Two-thirds of American adults find that fake news contributes significantly to confusion about current political issues. The sense of disruption occurs across all social classes, income and education levels, and party lines. At the same time, most adults are confident that they themselves are good at spotting fake news. However, someone has to fall prey to fake news; otherwise, fake news would not spread so widely and would not be such a problem. Which social groups have particular difficulty identifying fake news as such? Who mainly contributes to the spread of fake news? Furthermore, does this happen intentionally or by mistake? Fake news refers to any dissemination of false or at least misleading information, regardless of whether it was put into the world accidentally or to deliberately mislead people.

In a recent research paper, we used large-scale surveys in Germany and the UK to show that older, male, high-income, and politically left-leaning respondents are particularly good at recognizing fake news. Accidental sharing of fake news is much more common than deliberate sharing regarding further dissemination. Accidental sharing of fake news is significantly more common in the politically right-wing spectrum and decreases with age. Deliberate sharing is found more frequently than average among younger UK respondents.

For the research project, close to 2400 people in Germany and the UK were studied in December 2021 about their fake news behavior. In the online survey, we presented the respondents with ten headlines published online, of which five were fake news and the other five described correct facts. The fake news headlines included, for example, "Pope Francis shocks the world: support for Donald Trump as president." An accurate headline was "Donald Trump nominated for Nobel Peace Prize 2021.” The headlines were presented as plain text, without source citation or illustration, to avoid influencing participants with further information or signals. We asked subjects to rate whether they were more likely to think the headline was true or false. From these assessments, we constructed an index of individual ability to identify fake news for each subject. Those who could correctly identify all headlines as accurate or fake news and were confident in their judgment would have reached the maximum index value of 40. Those who misidentified all headlines but still felt confident in their judgment would have achieved an index value of zero. As Fig.1 shows, the extreme values did not occur in either Germany (left) or the UK (right). On average, participants scored 23 points - marginally more in Germany than in the UK. As the figure also shows, there is considerable variation in identifying fake news.

Then we were interested in whether people shared fake news intentionally or accidentally and to what extent. For this, the participants were asked for each headline whether they would share the respective story online, e.g., on Facebook or via Twitter. On the one hand, we then measured deliberate sharing, i.e., how many of the headlines a participant had correctly identified as fake news they still wanted to share. On the other hand, we captured accidental sharing by the number of fake news statements that were not recognized as fake and that the participant still wanted to share. Fig. 2 shows the percentage of respondents who wanted to share at least one fake news statement. The rates are shown separately for deliberate and accidental sharing and the two countries. There was significantly more accidental than deliberate sharing of fake news. About 12 percent of respondents wanted to share at least one news item correctly identified as fake news. The difference in accidental sharing between the two countries is also striking. In Germany, only 6 percent accidentally shared three or more fake news headlines, while the ratio in the UK was 13 percent.

The survey also elicited much socioeconomic information from the participants. Within the regression analysis framework, both the ability to identify fake news and the behavior in further disseminating fake news can be explained by these factors. This provides answers to the question of which groups mainly contribute to the spread of fake news. Table 1 shows these socio-demographic factors and their influence on the identification of fake news (columns 2 and 3) and deliberate (columns 4 and 5), and accidental sharing (columns 6 and 7). Only the statistically significant influencing factors are marked, with a red field indicating a negative and a green area a positive influence. The respective comparison categories are given in brackets. Above-average identification of fake news headlines came from participants who were older and earned a higher income. Women and participants who placed themselves politically in the center performed worse in comparison. In Germany, education positively influenced the identification of fake news; in the UK also having a middle income. When it comes to deliberately spreading fake news, we only found significant results in the UK – likely due to the larger number of cases. Deliberate sharing of fake news occurred less among women, older and more educated people, and more by higher income earners and those on the political right. For the accidental spreading of fake news, similar patterns were found for both countries': older people contributed less to the unintended spread of fake news, and politically right-wingers significantly more.

Our findings help identify social groups that particularly contribute to the spread of fake news - be it through deliberate sharing or a lack of identification of fake news. This leads to the question of which measures could at least mitigate the rapid spread of fake news. A comparison of the two countries could be instructive here; participants in Germany were only moderately better at identification but significantly more cautious in mass disseminating fake news headlines. Although it is difficult to establish a causal link, the political debate (and related legislative initiatives) against hate speech on social media in recent years may have sensitized people in Germany to the problem of fake news and made them more cautious about sharing dubious information. The political debate about communication in social media may thus have had an effect similar to that of the "Accuracy Prompts," which is currently being studied in social media research. This involves reminding people that the accuracy of the information shared is vital to other participants. Crucially, participants will not be told what is right and wrong but will be prompted to think about the importance of correct information.