.92).
Control variables. According to previous studies, party identification is connected to our
main dependent variable, internal political efficacy (see, for example, Baumgartner and
Morris, 2006; Kenski and Stroud, 2006). We therefore controlled for ‘strength of parti-
sanship’ (W1, M = 2.10; SD = 1.88). For statistical analysis, non-responses were treated
as missing values. Given that different types of social and institutional trust are related
to political discussion (Mou et al., 2011) and to internal efficacy (Torney-Purta et al.,
2004), we used ‘trust in the media’ as a proxy for social trust (4-item averaged scale;
W1, Cronbach’s α = .72; M = 4.28; SD = 1.72). Political knowledge has been both theo-
retically and empirically linked to internal political efficacy (Balch, 1974; Prior, 2003).
To measure respondent’s ‘political knowledge’, we created an index consisting of eight
questions. Some of them assessed their level of awareness of current policy issues, while
others quantified their degree of knowledge of the US political system and its institu-
tional rules (see Delli Carpini and Keeter, 1993) (W1: Cronbach’s α = .75; M = 4.58; SD
= 2.17; W2: Cronbach’s α = .72; M = 4.23; SD = 2.07). According to previous literature,
the size of political discussion networks has an impact on the frequency of political
discussion (Lee, 2012; Mutz, 2002) and on internal efficacy (Moy and Gastil, 2006). Our
measure of ‘discussion network size’ was created using an additive index of two open-
ended questions (adapted from Gil de Zúñiga et al., 2016). The resulting variable was
highly skewed (W1, M = 4.36; median = 1.00; SD = 16.89; skewness = 10.86). To address
this problem, the natural logarithm was used (W1, M = 0.33; median = 0.24; SD = 0.37:
skewness = 1.32). An individual’s level of interest in politics is strongly connected to
perceptions of political self-competence (Kenski and Stroud, 2006). Accordingly, this
study controls for the effects of ‘political interest’ to isolate potential confounding effects
(2-item averaged scale; W1, Spearman–Brown coefficient = .97; M = 6.67; SD = 2.70).
Finally, all our regression-based models included five demographic variables that have
been related to individual levels of internal efficacy (Hayes and Bean, 1993; Lee, 2006;
Moeller et al., 2013; Vecchione and Caprara, 2009) or to the frequency of political dis-
cussion (Gil de Zúñiga and Valenzuela, 2011). Thus, we considered in our models the
respondent’s gender (49.7% females), age (M = 52.71 years; SD = 14.72 years), and
race (77.9% Whites). We also controlled for education, highest level of formal educa-
tion completed (M = 3.61, median = some college), and household income (M = 4.46,
median = US$50,000–US$59,999).
In order to test the hypotheses and research questions, and to examine the mediating role
of political discussion in the relationship between news use, incidental exposure, and
internal political efficacy, we employed a series of autoregressive analyses as well as
structural equation modelling (SEM) test. Path analysis with SEM allows for the simul-
taneous accounting of direct and indirect effects in a single, comprehensive model
(Geiser, 2010). As an additional advantage, this technique does not limit the number of
mediating variables, so that we can compare the fit of different models, including the
two discussion variables (strong and weak ties) that mediate the relationship between
news exposure and internal efficacy (see Geiser, 2010; Muthén, 2002). To test the statis-
tical significance of indirect effects, we used asymmetric confidence intervals based on
bootstrapping methods (Geiser, 2010). Analyses were conducted using SPSS version
21.0 and Mplus version 7.0.
The first goal of this article was to identify the antecedents of internal political efficacy.
Second, employing the O-S-R-O-R framework, the study tests the direct and indirect
effects of news media use, incidental news exposure, and political discussion with weak
and strong ties on internal efficacy. Our first research question asked about the main
predictors of internal efficacy, considering demographics and social orientations.
Ardèvol-Abreu et al. 9
Autoregressive models presented in Table 1 (Model 2) show that social orientations have
a greater influence on internal efficacy than demographics: While the former accounts for
34.0% of the variance of internal efficacy, the latter only explains 12.2%. Among social
orientations, strength of partisanship (β = .060, p < .01), political knowledge (β = .066, p
< .05), and, most importantly, political interest (β = .134, p < .001) were positively associ-
ated to internal efficacy.
According to our second hypothesis, we expected news use to be directly, and posi-
tively, associated with internal efficacy. We did not find empirical support for H2.
Autoregressive models in Table 1 show that the effect of news media use on internal
efficacy is not statistically significant, either before (Model 1, β = .030, n.s.) or after con-
trolling for political discussion (Model 2, β = .027, n.s.).1 Similarly, RQ3 addressed the
relationship between incidental exposure to news and internal political efficacy. Results
(Table 1) show that incidental news exposure is not directly connected to self-perceptions
of political competence, either in the first (β = −.007, n.s.) or in the second model
(β = −.019, n.s.). However, these non-significant results do not exclude the possibility of
a fully mediated effect of news use and incidental news exposure on internal efficacy
through political discussion, which is tested below.
Table 1. Autoregressive regression models predicting internal efficacy.
Model 1 Model 2
Block 1 – demographics (W1)
Age −.023 −.019
Gender (female) −.020 −.016
Education .005 .013
Income .012 .013
Race (White = 1) −.007 −.002
Block 2 – social orientations (W1)
Partisanship .058** .060**
Trust in the media −.020 −.028
Political knowledge .062* .066*
Discussion network size .045 .028
Political interest .129*** .134***
Block 3 – news media use and incidental news exposure (W1)
News media use .030 .027
Online incidental news exposure −.007 −.019
Block 4 – autoregressive term (W1)
Internal efficacy .629*** .605***
Block 5 – political discussion (W1)
Political discussion (strong ties) – .005
Political discussion (weak ties) – .069*
Total R264.9% 64.8%
OLS: ordinary least squares.
N = 724. Cell entries are final-entry OLS standardized β coefficients.
*p < .05; **p < .01; ***p < .001 (two-tailed).
RQ4 asked about the effects of political discussion with weak (RQ4a) and strong ties
(RQ4b) on internal efficacy. Autoregressive models show that discussing politics with
weak ties (RQ4a) predicts higher levels of internal efficacy (β = .069, p < .05) (Table 1,
Model 2). This positive effect was not observed for discussion with strong ties (RQ4a)
(β = .005, n.s.).
In order to propose a comprehensive model, testing direct and indirect effects in a
theoretical structure comprising news consumption, political discussion (with both strong
and weak ties), and internal efficacy (RQ5), an SEM test was conducted (Figure 1) (boot-
strapped 1000 iterations; χ2 = 8.90; df = 4; p = .06; root mean square error of approxi-
mation (RMSEA) = .039, comparative fit index (CFI) = 0.990, Tucker–Lewis index
Figure 1. Autoregressive effects structural equation model of online incidental exposure, news
use, and political discussion (strong and weak ties) on internal efficacy.
Sample size = 798. Continuous path entries are standardized SEM coefficients (Betas). The effects of de-
mographic variables (age, gender, education, income, and race) and sociopolitical antecedents (strength of
partisanship, trust in the media, political knowledge, discussion network size, and political interest) have been
residualized in all endogenous variables. The model includes indirect effects of online incidental exposure and
news use on internal efficacy (represented in Table 2). Model goodness of fit: χ2 = 8.90; df = 4; p = .06; root
mean square error of approximation (RMSEA) = .039, comparative fit index (CFI) = 0.990, Tucker–Lewis
index (TLI) = 0.969, standardized root mean square residual (SRMR) = .02). Explained variance of criterion
variables: Political Discussion (Strong Ties) R2 = .035; Political Discussion (Weak Ties) R2 = .030; and Internal
Efficacy R2 = .163. This model was bootstrapped based on the standard errors with 1000 iterations.
*p < .05; **p < .01; ***p < .001.
Table 2. Indirect effect of incidental news exposure (W1) and news use (W1) on internal
Indirect effects path Point estimate Two-tailed p-value
Incidental news exposure (W1) → political discussion
(weak ties) (W2) → internal efficacy (W2)
0.013 (0.006) p < .05
News Use (W1) → political discussion (weak ties)
(W2) → internal efficacy (W2)
0.014 (0.006) p < .05
Standardized coefficients (standard error in parentheses).
Ardèvol-Abreu et al. 11
(TLI) = 0.969, standardized root mean square residual (SRMR) = .02) with strong ties
(R2 = .035), weak ties (R2 = .030), and internal efficacy (R2 = .163) as criterion variables.
The model controlled for the effect of internal efficacy in W1, as well as for demographics
and social orientations, as explained in detail in Figure 1.
Our first hypothesis stated a positive relationship between news media use and politi-
cal discussion with both weak (H1a) and strong ties (H1b). We confirmed both H1a and
H1b. SEM coefficients in Figure 1 show that the effect of news media use on political
discussion is direct and strong: Those scoring high in news media use reported higher
frequency of political discussion with weak (H1a) (β = .093, p < .05) and strong ties
(H1b) (β = .087, p < .05). Likewise, RQ2 addressed the effect of incidental news exposure
on political discussion with both weak and strong ties. The SEM in Figure 1 shows that
incidental exposure is positively associated to weak ties (RQ2a) (β = .109, p < .01) and
strong ties (RQ2b) (β = .129, p < .01). Taken together, news consumption, whether inten-
tional or incidental, seems to foster political discussion with both weak and strong ties.
The SEM also showed statistically significant indirect effects of news media stimuli
on internal efficacy via discussion with weak ties (see indirect effects in Table 2). On the
one hand, news media use is indirectly related to internal efficacy through weak ties (b =
.014, p < .05). Similar results were found for incidental news exposure, which was shown
to have an indirect effect on internal efficacy through weak ties (b = .013, p < .05). Results
suggest that political discussion, specifically with those in more ‘distant’ relationships, is
an important path to internal efficacy. Results also allow highlighting the importance of
news consumption, whether intentional or incidental, as a trigger of political discussion.
This study is part of a growing area of research concerned with the causes and precursors
to internal political efficacy. Despite the significance of internal political efficacy in creat-
ing active, better informed citizens, few studies have explored the antecedents of internal
efficacy in such an exhaustive and systematic manner. This study also makes it possible
to ascertain the effects of different uses of news media – deliberate versus incidental
exposure to news – on internal efficacy. In addition, we aimed to find out more about the
direct and indirect influence of political discussion (with both weak and close ties) on
internal efficacy. Last, the model proposed in this article provides additional empirical
evidence of an O-S-R-O-R model of media effects, responding to previous calls for fur-
ther exploration of political orientation outcomes (Cho et al., 2009; Jung et al., 2011).
Results show that both intentional news media use and incidental exposure have a fully
mediated effect on internal efficacy. Political discussion with weak ties is the mediating
variable that explains the effects of both types of news consumption patterns on internal
political efficacy. Consistent with scores of studies, news media use predicts political dis-
cussion (Jung et al., 2011; Xenos and Moy, 2007). Interestingly, this positive effect on
discussion was found not only for intentional but also for unintentional news exposure.
The theoretical explanation for this seems to be simple: People access new information
from traditional and online media and gain knowledge of current affairs, regardless of
whether exposure was intentional or incidental. That is, individuals may learn about poli-
tics either way. This new information fosters political discussion about these issues, as
people may seek to either exchange views, or just gain a better understanding on the topic.
Both patterns of media use (intentional and incidental) stimulate discussion with both
weak and strong ties. Somewhat unexpectedly, indirect effects of news media stimuli on
internal efficacy are fully mediated through discussion with weak discussion network
ties, but not strong ties. To our knowledge, this is the first study to objectively reveal this
important path. One explanation we can offer for this is that interaction with weak ties
may expose individuals to a greater number of arguments, views, and ideas, different
from those that are usually provided by their habitual circle of relationships (Gil de
Zúñiga and Valenzuela, 2011). This, in turn, may also lead to an increased self-perception
of competence in understanding political issues and participating in the democratic pro-
cess in general.
The findings of this study invite optimism, as they open new avenues to better under-
stand the connection between news exposure, political discussion, and internal political
efficacy. While it is true that traditional news media use has declined dramatically over
the past decade, it is also a fact that the Internet and online media use has risen exponen-
tially over the same period (Pew Research Center for the People and the Press, 2016),
providing greater opportunity for incidental exposure to news, even for those who do not
search for it. Our findings also reaffirm the substantive importance of political discus-
sions with weak ties, as they positively predict internal efficacy. These results are in line
with previous studies that underscore the beneficial effects of this type of discussion,
which has been shown to be a stronger predictor of civic participation than political talk
with strong ties (Gil de Zúñiga and Valenzuela, 2011).
There are, however, several limitations in the design of this research that are worth
noting. First, our findings need to be interpreted in the context of a non-election-year
news cycle. When compared to an election year, political issues receive less attention
from the media during these periods. Although 2014 was a midterm election year, the
election was not held until November 4, while the data collection finished in early
March. Further research should validate the model in the context of a presidential elec-
tion news cycle, when media effects might be stronger, more immediate, and perhaps
direct instead of fully mediated via discussion. Another limitation concerns the rela-
tively short time lag between waves (3 months). If we consider the influence of news use
and discussion to be cumulative, a longer time frame between waves could have resulted
in larger effects on our dependent variable. However, long time spans between survey
waves often cause the response rate to fall dramatically (for a more detailed discussion,
see Kessler and Greenberg, 1981), thereby lowering the quality of the data. Our analyses
show significant causal effects of news use and discussion on efficacy, even when con-
trolling for prior individual levels of internal efficacy (autoregressive models). Further
research could, however, try to alter the time frames between waves to check whether
the effects are maintained, increased, or reduced. Finally, although incidental exposure
to news takes place mainly in the online arena (see Kim et al., 2013; Tewksbury et al.,
2001), it is obvious that it can also happen via the traditional media. Future studies may
include the traditional media in the models in order to assess potential differences.
Despite these limitations, this article reinforces previous findings on the potential
value of incidental exposure to news media and political discussion, especially with weak
ties, for developing more effective and participatory citizens. All in all, our findings con-
tribute to the understanding of how news use fuels a cycle in which certain types of politi-
cal discussions explain individual differences regarding self-perception of ability to
participate in decision-making processes and, consequently, possibly contribute to the
development of alternative ways to strengthen a healthier democracy.
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Ardèvol-Abreu et al. 13
1. To better establish causality, and to explore the possibility that the effect of news use on internal effi-
cacy moves in both directions (i.e. a virtuous circle of influence between news use and internal effi-
cacy), we tested alternative regression-based models in which the relationships between variables were
reversed (from internal efficacy to news use, reverse to those shown in Tables 1 and 2 and Figure 1).
Autoregressive ordinary least squares (OLS) models predicting news use in time showed that internal
efficacy does not predict news use, both before (β = .022, n.s.) and after (β = −.002, n.s.) controlling
for political discussion with strong and weak ties (models not reported). To explore possible mediating
effects via discussion with either strong or weak ties, we also constructed an autoregressive structural
equation modelling (SEM) of internal efficacy in W1 predicting news use in W2. All model of fit indexes
suggested a very poor fit of this reversed model to our data (χ2 = 43.38; df = 3; p < .001; root mean square
error of approximation (RMSEA) = .139, comparative fit index (CFI) = 0.934, Tucker–Lewis index
(TLI) = 0.802, standardized root mean square residual (SRMR) = .06), and none of the indirect effects
reached statistical significance. This is also consistent with our theoretical O-S-R-O-R model: the effects
run from intentional and incidental news use (Stimulus) to internal efficacy (Subsequent Orientation),
but not the other way around.
Balch GI (1974) Multiple indicators in survey research: The concept “sense of political efficacy.” Political
Methodology 1(2): 1–43.
Bandura A (2002) Social cognitive theory in cultural context. Applied Psychology 51(2): 269–290.
Baumgartner J and Morris JS (2006) The Daily Show effect: Candidate evaluations, efficacy, and American
youth. American Politics Research 34(3): 341–367.
Bode L, Vraga EK, Borah P and Shah DV (2013) A new space for political behavior: Political social network-
ing and its democratic consequences. Journal of Computer-Mediated Communication 19(3): 414–429.
Brady HE, Verba S and Schlozman KL (1995) Beyond SES: A resource model of political participation.
American Political Science Review 89(2): 271–294.
Campbell A, Gurin G and Miller WE (1954) The Voter Decides. Evanston, IL: Row, Peterson and Company.
Cho J, Shah DV, McLeod JM, McLeod DM, Scholl RM and Gotlieb MR (2009) Campaigns, reflection, and
deliberation: Advancing an O-S-R-O-R model of communication effects. Communication Theory 19(1):
Coleman KM and Davis CL (1976) The structural context of politics and dimensions of regime performance:
Their importance for the comparative study of political efficacy. Comparative Political Studies 9(2):
Craig SC (1979) Efficacy, trust, and political behavior: An attempt to resolve a lingering conceptual dilemma.
American Politics Research 7(2): 225–239.
Delli Carpini MX and Keeter S (1993) Measuring political knowledge: Putting first things first. American
Journal of Political Science 37(4): 1179–1206.
Dryzek J (1994) Discursive Democracy: Politics, Policy, and Political Science. New York: Cambridge
Eveland WP Jr (2001) The cognitive mediation model of learning from the news: Evidence from non-election,
off-year election, and presidential election contexts. Communication Research 28(5): 571–601.
Eveland WP, Shah DV and Kwak N (2003) Assessing causality in the cognitive mediation model: A panel study
of motivations, information processing, and learning during campaign 2000. Communication Research
Finkel SE (1985) Reciprocal effects of participation and political efficacy: A panel analysis. American Journal
of Political Science 29(4): 891–913.
Garret K, Bimber B, Gil de Zúñiga H, Heinderyckx F, Kelly J and Smith M (2012) New ICTs and the study of
political communication. International Journal of Communication 6(1): 214–231.
Gastil J and Xenos M (2010) Of attitudes and engagement: Clarifying the reciprocal relationship between civic
attitudes and political participation. Journal of Communication 60(2): 318–343.
Geiser C (2010) Data Analysis with Mplus. New York; London: The Guilford Press.
Gil de Zúñiga H and Valenzuela S (2011) The mediating path to a stronger citizenship: Online and offline net-
works, weak ties, and civic engagement. Communication Research 38(3): 397–421.
Gil de Zúñiga H, Barnidge M and Scherman A (2016) Social media social capital, offline social capital, and
citizenship: Exploring asymmetrical social capital effects. Political Communication. Epub ahead of print
26 October. DOI: 10.1080/10584609.2016.1227000.
Gil de Zúñiga H, Valenzuela S and Weeks B (2016) Motivations for political discussion: Antecedents and con-
sequences on civic participation. Human Communication Research 42(4): 533–552.
Hayes BC and Bean CS (1993) Political efficacy: A comparative study of the United States, West Germany,
Great Britain and Australia. European Journal of Political Research 23(3): 261–280.
Holbert RL and Stephenson MT (2003) The importance of indirect effects in media effects research: Testing for
mediation in structural equation modeling. Journal of Broadcasting & Electronic Media 47(4): 556–572.
Huckfeldt R and Sprague J (1991) Discussant effects on vote choice: Intimacy, structure, and interdependence.
The Journal of Politics 53(1): 122–158.
Jung N, Kim Y and Gil de Zúñiga H (2011) The mediating role of knowledge and efficacy in the effects of com-
munication on political participation. Mass Communication & Society 14(4): 407–430.
Kenski K and Stroud NJ (2006) Connections between internet use and political efficacy, knowledge, and par-
ticipation. Journal of Broadcasting & Electronic Media 50(2): 173–192.
Kessler RC and Greenberg DF (1981) Linear Panel Analysis: Models of Quantitative Change. New York, NY:
Kim J, Wyatt RO and Katz E (1999) News, talk, opinion, participation: The part played by conversation in
deliberative democracy. Political Communication 16(4): 361–385.
Kim Y, Chen HT and Gil de Zúñiga H (2013) Stumbling upon news on the internet: Effects of incidental news
exposure and relative entertainment use on political engagement. Computers in Human Behavior 29(6):
Kobayashi T and Inamasu K (2014) The knowledge leveling effect of portal sites. Communication Research
Lake R and Huckfeldt R (1998) Social networks, social capital, and political participation. Political Psychology
Lee H (2012) Communication mediation model of late-night comedy: The mediating role of structural fea-
tures of interpersonal talk between comedy viewing and political participation. Mass Communication
and Society 15(5): 647–671.
Lee KM (2006) Effects of Internet use on college students’ political efficacy. CyberPsychology & Behavior
McLeod JM, Kosicki GM and McLeod DM (1994) The expanding boundaries of political communication
effects. In: Bryant J and Zillman D (eds) Media Effects, Advances in Theory and Research. Hillsdale, NJ:
Lawrence Erlbaum, pp.123–162.
McLeod JM, Scheufele DA and Moy P (1999) Community, communication, and participation: The role of mass
media and interpersonal discussion in local political participation. Political Communication 16(3): 315–336.
McLeod JM, Zubric J, Keum H, Deshpande S, Cho J, Stein S, et al. (2001) Reflecting and connecting: Testing
a communication mediation model of civic participation. Paper presented to the AEJMC annual meeting,
Washington, DC, 5–8 August.
Markus H and Zajonc RB (1985) The cognitive perspective in social psychology. In: Lindzey G and Aronson E
(eds) Handbook of Social Psychology. New York: Random House, pp.137–230.
Marx P and Nguyen C (2016) Are the unemployed less politically involved? A comparative study of internal
political efficacy. European Sociological Review. Epub ahead of print 16 May. DOI: 10.1093/esr/jcw020.
Moeller J, de Vreese C, Esser F and Kunz R (2014) Pathway to political participation: The influence of online
and offline news media on internal efficacy and turnout of first-time voters. American Behavioral Scientist
Morrell ME (2003) Survey and experimental evidence for a reliable and valid measure of internal political
efficacy. Public Opinion Quarterly 67(4): 589–602.
Morrell ME (2005) Deliberation, democratic decision-making and internal political efficacy. Political Behavior
Mou Y, Atkin D and Fu H (2011) Predicting political discussion in a censored virtual environment. Political
Communication 28(3): 341–356.
Moy P and Gastil J (2006) Predicting deliberative conversation: The impact of discussion networks, media
use, and political cognitions. Political Communication 23(4): 443–460.
Muthén BO (2002) Beyond SEM: General latent variable modeling. Behaviormetrika 2(9): 81–117.
Mutz D (2002) The consequences of cross-cutting networks for political participation. American Journal of
Political Science 46(4): 838–855.
Pew Research Center for the People and the Press (2013) July 2013 political survey methodology [online].
Available at: http://www.people-press.org
Pew Research Center for the People and the Press (2016) The modern news consumer: New attitudes and prac-
tices in the digital era [online]. Available at: http://www.journalism.org/2016/07/07/the-modern-news-
Ardèvol-Abreu et al. 15
Pinkleton BE, Austin EW and Fortman KJ (1998) Relationships of media use and political disaffection to politi-
cal efficacy and voting behavior. Journal of Broadcasting & Electronic Media 42(1): 34–49.
Prior M (2003) Any good news in soft news? The impact of soft news preference on political knowledge.
Political Communication 20(2): 149–171.
Rosenstone SJ and Hansen JM (1993) Mobilization, Participation, and Democracy in America. New York:
Macmillan Publishing Company.
Sax LJ, Gilmartin SK and Bryant AN (2003) Assessing response rates and nonresponse bias in web and paper
surveys. Research in Higher Education 44(4): 409–432.
Shah DV, Cho J, Nah S, Gotlieb MR, Hwang H, Lee N-J, et al. (2007) Campaign ads, online messaging,
and participation: Extending the communication mediation model. Journal of Communication 57(4):
Sinclair B (2012) The Social Citizen: Peer Networks and Political Behavior. Chicago, IL: University of Chicago
Son J and Lin N (2008) Social capital and civic action: A network-based approach. Social Science Research
Tewksbury D, Hals ML and Bibart A (2008) The efficacy of news browsing: The relationship of news con-
sumption style to social and political efficacy. Journalism & Mass Communication Quarterly 85(2):
Tewksbury D, Weaver AJ and Maddex BD (2001) Accidentally informed: Incidental news exposure on the
world wide web. Journalism & Mass Communication Quarterly 78(3): 533–554.
The American Association for Public Opinion Research (2011) Standard Definitions: Final Dispositions of
Case Codes and Outcome Rates for Surveys, 7th edn. Oakbrook Terrace, IL: AAPOR.
Torney-Purta J, Barber CH and Richardson WK (2004) Trust in government-related institutions and political
engagement among adolescents in six countries. Acta Politica 39(4): 380–406.
Vecchione M and Caprara GV (2009) Personality determinants of political participation: The contribution of
traits and self-efficacy beliefs. Personality and Individual Differences 46(4): 487–492.
Valenzuela S, Kim Y and Gil de Zúñiga H (2012) Social networks that matter: Exploring the role of politi-
cal discussion for online political participation. International Journal of Public Opinion Research 24(2):
Verba S and Nie NH (1972) Participation in America. New York: Harper & Row.
Watson N and Wooden M (2006) Modeling longitudinal survey response: The experience of the HILDA
survey. Paper presented at the ACSPRI social science methodology conference, Sydney, NSW,
Australia. Available at: https://old.acspri.org.au/conference2006/proceedings/streams/Modelling%20
Xenos M and Moy P (2007) Direct and differential effects of the internet on political and civic engagement.
Journal of Communication 57(4): 704–718.
Zukin C and Snyder R (1984) Passive learning: When the media environment is the message. Public Opinion
Quarterly 48(3): 629–638.
Alberto Ardèvol-Abreu (PhD, Universidad de La Laguna, 2013) is an assistant professor at the Universidad de
La Laguna (Spain), where he is also part of the ‘Laboratorio de Tecnologías de la Información y Nuevos
Análisis de Comunicación Social’ (LATINA) research group. His major research interests focus on political
communication, political participation, and new media. He is also interested in media representation of immi-
gration and ethnic minorities and its effects on public opinion.
Trevor Diehl is a research assistant and doctoral student at the Media Innovation Lab (MiLab) in the Department
of Communication (University of Vienna). He completed his undergraduate studies at the University of Nevada,
Reno, and earned an MA from the University of Texas at Austin’s School of Journalism. He previously worked
as a research assistant at the Annette Strauss Institute for Civic Life, where he studied the nature of civic partici-
pation in Texas. His research interests include digital media and journalism practice, political effects of social
media use, and the impact of technology on the public’s understanding of science.
Homero Gil de Zúñiga, PhD, is Doctor in Political Science at the Universidad Europea de Madrid and also
pursued a PhD in Mass Communication with a minor in Digital Media from the University of Wisconsin–
Madison. Currently, he holds the Medienwandel Professorship at University of Vienna, Austria, where he leads
the Media Innovation Lab (MiLab). In general, his research addresses the influence of new technologies and
digital media over people’s daily lives, as well as the effect of such use on the overall democratic process.
Internal political efficacy index: Questions 1 and 2
How much do you agree or disagree with the following statements about public life?
(from 1 = ‘strongly disagree’ to 10 = ‘strongly agree’)
1. I have a good understanding of the important political issues facing our country
2. I consider myself well qualified to participate in politics
Political discussion (weak ties) index: Questions 3–6
Political discussion (strong ties) index: Questions 7–11
How often do you talk about politics or public affairs online and offline with … (from
1 = ‘never’ to 10 = ‘all the time’)
5. Neighbours you don’t know well?
6. Co-workers you don’t know well?
7. Spouse or partner?
8. Family and relatives?
10. Neighbours you know well?
11. Co-workers you know well?
News media use index: Questions 12–20
People get news from various sources. To answer the questions, please use a scale
from 1 to 10, where 1 = ‘never’ and 10 = ‘all the time’. How often do you get news
from the following media sources?
13. National newspapers
14. Local newspapers
16. Cable news
17. Online news sites
18. Citizen journalism sites
Incidental news exposure index: Questions 21–26
Sometimes people encounter or come across news and information on current
events, public issues, or politics when they may have been using media for a purpose
other than to get the news. How often does that happen to you with the following
media? (from 1 = ‘never’ to 10 = ‘all the time’)
21. Online portals (e.g. MSN and Yahoo!)
22. Search engines
Ardèvol-Abreu et al. 17
25. Social networking sites
26. Microblogging sites (e.g. Twitter)
Strength of partisanship (folded): Question 27
27. Where would you place yourself on a scale of 1–10 (where 1 = Strong Republican, 6
= Independent, and 10 = Strong liberal)?
Trust in the media index: Questions 28–31
The following questions are about how much you trust news from various sources.
How much would you say you trust … (from 1 = ‘do not trust at all’ to 10 = ‘trust
28. News from mainstream news media (e.g. newspapers, TV)?
29. News from alternative news media (e.g. blogs, citizen journalism)?
30. News from social media sites?
31. News from news aggregators?
Political knowledge additive index: Questions 32–39
Here are some questions to which not everyone may know the answers. If there are
some you don’t know the answer to, just select ‘Don’t know’ and move on to the
next one. Please do not discuss these questions with others or look them up on the
32. What job or political office does Joe Biden currently hold? (open-ended)
33. What job or political office does John Roberts currently hold? (open-ended)
34. For how many years is a US Senator elected – that is, how many years are there in one
full term of office for a US Senator?
a) 2 b) 4 c) 6 d) 8 e) Don’t
35. On which of the following does the US Federal Government currently spend the least?
a) Foreign aid b) Medicare c) National defence d) Social
36. Do you happen to know whether the immigration bill before Congress was introduced by
a) A group of
b) A group of
c) A mix of Republican
and Democratic Senators
37. Do you happen to know what the ruling of the Supreme Court about Obamacare was?
c) Individual mandate
Appendix 1. (Continued)
38. Which organization’s documents were released by Edward Snowden?
a) FBI b) NSA c) IRS d) CIA e. Don’t
39. Recently, the United Nations (UN) and United States were in negotiations with the Syrian
government over the removal of:
c) Illicit drugs d) Al Qaeda
Discussion network size: Questions 40 and 41
40. During the past month, about how many total people have you talked to
face-to-face or over the phone about politics or public affairs (i.e. NOT via
the Internet)? (open-ended)
41. Still thinking about the people that you have talked to about politics or
public affairs during the past month, about how many total people would
you say you have talked to via the Internet, including email, chat rooms,
social networking sites and microblogging sites? (open-ended)
Political interest index: Questions 42 and 43
42. How interested are you in information about what’s going on in politics and
public affairs? (from 1 = ‘not at all’ to 10 = ‘a great deal’)
43. How closely do you pay attention to information about what’s going on in
politics and public affairs? (from 1 = ‘not at all’ to 10 = ‘a great deal’)
Socio-demographics: Questions 44–48
44. What is your gender?
45. What was your age on your most recent birthday? (open-ended)
46. What is your race or ethnicity?
a) Black or African American
b) White or Caucasian
c) Hispanic or Latino
d) Asian or Asian American
e) Native American
47. What is the highest level of education you have completed?
a) Less than high school
b) High school
c) Some college
d) Bachelors degree
e) Some graduate education
f) Professional certificate
Appendix 1. (Continued)
Ardèvol-Abreu et al. 19
g) Masters degree
h) Doctoral degree
48. Last year, what was your family’s total household income, before taxes?
a) Less than US$10,000
h) US$200,000 or more
Appendix 1. (Continued)