(C) PLOS One [1]. This unaltered content originally appeared in journals.plosone.org.
Licensed under Creative Commons Attribution (CC BY) license.
url:https://journals.plos.org/plosone/s/licenses-and-copyright

------------



A darkening spring: How preexisting distrust shaped COVID-19 skepticism

['J. Hunter Priniski', 'Department Of Psychology', 'University Of California', 'Los Angeles', 'Ca', 'United States Of America', 'Keith J. Holyoak', 'Brain Research Institute']

Date: 2022-02 

Prewave attitudes

To identify which beliefs shaped COVID-19 skepticism before the first wave of cases struck the U.S., we predicted COVID-19 skepticism as a function of all belief predictors measured in the prewave study and the participant’s political stance towards social issues. Because the predictors are strongly correlated (by design, as correlations between predictors is necessary to measure coherence), a high degree of collinearity between predictors may hinder interpretation of the findings. In Bayesian linear models, collinearity between predictors results in flat (uninformative) posteriors estimates, even if one of the predictors in fact is predictive [37]. But as will be seen, this is not the case in our full model (reported below), and hence is not a concern for these analyses.

Survey data and R scripts for reproducing the reported analyses and figures are available with open access at the Open Science Framework: https://osf.io/8xerq/?view_only=df620c3e86984668ae1a225028b5cd9b. For details of data analyses, see Supplemental Online Materials.

For the prewave survey, Fig 2 shows the effect for each of the predictors in the model on COVID-19 severity, where more negative values imply the belief had a stronger negative effect on COVID-19 severity judgments (i.e., increased COVID-19 skepticism). Distrust in the intentions of Democratic politicians and fear of a COVID-19 vaccine were the only two reliable predictors of taking the virus less seriously; foreign threat assessment was the only predictor of taking the virus more seriously.

Fig 3 shows the conditional effects of the three most reliable predictors of responses on the COVID-19 severity scale. These are conditional effects (i.e., the relationship between the two displayed variables is conditioned on the mean value for the remaining scales). Two of the predictors, distrust of Democratic politicians (Fig 3A) and COVID-19 vaccine fears (Fig 3B), predicted taking the virus less seriously; while one of the predictors, foreign threat (Fig 3C), predicted taking the virus more seriously. In addition to the model estimates, Fig 3 displays each participant’s averaged response on the scale, colored by political stance. This jittered data, along with the kernel density plots in the margins of the panels (a-c), highlight the complex relationship between political ideology and COVID-19 skepticism. The three density plots running horizontally at the top of each of the three panels show the degree of separation between how Democrats (blue) and Republicans (red) responded on the three measures, with Democrat responses being skewed to the right and Republican responses to the left. Note that political polarization (the degree of separation between Democrats and Republicans) is far larger for the three predictors than for the COVID-19 severity scale itself: the separation between the horizontal blue and red density functions at the top of each of the three panels is much greater than the corresponding separation for the vertical density functions on the right of the figure (i.e., for the COVID-19 severity scale).

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 3. Histograms of three strongest predictors of prewave COVID-19 skepticism and their relation to political ideology. Regression lines represent effect size estimates conditioned on the average response for the remaining predictors in the maximal model. Error regions represent 95% credible intervals (the interior 95% of the posterior distribution for the effect size). Participant-level averaged responses on the scales for those who identified as socially liberal (blue) and conservative (red) are jittered with kernel density plots representing the distribution of responses for each predictor. Responses of participants who identified as moderate were excluded from visualization but were included in the model. Horizontal density functions at top of each panel summarize distribution of responses by Democrats and Republicans on each predictor variable; vertical density functions at right summarize distribution of responses on the COVID-19 severity scale. https://doi.org/10.1371/journal.pone.0263191.g003

In fact, posterior estimates of the regression coefficients from the maximal model revealed that after other beliefs were accounted for, there was no direct effect of political affiliation on COVID-19 skepticism (b = 0.02, 95% CI [-0.04, 0.08]). This finding indicates that beliefs other than direct political attitudes were the most effective predictors of prewave COVID-19 skepticism. Foreign threat assessment (believing that the U.S. should halt all immigration, coupled with xenophobic attitudes, such as that contact with Chinese people should be avoided to reduce the risk of contracting the virus) was the sole predictor of taking the virus more seriously (b = .10, 95% CI [0.03, 0.17]), and was the belief most reliably correlated with conservative attitudes (r partial = .36). Pairwise partial correlations between all the predictors measured in the survey are shown in Fig 4. The stark political polarization surrounding foreign threat calculations, and its positive effect on rated COVID-19 severity, illuminates how political attitudes shaped COVID-19 attitudes indirectly through auxiliary beliefs.

The strongest predictor of undervaluing the health risks associated with COVID-19 was distrusting the intentions of Democratic politicians (e.g., believing Democrats exaggerated the COVID-19 health risks for political gain; b = -0.61, 95% CI [-0.71, -0.51]). Distrust of the media’s coverage of the virus was moderately correlated with distrust of Democratic politicians (r partial = .65); however, media distrust only weakly predicted COVID-19 severity (b = -0.09, 95% CI [-0.19, 0.00]). As these two predictors were uniquely correlated with one another, it suggests that these two beliefs represent a more general ontology of political distrust, which served to undermine propensity to consider the pandemic to be a serious public health threat prior to the onslaught of cases.

The second strongest predictor of COVID-19 skepticism was fearing future COVID-19 vaccination requirements (b = -0.17, 95% CI [-0.26, -0.07]). Even though vaccines had not yet been developed (indeed, COVID-19 had not yet impacted the U.S.), pre-seeded fears of the government requiring vaccination played a role in sowing early distrust about the virus. These fears moderately cohered with distrust of large medical organizations (r partial = .43), which was moderately predictive of COVID-19 skepticism (b = -0.11, 95% CI [-0.22, -0.01]). Vaccine fear also cohered with concerns about the origins of the virus (r partial = .28), which was not predictive in the full model (b = -0.02, 95% CI [-0.12, 0.07]), and with evaluations of foreign threat (r partial = .26). Given that distrust of large medical organizations and having concerns about the origins of the virus were also weakly correlated (r partial = .26), this overall pattern suggests these three beliefs formed a tightly connected triad acting as an ontology of medical science distrust, which may have operated independently of political distrust in shaping COVID-19 skepticism.

Two sources of evidence thus suggest that two separable ontologies of distrust independently shaped skepticism before the first wave of cases. First, partial correlations between the predictors reveal two clusters of predictors––distrust in Democratic politicians and media, versus COVID-19 vaccine fears, distrust in medical organizations, and beliefs about the origin of the virus. Each cluster is internally correlated, but the two clusters are relatively independent of one another. Second, comparing the leave-one-out (loo) cross-validation accuracy of the full model with subsets of that model reveals that the most parsimonious, best-predicting model includes only distrust of Democrats and COVID-19 vaccine fears as predictors of COVID-19 skepticism. As shown in Fig 5, which shows the loo cross-validation accuracy of the full model and multiple subset models, the model predicting COVID-19 severity as a function of distrust in democratic politicians and COVID-19 vaccine fears while excluding political attitudes (“Big 2—politics”) is the model with the smallest set of predictors with a predictive accuracy within the standard error of the most predictive model (the maximal model with all the predictors). Furthermore, the model that only includes participants’ political attitudes is by far the worst-fitting model. These results suggest that auxiliary beliefs, rather than political polarization per se, are critical in predicting COVID-19 skepticism.

[END]

[1] Url: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0263191

(C) Plos One. "Accelerating the publication of peer-reviewed science."
Licensed under Creative Commons Attribution (CC BY 4.0)
URL: https://creativecommons.org/licenses/by/4.0/


via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/