The Coronavirus, not the beer

An interesting point in the NYC data for those who think this only affects the elderly:

45-64, Deaths per 100,000: 83.75 (approximately one month)

And then here are the average monthly deaths in NYC that I pulled off the CDC website (ALL CAUSES, 2014-2016):

45-64, Deaths per 100,000: 42.33

So unless I messed up the data, Covid is killing people in this age range at close to twice the rate of literally everything else.
 
i guess trying to change the subject of being a liar and a failure on covid19 response

the best course of action is to make civil unrest in states that have rejected you at the ballot box
 
Likewise for the 18-44 demo.

COVID
18-44, Deaths per 100,000: 10.21 (approximately one month)

ALL DEATHS
20-44, Deaths per 100,000: 7.1 (CDC, monthly average)

So even for the "safe" folks, we are losing more people just to this than we normally do for everything.
 
Last edited:
Metaphysicist said:
Likewise for the 18-44 demo.

COVID
18-44, Deaths per 100,000: 10.21 (approximately one month)

ALL DEATHS
20-64, Deaths per 100,000: 7.1 (CDC, monthly average)

So even for the "safe" folks, we are losing more people just to this than we normally do for everything.
Click to expand...

We have lost. It was a spike for the vulnerable portions of each age grouping. This is not projectable to the future.
 
Metaphysicist said:
An interesting point in the NYC data for those who think this only affects the elderly:

45-64, Deaths per 100,000: 83.75 (approximately one month)

And then here are the average monthly deaths in NYC that I pulled off the CDC website (ALL CAUSES, 2014-2016):

45-64, Deaths per 100,000: 42.33

So unless I messed up the data, Covid is killing people in this age range at close to twice the rate of literally everything else.
Click to expand...

I think its relatively obvious that a bad illness is going to be more dangerous than just about anything else.

But you're still talking about something that 99.999% of population doesnt die from... and that's just age 45+
 
If we can get one or two more studies that show high spread rates then we should do a full reopen but restraint for at risk.
 
nsacpi said:
Did you read this? Prevalence estimate of 2.5-4.2%.
Click to expand...

I looked at the actual paper. Looks like the real positive rate was 1.5% which they adjust up which is theoretically fine, except...

With a test that has specificity confidence interval of (98.3-99.9%), wouldn't you expect ~.1-1.7% false positives? Is this outcome even statistically significant? Would love for them to be right, but this data seems... not amazing. Kind of bumping up against the limits of my stat comfort zone, though so who knows. If it makes it through peer review, somebody who understands this stuff better than me will have signed off on it, at least.
 
thethe said:
What does a longer plateau of infections and death tell us?
Click to expand...

That trumps an idiot and it didn’t just go away like a miracle in April and he wasted a bunch of time golfing and having rallies and we aren’t down to zero very soon like he said we would be
 
Metaphysicist said:
I looked at the actual paper. Looks like the real positive rate was 1.5% which they adjust up which is theoretically fine, except...

With a test that has specificity confidence interval of (98.3-99.9%), wouldn't you expect ~.1-1.7% false positives? Is this outcome even statistically significant? Would love for them to be right, but this data seems... not amazing. Kind of bumping up against the limits of my stat comfort zone, though so who knows. If it makes it through peer review, somebody who understands this stuff better than me will have signed off on it, at least.
Click to expand...

My reaction was the paper showed the opposite of what Shapiro seemed to think it did.
 
Metaphysicist said:
I looked at the actual paper. Looks like the real positive rate was 1.5% which they adjust up which is theoretically fine, except...

With a test that has specificity confidence interval of (98.3-99.9%), wouldn't you expect ~.1-1.7% false positives? Is this outcome even statistically significant? Would love for them to be right, but this data seems... not amazing. Kind of bumping up against the limits of my stat comfort zone, though so who knows. If it makes it through peer review, somebody who understands this stuff better than me will have signed off on it, at least.
Click to expand...

Why would you extrapolate a sample rate to non homogeneous populations?

Couldnt you also expect false negatives (serious question).
 
thethe said:
Why would you extrapolate a sample rate to non homogeneous populations?

Couldnt you also expect false negatives (serious question).
Click to expand...

Don't understand what the first question is asking.

But yes, you would expect false negatives (I think the article had something like a 80-90% sensitivity), but you are much more concerned about false positives when looking at very trace results. For example, imagine a very simple scenario:

Disease X occurs 1 in 1000 people. A test is 99% accurate, and you give it to all 1000. You will probably get something lkike

- Number of true positives: 0-1
- Number of false positives: 9-10
- Number of true negatives: ~990
- Number of false negatives: 0-1.

You can see that you are much more likely to get a false positive than a false negative, even if they are equally likely % wise (and this is an oversimplification).
 
The sample population is a mix of residents from various geographic areas within Santa Clara. The virus spreads based on a hosts patterns. People typically have movement patterns primarily around their living situation.

Therefore, its important to know rates for those areas as it's more precise. I think what they are doing is stratifying their sample population by doing a series of extrapolation using those sub group sample rates and then combining to a prevalence rate for the whole region.

Seems solid to me but I've been wrong before.


In response to your example. Why is that split a scenario we should expect?
 
thethe said:
The sample population is a mix of residents from various geographic areas within Santa Clara. The virus spreads based on a hosts patterns. People typically have movement patterns primarily around their living situation.

Therefore, its important to know rates for those areas as it's more precise. I think what they are doing is stratifying their sample population by doing a series of extrapolation using those sub group sample rates and then combining to a prevalence rate for the whole region.
Click to expand...

Okay, well that just has nothing to do with what I'm talking about. I'm talking about the original sample error of the tests before any adjustments. I have no problem population/age/whatever adjusting, which I explicitly said was "fine."

thethe said:
In response to your example. Why is that split a scenario we should expect?
Click to expand...

Now I don't know what this means. That "split" is just math to demonstrate a statistical phenomenon. If you are asking me why we are concerned about a scenario with low positive rates, that's just literally the kind of data from the study.
 
Last edited:
goldfly said:
[tw]1251221795123998720[/tw]
Click to expand...

Trump's messaging recently has been:

"It's up to the States"
"The States should have overstocked ventilators"
"The previous administration"

Remember, Pence promised 4 million tests a month ago. Today, Trump is bragging about just reaching 3.5 million tests, we are second to last in testing per capita.
 
You must log in or register to reply here.
Back
Top