Jump to content

Talk:Statistics of the COVID-19 pandemic in Germany

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

"Daily infection rate" chart

[edit]

Is that "daily infection rate" a well-sourced metric? As far as I can tell, it's not published like this by the RKI. We seem to be deriving it from other published data by means of WP:OR. It also appears to be not very informative at all, as beyond a regular weekly oscillation it shows next to no reflection of the changing infection trends. So if it is a metric, what is it a metric of? Should we get rid of the chart? Fut.Perf. 15:09, 15 November 2020 (UTC)[reply]

It's simply the infection rate, usually denoted as "r" in the context of exponential growth. In comparison to the effective reproduction number, it is actually a rate in the sense of "per time unit", here per day, as data summaries are released on daily basis by RKI. During the rise of the 1st wave this was the same as the daily change in percent given with many medical cases charts. Only when the majority of the overall cases weren't active cases anymore (and thus not infectuous), the figures of cause differ. As you could see there's no need for much of a difference in this rate to rise the number of cases, if you only give it enough time. I do not consider that original research but only a more accurate version of the infection rate that was intended to supply by the percentages in the medical cases chart. More or less the same goes for the death rate, only active cases can die, so what sense does it make to relate the number of deaths to those already dead or to the cumulated overall cases. To me this is the simple math of epidemics, not original research. One could improve it by passing it through a 7-day running average filter, but that might actually be original research. Anyways, what do you think (or anyone else bothering to read this talk)? -- Kohraa Mondel (talk) 17:44, 20 November 2020 (UTC)[reply]
I'm not sure I follow you here. Infection rate, according to our article on that topic, seems to be a different thing (number of infections relative to number of people at risk of infection; not number of infections relative to number of potential spreaders). In informal speech, in the media, "infection rate" seems to be also often used as a synonym for the basic reproduction number R0, but of course we are already reporting that separately, following the practice of the RKI and pretty much all the media that routinely report that figure. What added value does this "daily infection rate" offer, above these other figures (which seem to be significantly easier to interpret and read)? Fut.Perf. 18:04, 20 November 2020 (UTC)[reply]
It's the growth rate of Exponential growth, here for the growth (or decay/decline) of the number of those infected. It's a relative indicator for the effectiveness of infection control, as is the basic reproduction number, only that it's much more direct without all the assumptions required for R0. That's why the daily percentage rate is used in so many places, only that for the second wave at the latest the non-active cases need to be ignored to get true infection r. All simple growth rate math, you also don't hold births against all people that have ever lived, but against those still alive. Similarly the death rate is a relative measure for the medical system still being able to care for those seriously sick. If this simple application of exponential growth math is considered original research, well, so be it, though I don't consider myself having invented anything there, but thanks for the appreciation. -- Kohraa Mondel (talk) 20:56, 20 November 2020 (UTC)[reply]

Chart bug with dates in November

[edit]

It may be just me (using Safari Version 14.0.2) who is seeing this, but the chart seems to mess us with dates in November which don't have leaving zeros in the 'day' - i.e. 2020-11-1 would not work, but 2020-11-01 will. I've manually fixed some of the charts, and will try and fix the remaining, but if this is generated automatically (?) could whomever is generating it update their script to include leading zeros?

This is the same user (and I've realised I forgot to sign the last comment!) - I think the problem is broader than I thought and is for all dates without leading zeros! 2001:16B8:C1F2:500:9D44:B3E0:B179:3C4D (talk) 11:16, 21 January 2021 (UTC)[reply]

Making charts more meaningful for a general audience

[edit]

Data sources and processing for this page have been discussed thoroughly, but the link to it is almost hidden deep inside the more general article COVID-19 pandemic in Germany.

I like to see the number of active cases as a kind of "risk chart": Divided by the total population it can tell everybody how likely it is to run into an infected person, for example 1:200 during the peak of the December 2020 wave or 1:30 000 in the week of Angela Merkel's speech in March 2020, which was enough to make people enter voluntary shutdown, even before it was officially mandated. Of course, meeting an infected person does not mean that you become actually infected and local differences from the nationwide average may be significant, but such a risk chart could be a good guideline for how many protective measures are currently appropriate. Just for information, the risk today (beginning of July 2021) is 1:7 000, significantly higher than one year ago.

The second chart that I want to highlight is the effective reproduction number, you may regard it as the "behaviour chart". Given that no immunization effects are in place, there is a linear reduction factor from the basic reproduction number, caused by social distancing (the number of contacts, but also protective measures like masks or quarantine for the infected following successful contact tracing). Example: If an individual has met 100 people per week before the pandemic with a basic reproduction number for COVID-19 of 6 and he has reduced his effective contacts to 10 per week (effective means not just number of contacts, but factored down with protective measures), his personal reproduction number would be 0.6 and the effective reproduction number is just the average across the population.

Included in the effective reproduction number is also immunization, either for people who have already recovered from COVID-19 or who have received a vaccination. Adding this to the "behaviour chart" would make it even more meaningful, but it requires several difficult assumptions (worse than the calculation of the effective reproduction number itself) and I would consider this original research. It would be great if somebody finds a reliable source and if we could cite it, however.

TheDoktorchen (talk) 10:11, 3 July 2021 (UTC)[reply]