Lockdown blues in Berlin and Zurich
I live in Berlin and many of my friends and family live in Zurich. Like the rest of the world, we all felt the impact of Covid-19 in the past year but our experiences were somewhat different. The German and Swiss governments handeled the pandemic quite differently regarding regulations to contain the disease.
I have been wondering for a while, how the different lockdown situations affected the progress of infections. As a mother of two children, for example, I struggled with the closing of schools and often glanced at my Swiss friends somewhat enviously who sent their kids to school almost throughout the whole past year.
On the one hand I was simply interested in the incidences in both locations — were they similar or not? How did the temporal progress compare? On the other hand I wanted to find out which regulations had an effect on the incidences and what other variables such as vaccinations and temperature had an influence.
So, here is my take on the data.
When it came to the lockdown regulations, I skimmed through numerous press releases and newspaper articles to put together a data table containing information on when the following regulations took effect and when they were loosened: mask mandate, contact restrictions, closing of schools, closing of day care, closing of restaurants, closing of retail trade, closing of cultural facilities, closing of hotels.
Questions & Answers
- How do the statistics of the incidences in both locations compare?
The mean 7-day incidence between March 2020 and May 2021 was 76 for Berlin and 113 for Zurich. Looking at the boxplots shows that the median for both locations was similar, while there were large outliers in Zurich.
2. How does the progress of incidences in both locations compare?
This plot makes clear that while the shape of the incidence curves is similar, the second wave has hit Zurich a lot more strongly. The largest difference was on 31st October 2020 with 265 — more that 2 times higher incidences in Zurich.
This brings me to my last question.
3. What variables have influenced the progress of incidences in Canton Zurich and Berlin?
Here is a visualisation of the variables that potentially influenced the incidences:
I selected 4 “lockdown variables” and used them as predictors in an ARIMA model together with the temperature.
Strikingly, the temperature had the largest effect in both locations. Covid-19 — in its current form — is a seasonal problem.
And I have to grudgingly admit that the handling of school closings in Berlin did seem to affect the incidences. Just like the mask mandate, closing of cultural facilities and retail trade.
In Zurich, the results are similar, but the way the closing of cultural facilities was regulated did not have a significant effect. This could — but does not have to — mean that these facilities should have been closed earlier/more often.
As interesting and somewhat eye-opening these numbers are, there is a lot that could still be improved. Of course, other or more regulations could be taken into account. For example home-office rules. This analysis could be adapted for other locations. A multilevel model could be calculated in order to explain the differences between the locations. Individual time periods could be analyised. Etc.
The good thing is, the data is there. The code is there. Check out the GitHub repository and feel free to suggest changes.