What influences COVID-19 infection rates: A statistical approach to identify promising factors applied to infection data from Germany

Moritz Mercker; Uwe Betzin; Dennis Wilken

With respect to β 1 , this approach selected the variables Infection density and Retail recreation. The subsequent GAM-analysis with these predictors reveals a negative regression coefficient for infection density (-74.6, 95 %-CI: [-177.8,17.5] ) suggesting that higher local infection densities lead to stronger current decays of infection rates. The psychological effect of perceived abstractness of the threat might be the reason for this: as long as the total number of infected in an area is low, it is highly probable that an individual does not know anyone infected personally and the threat is perceived to be very abstract. If in contrast an area is affected by a high number of infections, it his more probable that someone knows somebody who is infected (and/or the region is more present in the media) and the threat is suddenly perceived as very real, which may induce a much stronger change to a more preventive behaviour. Furthermore, there is a highly significant positive regression coefficient for Retail recreation (0.061, 95 %-CI: [0.02,0.1]) indicating that a high use of places like restaurants, cafes, shopping centres, theme parks, museums, libraries, and movie theatres prevents the recent decay of infection rates. Even when most of these places are closed now, a previous intense use of these places might still influence current rates due to the delay processes in infections.