Why do we fall for misinformation?

The processes by which people form opinions and beliefs are of evident public importance, especially when widely held convictions contradict established facts. Political communication holds a crucial influence on shaping public opinion. When politicians consistently spread inaccurate information, such misinformation can underpin political and social decisions that may conflict with the public interest. Politicians, intentionally or not, often serve as potent sources of misinformation (Lewandowsky et al., Reference Lewandowsky, Ecker, Seifert, Schwarz and Cook2012).

Once exposed to misinformation, individuals may integrate it into their cognitive framework, particularly when it aligns with existing beliefs or when social dynamics reinforce its credibility (Zhou and Shen, Reference Zhou and Shen2024). Such integration can have enduring effects, as misinformation often intertwines with values and identities. Initial information, despite being debunked, remains embedded in memory, continuing to influence reasoning and future judgments (Lewandowsky et al., Reference Lewandowsky, Ecker, Seifert, Schwarz and Cook2012). Several studies have demonstrated that people continue to refer to debunked information even after learning of its inaccuracy (Ecker and Ang, Reference Ecker and Ang2019). These specific types of biases occur when individuals draw inferences or adopt beliefs that are not adequately supported by logical reasoning or sufficient evidence. People assess the truthfulness of information by relying on peripheral cues. These cues include the familiarity of the message, the ease with which the information can be processed, and the internal coherence (Begg et al., Reference Begg, Anas and Farinacci1992).

The “confirmation bias” plays a pivotal role in misinformation, prompting individuals to seek, interpret, and retain information aligning with their preexisting beliefs, disregarding contradictory evidence (Lewandowsky et al., Reference Lewandowsky, Ecker, Seifert, Schwarz and Cook2012; Flynn et al., Reference Flynn, Nyhan and Reifler2017). This phenomenon is linked to motivated reasoning, where information processing is guided by personal desires and interests (Taber and Lodge, Reference Taber and Lodge2006). These motivations shape how people interpret, process, and react to information that challenges their worldview. Although people may aspire to accuracy, they often fail to overcome their preconceptions, even when encouraged to maintain an objective stance (Taber and Lodge, Reference Taber and Lodge2006).

In addition to the psychological and cognitive factors that lead individuals to believe in misinformation, the surrounding digital and social environments must also be considered. The network effects of social media accelerate the spread of both accurate and false information compared to traditional media. The surge in online content consumption has led businesses and political parties to share ambiguous or false content to influence decisions for financial or political gain. Additionally, repeated exposure to belief-aligned content, driven by confirmation bias, fosters the echo-chamber effect, increasing credibility and shareability (Muhammed and Mathew, Reference Muhammed and Mathew2022). This effect is further amplified by digital platforms’ algorithms, which personalize content delivery and limit the diversity of accessible information. These structural features of the media ecosystem intertwine with cognitive, social, and emotional factors, playing a crucial role in shaping how misinformation is shared and accepted (Ecker et al., Reference Ecker, Lewandowsky, Cook, Schmid, Fazio and Brashier2022). The alignment between false narratives and partisan identities poses significant challenges, especially in highly polarized societies where misinformation exploits existing divisions to gain credibility and persist (Flynn et al., Reference Flynn, Nyhan and Reifler2017).

Assessing the role of politicians and correction in political misinformation

A significant body of literature focuses on the circulation of misinformation online, including concerns about the roles of foreign actors, conspiracy theorists, or extremists, and how to counter its spread, influence, and societal impact (Pennycook et al., Reference Pennycook, Epstein, Mosleh, Arechar, Eckles and Rand2021). However, there has been less focus on the extent to which political elites contribute to misinformation. In the “post-truth era,” political leaders often manipulate data through alternative interpretations, make bold claims, and at times resort to deliberate deception to gain partisan advantage (Lilleker and Pérez-Escolar, Reference Lilleker and Pérez-Escolar2023).

Research on political misinformation has largely focused on partisan distortions and confirmation biases (Cohen, Reference Cohen2003; Flynn et al., Reference Flynn, Nyhan and Reifler2017; Jerit and Zhao, Reference Jerit and Zhao2020), often overlooking the crucial role of political leaders in disseminating misleading content (Bullock, Reference Bullock2011). Empirical studies shed light on why statements by political leaders resonate with the public. Research confirms that citizens are more likely to rely on source cues when evaluating politically complex issues (Gilens and Murakawa, Reference Gilens, Murakawa, Delli Carpini, Huddy and Shapiro2002). Ratneshwar and Chaiken (Reference Ratneshwar and Chaiken1991) showed that, when faced with messages that are difficult to understand, individuals predominantly rely on the perceived credibility of the source to determine their level of agreement with the content. Conversely, greater message comprehensibility significantly reduces the importance attributed to source credibility. In this context, as a resource-saving strategy, individuals tend to rely on trusted experts and political elites to form opinions on political issues, thereby avoiding the need to analyze the details independently (Matsubayashi, Reference Matsubayashi2013).

Indeed, politicians employ strategic signaling and media campaigns to legitimize false or misleading narratives, thereby reinforcing partisan identities (Jerit and Zhao, Reference Jerit and Zhao2020). While individuals may generally distrust what they see on social media, they are more likely to believe political leaders, even when these leaders convey messages that contradict verified information (Bisbee and Lee, Reference Bisbee and Lee2022). Politicians exploit these tendencies by discrediting opposition leaders, media, or fact-checkers, firmly depicting them as untrustworthy and shielding supporters from corrective information (Garrett and Poulsen, Reference Garrett and Poulsen2019). This approach perpetuates falsehoods and strengthens partisan identity, undercutting neutral or external corrections (Vraga and Bode, Reference Vraga and Bode2017). This dynamic is rooted in the influence politicians exert by virtue of their prestige and authority (Cohen, Reference Cohen2003). Various studies in political psychology and communication demonstrate that attributing a statement to a specific politician significantly impacts the level of agreement or disagreement expressed by recipients, regardless of the actual content of the statement itself (Taber and Lodge, Reference Taber and Lodge2006; Bullock, Reference Bullock2011). People tend to use the origin of the statement as a cognitive shortcut to assess its credibility and desirability, often more than they rely on a thorough analysis of the facts (Zaller, Reference Zaller1992).

Therefore, individuals may use a cue, that is, the fact that something is said by a political authority, to assess the credibility of the message, and this might affect an individual’s preference about a certain issue. If this mechanism holds, then there could be an effect due to the political authority, which we hypothesize as follows:

Fact-checking represents one of the most widespread innovations to combat the spread of misinformation. It is a practice that involves systematically verifying the accuracy of statements made by public figures and institutions, with the explicit goal of determining their veracity (Walter et al., Reference Walter, Cohen, Holbert and Morag2019). As “alternative facts” permeate modern democratic politics, fact-checking has become essential for verifying and challenging politicians’ claims (Barrera et al., Reference Barrera, Guriev, Henry and Zhuravskaya2020). The credibility and neutrality of fact-checkers should make their assessments more influential and effective, especially in situations with conflicting information that creates uncertainty about whom to trust (Walter et al., Reference Walter, Cohen, Holbert and Morag2019). Although fact-checking is a valuable tool against misinformation, a recent study revealed significant limitations in both its reach and coverage (Wack et al., Reference Wack, Duskin and Hodel2024). Users engaged with false content have only a slim chance of later encountering fact-checks. Even when misinformation is corrected, the audiences that would benefit most are often the least likely to see the corrections. Moreover, because false claims vary widely in form and salience, not every information is verified.

Despite the expansion of fact-checkers and their increasing integration into traditional journalism, suggesting a reduction in the impact of misinformation, the available empirical evidence is mixed. Some research shows that fact-checking can effectively limit the spread of erroneous information (Fridkin et al., Reference Fridkin, Kenney and Wintersieck2015), while others do not find significant effects (Garrett and Weeks, Reference Garrett and Weeks2013). The literature has extensively demonstrated that the presentation of corrective information can reduce agreement with statements initially influenced by misinformation (Nyhan and Reifler, Reference Nyhan and Reifler2010; Lewandowsky et al., Reference Lewandowsky, Ecker and Cook2017; Vraga and Bode, Reference Vraga and Bode2017). Lewandowsky et al. (Reference Lewandowsky, Ecker, Seifert, Schwarz and Cook2012) show that fact-based corrections can substantially limit the impact of misinformation on people’s beliefs. Porter et al. (Reference Porter, Velez and Wood2023) find that factual corrections by fact-checker members of the International Fact-Checking Network regarding COVID-19 reduced misinformation, highlighting the enduring effectiveness of corrective information.

The perceived credibility of the source plays a crucial role in determining the effectiveness of fact-checking. Although the impartial nature and authority of fact-checkers should make their judgments more influential, especially in cases where there is a large amount of conflicting information that complicates the distinction between true and false (Walter et al., Reference Walter, Cohen, Holbert and Morag2019), distrust in institutions and experts continues to push many people toward alternative, often less reliable sources (Bash et al., Reference Bash, Meleo, Fera, Jaime and Basch2021). Corrections from authoritative sources such as scientific institutions tend to be perceived as more credible and effective (Vraga and Bode, Reference Vraga and Bode2017; Seo et al., Reference Seo, Xiong, Lee and Lee2022). As highlighted by Shin (Reference Shin2024), the treatment of scientific topics by specialized sources significantly increases the perception of credibility. Conversely, subsequent exposure to post-truth comments, even if corrected from less authoritative sources like bloggers, reduces the effectiveness of the correction (Stedtnitz, Reference Stedtnitz2020). We expect a correction effect to exist. However, the effectiveness of correction is stronger when a scientific/official source is presented compared to when a fact-checking website is used as a source of correction. Therefore, we hypothesize that:

Misinformation and partisanship

Existing research indicates that even individuals who possess substantial political knowledge frequently align their policy views with those advocated by their preferred party (Cohen, Reference Cohen2003). Partisanship functions as a social identity, a subjective perception of belonging to a group, that remains remarkably stable even amidst shifts in political platforms, party leadership, or governance performance (Mason, Reference Mason2015). Partisan identity encourages individuals to defend the preferred party, even when confronted with negative information, and generates strong emotions driving political actions to preserve and amplify the party’s electoral influence (Taber and Lodge, Reference Taber and Lodge2006).

Individuals categorize themselves into distinct social groups, fostering a sense of belonging that significantly shapes their cognitive and behavioral processes. Identification with one’s own group (in-group) often results in a preference for information that reinforces shared identity and values, while information from external groups (out-groups) is more likely to be dismissed or devalued (Tajfel, Reference Tajfel1978). This dynamic can amplify confirmation bias and motivated reasoning, as individuals are inclined to accept and propagate information aligning with their social group’s beliefs, thereby perpetuating and solidifying misperceptions. For instance, Swire-Thompson et al. (Reference Swire-Thompson, Berinsky, Lewandowsky and Ecker2017) investigated the influence of source credibility on how individuals assess the veracity of information originating from a political authority. Their results indicate that when information was attributed to Donald Trump, Republican supporters were more likely to regard it as credible compared to unattributed information, whereas Democrats displayed the opposite response. Another study finds that users with the strongest partisan identities are the most prolific spreaders of misinformation (Osmundsen et al., Reference Osmundsen, Bor, Vahlstrup, Bechman and Petersen2021). Indeed, their antipathy toward the out-party outweighs in-group affection, and within this ecosystem, political elites could play both as spreaders and prime targets of the false narratives. Identity-driven social pressure turns the sharing of false news into both a badge of loyalty to one’s side and a weapon against the out-group.

Directional motives contribute to misinformation insofar as they bias the ways in which people seek out and evaluate political information (Lodge and Taber, Reference Lodge and Taber2013). Additionally, research on motivated reasoning and cognitive biases demonstrates that individuals select messages that reinforce their preexisting beliefs. Within this context, partisanship emerges as a critical driver of political misinformation, as group affiliation shapes not only individuals’ opinions and preferences but also how they interpret and respond to information (Van Bavel and Pereira, Reference Van Bavel and Pereira2018). Two theories help explain the adoption of inaccurate political beliefs (Vegetti and Mancosu, Reference Vegetti and Mancosu2020). Partisan motivated reasoning (PMR) suggests that individuals tend to believe falsehoods that confirm their existing partisan leanings (Flynn et al., Reference Flynn, Nyhan and Reifler2017). This leads them to embrace the positions of affiliated political leaders, propelled by motivated reasoning and partisan identification. Dual process reasoning, on the other hand, distinguishes between intuitive, automatic processing (System 1) and more deliberate, reflective thinking (System 2), emphasizing that political judgments often emerge from initial affective responses, later rationalized (Lodge and Taber, Reference Lodge and Taber2013). Partisan attachment exacerbates the acceptance of misinformation, as individuals are more likely to dismiss corrections or arguments from opposing groups while embracing narratives that align with their identities (Wagner et al., Reference Wagner, Tarlov and Vivyan2014).

The interplay between misinformation, partisanship, and elite influence underscores that the spread and persistence of false information are not solely due to a lack of accurate information but are connected to individuals’ political identity. Misinformation often aligns with partisan identities, which function as social frameworks that cultivate positive attitudes toward one’s in-group while amplifying negative perceptions of out-groups. Against this background, we hypothesize that:

Correction of misinformation and partisanship

Do corrections work in the context of partisanship? Studies on political misinformation have shown that fact-checking can significantly diminish belief in false claims (Prike et al., Reference Prike, Reason, Ecker, Swire-Thompson and Lewandowsky2023). This effect has been observed across the political spectrum, as both supporters and nonsupporters of Trump and Sanders adjusted their beliefs about these politicians’ statements following fact-checks (Swire-Thompson et al., Reference Swire-Thompson, Ecker, Lewandowsky and Berinsky2020). However, a significant challenge stems from individuals’ political affiliation, which can lead to motivated reasoning, making people resistant to updating their beliefs when misinformation from politicians they support is corrected. When a correction challenges these beliefs, partisans may reject it, perceiving it as a threat to the coherence of their values system (Flynn et al., Reference Flynn, Nyhan and Reifler2017).

What happens, however, when the correction is not accepted? Another relevant mechanism is the so-called backfire effect, which occurs when a correction not only fails to reduce belief in misinformation but strengthens it (Nyhan and Reifler, Reference Nyhan and Reifler2010). This effect is particularly evident in highly polarized issues, where the corrective message is perceived as a direct attack on one’s worldview. However, studies have downplayed the frequency of this effect, suggesting that resistance to correction may be better explained by the outright rejection of the message (Wood and Porter, Reference Wood and Porter2019).

Partisans are likely to consider not only the content of a message but also its source when interpreting new information. Before processing the message, they assess the credibility and authority of the source (Benegal and Scruggs, Reference Benegal and Scruggs2018). Consequently, misinformation tends to persist more among individuals with strong partisan identities, despite attempts to correct or debunk false claims and dismiss corrections that challenge the position of a politically aligned figure (Thorson, Reference Thorson2015; Amazeen et al., Reference Amazeen, Thorson, Muddiman and Graves2018). Evidence also indicates that partisan attachments can prompt people to discount fact-checking efforts contradicting statements made by politicians from their own party (Barrera et al., Reference Barrera, Guriev, Henry and Zhuravskaya2020). Benegal and Scruggs (Reference Benegal and Scruggs2018) have shown that climate change corrections are more effective for Republicans when coming from Republican-affiliated sources rather than from scientific sources. Similarly, Berinsky (Reference Berinsky2017) showed that corrections perceived as coming from an individual’s own political affiliation are more effective in rectifying misinformation than corrections from nonpartisan sources (e.g. academic institutions) or sources aligned with the opposing political camp. Another study finds that journalistic fact-checks enhanced the accuracy of respondents’ factual beliefs, including those held by partisan supporters of Trump. Nevertheless, the fact-checks did not affect individuals’ evaluations of the candidate or their voting intentions (Nyhan et al., Reference Nyhan, Porter, Reifler and Wood2020). These findings suggest that the impact of fact-checking efforts on correcting misinformation varies depending on the source (Liu et al., Reference Liu, Qi, Wang and Metzger2023). Indeed, the intrinsic nature of the source could itself significantly shape whether individuals accept or reject corrective information.

For instance, in the health domain, corrections issued by healthcare institutions or universities have proven more effective than those from user-generated content in diminishing false health beliefs (Walter et al., Reference Walter, Brooks, Saucier and Suresh2021). Nevertheless, this may not always be the case. While scientists are typically seen as credible and neutral authorities on issues like climate change, capable of persuading individuals through scientific evidence, partisanship increasingly shapes perceptions of their credibility (Motta, Reference Motta2017). Consequently, partisanship may undermine the effectiveness of corrective messengers. Accordingly, our last hypothesis is:

Research design

Findings

Figure 1 shows the predicted differences in agreement with the statement about the effects of the introduction of the minimum wage, the efficacy of the COVID-19 vaccination campaign, and the amount of weekly working hours between the control group (indicated by the dashed line) and the EC.Footnote ³ The figure indicates that the differences for the minimum wage and the vaccination campaign are not statistically significant. In contrast, the effect can be found for the working hour statement. Indeed, there is a difference of about 0.90 points, from about 5.19 in the control group to 6.08 in the EC 1 group. Figure 1 also shows the differences between ECs 2 and 3 – those in which the politicians’ misinformed statements are corrected by a scientific/official source or a fact-checking website – and the control group. Compared to when the misinformed statement by the politician is simply presented, correction occurs in two cases, namely the minimum wage and the working hours statements. When presented with the evidence that the misinformed statements about the effects of the minimum wage or the amount of worked weekly hours Italians work are incorrect, respondents tend to agree less with the fact that the introduction of the minimum wage increases wages not only for those earning less than the minimum but also for other workers, or that, in the EU, Italy is the country where people work the most, and this applies to both the correction by a scientific/official source and the correction by a fact-checking website. Those presented with the corrected statement by a scientific/official source or a fact-checking website about minimum wage have an average level of agreement that is, respectively, about 0.7 and 1.13 points lower than when the misinformed statement by Elly Schlein is not corrected. Similarly, those presented with the corrected statement about working hours have an average level of agreement that is, respectively, about 1.06 and 0.88 points lower than the misinformed statement by Giuseppe Conte is not corrected.

Figure 1. Predicted differences with 95% confidence intervals between the control group (dashed line) and the experimental conditions (EC) of the agreement with the statements about the introduction of the minimum wage, the efficacy of the COVID-19 vaccination campaign, and the amount of worked weekly hours.

Therefore, agreement with the statements about the minimum wage and the amount of working hours decreased in the ECs 2 and 3 groups, that is, when respondents are presented with evidence contrary to what the politicians, showing that respondents react to the correction. Regarding the effectiveness of Italy’s vaccination campaign, participants exposed to EC 1 exhibited responses comparable to those of the control group, indicating no significant effect of political authority alone. In contrast, the scientific/official correction (EC 2) yielded a larger level of agreement (about 0.50 points), whereas the fact-checker-based correction (EC 3) yielded different levels of agreement with respect to the control group. This suggests that the scientific/official correction increased participants’ propensity to believe that the vaccination campaign had a significant impact on mitigating COVID-19 mortality.

We now evaluate the hypothesis that respondents who are partisans, that is, who express positive feelings toward the party of the politician making the statements, are more inclined to agree with the statements. Figure 2 shows the predicted differences in agreement with the statements between the ECs and the control group by levels of feeling toward the party. The figure shows that expressing a positive feeling toward the party of the politicians making the statements significantly increases the level of agreement with the statements about the role of minimum wage and the amount of working hours in Italy, but not in the case of the vaccination campaign. In the case of EC1, when the feeling for PD is completely negative (at 0), respondents have an agreement score that is about 0.6 points lower than in the control group. As feeling improves, for instance, at 5 and 10, agreement with the statement becomes higher than the control group, about 0.4 and 1.45 points higher, respectively. A similar pattern can be noticed by looking at the statement about the number of working hours. Therefore, it appears that having a more positive feeling toward the party of the politician involved in the statement makes the statement more credible. This implies that misinformation spreads more among those who feel close to a party than to another, at least for the two issues examined. These results confirm that the spread of misinformation is facilitated by partisan identification which explains how attachment to one’s political group leads individuals to accept and rationalize statements that align with their partisan identity, regardless of their veracity (Lodge and Taber, Reference Lodge and Taber2013; Flynn et al., Reference Flynn, Nyhan and Reifler2017), providing support for H3, that partisan respondents would show stronger agreement with the statement.

Figure 2. Predicted differences with 95% confidence intervals between the control group (dashed line) and the experimental conditions (EC) of the agreement with the statements about the introduction of the minimum wage, the efficacy of the COVID-19 vaccination campaign, and the amount of worked weekly hours, conditional on the feeling toward the PD, FDI, and M5S.

Next, are those expressing more positive feelings toward the party of the politician making the statement less receptive to the correction made by a scientific/official source or a fact-checking website? Figure 2 shows that respondents “react” to the correction, although depending on the source and the topic/politician. As far as the minimum wage experiment is concerned, the correction by a scientific/official source seems to have occurred. Yet, agreement with the statement seems to depend on the feeling toward the PD. In fact, the lower the feeling, the less individuals agree with the statement. In this case, those who are very close to PD (have a score of 10) accept the correction, but to a lesser extent with respect to individuals who have a negative feeling (have a score of 0). In fact, the confidence intervals do not overlap. Correction also occurs when this is made by the fact-checking source. The pattern is similar to the one seen before, although less marked. The second experiment shows that the level of feeling does not affect the agreement with the statement when this is corrected by a scientific/official or a fact-checking source, which is no different than in the control group. In the last experiment, we observed that individuals with completely positive feelings toward the M5S were more inclined to agree with the statement, although they were presented with a correction by an academic source (EC 2). In contrast, when feeling is completely negative, the correction occurs, as the level of agreement is lower than in the control group. Eventually, a correction seems to occur when individuals are presented with evidence from a fact-checking website, as there are no differences compared to the control group.

All in all, these results seem to contrast with H4, that partisans would “resist” correction. In fact, we can notice this pattern only in the third experiment, when partisan individuals agree with the statement, although they are exposed to the evidence of a scientific/official source. In the other cases, it seems that positive feelings toward the party of the politician involved do not translate into a stronger agreement with the statement to be evaluated. It might be that the effectiveness of corrections often depends on the type of political issue (Jerit and Zhao, Reference Jerit and Zhao2020). When a topic is closely tied to a person’s political or ideological identity, they tend to interpret information in a way that protects those ties. For this reason, it is challenging to correct misinformation on controversial topics or prominent political figures, as such corrections can challenge the worldview of the recipient (Ecker and Ang, Reference Ecker and Ang2019).