A New Look at Campaign Advertising and Political Engagement

Abstract

Exposure to cross-cutting versus like-minded political advertising is highly relevant in terms of deliberative democratic theory. However, few efforts have been made to shed light on the effects of such opinion-incongruent and -congruent political advertisements. By analyzing data from a representative panel survey, and hence identifying effects over time, we found that exposure to opinion-congruent advertising enhanced political participation. Opinion-congruent advertising also accelerated the timing of voting decisions when citizens were low in ideological strength. However, contrary to our expectations, exposure to opinion-incongruent political advertising had no effects on political participation and the timing of voting decisions. These findings suggest that opinion-congruent advertising is a strong mobilizer, whereas opinion-hostile advertising is a weak cross-pressure. Implications of these findings for the study of political advertising effects are discussed.

Keywords

cross-cutting advertising participation timing of voting decisions campaigns

Proponents of deliberative democracy have argued that exposure to opinion-incongruent—that is, cross-cutting—information is a cornerstone of a healthy society (see Fishkin, 1995; Thompson, 2008). The idea is that democratic citizens should be confronted with arguments that oppose and challenge their views. This, in turn, is expected to foster political tolerance, increase the awareness of oppositional opinions, and encourage political engagement. In line with this idea, the question of how individuals react to cross-cutting opinions has piqued the interest of many public opinion scholars. Most of their research is concerned with the effects of disagreement in political conversations, demonstrating that talking to people who express opposing views can pull citizens away from various forms of democratic engagement, such as political participation (e.g., Kwak, Williams, Wang, & Lee, 2005; Matthes, Morrison, & Schemer, 2010; McClurg, 2006; Mutz, 2002; Nir, 2005, 2011a, 2011b; Valenzuela, Kim, & Gil de Zúñiga, 2012; Wojcieszak, Baek, & Delli Carpini, 2010), or may slow down the timing of citizens’ voting decisions (Mutz, 2002; Nir, 2005).

In contrast to this prominent area of scholarship, there is little research on the democratic outcomes of other forms of cross-cutting exposure, such as mass-mediated cross-pressures. Even though recent studies have explored the effects of cross-cutting news exposure (see Feldman, Stroud, Bimber, & Wojcieszak, 2013; Kam, 2006; Matthes, 2012a; Nir & Druckman, 2008), no research to date has examined the effects of citizens’ exposure to cross-cutting advertising. This is surprising because one could make a strong case that political advertising is much more pervasive and inescapable compared to other potential arenas of disagreement, such as political talk or journalistic news. As there is no direct journalistic intervention, campaign managers regard political advertising as an essential campaign strategy to reach not only their own but also the opposing political camp. In the 2012 U.S. presidential election, for instance, campaign spending on political advertising has crossed the one-billion-dollar mark (Associated Press, 2012). It is therefore unsurprising that political advertising is frequently regarded as one of the most important tools for influencing the electorate. Yet the role of political advertising in exposing people to cross-cutting information has remained virtually unexplored. Furthermore, no studies have compared the mobilizing effects of cross-cutting advertisements with those of opinion-congruent ads. Whereas an abundance of research on the effects of negative political advertising on various forms of political engagement exists (e.g., Ansolabehere & Iyengar, 1995), negative advertising cannot be equated with cross-cutting versus opinion-friendly advertising. Because all political advertising is one-sided in promoting the views of a party or candidate, it is also potentially cross-cutting. However, that does not necessarily imply that it is negative.

In summary, we identify three research gaps in the existing literature on campaign advertising and political engagement. First, we have no knowledge about the consequences of cross-cutting political advertising for democratic citizenship. Second, and related to this, public opinion scholarship retains a limited understanding of the effects of opinion-friendly political advertising, that is, advertising that is in line with one’s political views. Third, almost all of the existing research on the effects of cross-cutting information in political campaigns is based on cross-sectional data, which are vulnerable to spuriousness and unable to prove any effects that unfold over the course of a campaign (see for this critique, Nir & Druckman, 2008; Rojas et al., 2005). To address these three research gaps, the present study examines the effects of opinion-congruent and -incongruent political advertising on political participation and, additionally, on the timing of citizens’ voting decisions. We use representative panel data from the German Longitudinal Election Study (GLES), which provides extensive and excellent measures for cross-cutting and opinion-friendly advertising exposure, political engagement, and people’s voting decisions.

Mechanisms Underlying the Effects of Cross-Cutting and Opinion-Friendly Information Environments on Political Engagement

Effects on Participation

In line with normative ideals expressed by democratic theorists (see Thompson, 2008), there is a tendency in the literature on political engagement to focus on the effects of cross-cutting rather than opinion-friendly information. A large quantity of research deals with political conversations as a key deliberative activity. In this body of work, cross-cutting exposure refers to the extent to which people’s discussion partners confront them with political views that are different from their own. Unlike of what would be ideal from the perspective of deliberative democratic theory, this stream of scholarship finds that cross-cutting discussion tends to depress political participation and engagement (Matthes, 2013; McClurg, 2006; Mutz, 2002; Nir, 2005, 2011a; Nir & Druckman, 2008; Valenzuela et al., 2012; Wojcieszak, 2011). Such a demobilizing effect is usually explained in one of two ways (see McClurg, 2006; Mutz, 2002; Nir, 2005): First, according to the ambivalence explanation, networks that challenge individuals’ views are likely to induce attitudinal ambivalence. This ambivalence, in turn, causes individuals to become uncertain of their own position and thus makes them less likely to take political action but to delay their voting decisions. A similar line of reasoning has been addressed in Spiral of Silence Theory (see Noelle-Neumann, 1974). Second, the social accountability explanation holds that adverse political networks discourage participation because individuals are afraid to put their social relationships at risk. Therefore, they refrain from participation in cross-cutting political talk to maintain social harmony. On the other hand, positive outcomes exist as well, as disagreement might also increase the awareness of alternate viewpoints and lead to more informed judgments (Cappella, Price, & Nir, 2002; McLeod, Sotirovic, & Holbert, 1998; Rojas, 2008; Rojas et al., 2005).

Compared to this substantial body of research, there is hardly any work on the role of opinion-friendly information in political participation. Some studies on political talk concentrate exclusively on discussion diversity (e.g., Sotirovic & McLeod, 2001; Wojcieszak et al., 2010) or discussion disagreement (e.g., Jang, 2009; Mutz, 2002; Pattie & Johnston, 2009) without explicitly modeling the effects of discussion homogeneity or agreement. For instance, Wojcieszak et al. (2010) asked their participants whether their discussions were “not diverse at all” or “very diverse” (p. 163). Along similar lines, Jang (2009) defined network disagreement as the number of discussion partners with different electoral preferences. Others, however, directly accessed the level of agreement in communication networks, yet did not focus on political participation (e.g., Huckfeldt, Ikeda, & Pappi, 2005). It should be noted that the lack of diversity or disagreement―that is, the lack of “dangerous discussion” (Eveland & Hively, 2009)―does not imply the presence of agreement. As Eveland and Hively (2009) observed, “What is missing [ . . . ] in most research is the inverse of dangerous discussion” (p. 208). Rephrased, discussion agreement and disagreement are “not polar opposites” (Eveland & Hively, 2009, p. 209; see also Valenzuela et al., 2012), and as a consequence, the influence of opinion-friendly discussion on political participation is unclear. In two notable exceptions, Eveland and Hively and Valenzuela et al. found a strong effect of opinion-friendly or “safe” discussion on political participation.

When it comes to political advertising, there is a great deal of research demonstrating that “the more one party dominated the campaign, the greater the proportion of its supporters who went to the polls” (McGhee & Sides, 2011, p. 329; see also Holbrook & McClurg, 2005; Nickerson, 2005). Moreover, research suggests that personal partisan contacting can mobilize voters as well (Arceneaux & Nickerson, 2010). However, most of this research does not distinguish carefully between ideology-friendly and ideology-hostile campaign contacts. This whole line of research deals with voter turnout, leaving general political engagement aside (as in Kim, Scheufele, & Han, 2011; Matthes, 2013; McClurg, 2006; Nir, 2005; Sotirovic & McLeod, 2001). Nevertheless, opinion-friendly advertising may, in fact, foster participation for two main reasons: As McGhee and Sides (2011) have argued, like-minded campaign contacts remind citizens of the psychological benefits and possible gains of voting. They also signal a majority opinion climate (Nickerson, 2005). In this respect, voters, as McGhee and Sides have put it, resemble sport fans, “When things are going well for their party, these partisans will be more enthused, and enthusiasm foments participation (Marcus et al., 2000). By contrast, when things are going poorly, they will become fair-weather fans, retaining their party identification but failing to act on it” (McGhee & Sides, 2011, p. 316). As far as cross-cutting advertising is concerned, it can thus be argued that, on the one hand, cross-cutting ads demobilize because they foster ambivalence in people’s thinking. However, it remains unclear whether the mechanisms underlying exposure to cross-cutting talk― where the demobilization hypothesis derives from―are similar for cross-cutting ads. On the other hand, one may argue that such ads may be seen as having a manipulative intent because they come from an opposing party. The activation of such persuasion knowledge however requires cognitive resources that might not be available when people are exposed to these ads.

Effects on the Timing of Voting Decisions

The timing of voting decisions is often regarded as one of the key mediating variables for campaign effects (Chaffee & Rimal, 1996; Fournier, Nadeau, Blais, Gidengil, & Nevitte, 2004; Lazarsfeld, Berelson, & Gaudet, 1944). Although delaying one’s voting decisions is not a direct indicator of low political participation, it signals uncertainty or lack of determination in taking political action. As Mutz (2002) argued, “It seems inevitable that the later one makes up his or her mind, the less time there is for actively promoting one’s political preferences” (p. 845). Based on this reasoning, many authors have treated both participation and decision timing as equally important outcomes of political disagreement (Matthes, 2012a; Mutz, 2002; Nir, 2005; Nir & Druckman, 2008). Understanding the processes that predict the timing of voting decisions is thus of fundamental importance to political communication scholarship (e.g., Chaffee & Rimal, 1996; Gronke & Krantz Toffey, 2008; Nir, 2005; Nir & Druckman, 2008). However, despite its relevance (see Chaffee & Rimal, 1996), few attempts have been made to unravel the mechanisms impacting decision-timing (e.g., Fournier et al., 2004; Matthes, 2012a; Nir, 2005; Nir & Druckman, 2008). When it comes to opinion-friendly information environments, research by Dilliplane (2011) determined that exposure to partisan news accelerates the timing of voting decisions. In addition, some studies have explored the role of cross-cutting exposure: Findings by Mutz, for instance, suggest that “exposure to dissonant views encourages people to make up their minds later in the campaign,” which “limits their opportunities to participate in an actively partisan fashion during the campaign” (p. 845). In another study, Nir and Druckman (2008) demonstrated that voters delay their decisions when they are both ambivalent about an issue and confronted with a cross-cutting news environment. Along similar lines, Matthes (2012a) found that exposure to counter-attitudinal news information delayed voting decisions for highly ambivalent individuals, but not for people with strong attitudes. The reason for this demobilizing effect is that cross-cutting information attacks people’s existing beliefs; it forces them to rethink what they already know, reflect upon it, and potentially revise their original opinions.

Exposure to Cross-Cutting and Opinion-Friendly Political Advertising

As we argue throughout this article, the demobilizing effect found in research on political discussions may also be expected for cross-cutting political advertising. Moreover, based on prior research, one would expect opinion-friendly advertising to foster participatory outcomes. However, before testing these assumptions, we need to ask whether citizens are frequently exposed to opinion-friendly and opinion-hostile ads. As explained above, little research exists on the consequences of cross-cutting political advertising. This is surprising as political advertising may be a major source of cross-cutting information. Nir and Druckman (2008) have generally argued that insights from interpersonal cross-pressures can “be applied even more appropriately to mediated contexts [ . . . ] in which individuals have a much higher chance of exposure to experiences and points of view unlike their own” (p. 329). In fact, campaign managers invest great efforts in making sure their advertising messages are received by both followers and opponents. As news “media became less predictable politically” (Chaffee, Saphir, Graf, Sandvig, & Hahn, 2001, p. 263), citizens might not always be able to choose information outlets according to their political predispositions. And even if they do, they might not be able to select those advertising messages that are in line with their political predispositions. It is rather difficult to know beforehand when and where one will encounter political advertising of one’s preferred or opposing party; accordingly, active (selective) avoidance is unlikely to occur.

Thus, exposure to ads by both one’s preferred and opposing party is inevitable. This is especially true for the public-service news media of many European countries. In Germany, for instance, prior to an election, public-service television emits the ads of all political parties during evening prime time. Furthermore, citizens find leaflets of various parties in their post boxes, or they encounter costly ads in newspapers or on the radio. Parties place posters in public areas and send messages via email, and campaigners approach people on the streets offering flyers and brochures. In addition to that, political ads have long gone online targeting younger voters by placing political ads on youtube, facebook, or even in online games (see CU News Corps and Buccholz, 2012).

Advertising reception is a comparatively passive process where citizens have, as one could put it, nothing to lose. As Mutz and Martin (2001) have argued, people may circumvent cross-cutting political discussions to avoid normative social pressures. Conversely, the “same people may be willing to expose themselves to media presentations [ . . . ] precisely because there is no social interaction” (p. 99). In other words, when watching or reading advertising messages, people need not react to opposing views, as they do not need to justify themselves, argue for their position, or feel attacked. We conclude that both opinion-congruent and -incongruent political advertising is hard to avoid during the course of a campaign, and the opportunities for selective exposure are limited.

Hypotheses and Research Questions

As should be apparent, the effects of cross-cutting and opinion-friendly political advertising on citizen engagement are a potentially fertile yet neglected area in political communication research. In the present study, we have examined these effects for two prominent outcomes that were suggested in prior research: political participation (see Matthes, 2013; McClurg, 2006; Sotirovic and McLeod, 2001), and the timing of voting decisions (Matthes, 2012a; Mutz, 2002; Nir, 2005; Nir & Druckman, 2008). Both outcomes indicate people’s engagement in the political process.

It can be theorized that opinion-hostile (cross-cutting) advertising erodes engagement: As explained by Mutz (2002) and outlined in detail above, exposure to such campaign information might foster ambivalence in people’s thinking, thereby impeding their political participation and delaying voting decisions. On the other hand, the social accountability explanation (Mutz, 2002) should be less relevant in the context of political advertising, because social relationships are not at stake (as they are in political conversations with conflicting opinions).

Opinion-friendly political advertising, in contrast, should foster political participation and accelerate the timing of voting decisions. There are two possible explanations for that. First, there are grounds to argue that opinion-friendly advertising reminds people of the benefits of their party alignment (McGhee & Sides, 2011), thereby prompting political action and speeding up their decision making. Second, as predicted by Spiral of Silence Theory (Noelle-Neumann, 1974), such advertising can be processed as a cue signaling that the majority of citizens hold similar views. As a result, “partisan campaigns may cause some voters to feel that they belong to a larger movement” (Nickerson, 2005, p. 11). Such an indication of a majority opinion climate may strengthen one’s viewpoint, or be a signal for normative pressure to do the right thing.

These considerations lead us to the following four hypotheses:

Hypothesis 1 (H1): Exposure to cross-cutting political advertising has a negative effect on political participation.

Hypothesis 2 (H2): Exposure to opinion-friendly political advertising has a positive effect on political participation.

Hypothesis 3 (H3): Exposure to cross-cutting political advertising delays the timing of voting decisions.

Hypothesis 4 (H4): Exposure to opinion-friendly political advertising accelerates the timing of voting decisions.

As outlined above, some indications in the literature suggest that the effects of hostile opinion environments on political engagement may depend on a person’s ideological strength (Matthes, 2012a; Nir, 2005; Nir & Druckman, 2008). People with strong convictions are certain that their attitudes are correct, and they are more resistant to persuasion (e.g., Fazio & Zanna, 1978). In contrast, people with ambivalent attitudes may have a preference, but they are not completely convinced. It follows that cross-pressures may only dampen participatory outcomes when people hold their ideological views with low or moderate conviction. A similar effect could be theorized for opinion-friendly advertising. At this point in time, research is inconclusive as to whether ideological strength moderates the effects of cross-pressures. Some studies find a moderating effect (Matthes, 2012a; Nir, 2005; Nir & Druckman, 2008; Wojcieszak, 2011), whereas others do not (McClurg, 2006; Mutz, 2002). For friendly opinion environments, no studies have looked at the moderating role of ideological strength so far. Therefore, we explore the role of ideological convictions in four research questions:

Research Question 1: Does ideological strength moderate the effects of cross-cutting political advertising on political participation?

Research Question 2: Does ideological strength moderate the effects of opinion-friendly political advertising on political participation?

Research Question 3: Does ideological strength moderate the effects of cross-cutting political advertising on the timing of voting decisions?

Research Question 4: Does ideological strength moderate the effects of opinion-friendly political advertising on the timing of voting decisions?

Method

Data

The data were obtained from the short-term campaign panel of the GLES (Component 3).¹ This data set comprises a sample of German respondents aged 18 and older who participated in an online-based survey over a period of 12 weeks preceding the 2009 German parliamentary elections. A total of 3,771 respondents were interviewed in the first wave of the panel-survey (age 18-104, M = 41.91, SD = 14.71, 50% male), with 1,462 people completing all seven waves of the panel. Participants were acquired from an online-access panel and invited via email to answer the survey; response rates for participation varied between 41.85% (Wave 1) and 85.19% (Wave 3; Rattinger, Roßteutscher, Schmitt-Beck, & Weßels, 2012). Because participation was only measured in Waves 3 and 7, we included responses from the third (n = 2.455) and seventh (n = 1.462) waves of the panel (08/07/09-08/17/09 and 09/29/09-10/07/09, respectively), with Wave 7 being conducted after the election. Panel attrition for Wave 7 was 31.56%. The online-based data cannot be considered as representative of the entire population. More details about the panel and sampling can be found in Rattinger et al. (2012).

Measures

Dependent variables

Participation is a concept formed by causal (formative) indicators, not by effects or reflective indicators. Formative scales are used when a construct is defined as a total score across all the items. That is, there is no underlying construct that explains the correlations among those indicators, and each indicator represents an independent dimension in its own right (see for this causal indicator approach, McClurg, 2006; Rojas, 2010; Sotirovic & McLeod, 2001). Following prior research (Kim et al., 2011; Matthes, 2013; McClurg, 2006; Nir, 2005; Sotirovic & McLeod, 2001), we measured political participation by forming a summary index generated from questions in both Waves 3 and 7, asking participants whether or not (1 = yes, 0 = no) they took part in various participatory activities. Respondents were asked whether they (1) signed a petition, (2) supported a politician/party in the campaign, (3) tried to gain support from a politician, (4) participated in a citizen’s initiative, (5) participated in an authorized demonstration, or (6) participated in a public protest such as a traffic blockade (M = .62, SD = 0.81 in Wave 3; M = .46, SD = 0.74 in Wave 7). These activities can be considered to be driven by party positions and are thus politically motivated, even though they do not compare to the campaign-related activities employed in studies on political participation in the U.S. context. The German Longitudinal Election Study was conducted during the major phase of the election, for example, when participating in a demonstration was almost certainly connected to campaign activities. Therefore, even though these items may not refer to participation with respect to a particular party, they do refer to activities that can be considered political.

The second dependent variable, the timing of voting decisions, was operationalized in Wave 7 with two single items asking when respondents had made the decision to vote or to abstain from voting all together. Both the voters’ and non-voters’ decisions were transferred into a single variable stating the timing (slowness) of the voting decision (M = 2.50, SD = 1.34).² This question is similar to those used in previous research (see Fournier et al., 2004; Matthes, 2012a; Nir and Druckman, 2008). In fact, there is strong evidence that such standard items are reliable and valid (Fournier et al., 2004). Participants were asked precisely when they had decided either to vote for a specific party or to renounce their voting right. It can be argued that both voting and abstention are quite stable decisions (Górecki, 2013) and indicate a strong position. If someone decides not to vote, this indicates a clear position against the parties that are running for the election. The question of whether voters and non-voters were influenced in the acceleration of this decision due to political advertising is thus very important. A total of 34.6% of the respondents indicated that their decision was made a long time ago, 19% said a couple of months ago, 20% some weeks before voting day, 14.6% some days before voting day, and 11.8% said that the decision was made on the voting day itself. When it comes to the actual numbers, n = 251 people indicated the timing of their non-voting decision. In contrast, n = 2,283 people indicated the timing of their decision to vote.

Independent variables

To capture exposure to hostile and friendly political advertising, an extensive number of variables were summarized for each individual respondent. In both Waves 3 and 7, the survey asked participants whether they had been exposed to different types of ads. Respondents were asked to name (a) the type of advertising they were exposed to and (b) the political parties responsible for this advertising message. Individuals were thus asked whether they had been exposed to (1) TV spots and (2) radio ads, (3) emails and short messages, (4) leaflets and brochures, (5) ads in newspapers and magazines as well as (6) political posters, and whether they had received (7) telephone calls from campaigners or (8) visits at home.³ If they affirmed the respective exposure, respondents were asked in a second step from which parties they had received the advertisements mentioned, naming all main political parties in the German parliament (CDU/CSU, SPD, FDP, the Greens, the Left), as well as “others.” For instance, respondents could state if they had received leaflets and brochures from the CDU/CSU, SPD, FDP, and the Greens, but not from the Left, or if they had been exposed to radio ads by the CDU/CSU and FDP, but not by all others. Based on this rich and unique list of variables, we created a list of advertising exposures from all political parties (in total: the number of advertisements multiplied with the number of parties) for each respondent. A summary index was then formed giving the total number of advertisements each respondent had received from every single party.

Next, affiliations of the different parties in possible coalitions, as well as respondents’ own political orientation, were considered to determine whether the ads people were exposed to in both waves were to be understood as being either opinion-friendly or opinion-hostile (i.e., cross-cutting). As coalitions in the German parliament are traditionally formed on the basis of political affiliations, this categorization resulted in two main political “camps” and two minor “camps”: The main camps were (a) CDU/CSU and FDP as well as (b) SPD and the Greens. As for the minor camps, we had (c) the Left, which although having grown to become a considerable electoral force, has not yet gained entrance into government at the national level. Not being ideologically closely affiliated with any of the other parties, the Left thus forms a third political “camp.” Finally, as participants in the survey were also given the opportunity to state “other” parties, this open category formed a fourth camp, (d) other.

The logic then went as follows: Based on a standard party identification measure (“Do you lean toward a party? If yes, which one?”), we categorized both respondents and ad exposures as belonging to one of the four camps. As a consequence, we were then able to say that Person X belongs to Camp (1) and was exposed to a certain number of ads by Camps (1) and (3), and so forth. Such a categorization reflects the lines of conflict in multi-party elections. Concededly, respondents identifying with the CDU would also feel positively toward political ads by the FDP (because both parties ran as coalition partners), whereas they would perceive advertisements from all other parties as hostile. Respondents who identified with another party (“other”) would perceive all of the big five parties’ advertising as hostile. This makes sense considering that parties categorized as “other” are in clear opposition to all five major parties.

Following this scheme, the total amount of both friendly and hostile political ads was calculated for each single person (depending on her/his party affiliation), allowing us to compute a friendly and hostile score of political advertising for Wave 3 (friendly: M = 6.16, SD = 1.57; hostile: M = 9.82, SD = 2.64) and Wave 7 (friendly: M = 4.49, SD = 2.84; hostile: M = 7.37, SD = 4.99). These indices indicate the sum of all hostile or friendly exposures. Naturally, receiving hostile advertisements was more likely than receiving friendly ads, because the “own” political camp had to compete with three others.

Controls

To avoid underspecified regression models, a number of statistical controls were included that had proven to be of substantial importance in prior research. In addition to the standard social-demographic questions (gender 54.4% male, age 18-80 years, M = 42.37, SD = 14.11, and education, 44.4% A levels), we included summary indices of politics-specific use of different TV news programs (M = 12.96, SD = 6.22) and various newspapers (M = 12.12, SD = 5.12) as well as internet use (M = 3.33, SD = 2.32) from Wave 3 (“On how many days during the last week did you use TV news/newspapers/the internet to gain information about politics?” see Nir, 2005; Nir & Druckman, 2008; Rojas, 2008; Sotirovic & McLeod, 2001). Moreover, we included measurements of political interest (see Nir & Druckman, 2008; Rojas, 2010; “How much are you interested in politics in general?” 1 = very much, 5 = not at all, M = 3.63, SD = 0.84) and strength of party identification (see McClurg, 2006; Nir, 2005; Nir & Druckman, 2008; Pattie & Johnston, 2009; “How strongly do you—in general—identify with your party?” 1 = very weakly, 5 = very strongly, M = 3.86, SD = 0.70). Participants were also asked about their dissatisfaction with the way democracy works in Germany (“How satisfied or unsatisfied are you—all in all—with the German democracy?” 1 = very satisfied, 5 = not at all satisfied, M = 2.6, SD = 0.81). This variable was included because dissatisfaction can be theorized to be a powerful predictor of engagement (see Nir, 2012, for a similar argument). As another indispensable standard measure (McClurg, 2006; Nir, 2005; Rojas et al., 2005), political ideology was assessed using a left to right scale (“Where would you put yourself on a left to right range?” 1 = left, 11 = right, M = 5.54, SD = 2.47). As an obvious predictor of participation, interest in the outcome of the election was incorporated from pre-election Wave 3 (“How important is the election outcome for you?” 1 = not important at all, 5 = very important, M = 4.09, SD = 0.81). Finally, we included political efficacy as one of the key predictors of participation (e.g., Pattie & Johnston, 2009). Efficacy was measured using three standard items (“People like me do not have any influence on the government (recoded),” “Citizens have hardly any possibilities to influence politics (recoded),” “Politicians care about us normal people,” 1 = strongly disagree, 5 = strongly agree, M = 2.69, SD = 0.77, α = .66). If not otherwise stated, all of these variables were collected during Wave 7 to exert the strongest impact as controls.

Data Analysis

Data were analyzed by running an autoregressive panel path analysis using the MPlus program. As one of the dependent variables, political participation, was a count variable that showed signs of over-dispersion, ordinary least-squares (OLS) regression was not appropriate (see Gardner, Mulvey, and Shaw, 1995). Thus, negative binominal regression was used (see Gardner et al., 1995) for this variable. In contrast to crossed-lagged correlational analyses, all variables were entered simultaneously in one path model (see Figure 1). We have four dependent variables measured during Wave 7, the autoregressive effects of the variables in Wave 3, and the controls. To estimate the interaction effects, we ran additional models with conditional effects (see Hayes, Glynn, and Huge, 2012).

Figure 1.

Conceptual autoregressive path model.

Results

Theoretical Model

Our panel design has clear advantages over the cross-sectional designs employed in most prior research. Because of the complexity of the data, we summarize the four hypotheses in the autoregressive panel model depicted in Figure 1. When it comes to the effects of opinion-hostile and opinion-friendly ads, we have two path coefficients for each outcome variable. First, the cross-lagged effects from advertising exposure at t₁ on participation or decision timing at t₂ were used to test our hypotheses. These are marked in bold in Figure 1. The major premises for the interpretation of an effect path are the temporal order of the variables and the control of other influencing variables. The autoregressive effect of participation at t₁ is controlled, and there is a temporal order, which is one (albeit not sufficient) precondition for causal effects. Second, the cross-sectional effects of advertising exposure on participation and decision timing are of no interest as no temporal order exists.

Effects on Participation

The cross-lagged and correlational effects are shown in Table 1. We stated in our first hypothesis (H1) that opinion-hostile ads have a negative effect on political participation. Although the effect was negative and points in the expected direction (b = −.04, ns), it was not significant by conventional criteria (p = .17). Thus, H1 was not supported. However, confirming our second hypothesis, there was a significant positive effect of friendly advertising at t₁ on political participation at t₂ (b = .09, p < .05). In the next step, we tested whether the paths of advertising exposure on participation were moderated by ideological strength (Research Questions 1 and 2). We found no such interaction, either for exposure to friendly (b = −.06, ns) or to hostile ads (b = .00, ns). Beside this analysis of cross-lagged effects, we found significant effects of left-right political ideology (b = −.04, p < .05) and dissatisfaction with democracy (b = .15, p < .05). As can be expected, the autoregressive effect of participation measured at Wave 3 was very strong and highly significant (b = .57, p < .001).

Table 1.

Unstandardized Path Coefficients Explaining Timing of Voting Decisions, Political Participation, Exposure to Friendly Ads, and Exposure to Hostile Ads at Wave 7.

	Participation t₂	Exposure to friendly ads t₂	Exposure to hostile ads t₂	Timing of voting decision (slowness)
Gender	.069 (.104)	.076 (.199)	.446 (.369)	.186 (.096)
Age	−.002 (.004)	.032 (.007)***	.055 (.014)***	−.014 (.004)***
Education	.021 (.106)	.059 (.20)	−.422 (.371)	−.038 (.096)
Print media use	−.005 (.009)	−.003 (.021)	−.061 (.038)	−.008 (.010)
TV News use	−.008 (.009)	.031 (.018)	.093 (.033)**	.003 (.009)
Internet use	.032 (.026)	−.083 (.049)	.010 (.092)	−.017 (.024)
Political interest	.115 (.08)	.309 (.146)*	.741 (.271)**	.093 (.071)
Strength of party ideology	.068 (.069)	−.258 (.14)	−.450 (.26)	−.584 (.068)***
Left-right ideology	−.044 (.021)*	.125 (.04)**	−.059 (.075)	.023 (.02)
Efficacy	.048 (.075)	.324 (.137)*	−.038 (.254)	−.111 (.066)
Dissatisfaction with democracy	.146 (.067)*	−.612 (.13)***	−.235 (.242)	.078 (.064)
Interest in election outcome	−.039 (.075)	.004 (.133)	−.149 (.246)	−.129 (.064)*
Participation t₁	.568 (.047)***	.107 (.116)	.405 (.215)	−.019 (.056)
Exposure to friendly ads t₁	.085 (.041)*	.250 (.093)**	−.227 (.173)	.008 (.045)
Exposure to hostile ads t₁	−.039 (.029)	.101 (.065)	.549 (.121)***	−.012 (.032)
Friendly ads t₂	.03 (.026)			.016 (.024)
Hostile ads t₂	.003 (.014)			.004 (.013)

Interaction friendly ads × Strength of party ideology	−.064 (.041)			.102 (.05)*
Interaction hostile ads × Strength of party ideology	−.002 (.032)			−.020 (.037)

Note: Values represent parameter estimates; numbers in parentheses show standard errors.

p < .05. **p < .01. ***p < .001.

Effects on the Timing of Voting Decisions

Contrary to H3 and H4, we also found no effects of exposure to opinion-friendly (b = .01, ns) or opinion-hostile (b = −.01, ns) advertising on the timing (slowness) of voting decisions. The impact of hostile ads on decision timing was not moderated by ideological strength either (Research Question 3; b = −.02, ns). However, in answering Research Question 4, there was a significant interaction between ideological strength and exposure to opinion-friendly ads (b = .10, p < .05). The positive sign of this interaction signals that opinion-friendly ads may accelerate (slow down) voting decisions for decreasing (increasing) levels of ideological strength. To further interpret this interaction, it was probed using the modprobe macro by Hayes and Matthes (2009) that applies the Johnson–Neyman technique. This technique identifies regions in the range of the moderator variable where the effect of the focal predictor on the outcome variable is either statistically significant or not significant. That is, we estimated the effects of opinion-friendly ads for all levels of ideological strength. For low levels of ideological strength (below 1.4), the effect of opinion-friendly ads on slowness of voting decisions was significantly negative (from b = −.22 to b = −.18, p < .05). For levels below 2.2, the effect was marginally significantly negative (from b = −.17 to b = −.12, p < .09). That is, when people were uncertain in their party conviction, opinion-friendly advertising accelerated their voting decisions. However, for people with moderate and high levels of ideological strength, there were no effects whatsoever on decision timing.

When it comes to the other predictors in Table 1, we found that older people arrived at their voting decisions more quickly (b = −.01, p < .001), as did people high in ideological strength (b = −.59, p < .001) and those with a strong interest in the outcome of this election (b = −.13, p < .05). The effects of gender (b = .19, p = .05) and efficacy (b = −.11, p = .09) were marginally significant suggesting that there is a tendency for female voters and those with low efficacy to be slower in arriving at their voting decisions.⁴

Discussion

The aim of this article was to offer a new look at campaign advertising and political engagement. Starting with the observation that exposure to cross-cutting versus like-minded political advertising is highly relevant in terms of deliberative democratic theory, we bemoaned a lack of research that relates these important phenomena to political participation.

In answering the call by Nir and Druckman (2008) who encouraged “future work to test findings in alternative settings with distinct designs such as employing panel data” (p. 339; similarly Rojas et al., 2005), we used the GLES to test whether opinion-hostile or opinion-friendly ads thwart or foster engagement. The GLES short-term campaign panel offers—to our knowledge—the most extensive survey measures of advertising exposures that are available in public opinion research. We found―for the first time in the existing research―that exposure to opinion-friendly advertisements exerts a significantly positive effect on people’s political participation activities. When interpreting this effect, it is important to note that we employed an autoregressive model that is very conservative when it comes to the explanatory power of the independent variables. As can be seen in all of our analyses, the autoregressive effects of a former state (t₁) on a subsequent state of the same variable (t₂) usually explain most of the variance, leaving little to be explained by the independent variables. Furthermore, the cross-lagged effect of the independent variable also competes with all cross-sectional effects at t₂.

We did not interpret these cross-sectional effects at t₂ because we cannot claim exposure to drive participation at a single point in time. In contrast, cross-lagged effects allowed us to model a temporal order, which is superior to modeling simple cross-sectional relationships. In addition to that, we have theorized that exposure to ads at a given time point will affect participation in the future. This idea is expressed by cross-lagged effects (see Burkholder and Harlow, 2003; Jöreskog, 1979; Pitts, West, and Tein, 1996).

Contrary to our expectations, however, opinion-hostile ads had no effects whatsoever. This finding sharply contrasts with most prior research on the effects of hostile opinion environments on political participation (e.g., Mutz, 2002; Nir, 2005). The most likely explanation is that hostile opinion environments differ in the degree to which people can discount the cross-cutting information they receive. Unlike interpersonal conversations, people need not react to opposing views when they are exposed to a political ad; they do not need to defend themselves, fight for their position, or feel criticized. They can simply ignore what does not fit with their political belief system. This reasoning has a strong grounding in research on attitude formation and decision making (see Lord, Ross, & Lepper, 1979). To maintain cognitive balance, opinion-hostile information may simply be rejected or discounted. In addition, people might put less trust in advertising messages that run counter to their opinions (cf. Tsfati, 2004). As Meijnders et al. (2009, p. 1118) have observed, “Message receivers use their assessment of the message source’s similarity as a basis for their trust judgment.” That is, compared to cross-cutting talk, cross-cutting ads may be seen as having a strong manipulative intent because they stem directly from the opposing party. Along similar lines, citizens also do not need to be afraid of risking their social relationships, as the social accountability explanation predicts (see Mutz, 2002). In short, exposure to opinion-hostile ads may simply be a weak form of cross-pressure.

Yet we found that opinion-friendly ads accelerated voting decisions for citizens low in ideological strength. So why did ideology strength moderate the effect on decision timing, but not the one on participation? We believe that participation is a high-effort form of engagement requiring a lot of energy and resources. Opinion-congruent ads seem to be helpful in mobilizing these resources on a regular basis. On the other hand, the timing of voting decisions does not require much effort. People with strong ideological convictions make up their minds early on in the campaign, no matter how much friendly advertising they are exposed to in the later course of the election contest (Fournier et al., 2004; Nir & Druckman, 2008). As Gronke and Krantz Toffey (2008, pp. 506-507) explain, “A strongly partisan and ideological individual can overcome the uncertainty threshold by simply referring to the cue provided by a candidate’s party identification.” In other words, strong partisans do not need to see reinforcing, opinion-friendly ads to come to their final voting decision. However, while they have already made up their minds, they might still need ideological support and reinforcement to participate in activities besides voting itself, that is, to actively do something.

The data also revealed some other interesting patterns: Citizens who were dissatisfied with democracy were more likely to participate, arguably because dissatisfaction is a strong motivating force for action (see Klandermans, 2010). Moreover, the finding that age, ideological strength, and campaign interest accelerated voting decisions validates insights gained from prior studies (e.g., Matthes, 2012a; Nir, 2005).

Limitations

There are, of course, some noteworthy limitations. To begin with, the participation activities used in the German Longitudinal Election Study cannot be compared to the campaign-related activities that are employed in most of the election research in the United States. However, it is important to stress that the activities we used are clearly driven by party positions. The German Longitudinal Election Study was conducted during the hot phase of the campaign. Signing a petition during that time was almost certainly connected to campaign activities. Therefore, even though these items may not refer to participation with respect to a particular party, they do refer to activities that can be considered political. Even though we cannot comment on the political activities of a particular party, we believe that the effects we identified may have been even stronger had we used a more straightforward operationalization of party-related participation activities (e.g., displaying a campaign sticker). Because of these items, and as this is the first study on the effects of cross-cutting and opinion-friendly ads, our findings need to be confirmed by additional research.

Furthermore, the ad exposure recall items used in the GLES are prone to measurement errors. We can, however, offer several arguments in defense of these measures. First, we have controlled a number of important variables that may explain biases in exposure recall, such as political ideology, ideology strength, education, or political interest. Second, there is no evidence in the research literature that opinion-friendly ads are more or less likely to be recalled than opinion-hostile ads. In fact, a meta-analysis by Eagly, Chen, Chaiken, and Shaw-Barnes (1999) suggests that there is no difference in memory between congenial and uncongenial information. That means that if a measurement error for ad recall emerges, it applies equally to both opinion-friendly and opinion-hostile ads. Finally, almost the entire body of literature on cross-cutting information has relied on recall measures (e.g., Jang, 2009; Matthes, 2013; Mutz, 2002; Pattie & Johnston, 2009; Sotirovic & McLeod, 2001; Wojcieszak et al., 2010).

Even though the GLES provides the best and most extensive information available for German national elections, it does draw its sample from an online panel. Therefore, our sample is not strictly representative of the German population, and limitations to the generalization of our findings are at hand (see Rattinger et al., 2012).

We did not implement an experimental design to test for the direct causal effects assumed in our hypotheses. Possible intervening variables that we did not account for might therefore interfere with the relationships we proposed. At least, we found no evidence for reverse causation (i.e., ad exposure at t₂ impacts participation at t₁). In addition to that, no effect of participation at t₁ on ad exposure at t₂ emerged (see table 1). However, the use of panel data allowed us to check for cross-lagged effects while controlling for various possible control variables at the same time.⁵ It is also important to note that Waves 3 and 7 of the survey were carried out with a time lag of only 2 months. Although this time lag may be considered as somewhat arbitrary, it is typical for media effects research (see Watt, Mazza, & Snyder, 1993). Therefore, we think it safe to assume that our effects would stand the test of an experimental survey or a controlled laboratory experiment.

It is also worth noting that the panel data we used are limited in so far as political participation was only measured in two of the waves. This meant that we could not investigate the dynamic effects over the full course of this presidential campaign. Related to that, decision timing was only measured once in the post-election wave, so the autoregressive effect could not be controlled.

Finally, we cannot account for the actual content of the political advertisements that respondents were exposed to. An extensive content analysis covering the campaigns of all major and minor German parties during the whole period would be necessary to provide such information. This, unfortunately, is beyond the scope of the present article. Therefore, no statements are possible about the negativity or positivity of either friendly or hostile advertisements or their potential effects

Practical Implications

Despite these limitations, our findings have implications for campaigners and campaign managers. Quite simply, campaigners can profit from reaching their own camp with political ads. There is no doubt that mobilizing partisans with political ads is one precondition for electoral success. Furthermore, because opinion-friendly ads can accelerate voting decisions, they may also protect a party’s camp against attacks from other camps. Even more importantly, opinion-hostile ads, as our findings suggest, do no harm by demobilizing the electorate. This seems to be an important difference compared to cross-cutting talk. Cross-cutting ads might evoke other reactions and effects mechanism than cross-cutting talks in political conversations.

Future Research

Now that the effects of opinion-friendly and -hostile political ads have been documented, a number of follow-up questions can be addressed. First and foremost, the exact process underlying the effect of opinion-congruent ads remains an important research issue. We can offer three possible explanations. First, opinion-friendly ads signal to citizens a majority opinion climate which, in turn, prompts reinforcement and motivation (Noelle-Neumann, 1974). Second, one could argue that exposure to like-minded ads strengthens group identity, which is arguably a very strong motivator for action (van Zomeren, Postmes, & Spears, 2008). Third, there are grounds to theorize that positive ads spark positive emotions such as enthusiasm that mediate the relationship between exposure and participation (Marcus & Mackuen, 1993). Clearly, further research is needed to empirically test these potential mediating mechanisms.

Related to that, we would also encourage future research to test the impact of opinion-congruent and -incongruent political advertisements in an experimental design. Experiments would also allow for a straight test of mediating mechanisms. Future studies should also take a closer look at ad content, potentially controlling for the negativity or positivity of political advertisements. The relative strength of several cross-pressures (i.e., talk, news, or ads) should be compared in future research as well.

Finally, this study focused on the distinction between opinion-congruent and opinion-incongruent ads. However, exposure to opinion-incongruent information is only one facet of cross-pressures (Nir, 2005; Eveland & Hively, 2009; Eveland, Morey, & Hutchens, 2011). In fact, Lazarsfeld et al. (1944) distinguished two forms of cross-pressures: counter-attitudinal information (i.e., disagreement) and heterogeneity. Heterogeneity would suggest that people are exposed to a diverse set of ads, that is, seeing both opinion-congruent and incongruent ads. In the present study, we looked at the effects of opinion-congruent ads while controlling exposure to opinion-incongruent ads, and vice versa, as we were interested in the unique effects of both types of ads. However, it may be worthwhile for future research to examine the effects of ad-heterogeneity, that is, the question of how opinion-congruent and -incongruent ads interact in their effects on various democratic outcomes.

Footnotes

Declaration of Conflicting Interests

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

Funding

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

Notes

Author Biographies

Jörg Matthes (PhD, University of Zurich) is professor of communication science at the Department of Communication, University of Vienna, Austria, where he chairs the division of advertising research and media effects. His research focuses on advertising effects, the process of public opinion formation, news framing, persuasion, and empirical methods.

Franziska Marquart (MA, University of Erfurt) is a PhD student and junior researcher at the Department of Communication, University of Vienna. Her research focuses on right-wing populism, political advertising and visual communication.

References

Ansolabehere

Iyengar

(1995). Going negative: How campaign advertising shrinks and polarizes the electorate. New York, NY: Free Press.

Arceneaux

Nickerson

D. W.

(2010). Comparing negative and positive campaign messages: Evidence from two field experiments. American Politics Research, 38, 54-83. doi:10.1177/1532673X09331613

Associated Press. (2012, October 15). $1.1 BILLION in political ads bombard America during presidential campaign but do they even work? DailyMail. Retrieved from http:// http://www.dailymail.co.uk/news/article-2218070/U-S-Presedential-election-2012–1-1-BILLION-political-ads-bombard-America-presidential-campaign.html

Burkholder

Harlow

(2003). An illustration of a longitudinal cross-lagged model design for larger structural equation models. Structural Equation Modeling, 10, 465-486. doi:10.1207/S15328007SEM1003_8

Cappella

J. N.

Price

Nir

(2002). Argument repertoire as a reliable and valid measure of opinion quality: Electronic dialogue during campaign 2000. Political Communication, 19, 73-93. doi:10.1080/105846002317246498

Chaffee

Rimal

R. N.

(1996). Time of vote decision and openness to persuasion. In Mutz

Sniderman

Brody

(Eds.), Political persuasion and attitude change (pp. 267-291). Ann Arbor: University of Michigan Press.

Chaffee

Saphir

M. N.

Graf

Sandvig

Hahn

K. S.

(2001). Attention to counter-attitudinal messages in a state election campaign. Political Communication, 18, 247-272. doi:10.1080/10584600152400338

CU News Corps& Buccholz, K. (2012, October 27). In swing state Colorado, even online there’s no escape from political ads. Retrieved from http://www.kunc.org/post/swing-state-colorado-even-online-theres-no-escape-political-ads

Dilliplane

(2011). All the news you want to hear: The impact of partisan news exposure on political participation. Public Opinion Quarterly, 75, 287-316. doi:10.1093/poq/nfr006

10.

Eagly

A. H.

Chen

Chaiken

Shaw-Barnes

(1999). The impact of attitudes on memory: An affair to remember. Psychological Bulletin, 125, 64-89. doi:10.1037//0033-2909.125.1.64

11.

Eveland

W. P.

Hively

M. H.

(2009). Political discussion frequency, network size, and “heterogeneity” of discussion as predictors of political knowledge and participation. Journal of Communication, 59, 205-224. doi:10.1111/j.1460-2466.2009.01412.x

12.

Eveland

W. P.

Jr. Morey

A. C.

Hutchens

M. J.

(2011). Beyond deliberation: New directions for the study of informal political conversation from a communication perspective. Journal of Communication, 61, 1082-1103. doi:10.1111/j.1460-2466.2011.01598.x

13.

Fazio

R. H.

Zanna

M. P.

(1978). Attitudinal qualities relating to the strength of the attitude-behavior relationship. Journal of Experimental Social Psychology, 14, 398-408. doi:10.1016/0022-1031(78)90035-5

14.

Feldman

Stroud

N. J.

Bimber

Wojcieszak

(2013). Assessing selective exposure in experiments: The implications of different methodological choices. Communication Methods and Measures, 7(3-4), 172-194. doi:10.1080/19312458.2013.813923

15.

Fishkin

(1995). The voice of the people: Public opinion and democracy. New Haven, CT: Yale University Press.

16.

Fournier

Nadeau

Blais

Gidengil

Nevitte

(2004). Time-of-voting decision and susceptibility to campaign effects. Electoral Studies, 23, 661-681. doi:10.1016/j.electstud.2003.09.001

17.

Gardner

Mulvey

E. P.

Shaw

E. C.

(1995). Regression analyses of counts and rates: Poisson, overdispersed poisson, and negative binomial models. Psychological Bulletin, 118, 392-404. doi:10.1037/0033-2909.118.3.392

18.

Górecki

M. A.

(2013). Election closeness, habit formation and voter turnout: Evidence from sixteen Swedish elections. Political Studies, 61, 234-248. doi:10.1111/1467-9248.12017

19.

Gronke

Krantz Toffey

(2008). The psychological and institutional determinants of early voting. Journal of Social Issues, 64, 503-524. doi:10.1111/j.1540-4560.2008.00575.x

20.

Hayes

A. F.

Glynn

C. J.

Huge

M. E.

(2012). Cautions regarding the interpretation of regression coefficients and hypothesis tests in linear models with interactions. Communication Methods and Measures, 6, 1-11. doi:10.1080/19312458.2012.651415

21.

Hayes

A. F.

Matthes

(2009). Computational procedures for probing interactions in OLS and logistic regression: SPSS and SAS implementations. Behavior Research Methods, 41, 924-936. doi:10.3758/BRM.41.3.924

22.

Holbrook

T. M.

McClurg

S. D.

(2005). The mobilization of core supporters: Campaigns, turnout, and electoral composition in United States presidential elections. American Journal of Political Science, 49, 689-703. doi:10.1111/j.1540-5907.2005.00149.x

23.

Huckfeldt

Ikeda

Pappi

F. U.

(2005). Patterns of disagreement in democratic politics: Comparing Germany, Japan, and the United States. American Journal of Political Science, 49, 497-513. doi:10.1111/j.1540-5907.2005.00138.x

24.

Jang

S.-J.

(2009). Are diverse political networks always bad for participatory democracy? Indifference, alienation, and political disagreements. American Politics Research, 37, 879-898. doi:10.1177/1532673X09332790

25.

Jöreskog

K. G.

(1979). Statistical estimation of structural models in longitudinal-developmental investigations. In Nesselroade

J. R.

Baltes

P. B.

(Eds.), Longitudinal research in the study of behavior and development (pp. 303-351). New York, NY: Academic Press.

26.

Kam

C. D.

(2006). Political campaigns and open-minded thinking. Journal of Politics, 68, 931-945. doi:10.1111/j.1468-2508.2006.00481.x

27.

Kim

Scheufele

Han

J. Y.

(2011). Structure or predisposition? Exploring the interaction effect of discussion orientation and discussion heterogeneity on political participation. Mass Communication & Society, 14, 502-526. doi:10.1080/15205436.2010.513469

28.

Klandermans

(2010). Peace demonstrations or anti-government marches? The political attitudes of the protesters. In Walgrave

Rucht

(Eds.), The world says No to war: Demonstrations against the war on Iraq (pp. 98-118). Minneapolis: The University of Minnesota Press.

29.

Kwak

Williams

A. E.

Wang

Lee

(2005). Talking Politics and engaging politics: An examination of the interactive relationships between structural features of political talk and discussion engagement. Communication Research, 32, 87-111. doi:10.1177/0093650204271400

30.

Lazarsfeld

Berelson

Gaudet

(1944). The people’s choice. New York, NY: Columbia University Press.

31.

Lord

C. G.

Ross

Lepper

M. R.

(1979). Biased assimilation and attitude polarization: The effects of prior theories on subsequently considered evidence. Journal of Personality and Social Psychology, 37, 2098-2109. doi:10.1037/0022-3514.37.11.2098

32.

Marcus

G. E.

Mackuen

M. B.

(1993). Anxiety, enthusiasm, and the vote: The emotional underpinnings of learning and involvement during presidential campaigns. American Political Science Review, 87, 672-685. doi:10.2307/2938743

33.

Marcus

G. E.

Russell Neuman

Mackuen

(2000). Affective intelligence and political

34.

judgment. Chicago: University of Chicago Press.

35.

Matthes

(2012a). Exposure to counterattitudinal news coverage and the timing of voting decisions. Communication Research, 39, 147-169. doi:10.1177/0093650211402322

36.

Matthes

(2013). Do hostile opinion environments harm political participation? The moderating role of generalized social trust. International Journal of Public Opinion Research, 25, 23-42. doi:10.1093/ijpor/eds006

37.

Matthes

Morrison

K. R.

Schemer

(2010). A spiral of silence for some: Attitude certainty and the expression of political minority opinions. Communication Research, 37, 774-800. doi:10.1177/0093650210362685

38.

McClurg

S. D.

(2006). Political disagreement in context: The conditional effect of neighborhood context, disagreement and political talk on electoral participation. Political Behavior, 28, 349-366. doi:0.1007/s11109-006-9015-4

39.

McGhee

Sides

(2011). Do campaigns drive partisan turnout? Political Behavior, 33, 313-333. doi:10.1007/s11109-010-9127-8

40.

McLeod

J. M.

Sotirovic

Holbert

R. L.

(1998). Values as sociotropic judgments influencing communication patterns. Communication Research, 25, 453-485. doi:10.1177/009365098025005001

41.

Meijnders

Midden

Olofsson

Öhman

Matthes

Bondarenko

. . .Rusanen

(2009). The role of similarity cues in the development of trust in sources of information about GM food. Risk Analysis, 29, 1116-1128. doi:10.1111/j.1539-6924.2009.01240.x

42.

Mutz

D. C.

(2002). The consequences of cross-cutting networks for political participation. American Journal of Political Science, 46, 838-855. doi:10.2307/3088437

43.

Mutz

D. C.

Martin

P. S.

(2001). Facilitating communication across lines of political difference: The role of mass media. American Political Science Review, 95, 97-114.

44.

Nickerson

D. W.

(2005). Partisan mobilization using volunteer phone banks and door hangers. Annals of the American Academy of Political and Social Science, 601, 10-27. doi:10.1177/0002716205278200

45.

Nir

(2005). Ambivalent social networks and their consequences for participation. International Journal of Public Opinion Research, 17, 422-442. doi:10.1093/ijpor/edh069

46.

Nir

(2011a). Disagreement and opposition in social networks: Does disagreement discourage turnout? Political Studies, 59, 674-692. doi:10.1111/j.1467-9248.2010.00873.x

47.

Nir

(2011b). Motivated reasoning and public opinion perception. Public Opinion Quarterly, 75, 504-532. doi:10.1093/poq/nfq076

48.

Nir

(2012).Cross-national differences in political discussion: Can political systems narrow deliberation gaps? Journal of Communication, 62, 553-570. doi:10.1111/j.1460-2466.2012.01648.x

49.

Nir

Druckman

J. N.

(2008). Campaign mixed-message flows and timing of vote decision. International Journal of Public Opinion Research, 20, 326-346. doi:10.1093/ijpor/edn035

50.

Noelle-Neumann

(1974). The spiral of silence: A theory of public opinion. Journal of Communication, 24, 43-51. doi:10.1111/j.1460-2466.1974.tb00367.x

51.

Pattie

C. J.

Johnston

R. J.

(2009). Conversation, disagreement and political participation. Political Behavior, 31, 261-285. doi:10.1007/s11109-008-9071-z

52.

Pitts

S. C.

West

S. G.

Tein

J.-Y.

(1996). Longitudinal measurement models in evaluation research: Examining stability and change. Evaluation and Program Planning, 19, 333-350. doi:10.1016/S0149-7189(96)00027-4

53.

Rattinger

Roßteutscher

Schmitt-Beck

Weßels

(2012). German Longitudinal Election Study–Wahlkampf-Panel, 10.07.-07.10.2009 (ZA5305, Version 3.0.0). Köln, Germany: GESIS.

54.

Rojas

(2008). Strategy versus understanding: How orientations towards political conversation influence political engagement. Communication Research, 35, 452-480. doi:10.1177/0093650208315977

55.

Rojas

(2010). “Corrective” actions in the public sphere: How perceptions of media and media effects shape political behaviors. International Journal of Public Opinion Research, 22, 343-363. doi:10.1093/ijpor/edq018

56.

Rojas

Shah

D. V.

Cho

Schmierbach

Keum

Gil de Zúñiga

(2005). Media dialogue: Perceiving and addressing community problems. Mass Communication & Society, 8, 93-110. doi:10.1207/s15327825mcs0802_2

57.

Shah

Cho

Eveland

W. P.

Kwak

(2005). Information and expression in a digital age modeling internet effects on civic participation. Communication Research, 32, 531-565. doi:10.1177/0093650205279209

58.

Sotirovic

McLeod

J. M.

(2001). Values, communication behavior, and political participation. Political Communication, 18, 273-300. doi:10.1080/10584600152400347

59.

Thompson

D. F.

(2008). Deliberative democratic theory and empirical political science. Annual Review of Political Science, 11, 497-520. doi:10.1146/annurev.polisci.11.081306.070555

60.

Tsfati

(2004). Exploring possible correlates of journalists’ perceptions of audience trust. Journalism & Mass Communication Quarterly, 81, 274-291. doi:10.1177/107769900408100204

61.

Valenzuela

Kim

Gil de Zúñiga

(2012). Social networks that matter: Exploring the role of political discussion for online political participation. International Journal of Public Opinion Research, 24, 163-184. doi:10.1093/ijpor/edr037

62.

van Zomeren

Postmes

Spears

(2008). Toward an integrative social identity model of collective action: A quantitative research synthesis of three socio-psychological perspectives. Psychological Bulletin, 134, 504-535. doi:10.1037/0033-2909.134.4.504

63.

Watt

J. H.

Mazza

Snyder

L. B.

(1993). Agenda-setting effects of television news coverage and the effects decay curve. Communication Research, 20, 408-435. doi:10.1177/009365093020003004

64.

Wojcieszak

(2011). Pulling towards or pulling away: Deliberation, disagreement and opinion extremity in political participation. Social Science Quarterly, 92, 207-225. doi:10.1111/j.1540-6237.2011.00764.x

65.

Wojcieszak

Baek

Y. M.

Delli Carpini

(2010). Deliberative and participatory democracy? Ideological strength and the processes leading from deliberation to political engagement. International Journal of Public Opinion Research, 22, 154-180. doi:10.1093/ijpor/edp050