The psychology of fear and health caused by COVID-19 was more influenced by “micro-level” psychological factors than “macro-level” environmental factors. These findings, from a machine-learning analysis of survey responses, were published in PLOS One.

Adults (N=533) living in Austria (n=190), Poland (n=136), Spain (n=107), the Czech Republic (n=56), or Germany, the United Kingdom, Ireland, Italy, or Pakistan (n=43) were repeatedly surveyed over a 7-week period (March 16 to May 3, 2020). The surveys assessed demographics, psychological and social features, immediate personal effects from the pandemic, characteristics about local and national government policies (eg, quarantine, infection rates), the Perceived Vulnerability to Disease Scale, and Experiences in Close Relationships Revised. Responses were fit with a least absolute shrinkage and selection operator (LASSO) and extremely randomized trees (ERT) machine learning models.

The study participants were aged mean 30.48 (standard deviation [SD], 12.18) years and most (n=345) were women.

Continue Reading

The models were able to predict 35% of the variance of fear (LASSO: R2, 0.35; P =<.001; ERT: R2, 0.32; P =.001) and <10% of the variance of health (LASSO: R2, 0.09; P =<.001; ERT: R2, 0.05; P <.001).

Important predictors for fear of the virus included worrying about resource shortages (30.45%), perceived infectability (16.33%), germ aversion (12.21%), and infection in the social sphere (7.08%). Predictors of perceived health were infectability (32.21%), exercise (17.50%), adult attachment security (6.67%), and younger age (6.20%). The spread of the virus and government measures were not important predictors for either fear or perceived health.

This study found little evidence (~2% importance) of gender bias among responses, which contradicts results from similar studies.

This study was limited by not assessing baseline psychiatric conditions which may have significantly impacted participant responses.

These data indicated micro, personal factors were more important contributors to fear about COVID-19 than macro, broad factors, such as the government’s involvement.


Eder SJ, Steyrl D, Stefanczyk MM, et al. Predicting fear and perceived health during the COVID-19 pandemic using machine learning: A cross-national longitudinal study. PLoS One. 2021;16(3):e0247997. doi:10.1371/journal.pone.0247997.