Population density once again invoked. Data say 'No'.

On Twitter recently, I see with increasing frequency once again, Population Density (persons/ Km²) invoked as the cause of increased cases or increased deaths due to COVID-19 infecting a country. Or sometimes, a part of a country.

That might take the form of "More cases and deaths in Flanders (or Brussels) due to an elevated Population density."

So let's see what we can find in Germany.

Why Germany?

Well they seem to have a pretty good outcome at this time (May 27, 2020) from the infestation of that country by SARS-CoV-2. So with such clear and positive and complete data, it should be easy to identify that Population Density effect.

Here is a summary panel (click to enlarge)

16 regions in Germany :

- - accounting for 98.99% of the total population
  - 179 365 COVID-19 cases, a median of 5152.5 cases per region
  - an enviable death rate (CFR) of 4.333±0.9756%, median 4.62% and a variance of only 0.010%
  - 90.206±2.4966% of these cases have now Recovered.
  - leaving only 5.461±2.5950% Actively ill.
  - the mean population of these 16 regions is 5 185 340,13±4 938 927,85 and a median of 3 270 769, with a mean population density of 709.26±1 1 11.90 persons per km². Amedian of 213.05 persons per km² and a large variance of 1 318 732.73 persons per km². With such a large variation in the population density, if cases and/ or deaths from COVID-19 are impacted by population density, this is a good sample to study to find such effects if they exist.

Data by Region

We add the Population Density

How do these look, sorted?

by Population density, since that is the hypothesis ...

by Cases of COVID-19

If you see a link between cases and deaths, that makes sense since in this sample, 4.33±0.98% perish.

But is the population density value tracking right along? i think not.

Sorted, by Deaths from COVID-19

Cases fit right along, but not that Population density column.

And since we used the word "fit."

How strong a correlation is there between the Population density as determining factor, and the Cases found? Clearly, if people are milling about, one squeezed against another, there just must be more Cases found. Right?

It seems like such a charmingly logical idea...

Well not.

A linear correlation coefficient of R² = 0.0512 is pretty darn weak.

Translation: Increased population density does NOT associate with increased COVID-19 case load.

How about for deaths?

If some of these regions do not have enough hospital resources to treat the very sick in a high population density region, there just have to be more deaths. No doubt about it.

Well not.

There is great doubt. In fact, an R² value of 0.000151 says: "Please just forget about that hypothesis. It doesn't even merit the name, hypothesis."

That movie image of people packed in like rats in a medieval city, and someone screaming "Bring out your dead!" as he passes. Well if it fit in the 14th century (and others) for bubonic plague, it does not fit for COVID-19.