This preprint explores important issues regarding right-wing political movements and their impacts on COVID-19 cases, however, both reviewers raise concerns about the theoretical and analytic approaches used.
RR:C19 Evidence Scale rating by reviewer:
Potentially informative. The main claims made are not strongly justified by the methods and data, but may yield some insight. The results and conclusions of the study may resemble those from the hypothetical ideal study, but there is substantial room for doubt. Decision-makers should consider this evidence only with a thorough understanding of its weaknesses, alongside other evidence and theory. Decision-makers should not consider this actionable unless the weaknesses are clearly understood and there are other theory and evidence to further support it.
The paper studies the correlation between the policy stand of the Austrian right-wing populist freedom party (FPÖ) and the number of Covid-19 deaths and cases per capita in Austria. The author shows that the vote share of the FPÖ can only predict the number of COVID-19 deaths per capita but not the number of cases after the policy switch.
The first part of the paper with OLS regressions is less convincing than the second part of the paper where the author uses SIRD models to show under what conditions their OLS results can be possible. The motivation for the use of SIRD models is the regression results presented in the first part. Therefore, the OLS regression issues have weakened the author’s argument.
Following the RR:C19 guidelines, I classify this paper as potentially informative. Here are my reasons:
The before and after policy-switch OLS regressions use different variables making them not comparable. Some of the variables that are present in after-policy switch regressions are not included in the before policy-switch regressions and vise versa. It would strengthen the paper’s argument if the variables included in before and after policy-switch regressions were the same. That way, comparing the coefficients of FPÖ vote share before and after policy-switch would have been more informative.
The logic behind choosing some of the variables is not very clear. For example, why the share of the population born in Turkey is included but not the share of the population born in any other country. If the reason is to control for ethnic backgrounds, it might be beneficial to include other groups as well. The same can be said about education. In Table 1, the author controls for education by using the share of university graduates but in Table 3, the share of the population who completed the most compulsory education is used to control for the education level.
A non-Austrian resident would not be familiar with the demographic trend of FPÖ supporters. Adding more information on age, education, and income distribution in FPÖ strongholds can provide a clearer picture of the characteristics of corona-sceptics in Austria.
My last point is about the second portion of the paper. In the SIRD models, parameter h (homophily of social contacts) plays an important role in explaining why it is possible to have a high number of deaths per capita and a low number of cases per capita at the same time. It is plausible to assume that people choose to spend their leisure hours with like-minded people. However, there are many instances that one cannot choose the group of people with who they interact with. For example, when one uses public transportation, or go to schools or universities or during a work meeting, or when one shops. These inevitable daily interactions make it difficult to justify the assignment of a high value to parameter h.