Rolf Degen summarizing... Large-scale study of information behavior in times of coronavirus provides the umpteenth evidence that people do not entrench themselves in echo chambers

COVID-19 Echo Chambers: Examining the Impact of Conservative and Liberal News Sources on Risk Perception and Response. Kenneth A. Lachlan, Emily Hutter, and Christine Gilbert. Health Security, Jan 19 2021. https://doi.org/10.1089/hs.2020.0176

Rolf Degen's take: Rolf Degen on Twitter: "Large-scale study of information behavior in times of coronavirus provides the umpteenth evidence that people do not entrench themselves in echo chambers. https://t.co/8wZpV5otWk https://t.co/5f8eFQe3Co"

Abstract: The coronavirus disease 2019 (COVID-19) pandemic has created substantial challenges for public health officials who must communicate pandemic-related risks and recommendations to the public. Their efforts have been further hampered by the politicization of the pandemic, including media outlets that question the seriousness and necessity of protective actions. The availability of highly politicized news from online platforms has led to concerns about the notion of “echo chambers,” whereby users are exposed only to information that conforms to and reinforces their existing beliefs. Using a sample of 5,000 US residents, we explored their information-seeking tendencies, reliance on conservative and liberal online media, risk perceptions, and mitigation behaviors. The results of our study suggest that risk perceptions may vary across preferences for conservative or liberal bias; however, our results do not support differences in the mitigation behavior across patterns of media use. Further, our findings do not support the notion of echo chambers, but rather suggest that people with lower information-seeking behavior may be more strongly influenced by politicized COVID-19 news. Risk estimates converge at higher levels of information seeking, suggesting that high information seekers consume news from sources across the political spectrum. These results are discussed in terms of their theoretical implications for the study of online echo chambers and their practical implications for public health officials and emergency managers.

Discussion

The findings for Hypothesis 1—that information seeking does not motivate general risk perception—were somewhat unexpected given a lengthy history of research connecting information seeking to perceptions of risk severity. The findings for Hypothesis 2—that information seeking does not motivate mitigation—are also puzzling given a long history of research connecting risk information to protective action. This may be a product of the relative simplicity of CDC guidelines, and the fact that suggestions like wearing a facemask and washing one's hands do not require a great deal of effort. While CDC guidelines have shifted over the course of the pandemic, the individual-level recommendations captured here were fairly consistent during the time data was collected (April through June 2020). The mean across this outcome variable was fairly high for the entire sample (M = 6.11; SD = 1.36 on a 7-point scale), suggesting there may simply have been little variance in the outcome. Hypothesis 3—that information seeking would predict specific estimates of risk—was supported by our analysis. This can perhaps be traced to the underlying processes surrounding information seeking through websites. Whether seeking information from conservative or liberal sources, information seeking requires some degree of active processing. It may be that heavy information seekers, by definition, engage in active processing and are, therefore, better able to encode risks and process them into specific estimates of infection, health risk, and mortality.

Findings for the research question—concerning the moderating effect of reliance on conservative and liberal websites—may shed further light on the findings for all 3 hypotheses and are the most interesting and impactful findings in this study. With regard to echo chambers, our findings for the research question largely indicate that higher information seekers did not experience attitudinal polarization; in fact, across all 3 outcome variables the risk estimates for those reliant on liberal and conservative news content converged at higher levels of information seeking. In other words, lower information seekers, those reliant on conservative sources, reported the lowest levels of risk probability, whereas those reliant on liberal sources reported the highest (Figures 1, 2, and 3). At high levels of information seeking these differences disappear. Accordingly, the impact of information seeking on risk estimates is higher among those reliant on conservative websites, since they have further to go to converge; this is evident in the standardized conditional effects (Tables 1, 2, and 3).

In short, these findings run counter to the notion of echo chambers, and more closely approximate the argument of Messing and Westwood35—that those who engage in high levels of information seeking likely gather information from a range of sources. It may also be the case that high information seekers draw from a range of platforms and may be more open to information that does not align with (or challenges) existing attitudes and beliefs. This would also explain the failure to find a relationship between information seeking and both general risk perception and mitigation (Hypotheses 1 and 2). If high information seekers draw from liberal and conservative news sources, they would likely be exposed to or open to a range of perspectives, including those suggesting high risk and the need to take protective action. This exposure could potentially weaken direct effects between overall information seeking and the variable outcomes.

If politicized underreporting of the threats associated with COVID-19 is a concern, the lower information seekers may be more at risk, as this is where clear differences are evident in risk estimation by source preference. This finding is particularly alarming when considering Slater's25 arguments concerning polarization spirals; if low information seekers with polarized conservative opinions consume congenial information about the pandemic, and only congenial information, they may be likely to double down on their positions concerning specific risk estimates and become even more inclined to seek information that affirms those positions.

Although this is a single study in a highly specified context, health officials may wish to consider these findings when countering misinformation and understatements of risk. The most impressionable audiences may be those who seek the least amount of information and are, therefore, susceptible to information that confirms their biases. Identifying and segmenting these audiences along media preferences and demographic and social strata may enable health officials to target risk messages to those least likely to actively seek information.