VWS corrects graph, interpretation remains tendentious

Last weekend I blogged that the Coronadashboard has been pretended for 4 months as if it was not too bad with the strikingly increased mortality (even recurring over -mortality) in last summer.
Great if you notice that prominent figures respond interested. Maurice de Hond paid attention to it the next day on his site and yesterday this message appeared on the Coronadashboard. It would be a technical error, which made the expected mortality in the summer struggled, also in the winter something and spring and autumn caused a smooth transition.

The graph looked perfect and was also 100% identical to that on the CBS site; There was only a slightly different data set behind it. This is only possible with a conversion stroke that increases the expected death in accordance with the description of increase and reduction. You can view an example of the dataset with which the graph is made below. (It must be test dates; the file is from April 16, 2021 and the death dates were already in it until the end of that year.)

Iemand suggereerde dat er misschien een kommaatje verkeerd had gestaan... Dan hebben we het dus over een bug. Een bug die de eerste waarde behoudt, de tweede verlaagt en de derde verhoogt – en dan nog ‘s winters op een andere manier dan ‘s zomers.

If a technical error gets an assignment if it gets so perfectly done, the question arises as to why updating something like the Corona app should cost 49 million.

The disinformation is still there: there was excess in March

At the end of March/beginning of April, the point immediately makes clear: the previous graph showed no excess mortality and the text then responds conveniently. In the correct graph you can see that the text is not completely correct. In any case, not at all.

Take a look, first the explanatory text as it is on both the VWS and the CBS site. Then the virus varia version.

VWS version (Coronadashboard)

An estimated 3600 people died in the last week of March (week 12, 21 to March 27, 2022). That is slightly more than died in the week before (3567) and about 400 more than expected. About 500 more people died throughout March than expected. The mortality was especially higher among Wlz care users and people over 65. Statistics Netherlands reports this on the basis of the provisional death rates per week.

Virus varia version

March: Increasing mortality leads to overdiction

An estimated 3,600 people died in the last full week of March, around 400 more than the week before. In the two weeks before 306 people more than expected, together around 700 more people than expected.
The mortality was especially higher among Wlz care users and people over 65. The overflow to April shows over mortality, while it is unclear where it comes from. Virusvaria.nl reports this on the basis of this week's provisional death figures.

28, 29, 30 en 31 maart zijn toch echt dagen in maart. In die week is er al oversterfte - althans volgens de CBS-norm, niet volgens de propagandanorm van VWS. Wat er in de kop staat is dus gewoon niet waar. In de Telegraaf, allà, maar op een overheidssite...

Met een handige definitie van "laatste week", waardoor alles na 27 maart niet meer bij maart hoort, lijkt het net alsof er in maart geen oversterfte was. Dat klopte wel met de vorige gemanipuleerde grafiek maar niet meer met de afbeelding die de correcte onzekerheidsmarge weergeeft. De laatste week van maart is volgens VWS week 12, 21 tot en met 27 maart 2022. Ze bedoelen misschien de laatste 'volledige' week? Als 'de eerste week van april' straks dan ook de eerste volledige week is, die op 4 april begint, zit er toch weer één oversterfteweekje minder tussen om uit te leggen.

En dan dat zinnetje: "Dat meldt het CBS op basis van de voorlopige sterftecijfers per week." Dat staat ook in het bericht op de CBS-site...!? Ik snap dat het zo op de VWS/rivm-site staat maar dat zet je toch niet op je eigen site? Alsof het CBS het hele berichtje van de rivm/VWS-versie heeft overgenomen, inclusief het nu naar zichzelf verwijzend zinnetje. CBS zet toch ook niet op elke pagina "Bron: CBS"?

Maybe it's just press releases speak on a public site. But my mind's eye sees the next Wob document appearing again:

From: zwartgelakt@minvws.nl
At: zwartgelakt@cbstatistiek.nl
Re: Streamlining communication excess mortality

Do you know what: we streamline the interpretations as follows: Send us the data, we will make something beautiful and then you also place our interpretation on your own site.