Tuesday, March 2, 2021

Math: Still A Thing


The above graphic is one that has been used frequently on the WRSA site whenever there's a story about the number finagling with COVID, esp. deaths.

I am loathe to point out the obvious, but there's a wee problem: it's total bullshit.

Don't believe me, use your own lying eyes.

Let's look at the tally, and see what we get.

Take the salmon pink "All causes" death percentages.

Note that the span goes from "Under 1 yr" to "85 years and over".

(Common Core grads: That would be "everyone".)

So those bars are all deaths, for all ages, according to the graph.

Under 1 yr: 1%

1-4 yrs: less than 1%

5-14 yrs: 0%

14-24 yrs: 0%

25-34 yrs: under 1%

35-44 yrs: under 1%

45-54 yrs: 1%

55-64 yrs: 1.5% (Don't worry about the fractions, we'll round up, to be fair.)

65-74 yrs: 2%

75-84 yrs: 5%

85 and up: 15%.


That's the percent of all people who died, who died. From "All causes".

So, according to this graphic, the other 71% of the people who died, didn't die.

It's right there on the graph, in living color. I'm not making this up.

I'm not a pathologist, but I'm pretty sure that's a slick bit of mathemagic, right there.

Explaining that no one died of "all causes" in 2020 between age 5 and 24 is also beyond magical.

The word you'd be looking for would be "recockulous". As in "so far beyond riDICulous that a bigger word is required".

Math hint: When you add the percentages, and it doesn't add up to 100 (give or take an infinitesimal fraction, over/under) what you're looking at is the same kind of math that got Biden 110% of eligible votes cast in certain precincts of Philadelphia in 2020. It's how Saddam Hussein was re-elected as dictator of Iraq with 125% of the votes, time after time.

Also known as what drops from the south end of a north-bound bull during a bowel movement.

Important safety tip:

If one is looking for graphics, selecting one that isn't more suitable for use as rose fertilizer is always a good policy.

You'll get no argument here that some unscrupulous entities have monkey-f**ked COVID deaths for nefarious ends, to a degree. (I doubt seriously that it's beyond 5%, either way. You can't pad them that much farther than that without it coming out and biting you in the ass. Ask Fredo Cuomo how that works out when you try it. It's also pretty tough to get thousands of random strangers to all collude randomly in such a vast conspiracy, accidentally or on purpose. If it were otherwise, Nigerian scammers would have the houses next to Bill Gates and Jeff Bezos. But I have no doubt at all that the numbers bandied are not 100% accurate.)

But as a general rule, it's probably not helping anyone to claim the numbers are fake, by posting a graphic that is itself self-evidently off by a factor of nearly 345%. This is like telling the cop you were only doing 29 when his radar gun reveals you were doing 100. Try selling that whopper to a judge in court, and tell us how it works out for you.

And let's let ABCNNBCBS corner the market concession on obvious bullshit and fake news.

I have no doubt that the graphic's use is an oversight, and that it was probably posted by someone originally with laudable intentions, but crap is still crap, and whoever made that graph is full of it. 

The only thing to do with such crap, is to flush it.


Watchman said...

If I read the chart correctly (I went and looked up the original, but maybe I'm missing something) it is a population fatality rate report. Thus each age bracket slows the percentage of that specific population that died. So less than 1% of those under one year old died and 15% of those 85+ died in 2020. The only way to get the number to add up to 100% would be if everyone died. I don't think the chart is in error.

Aesop said...

You may be absolutely correct.
When you need to look up the chart and find the missing information to get from A to Z, and that isn't part of the graphic, one is still lying with pictures.
It may be more correct for that information.
But absent that, the chart is a lie.
So the correct label wouldn't be what it is posted as, at the top of the chart.

And it's still a grossly misleading to outright falsehood as excerpted, because if 15% of 85s and up die annually, no one ever makes it past 92 (which ain't so), and 0% of those 15-24 isn't how many of them die each year. (Ask any car insurance agent about that.)

Also, the COVID deaths would correctly be expressed as a given portion of "all causes", not laid alongside them.

Again, it's lying with pictures.

It gets worse: Every one of those brackets is a different percent of the whole, And there are a lot more in some brackets than others. so it's by default comparing apples to oranges. They used to teach this sort of basic factual analysis in middle school, but now that actual thinking = "white supremacy", not so much.

Juggling the poop doesn't make it less poopy. It just flies better.

Doug Cranmer said...

If you look in the original report the graph is a _summary_ of a table with the exact numbers. None of the categories have zero deaths. The graph also follows text that makes clear every point you raise and also describes the graph as a summary. Just lifting the graph completely out of context and criticizing it alone is very misleading.

Greg said...

I have a little, tiny paperback on my shelf titled "How to Lie with Statistics". First published in 1954, and has never been out of print since. There's a reason for that.

ruralcounsel said...

It's not lying. It's taken out of context. Since the image is taken from Imgur, a secondary (or more) source, that should be an automatic alert. You needed the original document to tell you what the definitions of what they were charting. You made an assumption. The assumption was wrong. Admittedly, under your assumption, the graph appeared misleading. So there was no math errors, just confusion in defining the terms.

Much ado about nothing.

Aesop said...

I made no assumptions.
I interpreted the graph, as presented.
When someone only presents half the truth, they're lying, whether it was intentional or not.

I merely called out the naked magician because I can see the rabbit.

Doug Cranmer said...

"I interpreted the graph, as presented.
When someone only presents half the truth, they're lying, ..."

Exactly. It's been pointed out twice now that the graph was only part of a report that explains clearly what the graph summarizes. The graph is correct. You're taking it out of context (i.e., presenting half the truth) to push some narrative of yours and now you can't admit you made a mistake.

Unknown said...

Maybe it's Common Core math.


Aesop said...

No, Doug.
*I* didn't "take it out of context", it's been presented out of context, time after time after time.

I read the graph exactly as labelled.

It's still bullshit.
Not because it's out of context, but because it's wrong.

It's mislabelled, inaccurate as excerpted, and presents a false sense of reality.
When you fly the gauges and the map, and crash into a mountain anyways, it isn't the mountain that's in the wrong place.

If you still can't get that, I can't help you.

Robert Wood said...

There's the underlying inherent issue with statistics - you have to be very careful in determining WHAT EXACTLY is the question you're intending to answer when working with statistics (and wording it exactly how you want it to be), as it is extremely easy to make the data fit how you want it to fit. It's one of the first things I learnt in studying statistics, especially when using surveys to determine sets of data.

Robin Datta said...

It doesn't take a rocket surgeon to understand that "CDC 2020 Death Percent by Age" must add upto 100% of all deaths. Between 5 and 24 years of age, all the deaths are causeless. Since deaths from "COVID & Other Causes" in every age category add up to less than 100%, there are folks in every age category who die without cause: in most age categories, most deaths are causeless.

At least that's the way the graph is labelled. Not as a deaths per percentage of the porkpulation but as deaths percent by age.