Who encounters disinformation online? Combining survey and web tracking data to investigate predictors of disinformation exposure

Christner, Clara; Makhortykh, Mykola; Gil-Lopez, Teresa (2022). Who encounters disinformation online? Combining survey and web tracking data to investigate predictors of disinformation exposure (Unpublished). In: ECPR General Conference. Innsbruck. 22-26 August 2022.

Full text not available from this repository.

The rapid spread of disinformation online poses a challenge for democratic societies. Defined as intentionally or knowingly false statements that are disseminated to reach a certain goal, disinformation raises concerns, such as increasingly misinformed and polarized societies or declining trust in traditional journalism. However, certain studies (e.g., Altay et al., 2021) argue that the prevalence and impact of disinformation might be overstated and vary among individuals. Despite the importance of this issue, our empirical understanding of the degree to which individuals are exposed to disinformation, remains limited. While some studies (e.g., Allcott & Gentzkow, 2017) provide evidence for an overall high (perceived) exposure to disinformation, studies relying on passive measurements (e.g., Guess et al., 2018) found that exposure to disinformation is low on average, but highly concentrated within specific groups.
To address these diverging assessments, we introduce a novel approach to examine online disinformation exposure on the individual level and how it relates to individual characteristics and political attitudes which were identified as influential in this context by earlier research. While some of their findings are contradictory, earlier studies (e.g., Reuter et al., 2019) showed that demographic characteristics (i.e., age, gender, education) affect disinformation exposure. In line with the theory of selective exposure, we know that political attitudes determine information consumption and that it also affects exposure to disinformation (Guess et al., 2018). Following this theoretical argument, previous research that highlighted a conceptual affinity between populism and disinformation, and previous findings, we assume that populist radical-right (PRR) attitudes predict disinformation exposure online. Finally, a higher trust in non-traditional media which could result in a higher reliance on social media and right-wing alternative media for political information consumption is decisive, since recent studies (e.g., Guess et al., 2020) have highlighted the significant role of these platforms in the spread of disinformation.
We combined survey and tracking data of German participants’ online information behavior using a tracking tool based on the screen-scraping approach (N = 594). To detect disinformation, we trained artificial neural network-based classifiers on a large corpus of disinformation (N = 861 disinformation items retrieved from Germanophone fact checking projects) and true information (3k news stories scraped from news websites). By applying the classifiers to the tracking data (N of web pages = 144,404) and manually verifying the classified items, we make two contributions: first, we provide an empirical assessment of actual exposure to disinformation in a multiparty European context, which goes beyond the usual focus on a US context. Second, we introduce an automated method for disinformation detection for Germanophone textual data.
Our findings revealed that exposure to disinformation was low on average and exposure to disinformation was highly concentrated, as only a small portion of users (18%) was exposed to disinformation online. Further, results from a zero-inflated Poisson regression showed that individuals with lower levels of education, stronger PRR attitudes, a higher trust in non-traditional media and a stronger reliance on social and right-wing alternative media were more likely to be exposed to more disinformation online.

Item Type:	Conference or Workshop Item (Paper)
Division/Institute:	03 Faculty of Business, Economics and Social Sciences > Social Sciences > Institute of Communication and Media Studies (ICMB)
UniBE Contributor:	Makhortykh, Mykola
Subjects:	000 Computer science, knowledge & systems 300 Social sciences, sociology & anthropology
Language:	English
Submitter:	Mykola Makhortykh
Date Deposited:	17 Aug 2022 08:07
Last Modified:	05 Dec 2022 16:22
Uncontrolled Keywords:	disinformation, Germany, web tracking, neural networks, methodology, automated content analysis, mixed methods
URI:	https://boris.unibe.ch/id/eprint/171933