Visual polarisation: Examining the interplay of visual cues and media trust on the evaluation of political candidates

Abstract

Previous research in political communication has shown that visual news coverage can affect news consumers’ evaluations of political candidates. Yet, so far, the effects of subtle (positive/negative) visual background cues on candidate evaluation remain largely unclear. Also, the role of individuals’ media trust has not been explored in this context. That is, trusting individuals may interpret subtle visual cues in different ways compared with mistrusting individuals. Drawing from theory on visual communication, media trust, and the persuasion knowledge model a quota-based online experiment revealed that media trust moderates the relationship between the exposure to negative visual cues and candidate evaluation in the context of a political scandal. Mistrusting individuals showed more positive candidate evaluations, whereas trusting individuals showed more negative evaluations. Thus, visual background cues can have a polarising effect on citizens’ evaluations of political actors and influence the electorate in very different ways. Implications for journalism practice are discussed.

Keywords

Candidate evaluation media trust political scandal visual effects visual political polarisation

Political scandals¹ and attempts to scandalise political candidates are common features of the modern political media landscape (Allern and von Sikorski, 2018; Entman, 2012; Kepplinger, 2018; Kepplinger et al., 2012; Thompson; 2000; Tumber and Waisbord, 2019). Previous research revealed that exposure to political scandals generally has negative effects on citizens’ candidate evaluations (e.g. Kepplinger et al., 2012; see also Kepplinger, 2018; Tumber and Waisbord, 2019). However, research informed by the motivated reasoning perspective also showed that exposure to scandalous information about a liked political candidate may activate a defense mechanism that can actually improve partisans’ evaluations of scandalous political actors (Fischle, 2000; von Sikorski et al., 2020). Importantly, and similar other to domains of political communication, these previous insights on the effects of political scandals are almost exclusively based on verbal information (Barnhurst and Quinn, 2012; von Sikorski, 2018a). In contrast, the effects of visual information have remained largely unexplored in this context.

First, it thus remains unclear to what extent the effects of verbal scandal information also occur when visual scandal information is employed. Also, this neglect of visual information comes with some surprise, as scandalous news (similar to news about other political issues) is nowadays frequently accompanied by visual information. More precisely, it has been demonstrated that journalists, for instance, tend to use unfavorable images, camera angles, and visual cues (e.g. depicting a politician against an entirely black background) when reporting about political candidates involved in scandal (Kepplinger, 1982; Kepplinger, 2018; see also Kepplinger, 1987 for an overview). Sometimes a visual implicitly suggests an interpretation to news consumers that is distinctly worse compared to the explicit and factual verbal information (see Moriarty, 1996). Also, previous research more generally demonstrated that accounting for visuals is highly relevant because they can affect citizens’ emotional reactions (Grabe and Bucy, 2009; see also Kepplinger et al., 2012), their interpretation of political issues (e.g. Powell et al., 2015), and candidate evaluations independently of verbal information (Dan and Arendt, 2020; Kepplinger, 1982). Furthermore, Dan and Arendt (2020) recently demonstrated that the mere presence of subtle visual background cues (i.e. cues related to political ideology in the background of an image) had an effect on individuals’ voting intentions of a political candidate.

Second, previous research has not systematically examined if and how exposure to visual cues is conditional upon citizens’ particular prior attitudes and views (but see Domke et al., 2002). In this context, Tsfati and Cohen (2013) pointed out that citizens’ general level of media trust can be conceptualised as a key moderator variable influencing media effects because individuals with low levels of media trust generally tend to be less affected by mediated political information compared to individuals high in media trust (Miller and Krosnick, 2000; Tsfati, 2003). However, all of these findings are based on verbal forms of communication and it has not been examined if and how visual cues interact with an individuals’ level of media trust. Stated differently, do visual cues contribute to issue polarisation affecting individuals high in media trust in different ways compared with individuals with low levels of media trust (Kubin et al., 2021; Levendusky, 2013; Prior, 2013; Wojcieszak et al., 2018)?

In the present project, we aimed at closing these persisting research gaps mentioned above. Drawing from theory on visual communication, research on media trust, and the persuasion knowledge model (PKM, Friestad and Wright, 1994), we theorised that skeptical news consumers low in media trust may – as predicted by the PKM – interpret certain visual cues in political communication as persuasive attempt by the journalist or news outlet and will be less influenced compared with individuals high in media trust. To test our assumptions, we conducted a quota-based online experiment (N = 286) with three groups (between-subjects design) based on the demographic characteristics (age, gender, education) of the general population in Germany. Participants were either exposed to news coverage about a political scandal accompanied by a candidate image with an entirely black image background or otherwise identical news coverage with an entirely light image background. Control group participants were exposed to the identical news coverage without an image. Results revealed that participants were affected by the black cue condition (but not by the light cue condition) and that an individual’s level of media trust played a key role in this context. That is, exposure to the black cue condition affected low trusting individuals who showed more positive candidate evaluations. In contrast, individuals high in media trust showed more negative candidate evaluations after exposure to the identical visual cue. Moderately trusting individuals were not affected. These findings make an important theoretical contribution to research in journalism studies and political communication. We show that identical visual cues can affect candidate evaluation both in positive and negative ways and that assessments vary as a function of citizens’ level of media trust resulting in a phenomenon that we call visual polarisation.

Effects of visual information in political communication

In modern media environments news consumers are frequently exposed to large amounts of information; in classic news media but also via various online outlets and social media platforms. Yet, the human mind’s capacity is limited and is simply not able to process all mediated information available (Lang, 1995). Psychological research finds robust support for dual process theories, which differentiate fast automatic processes (Type 1) from effortful higher order reasoning processes (Type 2) (Kahneman, 2011). That is, individuals tend to be cognitive misers who prefer easy-to-evaluate mediated information, even if this information is less accurate (Kahneman, 2011). Thus, in order to manage these large amounts of mediated information news consumers frequently use Type 1 processing (and much less often Type 2 processing) and rely on a very limited number of available cues (three to five on average; Hastie and Dawes, 2009) such as party cues or ideology cues both when selecting and processing mediated political information (Garcia and Stark, 1991).

In this context, visual information generally plays an important role because visuals can be easily and quickly processed and news consumers tend to enter news (e.g. a webpage) through pictures or graphics and focus on visuals first before scanning or reading verbal information (Garcia and Stark, 1991). Furthermore, effects may be amplified when textual and visual information complement one another (Dan, 2018; Dan et al., 2020). Under certain conditions visuals may even yield stronger effects than verbal information (Powell et al., 2015) as visual information may be more persuasive than textual information because visual signs have a symbolic character, which can lead to a more immediate understanding compared to verbal information (Messaris and Abraham, 2001). ‘Photographs come with an implicit guarantee of being closer to the truth than other forms of communication are’ (Messaris and Abraham, 2001: 217). Previous research revealed relevant influences of visual information in various study contexts such as social protests (e.g. Arpan et al., 2006), science communication (e.g. McCabe and Castel, 2008), and political communication (e.g. Domke et al., 2002; Grabe and Bucy, 2009; see Barnhurst and Quinn, 2012).

Visual cues

Recently, researchers have started to examine the role of visual cues or tiny pieces of information embedded in visuals. Importantly, two types of visual cues – foreground and background cues – can be differentiated, for instance, regarding the portrayal of political candidates.

Foreground cues

Cues in the foreground of a picture directly relate to the portrayal of a political candidate. For instance, cues can relate to a candidate’s gestures and nonverbal behaviour (laughing, waving at supporters, thumbs-up gestures, but also unfavorable gestures or anger displays, etc.), clothing (formal attire, but also informal attire to symbolise ordinariness), symbols (flag pin, religious and ideological symbols, see Dan and Arendt, 2020), actions (physical activity to symbolise fitness or working physically, e.g. chopping wood, to symbolise ordinariness or holding children to symbolise family values, Grabe and Bucy, 2009).

Background cues

In contrast, background cues – that the present project focusses on – do not directly relate to a political candidate. For instance, cues can be specific symbols depicted next to or behind a candidate (flags, statues, religious or ideological symbols, but also signs of protestors with unfavorable or negative statements), other persons or crowds (family members, celebrities, prestigious persons like former presidents, athletes, typical citizens to symbolise ordinariness, cheering crowds and supporters (von Sikorski et al., 2012), but also the absence of crowds or supporters symbolising isolation; von Sikorski and Knoll, 2019; von Sikorski and Ludwig, 2018), places (Parliament, Wall Street, high-technology manufacturing plants to symbolise status and progress, Grabe and Bucy, 2009), camera angels and color (favorable/unfavorable camera angles; color cues: darkness cues to symbolise grief or guilt, light cues to symbolise innocence, Kepplinger, 1982, 1987; von Sikorski, 2018b).

Furthermore, visuals may consist of multiple cues in the image foreground and/or background and the size and placement of visual cues may vary considerably (for the role of picture size and placement as importance indicators for audiences see Coleman, 2010; Wanta, 1988).

Although, larger cues can be better detected by news consumers compared with smaller and less obvious cues, previous results indicated that the mere presence of subtle cues in the background of candidate images (e.g. flags or religious symbols) led citizens to infer the ideology of the politician, which affected participants’ voting intentions (Dan and Arendt, 2020). Also, visual cues can be used intentionally or unintentionally (i.e. persuasive attempt, by accident) by journalists and political candidates. For instance, a politician can try influence the presence of visual cues in media coverage by posing next to flag (background cue) and smiling (foreground cue) at a press conference. However, a journalist may not regard for certain image details (visual cue: flag) in postproduction and by selecting an image in which the politician’s display is rather neutral. Similarly, a journalist may depict a politician involved in scandal against an entirely black background (persuasive attempt) suggesting that a politician is guilty.

Thus, this line of research on the effects of visual cues goes beyond previous examinations that have traditionally analyzed how adding a visual (or certain types of visuals) to textual information changes individuals’ attitudes and issue perceptions (Domke et al., 2002; Powell et al., 2015; see Dan, 2018). That is, research on the effects of visual cues examines how identical images that are accompanied by specific visual cues affect individuals’ evaluative and decision-making processes.

Visual darkness and lightness background cues

Building on the results of Dan and Arendt (2020) and previous research on the effects of visual cues (Sullivan, 2015; von Sikorski, 2018b; von Sikorski et al., 2012), we theorised that exposure to darkness and lightness cues in the background of a candidate image will affect citizens’ candidate evaluation of a politician involved in scandal. In general, ‘black’ and darkness is negatively connoted in many cultures (e.g. in Germany, the context of the present study), whereas ‘white’ and lightness is frequently positively connoted in many countries around the world (Frank and Gilovich, 1988; Meier et al., 2004; Smith-McLallen et al., 2006; for an overview see Aslam, 2006). Also, darkness and brightness are regularly associated with negative and positive events and with negative (positive) evaluations. For instance, people in many cultures wear black at funerals and white at weddings. The night and darkness are connected to fears, the unknown, and negative evaluations, whereas daylight, the sun, and brightness is regularly connected to positive associations and attributes (Meier et al., 2004). Previous empirical research supports this theorising. For example, individuals evaluated black uniforms (compared to nonblack uniforms) of US sport teams more negatively and malevolent (Frank and Gilovich, 1988). More recently, Sullivan (2015) demonstrated that individuals who were exposed to an image of an open book evaluated the author of the book more positively when participants were exposed to the image with dark background cues compared to light background cues. Aslam (2006) examined the role of color cues in marketing and concluded: ‘Colours alter the meanings of the objects or situations with which they are associated and colour preferences can predict consumers’ behaviour’ (p. 15). Furthermore, von Sikorski (2018b) examined the effects of visual cues in a political communication context. Results revealed that students who were exposed to an image of a political candidate with a black background precived the politician more negatively compared to a condition without a visual. In summary, and based on these first results, we theorised that both black and light visual background cues will have an effect on citizens’ candidate evaluation of a political actor involved in scandal. That is, we theorised that exposure to verbal information (i.e. scandal news article) accompanied by the identical candidate image (image foreground) will negatively affect citizens’ candidate evaluations when the image background shows a black visual cue, and will positively influence candidate evaluation when citizens are exposed to a light visual background cue, as we formally articulated in our first hypothesis (H1). H1: Exposure to a candidate image with (a) a black background will result in more negative candidate evaluation, whereas (b) exposure to a candidate image with a light background will positively affect candidate evaluation.

The moderating role of media trust

If and to what extent citizens are affected by visual cues embedded in news coverage should depend on their individual level of media trust. That is, the lower one’s media trust is in general, the more skeptical a person should be when exposed to news coverage (Tsfati, 2003) and the less he or she should be willing to accept respective information. Especially, when information is controversial as it is oftentimes the case in political scandals (Thompson, 2000). These assumptions are in line with previous research showing that, for instance, media effects are stronger for individuals who score high in media trust (Miller and Krosnick, 2000). Furthermore, Tsfati (2003) showed that agenda-setting effects (i.e. the media’s power to define the most important problems a nation is facing) were weaker or absent for individuals low in media trust but present in individuals high in media trust (see also Tsfati and Cohen, 2013).

Theoretically, these results are also in line with research informed by the persuasion knowledge model (PKM, Friestad and Wright, 1994) showing that low levels of media trust activate persuasion knowledge in individuals, which makes them become more skeptical towards mediated information (Chen and Cheng, 2020). Put differently, we theorised that when a journalist uses specific visual background cues like an entirely black or light background to report about an (allegedly) scandalous behaviour of a politician, a skeptical news consumer may interpret such visual information as persuasive attempt by the journalist or news outlet. That is, a skeptic may perceive such visual background cues as an implicit attempt to influence and convince him or her, for instance, of the alleged responsibility (i.e. black background cue) or innocence (i.e. light background) of the politician. As abovementioned, individuals tend to react sensitively to color cues and black (Frank and Gilovich, 1988; Meier et al., 2004; Smith-McLallen et al., 2006) as well as light cues (Sullivan, 2015; von Sikorski, 2018b) have been shown to affect evaluative processes both in negative (black) and positive (white) ways. Such perceived persuasive attempts can then activate persuasion knowledge and may dampen the effects of visual background cues in mistrusting individuals (Chen and Cheng, 2020; Friestad and Wright, 1994). Based on this reasoning, media trust can be conceptualised as an important moderator variable influencing the effects of subtle visual cues. More precisely, we expected that citizens with low levels of media trust (compared to highly trusting individuals) will be less affected by both negative and positive visual background cues, as we articulated in our second hypothesis (H2). H2: The effects of (a) a black image background and (b) a light image background on candidate evaluation will be stronger for individuals with high levels of media trust.

Method

To test these hypotheses, we conducted an online experiment (between-subjects design) with three groups (scandal coverage accompanied by no image, candidate image with light visual background cue, candidate image with dark visual background cue).

Participants

A quota sample (N = 286) based on the demographic characteristics (age, gender, education) of the general population in Germany was used (51.7% female; M_age = 46.78, SD = 14.26; range: 18 to 71 years; 0.3% no school degree, 32.2% compulsory school, 30.8% vocational school degree,18.2% with a high school degree, 17.8% with an academic degree, 0.7% still in school). Data collection was realised by Survey Sampling International (SSI).

Procedure

First, participants gave their informed consent and completed a survey containing demographic variables, control variables, and the moderator variable (media trust). Second, participants were then randomly assigned to group 1 (article about a politician involved in scandal; control condition, n = 92), group 2 (identical article accompanied by an image of the politician with a light background cue, n = 96), and group 3 (same article accompanied by the identical image of the politician with a dark background cue, n = 98). The survey software was programed in a way that a participant was at least exposed to a news article for 20 seconds before being able to continue with the survey completing the dependent variable (candidate evaluation). Participants were then thanked and debriefed.

Stimulus material

Participants in the two experimental conditions and the control condition were exposed to a news article from the well-known German news platform sueddeutsche.de (sz.de). SZ is a rather liberal German quality news outlet and SZ regularly reports about political scandals. The article’s textual information was based on original scandal coverage and the layout of SZ was used (approx. 250 words, Appendix 1). According to the article, a district attorney was currently investigating allegations of corruption against a German European Union (EU) politician (i.e. Bettina Wimmer) of the Social Democratic Party of Germany (SPD). The article further described that the politician had allegedly taken bribes in return for assigning EU subsidies to infrastructure projects in Eastern Europe. The article reported about the allegations against the politician in a neutral manner and also mentioned that several aspects of the case were still unclear (as typical in new and emerging scandals) and that investigations were ongoing. Similar to previous research (e.g. Carlson et al., 2000; von Sikorski and Herbst, 2020), we used a fictitious case to prevent any prior knowledge of the scandal in participants, as knowledge about the candidate or scandal details (e.g. the outcome of the case) could have potentially undermined the purpose of the present study. All participants (in all of the three conditions) were exposed to the identical textual information. However, the visual information that accompanied the news article was systematically altered. That is, Group 1 participants were exposed to the article without an image (control condition, n = 92). Group 2 participants were exposed to the identical article. Yet, the article was accompanied by an image of the politician with an entirely light background (positive cue, n = 96). Group 3 participants were exposed to the identical article and image. However, while the foreground of the image remained unchanged (i.e. politician) the background was altered and showed the politician against an entirely black background (negative cue, n = 98; Appendix 1). Importantly, both images in the two experimental conditions were completely identical and only the background cues were altered. Also, all textual information in the three conditions was identical.

Measures

First, several demographic variables and the moderator variable (media trust) were measured. That is, general media trust was measured with five Likert-type items (1 = disagree, 7 = agree) based Kohring and Matthes (2007). Participants completed the following five statements: ‘Political media coverage in Germany (is). . .’: ‘fair’, ‘credible’, ‘unbiased’, ‘tells the whole story’, ‘trustworthy’. The reliability of the scale was excellent (Cronbach’s α = 0.966). The items were combined to an additive index (M = 3.70, SD = 1.55). Participants’ party identification was gauged using one item: ‘How strongly do you identify with the Social Democratic Party of Germany (SPD)?’ (1 = very weakly, 7 = very strongly; M = 2.99, SD = 1.66). Furthermore, political ideology was measured with the help of a scale running from 1 = politically left to 7 = politically right (M = 3.84, SD = 1.25). After stimulus exposure candidate evaluation of the scandalous politician was gauged using a semantic differential (1 = very negative, 7 = very positive) based on Carlson et al. (2000): ‘incompetent–competent’, ‘unqualified–qualified’, ‘not credible–credible’, ‘immoral–moral’, ‘untrustworthy–trustworthy’, ‘unprincipled–principled’, ‘not likeable–likeable’, ‘dishonest–honest’, ‘unfair–fair’, ‘arrogant–modest’, ‘cold–compassionate’. The reliability of the scale was excellent (Cronbach’s α = 0.961). The items were comprised to an additive index (M = 3.19, SD = 1.18).

Manipulation check

All participants in the two experimental groups correctly identified that they were exposed to a candidate image, whereas participants in the control group correctly reported that they were not exposed to a candidate image. Next, only participants who had indicated that they had been exposed to a candidate image were asked to complete the following item: ‘Please try to remember what the background of the candidate image looked like. The background of the candidate image was. . .’. Participants made their assessments using a scale with the endpoints ‘1 = rather dark’ and ‘7 = rather light’ (M = 3.98, SD = 1.84). As expected, participants exposed to a black visual cue (M = 2.86, SD = 1.65) assessed the background of the photograph to be significantly darker (t(192) = –8.247, p < 0.001, Cohen’s d = 1.25) compared with participants exposed to the light background condition (M = 4.83, SD = 1.50). Also, participants completed three items (semantic differential) to assess the presented news article: ‘not informative–informative’, ‘not credible–credible’, ‘not biased–biased’ (1 = negative, 7 = positive; Cronbach’s α = 0.912). The items were combined to an additive index (M = 4.82, SD = 1.35). An Analysis of Variance (ANOVA) revealed no significant group differences (F(2, 283) = 0.623, p = 0.537, partial η² = 0.004) indicating that participants assessed the stimuli as credible and the journalistic quality of the articles as similar. Overall, the manipulation thus worked as intended.

Data analysis

To test the hypotheses, a moderation analysis using the PROCESS macro in SPSS (Model 1) was conducted. First, the experimental condition was dummy coded with the control group as reference group. As depicted in Figure 1, candidate evaluation was modeled as the dependent variable and media trust was modeled as the moderator variable. Also, three important covariates were regarded for, as recommended by Hayes (2013). First, party identification served as a control variable because partisanship has been shown to affect political evaluations in the context of political scandals (e.g. Fischle, 2000). The news article stated that the politician involved in the scandal was a member of the Social Democratic Party (SPD) in Germany and we therefore controlled for SPD identification to ensure that candidate evaluation was gauged independently of prior identification with the SPD. Second, we controlled for political ideology, as liberals’ candidate evaluations of a politician affiliated to the liberal party SPD may generally differ from candidate evaluations of more conservative individuals in a multi-party system such as Germany. Third, previous research has shown that a participant’s gender may affect an individual’s perception of political actors involved in scandal (Cucchi and Cavazza, 2020). Since participants were exposed to a female candidate, we controlled for their self-reported gender to ensure that effects were assessed independently of participants’ gender. Media trust was mean centered prior to computing the product and 95% bias-corrected bootstrap confidence intervals based on 10,000 bootstrap samples were used for statistical inference (Hayes, 2013).

Figure 1.

Hypothesised theoretical model.

Results

First, the effects of exposure to scandal news coverage on participants’ candidate evaluation were examined (H1). Results revealed a non-significant effect of exposure neither to the scandal coverage accompanied by a candidate image with a dark background (b = −1.13, SE = 1.80, p = 0.530), nor the identical scandal coverage accompanied by the identical candidate image with a light background (b = −0.87, SE = 1.81, p = 0.631) compared with the control condition (Figure 2). Both H1a, which proposed that exposure to a candidate image with a black background will negatively influence candidate evaluation and H1b that assumed that exposure to a candidate image with a light background will positively affect candidate evaluation were thus not supported (see Table 1). Next, it was examined if exposure to scandalous news with a candidate image with a black (light) cue interacts with participants’ media trust. This was the case. Supporting H2a, the effect of exposure to the scandal coverage consisting of a candidate image with a black cue on candidate evaluation was moderated by media trust (b = −0.67, SE = 0.20, p < 0.001). Interestingly, two very different patterns regarding the moderating role of media trust were detected. First, the Johnson-Neyman technique of defining significance regions for the conditional effect of the black cue condition on candidate evaluation was used (see Hayes, 2013). This analysis revealed a significant moderation effect for individuals with low levels of media trust (see Table 2). That is, mistrusting individuals (boundaries of the zone of significance −13.50 and −9.56) showed a more positive candidate evaluation after exposure to the black cue condition; an effect that we refer to as visual cue reactance effect. Second, an opposite effect emerged for trusting individuals. As depicted in Table 2, an analysis using the Johnson-Neyman technique revealed that individuals with high levels of media trust showed a more negative candidate evaluation after exposure to the black background condition (boundaries of the zone of significance 3.99 and 16.50); an effect that we refer to as visual cue boosting effect.

Figure 2.

Unstandardised path coefficients.

Table 1.

Ordinary least squares path analysis. N = 286.

Variables	Candidate evaluation
Dark background (vs Control)	–1.13 (1.80)
Light background (vs Control)	–0.87 (1.81)
Media trust	0.36 (0.11)***
Party ID (Social Democratic Party. SPD)	0.96 (0.48)*
Gender	–0.64 (1.49)
Political ideology	1.00 (0.58)⁺
Dark background condition*	–0.67 (0.20)***
Media trust
Light background condition*	0.33(0.21)
Media trust
Adj. R ²	0.09

***p < 0.001; *p < 0.05; ⁺p < 0.08.

Table 2.

Moderator values defining Johnson-Neyman significance region.

Media Trust	Effect	se	t	p	LLCI	ULCI
–13.5000	7.8446	3.2629	2.4042	0.0169	1.4215	14.2677
–12.0000	6.8474	3.0226	2.2654	0.0243	0.8973	12.7976
–10.5000	5.8502	2.7924	2.0950	0.0371	0.3533	11.3472
–9.5599	5.2253	2.6544	1.9685	0.0500	0.0000	10.4506
–9.0000	4.8531	2.5749	1.8848	0.0605	–0.2157	9.9218
–7.5000	3.8559	2.3736	1.6245	0.1054	–0.8166	8.5284
–6.0000	2.8587	2.1930	1.3036	0.1935	–1.4583	7.1757
–4.5000	1.8615	2.0386	0.9131	0.3620	–2.1515	5.8745
–3.0000	0.8643	1.9167	0.4509	0.6524	–2.9087	4.6374
–1.5000	–0.1329	1.8338	–0.0724	0.9423	–3.7428	3.4771
0.0000	–1.1300	1.7954	–0.6294	0.5296	–4.6644	2.4043
1.5000	–2.1272	1.8043	–1.1790	0.2394	–5.6790	1.4245
3.0000	–3.1244	1.8597	–1.6800	0.0941	–6.7854	0.5366
3.9846	–3.7790	1.9197	–1.9685	0.0500	–7.5579	0.0000
4.5000	–4.1216	1.9579	–2.1052	0.0362	–7.9757	–0.2675
6.0000	–5.1188	2.0926	–2.4461	0.0151	–9.2382	–0.9993
7.5000	–6.1160	2.2575	–2.7091	0.0072	–10.5600	–1.6720
9.0000	–7.1132	2.4464	–2.9076	0.0039	–11.9290	–2.2973
10.5000	–8.1103	2.6542	–3.0556	0.0025	–13.3353	–2.8854
12.0000	–9.1075	2.8768	–3.1658	0.0017	–14.7707	–3.4443
13.5000	–10.1047	3.1111	–3.2480	0.0013	–16.2290	–3.9804
15.0000	–11.1019	3.3545	–3.3095	0.0011	–17.7054	–4.4984
16.5000	–12.0991	3.6053	–3.3559	0.0009	–19.1962	–5.0019

Conditional effect of candidate photo with black background on candidate attitude at different values of the moderator (media trust). Boundaries of the zone of significance for individuals with low levels of media trust: –13.500 and –9.56. Boundaries of the zone of significance for individuals with high levels of media trust: 3.99 and 16.500. Boundaries of the zone of significance for individuals with moderate levels of media trust (insignificant): –9.000 and 3.00. N = 286.

In contrast, no interaction effect between the light background condition and media trust could be detected (b = 0.33, SE = 0.21, p = 0.116). Thus, the data did not support H2b. Also, no effect on candidate evaluation could be detected for individuals’ gender that was modelled as a control variable (b = −0.64, SE = 1.49, p = 0.669). Political ideology showed a marginally positive and significant effect on candidate evaluation (b = 1.00, SE = 0.58, p = 0.088). Finally, Social Democratic Party (SPD) identification (the party of the scandalised politician) revealed a positive and significant effect on candidate evaluation (b = 0.96, SE = 0.48, p = 0.047).

Discussion

Journalists frequently report about political scandals. However, the role of visual cues and how they interact with news consumers’ prior evaluations of the media (i.e. media trust) has remained largely unexplored. Corroborating and extending previous research on the effects of visual cues (Dan and Arendt, 2020; Sullivan, 2015) the present results show that exposure to subtle visual background cues can affect a news consumer’s candidate evaluation of a political actor involved in scandal. In general, our findings are also in line with previous results on the effects of visual lightness and darkness cues (Sullivan, 2015; von Sikorski, 2018b; see also Aslam, 2006). Having said that, individuals’ level of media trust moderated effects on candidate attitude. This finding extends previous findings on the role of media trust (Miller and Krosnick, 2000; Tsfati, 2003; Tsfati and Cohen, 2013). Interestingly, while trusting individuals were negatively affected in their candidate evaluations, an opposite effect emerged for mistrusting individuals, who evaluated the scandalous politician more positively. This visual polarisation effect is a truly novel finding that has important theoretical and practical implications both for journalism studies and political communication.

First, the question arises why individuals react in such fundamentally different ways to the exact same news coverage depending on their particular level of media trust? One explanation is that the black background cue activated persuasion knowledge in mistrusting individuals. This is in line with previous research. That is, research informed by the persuasion knowledge model (PKM, Friestad and Wright, 1994) shows that low levels of media trust activates persuasion knowledge in individuals, which makes them become more skeptical towards mediated information (Chen and Cheng, 2020). Put differently, the black visual background cue may have been interpreted by a skeptical individual as an attempt by the journalist to influence and convince him or her of the alleged responsibility or guilt of the politician involved in the scandal. In line with the PKM, such perceptions then resulted in reactance (Brehm, 1966) as a coping strategy in mistrusting individuals. Reactance occurs when a person feels that he or she is pressured to accept certain views or attitudes and has been described as a threat to one’s attitudinal freedom. Reactance results in people adopting or strengthening views or attitudes that are contrary to what an individual perceives or thinks was originally intended. Thus, mistrusting individuals may have activated persuasion knowledge and perceived the black background cue as a persuasion attempt by the journalist, which then resulted in reactance and more positive candidate evaluations of the politician involved in scandal. In contrast, trusting individuals did not activate persuasion knowledge and therefore did not perceive the black background cue as a persuasive attempt by the journalist. Quite the contrary, the negative visual background cue amplified the negative verbal scandal information for trusting individuals and resulted in a more negative candidate evaluation of the politician involved in scandal. Importantly, these effects occurred independently of participants’ party identification with the party of the scandalous politician, participants’ political ideology or their gender, as we controlled for these variables.

Second, no effects on candidate evaluation could be detected for the light background condition, which is contrary to our original assumption (i.e. more positive candidate evaluation) and first previous findings (Sullivan, 2015). Also, we did not find an interaction effect between the condition (i.e. light cue) and media trust. One explanation for these findings is that the light visual cue was simply too weak to revert the effects of the negative textual information on candidate evaluation. Also, participants did obviously not interpret the combination of the negative textual information and the more positive visual background information as generally more balanced (this is also in line with the manipulation check that indicated that participants evaluated the journalistic quality of all three article versions similarly), nor as a persuasion attempt by the journalist, which therefore did not result in reactance. This is also in line, with previous research in visual communication showing that media effects are stronger when textual and visual information complement one another (i.e. negative textual scandal information and negative visual cue) compared to dual channel messages that is unrelated or mismatched (Dan, 2018; Dan et al., 2020; Lang, 1995).

Implications

These findings have several practical and theoretical implications and make a contribution to journalism studies, research in visual political communication, as well as research on media trust. First, journalists should generally be very sensitive and careful when using visual information and images with specific visual cues; especially, when covering politically controversial topics like scandals. As the present study exemplarily shows, subtle visual cues in the background of an image – independently of textual information as well as information in the image foreground – have the power to affect candidate evaluations and voting intentions (Dan and Arendt, 2020; see also Kepplinger, 1987). News reporting frequently is a team effort. In modern news rooms it is oftentimes not the journalist who gets to select an image that accompanies a news story but photo editors (sometimes working in other locations; Robinson, 2011). That is, images that include certain visual cues such as color cues or cues that refer to political ideology might thus be selected completely unintentionally or for illustrative and aesthetic reasons, for instance, to meet the color scheme of a news outlet. Nevertheless, these selection decisions can affect news’ consumers political evaluations. Thus, both journalists and photo editors should be sensitised and educated about the subtle effects of visual background cues. On the other hand, some journalists, for instance, working for partisan media outlets may also be interested in using specific unfavorable or provocative visual background cues more purposefully to raise awareness for a particular issue or to implicitly suggest certain issue interpretations and the attribution of responsibility towards a political actor (Kepplinger et al., 2012). Such practices might be beneficial for certain news outlets (attention, return on investment) but can at the same time damage the reputation of political actors and may increase issue polarisation among citizens, for instance, with different levels of media trust, which leads us to another important implication.

Second, the results extend previous findings on the polarising potential of the news media revealing how news can affect issue polarisation (Kubin et al. 2021; Levendusky, 2013; Prior, 2013; Wojcieszak et al., 2018). In fact, our findings suggest a novel phenomenon that we refer to as visual issue polarisation or visual polarisation. That is, the identical news image can affect audiences with different prior attitudes in very different ways thus contributing to polarised perceptions of political issues. Future studies should continue to examine which factors contribute to visual polarisation, e.g. the interplay of certain ways of processing multimodal information (e.g. verbal versus visual processing style; see Powell et al., 2019), particular individual features (political ideology, political interest, political knowledge) and certain visual (background and foreground) cues. Future studies may also analyze the downstream consequences of visual polarisation, for instance, on perceptions of a particular news outlet or journalists using negative visual cues in their reporting. That said, future research on the effects of visual cues should regard for other relevant individual differences or moderator variables such as individuals’ political interest and issue involvement (Prior, 2013) and should also employ other dependent variables such as citizens’ emotional reactions (Grabe and Bucy, 2009) or issue interpretations (Powell et al., 2015). Third, the present study extends knowledge on the effects of political scandals. Previous research largely neglected to account for the role of visual candidate portrayals on individuals’ evaluations of politicians involved in scandal (von Sikorski, 2018a). As the present study demonstrates, not accounting for visual information may result in misleading results and may also explain some of the inconsistent findings in previous research of the effects of political scandals. Fourth, the present study elucidates the importance of regarding for the role of media trust both as a depend and independent variable (Kohring and Matthes, 2007). That is, media trust can be affected by certain types of media coverage (i.e. media trust as dependent variable) but the way that individuals perceive mediated information is also affected by pre-existing levels of media trust (i.e. media trust as independent variable), as the present study revealed.

Limitations

This study has some noteworthy limitations. First, we examined the interaction of visual background cues and media trust with the help of a specific type of visual background cue (i.e. lightness and darkness cues). Future research should try replicate the present findings with other types of visual background cues (and combinations thereof), for example, examining interactions between media trust and specific visual ideology cues (Dan and Arendt, 2020). That is, citizens’ specific prior views and attitudes might be more or less important depending of the particular type of cue that they are exposed to. For instance, individuals’ level of national identity may be important when visual cues refer to nationality (i.e. flags), whereas citizens’ environmental attitudes may be particularly important when background cues refer to nature. Future research should test these assumptions. Second, future research should examine the effects of visual background cues in other study contexts, for instance, using other types of scandals (e.g. sex scandals) or less controversial political issues. Third, we conducted a quota-based experiment representative for the German population. Yet, the study should be replicated in other cultural contexts (e.g. North America, Asia) testing if the effects found in the present study can be generalised. It would be particularly interesting to examine if the effects are, for instance, reversed in some Asian cultures where (in contrast to Western societies) white is sometimes associated with death, mourning and unhappiness, whereas black is connoted to more positive associations like wealth and prosperity (see Aslam, 2006). Future research should test this. Fourth, we used a mainstream (non-partisan) quality news outlet (Levendusky, 2013) in our study. Future studies should try to replicate the present findings using, for instance, tabloid news outlets or partisan media. However, regardless of these limitations, our findings pose important implications for understanding effects of visual background cues.

Conclusion

Results from a quota-based online experiment show a novel finding triggered by subtle visual background cues in political communication. The mere presence of a visual cue in the background of a candidate image (i.e. entirely black background) showing a scandalous politician resulted in a visual cue boosting effect for individuals high in media trust amplifying negative effects on candidate evaluation. In contrast, the identical news coverage resulted in a visual cue reactance effect for individuals with low levels of media trust improving candidate evaluations despite scandalous textual information. Results show the polarising potential of subtle visual background cues in political communication.

Footnotes

Funding

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

ORCID iD

Christian von Sikorski

Notes

Author biography

Christian von Sikorski is an assistant professor (tenure track) of political psychology at the Department of Psychology, University of Koblenz-Landau (at Landau), Germany.

Appendix

References

Allern

von Sikorski

(2018) Political scandals as a democratic challenge: From important revelations to provocations, trivialities, and neglect. Introduction. International Journal of Communication 12: 3014–3023.

Arpan

Baker

Lee

, et al. (2006) News coverage of social protests and the effects of photographs and prior attitudes. Mass Communication and Society 9: 1–20.

Aslam

(2006) Are you selling the right color? A cross-cultural review of colour as a marketing cue. Journal of Marketing Communication 12: 15–30.

Barnhurst

Quinn

(2012) Political visions: Visual studies in political communication. In: Semetko

Scammell

(eds) The SAGE Handbook of Political Communication. London: Sage, pp.276–292.

Brehm

(1966) A Theory of Psychological Reactance. New York: Academic Press.

Carlson

Ganiel

Hyde

(2000) Scandal and political candidate image. Southeastern Political Review 28: 747–757.

Chen

Cheng

(2020) Consumer response to fake news about brands on social media: The effects of self-efficacy, media trust, and persuasion knowledge on brand trust. Journal of Product and Brand Management 29: 188–198.

Coleman

(2010) Framing the pictures in our heads. In: D’Angelo

Kuypers

(eds) Doing News Framing Analysis: Empirical and Theoretical Perspectives. New York, NY: Routledge, pp.233–261.

Cucchi

Cavazza

(2020) More guilty of woman: The role of gender and causal attribution in political scandals’ impact. The Journal of Social Psychology. Epub ahead of print 13 June 2020. DOI: 10.1080/00224545.2020.1779641.

10.

Dan

(2018) Integrative Framing Analysis: Framing Health Through Words and Visuals. New York, NY: Routledge.

11.

Dan

Arendt

(2020) Visual cues to the hidden agenda: Investigating the effects of ideology-related visual subtle backdrop cues in political communication. The International Journal of Press/Politics. Epub ahead of print 9 July 2020. DOI: 10.1177/1940161220936593.

12.

Dan

Grabe

Hale

(2020) Testing three measures of verbal–visual frame interplay in German news coverage of refugees and asylum seekers. International Journal of Communication 14: 3843–3865.

13.

Domke

Perlmutter

Spratt

(2002) The primes of our times? An examination of the ‘power’ of visual images. Journalism 3: 131–159.

14.

Entman

(2012) Scandal and Silence: Media Responses to Presidential Misconduct. Cambridge: Polity Press.

15.

Esser

Hartung

(2004) Nazis, pollution, and no sex: Political scandals as a reflection of political culture in Germany. American Behavioral Scientist 47: 1040–1071.

16.

Fischle

(2000) Mass response to the Lewinsky scandal: Motivated reasoning or Bayesian updating? Political Psychology 21: 135–159.

17.

Frank

Gilovich

(1988) The dark side of self- and social perception: Black uniforms and aggression in professional sports. Journal of Personality and Social Psychology 54: 74–85.

18.

Friestad

Wright

(1994) The persuasion knowledge model: How people cope with persuasion attempts. Journal of Consumer Research 21: 1–31.

19.

Garcia

Stark

(1991) Eyes on the News. St. Petersburg, FL: Poynter Institute for Media Studies.

20.

Grabe

Bucy

(2009) Image Bite Politics: News and the Visual Framing of Elections. Oxford: Oxford University Press.

21.

Hastie

Dawes

(2009) Rational Choice in an Uncertain World. Thousand Oaks, CA: Sage.

22.

Hayes

(2013) Introduction to Mediation. Moderation. and Conditional Process Analysis: A Regression-Based Approach. New York, NY: Guilford Press.

23.

Kahneman

(2011) Thinking, Fast and Slow. New York, NY: Farrar, Straus and Giroux.

24.

Kepplinger

(1982) Visual biases in television campaign coverage. Communication Research 9: 432–446.

25.

Kepplinger

(1987) Darstellungseffekte. Experimentelle Untersuchungen zur Wirkung von Pressefotos und Fernsehfilmen. [Effect of Depiction. Experimental analysis and effects of press photos and movies]. Munich, Germany: Alber.

26.

Kepplinger

(2018) Die Mechanismen der Skandalisierung: Warum man den Medien gerade dann nicht vertrauen kann. wenn es darauf ankommt [Mechanisms of scandalization. Why we cannot trust the media when it really matters]. Reinbek, Germany: Lau Verlag and Handel.

27.

Kepplinger

Geiss

Siebert

(2012) Framing scandals: Cognitive and emotional media effects. Journal of Communication 62: 659–681.

28.

Kohring

Matthes

(2007) Trust in news media: Development and validation of a multidimensional scale. Communication Research 34: 231–252.

29.

Kubin

Puryear

Schein

, et al. (2021) Personal experiences bridge moral and political divides better than facts. Proceedings of the National Academy of Sciences.

30.

Lang

(1995) Defining audio/video redundancy from a limited-capacity information processing perspective. Communication Research 22: 86–115.

31.

Levendusky

(2013) Why do partisan media polarize? American Journal of Political Science 57: 611–623.

32.

McCabe

Castel

(2008) Seeing is believing: The effect of brain images on judgments of scientific reasoning. Cognition 107: 343–352.

33.

Meier

Robinson

Clore

(2004) Why good guys wear white: Automatic inferences about stimulus valence based on brightness. Psychological Science 15: 82–87.

34.

Messaris

Abraham

(2001) The role of images in framing news stories. In: Reese

Gandy

Grant

(eds) Framing Public Life: Perspectives on Media and Our Understanding of the Social World. Mahwah, NJ: Lawrence Erlbaum, pp.215–226.

35.

Miller

Krosnick

(2000) News media impact on the ingredients of presidential evaluations: Politically knowledgeable citizens are guided by a trusted source. American Journal of Political Science 44: 301–315.

36.

Moriarty

(1996) Abduction: A theory of visual interpretation. Communication Theory 6: 167–187.

37.

Prior

(2013) Media and political polarization. Annual Review of Political Science 16: 101–127.

38.

Powell

Boomgaarden

De Swert

, et al. (2015) A clearer picture: The contribution of visuals and text to framing effects. Journal of Communication 65: 997–1017.

39.

Powell

Boomgaarden

De Swert

, et al. (2019) Framing fast and slow: A dual processing account of multimodal framing effects. Media Psychology 22: 572–600.

40.

Robinson

(2011) Convergence crises: News work and news space in the digitally transforming newsroom. Journal of Communication 61: 1122–1141.

41.

Smith-McLallen

Johnson

Dovidio

, et al. (2006) Black and white: The role of color bias in implicit race bias. Social Cognition 24: 46–73.

42.

Sullivan

(2015) Judging a book by its cover (and its background): Effects of the metaphor intelligence is brightness on ratings of book images. Visual Communication 14: 3–14.

43.

Thompson

(2000) Political Scandal: Power and Visibility in the Media Age. Malden, MA: Polity Press Blackwell.

44.

Tsfati

(2003) Does audience skepticism of the media matter in agenda setting. Journal of Broadcasting & Electronic Media 47: 157–176.

45.

Tsfati

Cohen

(2013) Perception of media and media effects: The third-person effect, trust in media, and hostile media perceptions. In: Scharrer

(ed.) Blackwell’s International Companion to Media Studies: Media Effects/Media Psychology. New York, NY: Wiley-Blackwell, pp.128–146.

46.

Tumber

Waisbord

(eds) (2019) The Routledge Companion to Media and Scandal. New York, NY: Routledge.

47.

von Sikorski

(2018a) The aftermath of political scandals: A meta-analysis. International Journal of Communication 12: 3109–3133.

48.

von Sikorski

(2018b) The effects of darkness and lightness cues in the visual depiction of political actors involved in scandals: An experimental study. Communication Research Reports 35: 162–171.

49.

von Sikorski

Heiss

Matthes

(2020) How political scandals affect the electorate. Tracing the eroding and spillover effects of scandals with a panel study. Political Psychology 41: 549–568.

50.

von Sikorski

Herbst

(2020) Not practicing what they preached! Exploring negative spillover effects of news about ex-politicians’ hypocrisy on party attitudes, voting intentions, and political trust. Media Psychology 23: 436–460.

51.

von Sikorski

Knoll

(2019) Framing political scandals: Exploring the multimodal effects of isolation cues in scandal news coverage on candidate evaluations and voting intentions. International Journal of Communication 13: 206–228.

52.

von Sikorski

Ludwig

(2018) The effects of visual isolation on the perception of scandalized politicians. Communications: European Journal of Communication Research 43: 235–257.

53.

von Sikorski

Schierl

Möller

, et al. (2012) Visual media framing and effects on recipients’ attitudes toward athletes with physical disabilities. International Journal of Sport Communication 5: 69–86.

54.

Wanta

(1988) The effects of dominant photographs: An agenda-setting experiment. Journalism Quarterly 65: 107–111.

55.

Wojcieszak

Azrout

de Vreese

(2018) Waving the red cloth: Media coverage of a contentious issue triggers polarization. Public Opinion Quarterly 82: 87–109.