COVID-19 risk perception measures: factoring and prediction of behavioral intentions and policy support

AbstractAlthough early concepts of risk perception measures distinguished cognitive from affective items, until recently multi-dimensional taxonomies were absent from risk perception studies, and even more from tests of their association with behavior or policy support. Six longitudinal panel surveys on U.S. COVID-19 views (n = 2004 February 2020, ending April 2021) allowed testing of these relationships among ≤ 10 risk perception items measured in each wave. Confirmatory factor analyses revealed consistent distinctions between personal (conditioning perceived risk on taking further or no further protective action), collective (U.S., global), affective (concern, dread), and severity (estimates of eventual total U.S. infections and deaths) measures, while affect (good-bad feelings) and duration (how long people expect the outbreak to last) did not fit with their assumed affective and severity (respectively) parallels. Collective and affective/affect risk perceptions most strongly predicted both behavioral intentions and policy support for mask wearing, avoidance of large public gatherings, and vaccination, controlling for personal risk perception (which might be partly reflected in the affective/affect effects) and other measures. These findings underline the importance of multi-dimensionality (e.g. not just asking about personal risk perceptions) in designing risk perception research, even when trying to explain personal protective actions.Keywords: behavioral intentionsCOVID-19policy supportRisk perceptiontaxonomy Disclosure statementNo potential conflict of interest was reported by the authors.Notes1 A corollary might be that the global risk perception measure also belongs in this cluster, particularly for duration, which does impose a geographical limit on the area where the pandemic “ends.” A separate analysis (unreported here) showed results similar to those for this fifth model.2 Backup exploratory factor analyses for Waves 2-6 identified six factors out of the 10 items: collective, severity (infection, deaths), personal, affect, duration, and dread. Concern loaded on both collective and personal factors (> .49 and > .41, respectively). The personal connection might be prompted by the measure’s reference to “where you live”; its association with collective measures is unclear. Models clustering personal, collective, and concern measures, including affect and duration as single-item factors, had poor fit (e.g. Wave 2: chi-square/df = 26.849; RMSEA = .127 [.118, .135]; CFI = .928; AIC = 42,991.443).Additional informationFundingThe work contributing to this article was funded by the United States National Science Foundation under Grant No. 2022216.