The Donor Map - regional distribution and statistics

Here we go: Our first intermediate result - The Regional Donor Map.

Prior to developing the anticipated fever map from aggregated heart rate information, we need to get a good understanding of the geographical distribution of our donors. Regional biases, for instance, need to be identified and considered in further analyses. Moreover, we must ensure that enough donors are registered in each district so that statistical evaluations are valid and the anonymization gained through the regional consolidation is guaranteed.

The donors in Germany

As of (16.06.2021), 538.174 donors have successfully registered their fitness wearables with our Corona Donation App. That is an average of 1.304 donors per district. Nationwide, we have a total participation of around 6 donors for every 1000 people, which is significantly more than we initially anticipated.

The particication in different districts

As mentioned above, we need to first check whether donors are geographically representative in our sample. Additionally, the method of aggregating from ZIP level to district level must be verified, since ZIP codes, which are used by donors to identify their location, in Germany do not necessarily merge perfectly into federal districts.

Using this approach we generated the following map showing the regional distribution of donors in Germany:

From the figure above, we can assume that we have a solid data basis, where each district contains a sufficient number of donors.

Distribution of donors per district

The number of donors per district varies quite a bit from the number of inhabitants per district. In some districts, only a few 100 people decide to donate, while in other districts, more than 20,000 donors have registered. It is expected that a higher population produces more registered donors. The figure below illustrates the frequency of participation per district. On average, a district has around 1.304 donors.

Since federal districts are populated to varying degrees, we anticipate a donor count that varies with inhabitant count. The relation between number of donors and inhabitants per county can be represented in a scatter plot like the one shown below. As expected, we observe a nearly linear relationship, indicating that participation does indeed increase with in proportion the the population size of a district.

Donors per capita

In total, 538.174 donors are currently participating with a nationwide population of around 83.02 million people. This means that about 0,56% of German people are donating their data, or about 6 donors per every 1,000 people.

Taking this into account, it still needs to be investigated whether differences and geographical heterogeneities exist in the per capita donor participation. In other words, how does this per capita participation vary across counties? The following map answers this question.

The number of donors per inhabitant is very important to consider in for further analyses. The fact that the per capita participation is not spread uniformly across the country needs to be taken into account in order to avoid any biases.

The question then arises whether the per capita participation is dependent on the inhabitant count. In other words: Does the population of a region determine the number donors per inhabitant? The scatterplot below depicts the per capita participation vs. the inhabitant count for each district. Despite regional differences, we can see that the relationship between the per capita participation and the population size of a district is weak and highly variable.

Regional donor distribution vs. COVID-19 prevalence

The important question now is whether the degree to which people agree to participate in the corona data donation project correlates with the local COVID-19 prevalence. In other words, is participation determined by the local and regional situation?

The following figure depicts the per capita participation as a function of COVID-19 cases per district. The correlation of these two quanties is rather weak and statistically independent.