Affiliations
AffiliationsItem in Clipboard
Changes in risk perception and self-reported protective behaviour during the first week of the COVID-19 pandemic in the United StatesToby Wise et al. R Soc Open Sci. 2020.
. 2020 Sep 16;7(9):200742. doi: 10.1098/rsos.200742. eCollection 2020 Sep. AffiliationsItem in Clipboard
AbstractEfforts to change behaviour are critical in minimizing the spread of highly transmissible pandemics such as COVID-19. However, it is unclear whether individuals are aware of disease risk and alter their behaviour early in the pandemic. We investigated risk perception and self-reported engagement in protective behaviours in 1591 United States-based individuals cross-sectionally and longitudinally over the first week of the pandemic. Subjects demonstrated growing awareness of risk and reported engaging in protective behaviours with increasing frequency but underestimated their risk of infection relative to the average person in the country. Social distancing and hand washing were most strongly predicted by the perceived probability of personally being infected. However, a subgroup of individuals perceived low risk and did not engage in these behaviours. Our results highlight the importance of risk perception in early interventions during large-scale pandemics.
Keywords: COVID-19; coronavirus; pandemic; protective behaviour; risk perception.
© 2020 The Authors.
Conflict of interest statementWe declare we have no competing interests.
FiguresFigure 1.
Location of subjects and distributions…
Figure 1.
Location of subjects and distributions of responses to items regarding risk perception and…
Figure 1.Location of subjects and distributions of responses to items regarding risk perception and protective behaviour (n = 1591). (a) Number of responses in each state in the mainland United States. (b) Correlation between number of responses and state population, indicating that the number of responses was in line with expectations based on population. (c) Distribution of responses to risk perception items and (d) distribution of responses to protective behaviour items. All responses were recorded on a visual analogue scale ranging from 0 to 100. Bar plots indicate mean responses to these items over the two time points where a subgroup of subjects was retested (n = 375), and p-values represent results from a repeated measures t-test between the two time points. Subject-level changes over the two time points are shown in electronic supplementary material, figures S9 and S10.
Figure 2.
Changes in protective behaviours and…
Figure 2.
Changes in protective behaviours and risk perception over time. ( a ) Perceived…
Figure 2.Changes in protective behaviours and risk perception over time. (a) Perceived probability of becoming infected for participants themselves and average people at different geographical scales in separate samples tested over 5 days. (b) Perceived likelihoods of infection in a subset of subjects followed up after 5 days. (c) Reported likelihood of attending events with a given number of other people in separate samples tested on 5 days in the early stages of the outbreak in the United States. (d) Reported probability of attending events of different sizes in a subset of subjects followed up 5 days after initially completing the survey.
Figure 3.
Results of linear regression predicting…
Figure 3.
Results of linear regression predicting self-reported engagement in hand washing and social distancing…
Figure 3.Results of linear regression predicting self-reported engagement in hand washing and social distancing (represented by responses to an item regarding staying home) from measures of risk perception, with validation in a subsample of 25% of subjects. (a) represents the discovery dataset and (b) represents results from the validation dataset. Regression coefficients represent standardized β values.
Figure 4.
Results of Bayesian Gaussian mixture…
Figure 4.
Results of Bayesian Gaussian mixture model (GMM) decomposing response distributions for protective behaviour…
Figure 4.Results of Bayesian Gaussian mixture model (GMM) decomposing response distributions for protective behaviour items into clusters. (a) Mean scores for each component in the GMM model on the four items used to generate clusters. (b) Number of subjects assigned to each group. Four components were rejected due to having negligible weights (less than 0.01). (c) Z-scored responses on other questionnaire items for the low engagement and high engagement clusters, demonstrating how they compare to the average individual. (d) Results of logistic regression predicting membership of the low engagement group, showing beta coefficients from the model.
RetroSearch is an open source project built by @garambo | Open a GitHub Issue
Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo
HTML:
3.2
| Encoding:
UTF-8
| Version:
0.7.3