This paper investigates the regional differences in the spread of COVID-19 infections in Germany. A machine learning selection procedure is used to reduce variables from a pool of potential influencing variables. The empirical analysis shows that both regional structural variables and regionally aggregated personality traits are significant for the different corona spread. The latter characteristics express differences in mentality between the federal states. The north-east of the country shows a lower degree of affectedness. Regions with a high proportion of migrants show a higher incidence than others. If personality traits are neglected, the migrants' influence is overestimated. With school education and the risk of poverty, two further important regional characteristics are identified. Federal states that have a disproportionately high share of the population with low school education tend to have fewer COVID-19 cases. With regard to poverty, no clear statement can be made. The more the population tends to be responsible towards fellow human beings, the higher is the risk of a more pronounced spread. Where there is a tendency towards altruism, which consists of helping other people, a higher level of COVID-19 infections is revealed. A significant positive correlation between infections and testing is shown by the estimates. The link between vaccinations and the number of infections is less clear. Across the three corona waves,significant changes emerge. This relates in particular to the proportion of migrants and the proportion of families at risk of poverty. The effects decrease over the course of the pandemic.