1Conspiratorial and populist attitudes are related to believing fake news.
2People who get their news from social media (Whatsapp, Facebook, Twitter) are more likely to believe fake news.
3Corrective messages that call into question more strongly held personal beliefs and convictions may backfire and increase fake news belief.
4No source of corrective messages (institutional or individual) is more effective than any other: different types of sources create similar effects.
Whereas 85% of respondents considered it to be true that human activity is causing extreme weather events, the level of belief in the other claims (all of them false) was highly varied. Half of the respondents believed that patent holders were limiting the supply of cancer drugs in order to boost their profits or that genetically modified foods were unsafe. Only around 10% believed that the Spanish government was planning to replace language classes with religion classes in schools.
Populism, conspiratorial thinking and fake news
In countries across the world, online misinformation – or ‘Fake News’ – is distorting elections with populist leaders and parties exacerbating this challenge, often employing conspiracy theories and other forms of political misinformation to attract voters. For instance, populist narratives attribute blame for adverse events or conditions on powerful and undeserving groups or elites who conspire to hide their influence from the general public (Hawkins 2009; Mudde and Rovira Kaltwasser 2012; Mudde and Rovira Kaltwasser 2013). Despite the growing importance of fake news in politics, we know little about which types of people are most susceptible to fake news and about how to effectively combat it.
To shed light on these questions, we conducted survey experiments in Spain with a sample of 8,000 participants that resembled the Spanish population in terms of gender, age, education, and region of residence. We focused in particular on the role of populist and conspiratorial attitudes, examining whether voters who hold these attitudes are more susceptible to fake news and/or resistant to fact checks.
Respondents began the experiment with a “practice question” in which they indicated their belief in the true claim that human activity is causing more extreme weather events. Respondents then read short news articles about several false claims, all of them related to fake news items that had circulated in Spain shortly before the experiment was conducted. For the purposes of this research, the items were divided into four groups of fake news with different degrees of populism and conspiratorial content (table 1). All respondents read one claim from each of the following four groups, and were then asked to indicate whether they considered that the information was very accurate, somewhat accurate, somewhat inaccurate or very inaccurate.
Figure 1 shows that 85% of respondents considered that the first news item, i.e., the story that linked human action to climate change, was true (as, indeed, is the case). However, belief in the other claims, all of which are false, varied. At the high end, more than half of respondents (52.1%) believe that patent holders are restricting the supply of cancer drugs in order to drive up prices.
At the other extreme, only around one-in-ten respondents (11.3%) believe that the central government is considering ways to reduce language classes and replace them with religious studies. In turn, 18.6% of respondents believed that vaccination increases the risk of autism, a false claim with potentially very serious implications for public health. Although this figure is comparatively low, it is still troubling that nearly one in five respondents holds a misperception that could lead them to eschew potentially life-saving vaccines.
Who is more likely to believe in fake news? And what role do populism and conspiratorial thinking play? Populist rhetoric emphasizes the division between ‘the people’ and ‘the elites’ embedded in a moral (Manichean) discourse, a ‘good-versus-evil’ understanding of the world and politics (Mudde and Rovira Kaltwasser 2013; Hawkins 2010).
Conspiratorial rhetoric, too, emphasizes secret plots by powerful actors, but reaches beyond the political sphere to include broader societal phenomena. In our survey, we measured respondents’ affinity towards populism and their degree of conspiracism using scientifically validated attitudinal scales (Castanho et al. 2019 and Bruder et al. 2013). For the analyses below, we divided our sample into different groups: respondents above and below median populism respectively, and respondents above and below median conspiracism.
Figure 2 reveals similar patterns between populist predispositions, conspiratorial thinking, and belief in false claims. More than 55% of respondents with high levels of both populist and conspiratorial predispositions are susceptible to misinformation involving genetically modified foods, a claim that does not involve any elite actor. In addition, the same type of respondent is more prone to accept news stories that do involve colluding elites: for example, the claims referring to pacts between parties to block the participation of the opposition, or those dealing with clandestine NATO operations or profit-seeking pharmaceutical patent holders. However, it is interesting to note that respondents with high levels populist or conspiratorial attitudes do not differ in their reaction from the rest of the respondents when it comes to the claims about changes in the school curriculum, even though both stories involved a clearly identified elite actor (in this case, central government).
Are factors other than populist or conspiratorial attitudes related to belief in false claims? The results of a series of regression models, shown in table 2, point to a striking correlation: respondents who reported seeking news on social media (Whatsapp, Facebook, Twitter) are more likely to believe all eight false claims included in our study. While these results cannot be taken as causal, they nonetheless suggest that social media use may increase exposure to and belief in false news stories.
Respondents we class as “populist” show a greater propensity to believe five out of the eight fake news items, particularly those involving colluding elites (political or otherwise). Respondents with a conspiratorial mentality, on the other hand, gave more credit to items that suggest broader conspiracies: claims about withheld medicines, clandestine fumigation or the risk posed by genetically modified foods.
Are there partisan differences in belief in fake news? In our study, Vox and Podemos voters stand out, as they appear inclined to believe claims that are consistent with their political preferences. This is consistent with a psychological process known as motivated reasoning, in which people unconsciously seek to validate their pre-existing beliefs when processing information, resulting in a cognitive bias (Flynn et al. 2017).
The results suggest that affective evaluations of the elites involved in the claim (PP for Podemos voters, PSOE for Vox voters), together with the salience of the issue depicted, played a fundamental role for respondents’ belief in the accuracy of the false claim. Similarly, conservative respondents appear inclined to lend credence to false claims when they align with preconceived ideological constructs.
1. Partisan differences in conspiracism, populism and fake news belief
The changing nature of the Spanish party system begs the question of whether voters differ in their level of populist and conspiratorial predispositions across the five main nation-wide parties. Our results show that both populism and conspiracism attitudes are relatively widespread among respondents, which is in line with previous studies (Van Hauwaert et al. 2016).
Our data show that Podemos and Vox supporters are on average more populist or conspiratorial than other voters (figure 3). However, results also highlight that we can indeed find populist and conspiratorial voters in all of the five analysed parties.
What patterns emerge if we simultaneously take into account political affinity, populist predisposition, conspiratorial mentality and use of social media? We repeate our previous regression analyses, this time separately for the four groups of respondents classified previously, namely those with high and low levels of populism (table 3) and those with high and low levels of conspiratorial thinking (table 4).
The results again highlight the strong correlation between the use of social media to seek news and the predisposition to believe false information. But in this case one specific characteristic stands out: those most affected by social media use when it comes to belief in fake news are voters with lower conspiratorial or populist tendencies. This troubling finding suggests that social media could promote fake news belief even among people who originally do not show other predispositions for doing so.
In relation to political affinity, the results lend further credence to the importance of motivated reasoning. Highly populist Vox, PP and Ciudadanos voters are more inclined to believe the false claim about a secret pact between PSOE and Podemos. This result also holds for highly conspiratorial Vox voters. However, Vox supporters showing low levels of populism or conspiratorial thinking are not more predisposed than other respondents to believe this fake news item. We find a similar pattern for populist and conspiracist Podemos voters when the opposite conspiring elite is comprised of PP, Ciudadanos and Vox.
2. Can fake news be corrected?
Our study also tested the effectiveness of fact checks, i.e., information that corrects the fake news by offering counterarguments. Thus, the experiment presented some randomly selected participants with items that refuted the central claim of the fake news story, while other participants (the control group), read only the fake news headline and a short description. For instance, this is the headline, description, and fact check used in the genetically modified foods story:
Debate on the safety of genetically modified foods
Anti-GM food activists argue that the changes made can alter the nutritional content and introduce toxins that could harm vital organs.
However, the vast majority of scientists believe that genetically modified foods are safe. Furthermore, scientific studies have not shown any negative side effects from the consumption of these foods.
The effectiveness of the corrections, i.e., of introducing fact checks into the experiment, varied dramatically across the tested statements (figure 4). Fact-checking reduced belief in some false claims, namely about GM foods, the secret plot by PP, Ciudadanos and Vox in Andalusia, and the collusion of pharmaceutical companies. Yet, in other cases, adding a fact check to the information did not alter the degree of belief in the fake news in comparison with the control group, i.e. respondents who were not given the correction for that same story. This is the case for the stories about substituting language classes with religious studies, a secret pact between PSOE and Podemos in Valencia, and clandestine NATO fumigations.
But what is especially troubling is that, in two cases, corrections “backfired”: providing arguments against the fake news item not only fails to offset belief in the information but actually reinforces the reader’s degree of belief in the story. We observe this effect for the items about the link between vaccination and autism, and the introduction of Islamic studies into the school curriculum. In both instances, correction strengthens the belief in the fake claim.
The backfire effects we document are particularly noteworthy for their counterintuitive nature. However, once again this reaction can be explained with the process of motivated reasoning, the cognitive bias triggering a reaction in the opposite direction in the face of facts that call into question the most strongly held personal beliefs and convictions. Thus, when presented with fact checks, readers call upon other considerations that justify their (incorrect) beliefs, which in this way are reinforced. While ours is the first study to document this backfire effect of fact checking in Spain, similar reactions have been documented in the United States (Nyhan and Reifler 2010, Porter and Wood 2019).
3. Are fact checks from certain sources more effective?
In the experiment, the corrections to the fake claims were attributed to a variety of individual and institutional sources. Our goal was to determine whether certain types of people or institutions are uniquely suited to combat misinformation. Thus, for each of the eight items we created a list of domain-experts that appeared as the sources of the fact checks. For instance, in the GM foods and vaccine safety experiments, we attributed corrections to doctors and existing professional medical associations. In other cases, we included corrections attributed to government institutions, experts, a fictional NGO, news outlets and fact-checking platforms such as Newtral and Maldita Hemeroteca.
The pattern that emerges from the results (table 5) is that the response to the fake news was similar regardless of the type of source to which the fact checks were attributed. For instance, in the GM foods experiment, corrections from all three sources reduce belief in the false claim. Similarly, in the Islamic studies experiment, both corrections result in backfire. In some cases, the involvement of a source that a priori would appear to represent an authority on the topic causes no significant change in the attitude to the fake news story and the effects are in the expected direction, and in no case do we find that corrections from different sources push respondents in opposite directions.
While in some cases corrections from authoritative sources do not cause a significant change in the attitude to the fake claim, effects are in the expected direction, and in no case do we find that corrections from different sources push respondents in opposite directions. (A recent study of fake news in India reaches similar conclusions about the general effectiveness of corrections, regardless of source, see Badrinathan et al. 2020).
Our study sheds new light on the extent and correlates of fake news belief in Spain. The results indicate that people with conspiratorial and populist attitudes are more likely to believe fake news stories about a range of topics. Furthermore, the results suggest that attempts to fact check false claims often fail and, in some cases, may even backfire. Our findings add to a growing number of studies that document the difficulty of effectively combating misinformation.
For instance, Nyhan et al. (2014) document backfire effects among parents in the United States: vaccine-skeptical parents who received corrective information about vaccine safety reporting being less likely to vaccinate their children than skeptical parents who did not receive this information. Carey et al. (2020) uncover a “tainted truth effect” in which fact checks intended to correct false claims about the Zika and yellow fever epidemics in Brazil cause people to become less certain about true claims about the epidemics.
Viewed in light of this research, our results suggest the need for caution when employing fact checks intended to combat misinformation or fake news. Specifically, they point to the use of experimental trials to test potential corrective messages before implementing them in the real world. If practitioners implement untested anti-misinformation campaigns, they may fail to achieve intended results or, more troubling, reinforcing the misinformation they seek to combat.
BADRINATHAN, S., S. CHAUCHARD and D.J. FLYNN (2020): «I Don’t Think That’s True, Bro! » An Experiment on Fact-checking WhatsApp Rumors in India. Working Paper.
BRUDER, M., P. HAFFKE, N. NEAVE, N. NOURIPANAH and R. IMHOFF (2013): «Measuring individual differences in generic beliefs in conspiracy theories across cultures: Conspiracy Mentality Questionnaire», Frontiers in Psychology, 4(225).
CAREY, J., V. CHI, D. FLYNN, B. NYHAN and T. ZEITZOFF (2020): «The effects of corrective information about disease epidemics and outbreaks: Evidence from Zika and yellow fever in Brazil», Science Advances, 6(5).
CASTANHO SILVA, B., I. ANDREADIS, E. ANDUIZA, N. BLANUSA, and MORLET CORTI, G. DELFINO, G. RICO, S.P. RUTH, B. SPRUYT, M. STEENBERGEN and L. LITTVAY (2018): «Public opinion surveys: a new scale», en K.A. Hawkins, R. Carlin, L. Littvay and C. Rovira Kaltwasser (eds.): The ideational approach to populism: theory, method and analysis, Londres: Routledge.
FLYNN, D. J., B. NYHAN and J. REIFLER (2017): «The Nature and Origins of Misperceptions: Understanding False and Unsupported Beliefs About Politics», Political Psychology, 38(S1).
HAWKINS, K. (2009): «Is Chavez populist? Measuring populist discourse in comparative perspective», Comparative Political Studies, 42(8).
HAWKINS, K. (2010): Venezuela’s Chavismo and populism in comparative perspective, Cambridge: Cambridge University Press.
MUDDE, C., and C. ROVIRA KALTWASSER (2012): «Exclusionary vs. inclusionary populism: comparing contemporary Europe and Latin America», Government and Opposition 48(02).
MUDDE, C., and C. ROVIRA KALTWASSER (2013): «Populism», en M. Freeden, L.T. Sargent y M. Stears (eds.): Oxford Handbook of Political Ideologies, Oxford: Oxford University Press.
NYHAN, B., and J. REIFLER (2010): «When corrections fail: the persistence of political misperceptions», Political behavior 32(2).
NYHAN, B., J. REIFLER, S. RICHEY and G. L. FREED (2014): «Effective messages in vaccine promotion: a randomized trial», Pediatrics 133(4)
PORTER, E. and T. J. WOOD (2019). False Alarm. The Truth about Political Mistruths in the Trump Era. Cambridge University Press
VAN HAUWAERT, S., C. SCHIMPF and F. AZEVEDO (2016): «Measuring populism as a demand-side phenomenon: a psychometric evaluation of populist attitudes across Europe», trabajo presentado en la conferencia de Team Populism «The Causes of Populism», Provo, Utah, 28-30 de enero.
Social welfare systems and inequality in Europe
Spain’s social protection system is less redistributive than those of other EU countries. What reforms could help reduce economic inequality in Spain?
Detection of workplace bullying and its negative impact on psychological well-being
11.2% of the population present a high probability of suffering a situation of harassment in their workplace and may develop generalised anxiety disorder. How can possible cases of psychological harassment be detected?
Technocratic attitudes in Spain during the pandemic
In times of crisis, do citizens prefer to adopt a more technical type of government? According to this study, technocratic attitudes among Spanish people increased during the pandemic, especially among right-wing voters.
How do people’s acquaintances shape their support for economic redistribution and social protection?
We analyse how opinions on economic redistribution and social protection depend not only on family incomes, but also on the wages earned by people in the immediate social environment.
Gender inequality in paid and unpaid work after the pandemic
Following the pandemic, 30% of men and 33% of women with children who are minors have been working from home at least one day a week. According to this study, this could favour greater equality in relation to family responsibilities.