Four Paths To Misperceptions: A Panel Study On Resistance Against Journalistic Evidence

ABSTRACT

Citizens’ misperceptions on critical issues such as climate change, migration, or health are viewed as a major problem in today’s democratic systems. A large body of literature shows how inaccurate information might lead to misperceptions despite of corrections and retractions. This study highlights individuals’ acts of resistance against journalistic reporting as a driver of misperceptions. Based on the framework of resistance strategies, we examine four processes which enable individuals to arrive at political realities that differ from the facts that are reported in the legacy media: 1) avoidance of the evidence; 2) biased evaluation of journalists’ expert opinion as a form of biased processing; 3) contesting the content and source of evidence; and 4) bolstering attitudes by seeking out like-minded discussions. We apply this theoretical model to explain misperceptions on the political “Ibiza scandal” and misperceptions about climate change policies in Austria. Findings from a two-wave panel study in the Austrian election context (N = 523) suggest that misperceptions stem in part from wrong inferences about journalistic expert opinion. Moreover, individuals that engage in source derogation of legacy media are able to uphold their misperceptions in the face of opposing evidence.

Misinformed citizens make potentially harmful decisions for themselves, for others, and for democratic societies. These detrimental effects of misperceptions have sparked a new wave of research on the phenomenon in recent years. Scholarship on misperceptions acknowledges that “facts” and “fake” are ideologically loaded terms (Kuklinski, Quirk, Schwieder, & Rich, 1998). To differentiate between inaccurate and accurate perceptions, scholars define misperceptions as “factual beliefs that are false or contradict the best available evidence in the public domain” (Flynn, Nyhan, & Reifler, 2017, p. 128). Thus, only when there is convincing and overwhelming evidence against it, a perception can be labeled as a misperception.

How do misperceptions form and how do they persist when expert opinion clearly contradicts them? On the one hand, misperceptions might stem from dysfunctional information systems that don’t provide citizens with trustworthy and reliable information. This involves unintended reporting mistakes (misinformation) and intentionally misleading news stories (disinformation; Wardle & Derakhshan, 2017). On the other hand, individuals’ biased information search might lead to misperceptions due to one-sided information and an incomplete exposure to facts (Garrett, Weeks, & Neo, 2016).

Moreover, individuals might hold on to misperceptions despite of being confronted with factual evidence. So far, studies have shown that individuals adhere to their misperceptions when confronted with counter-attitudinal news reports, providing support to the theory of motivated reasoning (Flynn et al., 2017). However, we still lack a systematic and theory-driven investigation into the exact strategies which allow individuals to stay misinformed despite of being exposed to correct information. This constitutes a pressing research gap, since such strategies might greatly determine which counter-strategies could debias misinformed citizens (Fransen, Smit, & Verlegh, 2015).

We fill this gap by building on the framework of resistance strategies (Fransen et al., 2015). Following this theoretical approach, we examine four main paths of resistance against journalistic evidence presented in the legacy media: 1) Avoidance of the journalistic evidence; 2) the biased evaluation of journalists’ expert opinion as a form of biased processing; 3) contesting the content and source of journalistic evidence; and 4) bolstering attitudes, e.g. by seeking out like-minded discussions. We propose that these strategies enable individuals to hold misperceptions even when they are confronted with factual evidence through the legacy media. We apply the theoretical model to explain misperceptions on the political “Ibiza scandal” (see Appendix A) and climate change policies in a two-wave panel in the context of the Austrian national election 2019.

Motivated Reasoning

According to the theory of motivated reasoning, any “wish, desire or preference” (Kunda, 1990, p. 480) – subsumed under the term motivation – may influence our memory search, attention to and evaluation of information. In other words, whether or not citizens seek out information in the first place and how they process this information, is a question of their motivations. While each individual can shift between different motivations, studies show that motivations to defend one’s ideology are prevalent in political information processing (Bolsen et al., 2014).

The paradigm of motivated reasoning takes a central role in the investigation of misperceptions (Flynn et al., 2017). So far, motivated reasoning provides a common framework to interpret why individuals of different ideological backgrounds arrive at different levels of acceptance of evidence (Garrett et al., 2016; Walter & Murphy, 2018). However, the exact mechanisms through which individuals reach their defense goal still remain an open question (Flynn et al., 2017).

This is surprising, as strategies, not outcomes, are at the center of the theory of motivated reasoning. As Leeper and Slothuus (2014) state: “Because motivations are desired end states and not outcomes per se, they manifest in strategies that individuals – consciously or unconsciously – employ in an effort to obtain those desired end states” (p. 139). This study complements previous studies on motivated reasoning in misperceptions research by investigating the exact strategies that might be used in the defense process. Holding misperceptions might motivate individuals to use a set of defensive strategies outlined by the framework of resistance strategies (Fransen et al., 2015). In return, the use of resistance strategies might help individuals to sustain and strengthen their misperceptions.

Resistance against Persuasive Attacks

Consistency theories have greatly improved our understanding of why messages fail to change recipients’ beliefs in message-consistent ways. In the 1960, the theory of cognitive dissonance (Festinger, 1957) and balance theory (Heider, 1958) stated that individuals strive for “a harmonious state, one in which the entities comprising the situation and the feelings about them fit together without stress” (Heider, 1958, p. 180). When cognitive elements contradict each other, a state of discomfort initiates actions that relieve individuals’ dissonance (Festinger, 1957).

Festinger’s theory of cognitive dissonance became prominent in political communication research, as it delivers a basis to understand why attitudinal changes are difficult to induce through media messages when inconsistent with citizens’ attitudes. It has informed research on selective exposure, selective attention, and selective processing, which have been suggested as central factors that explain how individuals might resist certain interpretations of political events (Knobloch-Westerwick, 2014). Simultaneously, also advertising and health communication research contributed to our understanding of resistance against messages, for instance by investigating different forms of avoidance (Speck & Elliott, 1997), and explaining in which situations individuals use their knowledge to counterargue a message (Friestad & Wright, 1994).

Building on these research traditions, Fransen et al. (2015) developed a framework in an attempt to pool and organize resistance strategies which have been established across different disciplines. Importantly, the framework doesn’t provide new theoretical foundations for explaining resistance. However, by providing a structured overview, it allows for a systematic investigation into different types of resistance. To date, this comprehensive approach is still lacking in misperceptions research.

The authors identify four overarching mechanisms: avoidance of the message, contesting different message features, biased processing of the message, and empowering one’s own position. Building on the theory of motivated reasoning and resistance strategies, we expect that individuals use resistance strategies against factual journalistic evidence presented in the legacy media in defense of their misperceptions. By legacy media, we mean outlets such as the public broadcaster, quality newspapers, the free tabloid press, as well as the commercial tabloid press. These four different types of legacy media ”are characterized by certain, often hierarchical, organizational structures and traditional publishing routines” (Holt, Ustad Figenschou, & Frischlichet al., 2019, p. 861). Legacy media still represents the most important source of political information in Austria (Gadringer, Holzinger, Nening, Sparviero, & Trappel, 2019). Therefore, it is of special interest to see how individuals might resist information by these sources.¹

Avoidance Strategies

One strategy of resisting the influence of messages on beliefs and attitudes is to avoid content (Fransen et al., 2015). Long prominent in the study of social psychology, cognitive dissonance theory posits that individuals avoid information in order to circumvent a psychologically uncomfortable state in which cognitions contradict each other (Festinger, 1957). Within the context of unfavorable political news, this strategy might take the form of intentional news avoidance, which is defined as “low news consumption over a continuous period of time caused […] by a dislike for news” (Skovsgaard & Andersen, 2020, p. 463). Importantly, this form of avoidance does not exclude the possibility that avoidant citizens are regular news users, but describes to which extent citizens find themselves actively avoiding certain genres and content (see also Kalogeropoulos, 2017). While there are different motivations to actively avoid news such as mood management, a lack of trust in news, or information overload (Skovsgaard & Andersen, 2020), the political communication literature on selective avoidance also highlights the importance of incongruence with prior political attitudes in this process (Hameleers & van der Meer, 2020; Knobloch-Westerwick & Meng, 2009). In line with these findings, we expect that news avoidance helps individuals to maintain their misperceptions about political issues by withdrawing from the evidence presented in the legacy media.

H1) Avoidance of evidence presented in the legacy media increases misperceptions of public issues over time.

Biased Processing

While avoidance of the mainstream interpretation of events is one possibility to hold on to misperceptions, it may not be the most likely strategy. Media consumption is heavily habitualized and only rarely subjected to sudden changes (Schnauber & Wolf, 2016). This raises the question of how some individuals form misperceptions despite of exposure to evidence through their legacy media consumption. The resistance strategy of biased processing offers a possible explanation.² This study defines biased processing as “strategies by which recipients selectively process or understand the message in such way that it favors their original attitudes or behavior” (Fransen et al., 2015, p. 2). In contrast to counterarguing or source derogation, this strategy describes that individuals form different interpretations and memories from the same body of information. To give an example, imagine a citizen who questions anthropogenic climate change. When presented with the information that 97% of scientists agree that climate change is caused by humans’ greenhouse gas emissions, this citizen might interpret this information as evidence that the science on climate change is not fully settled – after all, three percent of scientists still disagree with the mainstream opinion. On the contrary, a more environmentally concerned citizen might take this information as definite proof for anthropogenic climate change. Thus, by exploiting ambiguity in messages, people reach different conclusions drawing on the same piece of information (Dunning, 2014). To summarize, biased processing means that individuals might be exposed to evidence, but only selectively remember information and exploit ambiguity to arrive at their own version of what the evidence suggests. The more ambiguous the evidence is and the more information is consumed, the more biased processing might take effect. This phenomenon has already been observed in individuals’ interpretation of scientific consensus (Kahan, Jenkins‐Smith, & Braman, 2011) and political facts (Garrett et al., 2016; Jerit & Barabas, 2012).

One way through which biased processing manifests is the biased evaluation of expert opinion. Kahan et al. (2011) propose that not only individuals’ own perception, but already their perception of expert opinion is skewed toward their favored conclusion. Garrett et al. (2016) have pointed to the fact that interpreting expert opinion wrongly is critically different from getting experts' interpretation of reported facts right, but rejecting and disbelieving it. People who employ the strategy of biased processing might think that experts support their own, biased view on the evidence and adopt misperceptions based on this misinterpretation. In contrast, people who do not engage in biased processing might get the meaning of a message right, but choose to counterargue it or contest the source instead. Since individuals’ resistance against journalistic evidence is the main concern of this study, we investigate individuals’ biased processing in regard to evaluations of the expert opinion of journalists, hereafter referred to as “perceptions of journalistic evidence.”

Media Outlet-Specific and Contextual Differences in Biased Processing

The more individuals consume the news, the more they should find opportunities to selectively misremember and misinterpret information. Therefore, biased processing should become more pronounced with increased news consumption (Jerit & Barabas, 2012). However, as mentioned before, also the ambiguity of presented facts plays an important role, as it presents individuals with opportunities to interpret the information in an attitude-consistent way (Dunning, 2014; Lord, Ross, & Lepper, 1979). Therefore, it is highly important to differentiate between different news outlets which might vary in their level of ambiguity in presenting the facts. Ideology should play a larger role for the biased processing of news when news outlets present the facts in a balanced way and include multiple viewpoints. In contrast, news outlets that offer one-sided messages might minimize the potential for biased processing.

We therefore differentiate four different types of legacy media – the quality press, the free tabloid press, the commercial tabloid Kronen Zeitung, and the public broadcaster ORF. These types of media constitute the most important sources of political information for Austrian citizens (Gadringer et al., 2019). Moreover, based on previous studies, they might also systematically differ in their degree of balanced reporting and therefore ambiguity on the issues under study, that is, the Ibiza scandal and climate change.

In the case of the Ibiza scandal, which included the party leader of the right wing-party FPÖ, Heinz-Christian Strache, biased processing might vary between media outlets based on different evaluations of the candidate and reporting standards. Media outlets differ in their reporting about Heinz-Christian Strache (Eberl, Wagner, & Boomgaarden, 2016, 2018): A certain degree of fluctuation can be seen in the reporting about Strache’s political agenda in the quality news outlet die Presse (Eberl et al., 2016), leading to more opportunities for biased processing. Similarly, findings about the free tabloid press are mixed, suggesting both negative and positive reporting can co-occur (Eberl et al., 2016). In the case of the public broadcaster, there is a legal obligation to give room to a plurality of opinions (ORF Public Value, 2020), which leads to more balanced reporting. Based on these findings, we expect that the public broadcaster, the quality press, and the free tabloid press give room for the selective interpretation and memory of facts as a necessary precondition for biased processing. Therefore, individuals which are in favor of the party FPÖ will be more able to form skewed interpretations on the evidence for these outlets:

H2) FPÖ party preference moderates the effect of a) quality press news consumption, b) free tabloid press news consumption, and c) public broadcaster news consumption on accurate perceptions of journalistic evidence over time for the case of the Ibiza scandal, so that individuals high in FPÖ party preference will arrive at less accurate perceptions of the journalistic evidence.

The picture might, however, change in the case of the most influential commercial tabloid press, the “Kronen Zeitung. The traditional tabloid outlet reaches around 37% of Austrian citizens. Compared to other media, the Kronen Zeitung stands out for its particularly negative tone when reporting about Heinz-Christian Strache (Eberl et al., 2016). In addition, the Ibiza scandal directly uncovered plans of Heinz-Christian Strache to infiltrate and control the commercial tabloid Kronen Zeitung, which was met with heavy criticism by the newspaper. Therefore, there is a strong indication that the Kronen Zeitung is more one-sided and negative in their representation of the scandal, allowing less room for skewed interpretations in favor of Heinz-Christian Strache. Since it is unclear if biased processing can happen in this context, we pose a research question:

RQ1) How does FPÖ party preference affect the effect of commercial tabloid press news consumption on accurate perceptions of journalistic evidence over time for the case of the Ibiza scandal?

No content analytical data is available on climate change reporting in Austria. However, previous research suggests that overall, the amount of climate skeptics’ voices in media reporting that counter available evidence in Western media is in decline (Brüggemann & Engesser, 2017). Nevertheless, skeptical voices remain present in quality media reporting, in part due to a refutation of these claims (Kaiser & Rhomberg, 2016). Due to the reporting norm of balance, similar findings can be expected for the public broadcaster. The evidence is mixed for tabloid media (Boykoff, 2011; Schmid-Petri, 2017), therefore we pose a research question. Thus, we hypothesize the following interactions between individuals' environmentalism as a topically relevant preexisting attitude and news consumption:

H3) Environmentalism moderates the effect of a) quality press news consumption and b) public broadcaster news consumption on accurate perceptions of journalistic evidence over time for the case of climate change reporting, so that individuals low in environmentalism will arrive at less accurate perceptions of the journalistic evidence.

RQ2) How does a) free tabloid press media consumption and b) commercial tabloid press media consumption affect accurate perceptions of journalistic evidence over time for the case of climate change reporting based on different levels of environmentalism?

Moreover, once individuals are biased in their perceptions of journalistic evidence, they will also adapt their own beliefs accordingly, resulting in more pronounced misperceptions (Garrett et al., 2016). Therefore, we suggest that, in part, it is citizens’ faulty interpretation of expert opinion, not their defensive scrutiny against it, that drives misperceptions:

H4) Lower levels of accurate perceptions of journalistic evidence in the legacy media increase misperceptions of public issues over time.

Contesting Strategies

The next set of strategies outlined by resistance research are so-called contesting strategies (Zuwerink Jacks & Cameron, 2003). Despite of their knowledge of what is reported as the “truth,” citizens can form conclusions that differ from the journalistic evidence (Garrett et al., 2016). This indicates that citizens contest the presented information. An exploration of different resistance strategies identified the strategies of counterarguing (contesting the content) and source derogation (contesting the source) as the two most prevalent tools in the defense of partisan attitudes (Zuwerink Jacks & Cameron, 2003).

Source derogation serves as low effort strategy through which partisans can discount journalistic evidence as inaccurate or irrelevant (Kunda, 1990). In the context of misperception studies, questioning the credibility of the sender can critically undermine the persuasiveness of fact-checking messages (Garrett & Weeks, 2013). Although individuals can be unaware of this process, they question the credibility of experts when they present counter-attitudinal facts on contested issues such as the death penalty or abortion (Zuwerink Jacks & Cameron, 2003) as well as climate change or nuclear waste disposal (Kahan et al., 2011). Hence, discrediting the trustworthiness of factual journalistic evidence in the legacy media emerges as one of the key strategies by which individuals are able to hold on to their misperceptions.

In contrast to source derogation, counterarguing constitutes a more effortful process in which individuals pay close attention to a message and generate thoughts against the presented arguments. Researchers found that engaging in counterarguing is a fruitful strategy to shield one’s attitude from unwanted influences (Lord et al., 1979; Zuwerink Jacks & Cameron, 2003). Subsequently, we propose that contesting journalistic factual evidence through criticizing its arguments and contesting legacy media as a source of information are important mechanisms in explaining how misperceptions persist:

H5) Derogating the legacy media as a source of information on an issue increases misperceptions of public issues over time.

H6) Counterarguing the content of legacy media reporting on an issue increases misperceptions of public issues over time.

Like-minded Discussion as Empowering Strategy

Lastly, so-called empowering strategies can restore certainty about one’s belief. As counter-attitudinal evidence in the legacy media can pose a threat to attitudes which are central to their self-concept, individuals seek support for their position, for instance by talking to like-minded peers – a phenomenon that has already been described in cognitive dissonance theory in the early 1950ies (Festinger, 1957). Additionally, individuals may generate supportive thoughts for their own position and assert their self by stressing that nothing can change their stance (see also Zuwerink Jacks & Cameron, 2003). This study focusses on one specific way of empowerment, which is the assertion of one’s position by seeking out like-minded discussions. While the role of biased information search and selection is already well established in previous research on misperceptions (Feldman, Myers, Hmielowski, & Leiserowitz, 2014; Garrett et al., 2016), the role of interpersonal communication with citizens that share the same political position is still less understood. We expect that preexisting attitudes moderate this effect – only when individuals engage in discussions with likeminded peers that are motivated to reject the evidence, this effect will take place:

H7a) Like-minded discussions increase misperceptions of public issues over time.

H7b) The effect of like-minded discussions on misperceptions of public issues is moderated by preexisting attitudes.

Reinforcement

Building on the theory of motivated reasoning, we propose that holding strong misperceptions motivates the use of strategies that defend one’s standpoint (Flynn et al., 2017; Kunda, 1990). First, the more an individuals’ interpretation of the reality about a political issue conflicts with the legacy media reporting, the more likely this person is to avoid such news. This seems likely, as news avoidance serves as an important way to prevent cognitive dissonance (Festinger, 1957). Second, individuals that hold strong misperceptions are more motivated to read skewed interpretations into expert opinions (Kahan et al., 2011). Third, individuals might be motivated to derogate a source and counterargue the content of a message when it contradicts their position (Zuwerink Jacks & Cameron, 2003). Fourth, when one’s interpretation of facts largely differs from the view presented in the legacy media, it may become even more important for individuals to seek out support from their peers, resulting in more like-minded discussion. Therefore, the relationship between resistance strategies and misperceptions may be twofold: On the one hand, resistance strategies enable individuals to hold on to misperceptions in the face of opposing factual journalistic evidence. On the other hand, already existing misperceptions might motivate individuals to employ these resistance strategies. Thus, we suggest the following hypotheses:

H8) Misperceptions of pubic issues lead to a) increased news avoidance; b) decreased accurate perceptions of journalistic evidence; c) increased derogation of the legacy media as a source of information; d) increased counterarguing of the content of legacy media reporting; and e) increased like-minded discussion over time.

Method

We conducted a two-wave online panel survey during the campaign period of the Austrian national election 2019. The survey company Dynata recruited participants based on representative quotas for age (range 18 to 81, M = 49.43, SD = 15.27) and gender (49.5% female). The sample is heterogenous in regard to educational backgrounds (49.3% hold a high school diploma, 22.6% finished higher education). Out of the 1206 participants which started our online survey, 1105 completed the full survey. The second wave was scheduled to take place before the Austrian national election on September 29th, 2019. Data collection for the second wave started on September 13th and ended on September 21st, 2019. 564 participants finished the second survey (retention rate: 50%). To ensure data quality, we only included participants which took longer than 10 minutes to finish the 25-minute-long survey.³ In addition, one case was excluded because of missing values, resulting in a final sample of N = 523.

Both survey waves followed the same structure: After giving their informed consent, individuals first indicated their demographics; next they answered questions about their social media use. Then they answered a question block on media use, including items on news avoidance and like-minded discussions. Next, questions about accurate perceptions of journalistic evidence and misperceptions were asked, followed by questions about counterarguing and source derogation. The question block was preceded by an introduction that explicitly referred to legacy media (see Appendix C). In a next step, participants indicated their political preferences such as their party preference. Lastly, individuals were asked about their dispositions, including political interest, environmentalism, and knowledge questions, followed by a debriefing. The study followed all ethical guidelines proposed by WAPOR (2011).⁴

Measures

All variables were measured in wave 1 (W1), except for dependent variables, which were measured in both waves, and campaign knowledge, which was measured in wave 2 (W2). All items within one question block were asked in a randomized order. If not stated otherwise, items were measured on a 7-point scale (1 = lowest and 7 = highest level). All items are reported in Appendix C.

Misperceptions

Two highly salient cases in the Austrian national election were selected: The case of the so-called “Ibiza scandal” (IS), which involved politicians by the former governmental party FPÖ and sparked reelections, and facts about climate change (CC). Measurements followed the example of Garrett et al. (2016). A more detailed description of the selected facts can be found in Appendix A. Measures were recoded so that higher values indicate more pronounced misperceptions.

IS misperceptions (W1: α_w1 = .89, M = 2.48, SD = 1.60; W2: α_w2 = .90, M =2.48, SD =1.67) were measured by asking about individuals’ personal opinion on how accurate they think four statements on the Ibiza scandal are (1 = wrong statement, 7 = accurate statement). The statements build on verified facts based on released video material. The statements stated that, “Heinz-Christian Strache wanted to exert political influence on the Kronenzeitung in order to get votes”; “Heinz-Christian Strache wanted to award state contracts to the alleged oligarchic niece”; “Heinz-Christian Strache wanted to have Austrian water sold by a private company”; “Heinz-Christian Strache discussed with the alleged Russian oligarchic niece how she could support the FPÖ through illegal party donations.”

To tap CC misperceptions (W1: α_w1 = .75, M = 3.12, SD = 1.36; W2: α_w2 = .72, M = 3.16, SD = 1.33), individuals were asked about their own assessment of the accuracy of four statements about climate change in Austria. The statements were selected based on commonly reported accurate information across different media channels and indicated that: “People who want to protect the climate should avoid eating meat and air travel as far as possible”; “Austria’s CO2 balance is in the bottom quarter of the EU.”; “Austria has missed the climate targets”; and “Heat waves will become hotter and more frequent in Austria in the future.”

Accurate Perceptions of Journalistic Evidence

Accurate perceptions of journalistic evidence on the Ibiza scandal (IS accurate PJE, W1: α_w1 = .87, M = 5.48, SD = 1.49; W2: α_w2 = .90, M = 5.46, SD = 1.57) and on climate change (CC accurate PJE, W1: α_w1 = .68, M = 4.84, SD = 1.21; W2: α_w2 = .72, M = 4.86, SD =1.22) were measured in the same way (Garrett et al., 2016): Individuals indicated whether journalists who report on the topic and talk to experts think that this statement was accurate. Again, the same statements as those presented to measure misperceptions were shown. Higher values indicate more accurate perceptions of the journalistic evidence.

Resistance Strategies

News avoidance

We measured news avoidance (W1: α_w1 = .90, M = 2.71, SD = 1.62; W2: α_w2 = .89, M = 2.93, SD = 1.70) with three items asking participants how strongly they agreed, e.g., that “I want to see and read as few news and articles as possible” (Song, Jung, & Kim, 2017).

Source derogation

To measure source derogation of IS reporting (W1: α_w1 = .92, M = 3.53, SD = 1.60; W2: α_w2 = .94, M = 3.64, SD = 1.61) and of CC reporting (W1: α_w1 = .94, M = 3.91, SD = 1.57; W2: α_w2 = .93, M = 4.00, SD =1.51), we adapted a scale by Zuwerink Jacks and Cameron (2003) to fit the respective context. In a sematic differential, participants indicated if they find the reporting on climate change as well as on the Ibiza scandal i) not trustworthy or trustworthy, ii) not credible or credible, and iii) not biased or biased. Values were recoded so that high values indicate stronger source derogation.

Counterarguing

IS counterarguing (W1: α_w1 = .88, M = 3.10, SD = 1.66; W2: α_w2 = .91, M = 3.10, SD = 1.66) and CC counterarguing (W1: α_w1 = .90, M = 3.51, SD = 1.72; W2: α_w2 = .91, M = 3.62, SD = 1.70) were measured with three items asking individuals how strongly they have contested the reporting within the last six weeks on the respective issue (e.g., “I had to think about arguments which countered the reporting on climate change”; adapted from Asbeek Brusse, Fransen, & Smit, 2017).

Like-minded discussion

Like-minded discussion (W1: α_w1 = .94, M = 3.24, SD = 1.77; W2: α_w2 = .95, M = 3.43, SD = 1.83) was measured by asking participants how often they discussed political content with people that i) “support the same party,” ii) “share the same views and values,” and iii) “find similar topics important for the election” (Campbell & Kwak, 2011).

Media use

We measured media use of the quality press (W1: M = 4.59, SD = 2.75), the free tabloid press (W1: M = 3.05, SD = 2.24), the commercial tabloid “Kronenzeitung” (W1: M = 2.99, SD = 2.55), and the online and offline news of the public broadcaster “ORF” (W1: M = 5.11, SD = 2.53). Each type of media was measured with a single item, asking individuals on how many days a week they use the type of media to inform themselves about politics. Examples of media outlets were given in brackets, e.g. “free newspapers on- and offline (heute, Österreich, oe24.at, heute.at).”

Preexisting attitudes

Party preference for the FPÖ (W1: M = 3.45, SD = 3.34) was included in models on the Ibiza scandal and was measured using a single item, asking individuals how probable it is that they will vote for the FPÖ in future elections on a scale from 1 = “very unprobable” to 10 = “very probable” (van der Eijk, van der Brug, Kroh, & Franklin, 2006). Environmentalism (W1: α_w1 = .90, M = 5.22, SD = 1.44) was included in models on climate change misperceptions and measured with three items adapted from Schuhwerk and Lefkoff-Hagius (1995) (e.g., “I am willing to change my behavior in order to change the state of the environment”).

Control Variables

A number of controls was added to the statistical models. Alterative media use (wave 1: M = 1.53, SD = 1.28) and social media use (wave 1: M = 4.77, SD = 2.21) were measured using a single item asking how many days a week individuals used the respective media. Political interest was measured with two items (wave 1: α_w1 = .95, M = 4.55, SD = 1.90). We measured campaign knowledge (wave 2: M = 6.57, SD = 3.40, range: 0 to 13) by forming an additive index of correct answers to political knowledge questions. Lastly, we measured age, gender, and education.

Results

To test our hypotheses, we ran autoregressive regression models. To account for deviations from normal sampling distributions in our data we employed bootstrapping-technique based on 5,000 samples. Additional information on the coefficients of control variables can be found in the Appendix, Tables B1 to B5.

Resistance Strategies Hypotheses

News Avoidance

In a first step, we analyzed the impact of different resistance strategies on misperceptions on the Ibiza scandal (IS) and climate change (CC). H1 suggested that news avoidance increases misperceptions. As can be seen from Table 1, we could not confirm this hypothesis for the case of IS misperceptions, b = −0.01, SE = 0.03, p = .788, 95% CI [−0.07, 0.05], and CC misperceptions, b = 0.02, SE = 0.03, p = .433, 95% CI [−0.04, 0.09].

Table 1. Regression model predcting misperceptions on the Ibiza Scandal (IS) and Climate Change (CC) controlling for autoregressive effects

Predictor Variables	IS misperceptions^a (Wave 2)		CC misperceptions^a (Wave 2)
	b (SE)	95% CI	b (SE)	95% CI
Autoregressive effect
IS misperceptions (Wave 1)	0.55 (0.07)***	[0.42; 0.68]
CC misperceptions (Wave 1)			0.47 (0.05)***	[0.36; 0.57]
Resistance strategies
News avoidance (H1)	−0.01 (0.03)	[−0.07;0.05]	0.02 (0.03)	[−0.04; 0.09]
IS accurate PJE (H4)	−0.23 (0.07)***	[−0.37; −0.10]
IS source derrogation (H5)	0.09 (0.04)*	[0.01; 0.16]
IS counterarguing (H6)	0.05 (0.04)	[−0.02; 0.13]
CC accurate PJE* (H4)			−0.15 (0.06)*	[−0.27;-0.02]
CC source derrogation (H5)			0.08 (0.04)*	[0.006; 0.15]
CC counterarguing (H6)			−0.04 (0.03)	[−0.10; 0.02]
Like-minded discussion (H7a)	−0.004 (0.03)	[−0.06; 0.05]	0.02 (0.03)	[−0.03; 0.08]
Preexisting attitudes
Party Preference	0.03 (0.02)	[−0.01; 0.06]
Environmentalism			−0.13 (0.03)***	[−0.19; −0.06]
Interactions
Like-minded discussion x Party Preference (H7b)	0.004 (0.01)	[−0.01; 0.02]
Like-minded discussion x Environmentalism (H7b)			−0.02 (0.02)	[−0.05; 0.01]
Media use
Quality press	−0.01 (0.02)	[−0.05; 0.03]	−0.01 (0.02)	[−0.05; 0.02]
Free tabloid press	−0.01 (0.02)	[−0.05; 0.04]	0.01 (0.02)	[−0.04; 0.05]
Commercial tabloid press	0.01 (0.02)	[−0.03; 0.06]	0.05 (0.02)*	[0.02; 0.10]
Public broadcaster	0.002 (0.02)	[−0.04; 0.05]	−0.05 (0.02)*	[−0.09; −0.01]
Alternative media	−0.04 (0.04)	[−0.12; 0.04]	−0.005 (0.04)	[−0.09; 0.07]
Social media use	−0.003 (0.02)	[−0.05; 0.04]	0.001 (0.02)	[−0.04; 0.04]
Demographics
Age	0.000 (0.004)	[−0.01; 0.01]	0.004 (0.00)	[−0.00; 0.01]
Gender (dummy: female = 1)	0.04 (0.09)	[−0.14; 0.23]	−0.13 (0.09)	[−0.30; 0.04]
Intermediate education	0.03 (0.11)	[−0.20; 0.24]	−0.06 (0.10)	[−0.25; 0.13]
Higher education	−0.06 (0.14)	[−0.34; 0.21]	−0.05 (0.12)	[−0.28; 0.18]
Campaign knowledge^a	−0.03 (0.02)	[−0.06; 0.00]	−0.02 (0.01)	[−0.05; 0.01]
Political interest	0.04 (0.03)	[−0.03;0.10]	0.04 (0.03)	[−0.03; 0.10]
R^2 (R^2adjusted)	0.67 (0.65)		0.55 (0.53)

Note: ^aMarked Variables measured in wave 2, all other variables were measured in wave 1; SE and Confidence Intervals were calculated using bootstrap technique with 5,000 samples; IS misperceptions: F(20, 502) = 50.17, p < .001; CC misperceptions: F(20, 502) = 30.41, p <.001; * p < 0.05; ** p < 0.01; *** p < 0.001

Biased Processing

Next, we tested H2, H3, RQ1, and RQ2 by testing how different levels of preexisting attitudes, namely FPÖ party preference and environmentalism, influence effects of media use on IS accurate perceptions of journalistic evidence (IS accurate PJE) and CC accurate perceptions of journalistic evidence (CC accurate PJE). In addition, higher levels of accurate perceptions of journalistic evidence were hypothesized to lead to lower misperceptions (H4). Based on the non-significant interaction effect, we could not confirm that the effects of quality press news consumption (H2a), b = −0.001, SE = 0.01, p = .805, 95% CI [−0.01, 0.01], free tabloid press news consumption (H2b), b = −0.01, SE = 0.01, p = .220, 95% CI [−0.02, 0.01], or public broadcaster news consumption (H2c), b = −0.01, SE = 0.01, p = .052, 95% CI [−0.03, 0.00], on IS accurate PJE were moderated by individuals’ level of FPÖ party preference (R² = .49). In regard to RQ1, we did not find that the effects of individuals’ commercial tabloid press news consumption on IS accurate PJE was moderated by respondents’ FPÖ party preference, b = 0.01, SE = 0.01, p = .309, 95% CI [−0.01, 0.02] (see also Table B1, Appendix B).

Paralleling these findings, we could not support that environmentalism moderated the effect of quality press news consumption (H3a), b = 0.03, SE = 0.01, p = .062, 95% CI [−0.00, 0.05], and public broadcaster news consumption (H3b), b = 0.02, SE = 0.01, p = .214, 95% CI [−0.01, 0.05], for the case of CC accurate PJE (R² = .37). Furthermore, the results also did not suggest an interaction of environmentalism and free tabloid press media consumption (RQ2a), b = −0.02, SE = 0.02, p = .325, 95% CI [−0.05, 0.02], as well as commercial tabloid press media consumption (RQ2b), b = 0.02, SE = 0.01, p = .136, 95% CI [−0.01, 0.05] (see Table B2, Appendix B). However, H4 is supported in our data: IS accurate PJE, b = −0.23, SE = 0.07, p < .001, 95% CI [−0.37, −0.10], and CC accurate PJE, b = −0.15, SE = 0.06, p = .020, 95% CI [−0.27, −0.02], negatively predicted misperceptions (see Table 1).

Contesting Strategies

As predicted by H5, the more respondents engage in source derogation on IS reporting, b = 0.09, SE = 0.04, p = .029, 95% CI [0.01, 0.16], and CC reporting, b = 0.08, SE = 0.04, p = .036, 95% CI [0.01, 0.15], the higher they score on the misperception indices. In contrast, H6 is rejected: Counterarguing did not emerge as a significant predictor in our analysis on IS, b = 0.05, SE = 0.04, p = .147, 95% CI [−0.02, 0.13], and CC, b = −0.04, SE = 0.03, p = .226, 95% CI [−0.10, 0.02] (see Table 1).⁵

Like-minded Discussions

Against our expectations, there was no significant relationship between like-minded discussions and misperceptions on IS (H7a), b = −0.004, SE = 0.03, p = .879, 95% CI [−0.06, 0.05], and CC, b = 0.02, SE = 0.03, p = .385, 95% CI [−0.03, 0.08]. In addition, the moderation by preexisting attitudes, namely party preference for the FPÖ, b = 0.004, SE = 0.01, p = .633, 95% CI [−0.01, 0.02], and environmentalism, b = −0.02, SE = 0.02, p = .241, 95% CI [−0.05, 0.01], was not significant in our data (H7b, see Table 1).⁶ Further analysis revealed that there is a significant and positive correlation between environmentalism and like-minded discussion, meaning that individuals low in environmentalism in general tend to have less discussions about political matters, r =.19, n = 523, p < .001. There was no correlation between party preference and like-minded discussion, r = −.04, n = 523, p < .386.

Reinforcement Hypotheses

Lastly, we tested the effects of IS and CC misperceptions on news avoidance (H8a), accurate PJE (H8b), source derogation (H8c), counterarguing (H8d), and like-minded discussions (H8e). IS misperceptions significantly predicted news avoidance (R² = .51), b = 0.10, SE = 0.05, p = .042, 95% CI [0.01, 0.20]. However, CC misperceptions did not predict news avoidance, b = −0.03, SE = 0.05, p = .604, 95% CI [−0.14, 0.08], thus H8a is only partly accepted (see Table B5 in Appendix B). Accurate PJE showed a negative relationship with IS misperceptions (R² = .49), b = −0.20, SE = 0.07, p = .007, 95% CI [−0.35, −0.07], and CC misperceptions (R² = .37), b = −0.22, SE = 0.05, p < .001, 95% CI [−0.31, −0.12], supporting H8b (see Tables B1 and B2 in Appendix B). Misperceptions significantly predicted source derogation in the case of IS (R² = .52), b = 0.16, SE = 0.05, p = .001, 95% CI [0.07, 0.25], and CC (R² = .48), b = 0.13, SE = 0.05, p = .008, 95% CI [0.04, 0.23]. The same pattern was observed in the case of counterarguing for IS (R² = .52), b = 0.16, SE = 0.05, p = .001, 95% CI [0.07, 0.25], and CC (R² = .32), b = 0.17, SE = 0.06, p = .004, 95% CI [0.06, 0.28]. Thus, H8c and H8d are supported (see Tables B3 and B4 in Appendix B). H8e is rejected: Misperceptions on IS, b = 0.02, SE = 0.05, p = .664, 95% CI [−0.07, 0.12], and on CC, b = −0.10, SE = 0.06, p = .088, 95% CI [−0.07, 0.01], did not increase like-minded discussions (R² = .47, see also Table B5 in Appendix B).

Additional Results

To replicate previous findings from misinformation studies (Allcott & Gentzkow, 2017; Zimmermann & Kohring, 2020), the effects of legacy media use are of additional interest. Higher values of public broadcaster media use, b = −0.05, SE = 0.02, p = .016, 95% CI [−0.09, −0.01], reduced CC misperceptions, while commercial tabloid media use increased CC misperceptions, b = 0.05, SE = 0.02, p = .009, 95% CI [0.02, 0.10]. In contrast, traditional media use variables did not affect IS misperceptions (for detailed results, see Table 1). Furthermore, public broadcaster media use reduced CC source derogation, b = −0.05, SE = 0.03, p = .048, 95% CI [−0.10, −0.00], CC counterarguing, b = −0.08, SE = 0.03, p = .007, 95% CI [−0.14, −0.02], and IS source derogation, b = −0.05, SE = 0.03, p = .040, 95% CI [−0.11, −0.00]. Quality press media use decreased news avoidance, b = −0.08, SE = 0.02, p = .001, 95% CI [−0.13, −0.03], and increased like-minded discussion, b = 0.06, SE = 0.03, p = .018, 95% CI [0.01, 0.12]. News consumption of the commercial tabloid press increased CC counterarguing, b = 0.07, SE = 0.03, p = .028,, 95% CI [0.01, 0.13]. There was no effect of free tabloid press media use on resistance strategies and misperceptions (see Table 1 and Tables B1 to B5 in the Appendix B).

Discussion

Following the framework of resistance strategies (Fransen et al., 2015), the current panel study investigated four paths through which individuals can uphold misperceptions: 1) avoidance of journalistic evidence, 2) biased evaluation of journalists’ expert opinion as a form of biased processing; 3) contesting the source and content of journalistic evidence, and 4) bolstering one’s stance through like-minded discussions. In addition, we investigated how misperceptions retroact on the use of resistance strategies. We tested our hypotheses on two salient cases in the Austrian national election context, the “Ibiza scandal” and climate change beliefs.

First, our data suggests that one specific contesting strategy plays an important role in the formation of misperceptions: Results across both issues suggest that source derogation heightens misperceptions (see also Garrett & Weeks, 2013; Zuwerink Jacks & Cameron, 2003). In contrast, counterarguing – that is, criticizing the reported content – did not emerge as a significant predictor. These findings present preliminary evidence that low effort strategies of contesting might represent a more successful resistance strategy as compared to the high effort strategy of counterarguing (Zuwerink Jacks & Cameron, 2003).

Second, individuals’ perception of the journalistic evidence emerged as a robust predictor of misperceptions across both topics. In other words, some individuals were biased in their perception of the journalistic evidence, which subsequently heightened their misperceptions. This is distinct from other defensive processes, as citizens seemingly didn’t actively counter the evidence, but interpreted it in biased ways. This supports previous findings that skewed interpretations of expert opinion are an important source of misperceptions (Garrett et al., 2016: Kahan et al., 2011).

Against our expectations, we found no evidence for the process of biased processing, which would suggest that the more partisan citizens consume the news, the more they might diverge on their perception of journalistic evidence (see e.g. Jerit & Barabas, 2012). One possible explanation for this null-effect is that individuals’ perceptions might already have been consolidated before the onset of our study. The Ibiza scandal and climate change facts were salient topics before August 2019, which could explain why further reporting did not additionally change individuals’ perceptions. The lack of an interaction effect between media use and preexisting attitudes might also stem from the fact that news media offered a strong interpretation of events instead of presenting varied and balanced viewpoints on the issues, leaving less room for own interpretations. In addition, there are also other forms of biased processing which have not been assessed in this study such as the relative weighting of evidence (Fransen et al., 2015). Moreover, survey data is limited in providing insight into the exact processes underlying biased processing. Therefore, experimental or content analytic data would be needed to further improve our understanding of biased processing.

The pattern of results did not support the hypotheses that avoiding the news or engaging in like-minded discussions resulted in stronger misperceptions. In the case of avoidance, this null-finding could be due to the fact that citizens are confronted with news when they don’t actively seek out political information (Trilling & Schoenbach, 2013). Why didn’t like-minded discussions allow individuals to defend their misperceptions? Albeit speculative, in the case of the Ibiza scandal, like-minded discussions with other FPÖ-voters might have included criticism about the FPÖ party leader, thus undermining their potential to bolster attitudes. This assumption is supported by the fact that many former FPÖ voters withdrew their vote from the party (SORA Institute for Social Research and Consulting, 2019). For the case of climate change, additional analyses explain this null-finding: Individuals low in environmental values are less likely to engage in like-minded discussions and therefore might not resort to this strategy.

Vice versa, our analyses revealed that misperceptions motivate individuals to engage in counterarguing and source derogation and reduce accurate perceptions of journalistic evidence across both topics. In addition, we observed higher levels of news avoidance among participants with more pronounced misperceptions on the Ibiza scandal. Against our expectations, misperceptions did not lead to increased like-minded discussions. This could be due to the fact that individuals might fear that others hold different beliefs on the issues and therefore abstain from discussions. To investigate this possibility, future studies should include third-person perceptions as a possible moderator.

Although not at the center of this study, we would also like to highlight the role of media consumption in citizens’ formation of misperceptions. In contrast to earlier studies (Allcott & Gentzkow, 2017; Zimmermann & Kohring, 2020), legacy media use did not affect Ibiza scandal misperceptions. One possible explanation for this finding is that media effects on misperceptions are less likely to occur when topics have already been heavily reported on (Garrett et al., 2016). This is probable in the case of Ibiza scandal, as the video material which provided evidence of the study-relevant facts was released more than two months before the start of our data collection. In contrast, climate change misperceptions were reduced by consuming the public broadcaster and heightened by commercial tabloid media use. Further investigation is warranted to understand these patterns. They might stem from differences in the reporting of the issues, as tabloid media has also been found to insufficiently explain climate phenomena in content analytical studies (Boykoff, 2011).

In summary, our study highlights the role of individuals as defendants of their own realities. This does not mean that partisan or alternative news and disinformation are a neglectable source of misperceptions – especially given the weight of evidence from previous literature (Feldman et al., 2014; Garrett et al., 2016). However, controlling for media use, there are additional motivated processes that explain how citizens come to different versions of the truth.

Limitations

The limitations of the presented study include concerns about the generalizability of our findings to other national and electoral contexts. Regarding the limitations of our measurements, it could be argued that closed measures of counterarguing don’t capture the spontaneous thoughts generated in the actual process of real media consumption. Moreover, the measurement of biased processing has its limitations: First, a survey study does not allow to test the exact mechanisms which lead to a biased evaluation of expert opinion. Future studies should employ experimental data or use linkage analyses to validate the results. Second, a biased evaluation of expert opinion is only one form of biased processing and therefore, we can’t draw conclusions about other types of biased processing. Third, building on previous literature (Garrett et al., 2016), our measure of individuals’ perceptions of accurate journalistic evidence asked about perceptions of how journalists themselves would evaluate the facts. While this measure can be insightful for the context of the Ibiza scandal, where only journalists had full access to all relevant information about the video that sparked the scandal, it was less suited for evaluating perceptions about journalistic evidence on climate change, where scientific voices might play a more important role. Lastly, one of our news avoidance items could also have been misinterpreted by some participants, as it was not clearly stated that it concerned the avoidance of legacy media.

Moreover, the two-wave design of our panel study limited our possibility of testing mediations. Therefore, future studies with three-wave designs could validate our findings. Furthermore, future studies would profit from content analytical data in order to examine different message characteristics that could lead to misperceptions.

Acknowledgments

Special thanks to Sophia Hackel, who helped to code the articles for the small-scale content analysis (Appendix C).

Disclosure Statement

No financial interest or benefit has arisen from the direct applications of this research.

Data Availability Statement

Derived data supporting the findings of this study are available at the Open Science Framework (10.17605/OSF.IO/5EY8A).

Supplementary Material

Supplemental data for this article can be accessed on the publisher’s website

Additional information

Funding

This work was supported by the University of Vienna via the uni:docs fellowship and the Austrian Press Agency (APA) via the APA-UNIVIE Data Project;

Notes

1. Our argument builds on the assumption that the legacy media reported the facts correctly and did not report information that could have increased misperceptions. To validate this assumption, we conducted a small-scale content analysis on 96 articles from four different legacy media, in which the statements tested in our survey occur N = 131 times. 3.05% of these statements presented counterevidence that challenged the truth status of the factual evidence and might have heightened misperceptions in recipients. This estimate shows that the potential of legacy media to directly influence misperceptions by reporting conflicting facts is neglectable. Full information on the content analysis can be seen in Appendix D.

2. Note that in previous literature on motivated reasoning processes, the term biased processing has sometimes described different defensive reasoning strategies, including counterarguing or source derogation. In this study, however, biased processing follows the definition provided in the resistance strategy framework, which limits the term to the biased interpretation, memory, and weighting of facts and excludes active forms of contesting such as source derogation and counterarguing.

3. Individuals with above average response times were included in the sample, because this could also stem from the fact that they had clicked on the link, but started the survey at a later point. Therefore, a longer response time does not indicate inattention by default.

4. At the time of the study, there was no IRB board for standard research at The Department of Communication at the University of Vienna. However, the ethical conduct was ensured by several steps. Individuals signed an informed consent form before their participation in each survey wave which informed them about the purpose of this study and all their rights regarding their data and participation, such as the right to end the study at any point, the full anonymization of their responses, or their right to withdraw the consent for the usage of parts or the entirety of their data. Participants could contact the researchers at any time to request additional information or withdraw their consent. Since the data collection was conducted by the market research institute Dynata, there was an additional external screening of the questionnaire to ensure full compliance with the GDPR. This procedure ensured that no data is collected that can make respondents identifiable and that participants can refrain from answering sensitive question, e.g., in regard to their political affiliation.

5. Extreme speeders can introduce systematic biases into the data and compromise the validity of results (see e.g. Malhotra, 2008), therefore they were excluded from our analyses. However, in order to the get a full understanding of our results, the analyses were repeated including all speeders. A change in the pattern of the results could be observed in regard to H5 and H6. For the case of IS, the coefficient for counterarguing significantly predicted misperceptions, b = 0.07, p < .029, while the coefficient for source derogation dropped below the conventional level of significance, b = 0.06, p < .086. This is possibly due to the high correlation of both concepts, r = .54, n = 523, p < .001, by which patters of results can shift. For the case of CC, the effects of source derogation on misperceptions turned non-significant, b = 0.05, p = .095. Therefore, the results of H3 and H4 have to be interpreted with caution.

6. Preexisting attitudes could also moderate the effects of other resistance strategies on misperceptions, as suggested by one reviewer. We tested this assumption and found that counterarguing had a stronger effect on CC misperceptions for individuals low in environmentalism, b = −0.04, p = .027. Effects of source derogation on IS misperceptions were stronger for individuals high in FPÖ party preference, b = 0.03, p = .012. Since these findings were not predicted by our theoretical framework, they will not be further discussed in the paper and should be further investigated in future studies.