Does the platform matter? Social media and COVID-19 conspiracy theory beliefs in 17 countries. Yannis Theocharis et al. New Media & Society, October 9, 2021. https://doi.org/10.1177/14614448211045666
Abstract: While the role of social media in the spread of conspiracy theories has received much attention, a key deficit in previous research is the lack of distinction between different types of platforms. This study places the role of social media affordances in facilitating the spread of conspiracy beliefs at the center of its enquiry. We examine the relationship between platform use and conspiracy theory beliefs related to the COVID-19 pandemic. Relying on the concept of technological affordances, we theorize that variation across key features make some platforms more fertile places for conspiracy beliefs than others. Using data from a crossnational dataset based on a two-wave online survey conducted in 17 countries before and after the onset of the COVID-19 pandemic, we show that Twitter has a negative effect on conspiracy beliefs—as opposed to all other platforms under examination which are found to have a positive effect.
Keywords: Affordances, conspiracy theories, COVID-19, misperceptions, social media
Conclusion
While the Internet has always served as a meeting place for fringe groups and conspiracy theorists, social media have added a new layer to this reality. Aided by the platforms’ interactive and networking features, as well as their capacity to deliver different kinds of content to very different audiences, social media have become hotspots for unsubstantiated information and the diffusion of misperceptions. Nevertheless, not all social media platforms should be painted with the same brush as different architectural features and affordances of social media platforms have consequences for how users encounter content and others with whom they can interact and build relationships (Bossetta, 2018).
The outbreak of the COVID-19 pandemic gave rise to many conspiracy theories, providing us with a unique opportunity to study conspiracy theory proliferation in platforms during the initial phase of the pandemic. We theorized that variation across different features of social media platforms might make some platforms more likely to accommodate conspiracy theory beliefs than others. Our results support the hypothesis that Twitter has a negative effect on conspiracy beliefs, while use of Facebook, YouTube, Messenger, and WhatsApp were found to have positive effects. For Facebook and its private messaging counterpart Messenger, it should, however, be noted that coefficients are only significant for the pooled models, when country effects were not weighted. Although this suggests that effects might not be uniform across countries, scrutinizing country differences is beyond the scope of this study and could make for a valuable endeavor for future research.
Our study makes several contributions. We show that not all social media platforms are the same when it comes to conspiracy theory beliefs about COVID-19. Our findings resonate with the core theoretical tenets of affordances theory, that there is a multifaceted relational structure between a technological tool and the user which might enable (in the case of Facebook, YouTube, and Messenger services) or constrain (in the case of Twitter) behavioral outcomes in a particular context (Evans et al., 2017: 36). This has implications for theory-building. Understanding how the spread of conspiracy theories differs across social media platforms is key to developing strategies to correct misperceptions, as well as to theorizing about how different features lead to different information diffusion dynamics. Our design does not allow us to tease out the specific effect of different affordances which would help understand why precisely we observe these effects, but the main finding lays the ground for future research zooming into individual affordances (network features, type of content, etc.) and studying the particular dynamics they give rise to. Specifically, a number of platform-specific features that we discussed earlier may have a link to how CTB proliferate. Future research focusing on the micro-mechanisms of the effects tied to specific affordances could consider a number of possibilities when studying, for example, Twitter’s distinct effect in comparison to other platforms. Twitter’s users combine higher than average education with a greater tendency for news-seeking and engagement into political discussions than any of the platforms in our study. This could imply that a larger number of users with potentially high-quality information sources were there to create content which, due to the asymmetrical structure of connections on the platform, can reach very far very fast through retweeting that cuts across different types of networks. In this fashion, it is possible that conspiratorial content—when it appeared—could be debunked fast or possibly “drown out” with better quality information or the sheer volume of those willing to quickly jump in and correct misperceptions. This is in contrast to platforms like Facebook or Messenger services, where the networks are not only more homogeneous, but countering opinions may be harder to emerge for different reasons. For example, precisely because of Facebook’s more family-and-friends oriented connections, users might think twice before attempting to correct conspiratorial content, as they are more likely to have to face the cost of jeopardizing social relationships. Indeed, the topic of how to talk to friends and family sharing conspiracy theories on social media became the subject of many articles in reputable news sources once conspiracy theories started to emerge (Warzel, 2020). But while this might be very relevant to Facebook users, it is not a barrier to Twitter users who could have happened upon a conspiratorial content and decided to interject (or hijack) a discussion with their own evidence. The contrast between Twitter also holds for YouTube where, due to the platform’s architecture, upon watching a video users have to move to the comments to encounter any debate about the content, thus possibly having less exposure to corrective information other than those flagged by the platform (when that happenes).
Different approaches to platform governance might have also played a significant role, meaning that some platforms need more oversight than others. Twitter quickly put in place measures such as deprioritizing content that could pose risks to people’s health along with labels and warning messages to content that contained COVID-19 misinformation. Facebook, which eventually instituted similar measures, continues to face criticism for being too slow to act on groups profiting from COVID-19 conspiracy theories (Jackson et al., 2021) and it is possible that related content or groups were able to gain substantial audiences before they were blocked (Marchal and Au, 2020). Finally, as none of the two messenger services has the type of warnings or content removal practices implemented on Twitter or Facebook, letting private conversations (often among large groups) be unmoderated could have increased the likelihood for conspiratorial content to proliferate on these. Zooming into the specific affordance-related mechanisms would help identify which parts of the conspiracy theory diffusion chain platforms need to work harder on to make their products safer, especially when it comes to public health. While the COVID-19 pandemic should serve as an important case study on how platforms react when scientific consensus in relation to content previously labeled as misinformation shifts, our findings add to the mounting evidence enticing social media platforms to self-reflect on the information quality in their environments and its potential broader effects on citizen attitudes and beliefs that are essential for public choices.
While our study has focused on conspiratorial beliefs about COVID-19, there is a little theoretical reason to believe that the effects we uncover would not apply to other types of conspiracy theories. This reasoning implies that our study has applications to a far larger problem than COVID-19-specific conspiracy theories. Future research should re-examine the connection of platform type and conspiracy theory beliefs using other conspiracy theories.
Our study does not come without limitations. The most important one is selection bias, which our data and design do not allow us to settle definitively. We have provided a strong theoretical rationale that platforms’ diverse features might be responsible for the way in which each platform is connected to conspiracy theory beliefs. It is, however, also possible that users’ particular characteristics or motivations may shape their decision to use this or the other platform precisely because it can offer the type of environment that fits their individual or community needs best. In recognition of this limitation, we employed propensity score matching which enabled us to find users with comparable characteristics to non-users across platforms, thus providing an additional test that our assumptions about the role of platforms are robust. This is an important test as, compared to regression-based techniques, this technique at least does not rely on out of data range extrapolations. Nevertheless, the ideal design for disentangling the causal order in this puzzle is, ultimately, a randomized experimental design. Random assignment into platforms could come with significant challenges, however, given the social media saturated environment of our times, and the difficulties of compliance with being active on a singular platform (Theocharis and Lowe, 2016). Ultimately, one would have to trade external validity for internal validity, which is why no single study design can stand alone.
Despite these shortcomings, our study provides important evidence that future studies based on other designs can build upon to address an issue of increasing importance given the role of platforms in people’s media and socialization diets.
Posts
