I know math is hard. It is for me too.
1570 visits yesterday to this blog. Let's assume that each person in the US is exposed to the Coronavirus.
Approximately 15 people known to have the virus were reported in comments (at least one death, sadly)
So the anecdotal data shows about 1 in 100 people from 1570 responses.
Now let us use some numbers and some math.
For a number, lets say that each person responding knows....50 people in the parameters that I set. I don't know that that is a valid number, but it should work to give us a picture. 50 times 1500 gives us a sample size of 75,000
15 cases out of 1570 responses. Lets round that down to 1500 just to make the numbers neat.
15/1500= 0.01....1 percent.......1 percent of the people reading this blog knew of a death or a case of coronavirus in their circle of friends and/or acquaintances and chose to comment.
So far, most have or are recovering. 1 out 15. So of the confirmed (ish) cases, we have a current death rate of (1/15)....or 6.6666% Six percent of the cases died. Let's round that number to .07....seven percent.
So (so far!) we have 7 percent of 1% as a rate of death. 0.01 times 0.07 equals .0007. Do it in percentages......07%.
Is it gonna be a big number when extrapolated out to the 375 Million people in the US? Let us see....
0.0007 times 375,000,000 equals 262,500. Which is pretty close to the death total for Swine Flu. (and we all remember the dead piled up like cordwood outside of the hospitals in 2009-2010, right?).
This is more serious than the "Normal" flu (about 55,000 dead each year) but it isn't gonna end up killing us all. Even if this is low by a factor of TEN, it still isn't gonna end us all.
Now, please understand, this isn't a valid survey, it isn't really terribly good data (lots of assumptions made, and a poor sample, and it isn't corrected for location nor age of the sample or a lot of other things), and it isn't, in fact, designed to prove anything one way or the other. But as a sample, it does match the currently REPORTED cases vs deaths at least somewhat. The picture looks something like what I present here.
This is based on some pretty sketchy data and it is a snapshot, not total cases...., so take these numbers with a grain of salt...maybe the whole shaker....
Try to stay healthy. Do the Distancing Thing. Be smart about this.
Wash your goddamned hands. Often.
1570 visits yesterday to this blog. Let's assume that each person in the US is exposed to the Coronavirus.
Approximately 15 people known to have the virus were reported in comments (at least one death, sadly)
So the anecdotal data shows about 1 in 100 people from 1570 responses.
Now let us use some numbers and some math.
For a number, lets say that each person responding knows....50 people in the parameters that I set. I don't know that that is a valid number, but it should work to give us a picture. 50 times 1500 gives us a sample size of 75,000
15 cases out of 1570 responses. Lets round that down to 1500 just to make the numbers neat.
15/1500= 0.01....1 percent.......1 percent of the people reading this blog knew of a death or a case of coronavirus in their circle of friends and/or acquaintances and chose to comment.
So far, most have or are recovering. 1 out 15. So of the confirmed (ish) cases, we have a current death rate of (1/15)....or 6.6666% Six percent of the cases died. Let's round that number to .07....seven percent.
So (so far!) we have 7 percent of 1% as a rate of death. 0.01 times 0.07 equals .0007. Do it in percentages......07%.
Is it gonna be a big number when extrapolated out to the 375 Million people in the US? Let us see....
0.0007 times 375,000,000 equals 262,500. Which is pretty close to the death total for Swine Flu. (and we all remember the dead piled up like cordwood outside of the hospitals in 2009-2010, right?).
This is more serious than the "Normal" flu (about 55,000 dead each year) but it isn't gonna end up killing us all. Even if this is low by a factor of TEN, it still isn't gonna end us all.
Now, please understand, this isn't a valid survey, it isn't really terribly good data (lots of assumptions made, and a poor sample, and it isn't corrected for location nor age of the sample or a lot of other things), and it isn't, in fact, designed to prove anything one way or the other. But as a sample, it does match the currently REPORTED cases vs deaths at least somewhat. The picture looks something like what I present here.
This is based on some pretty sketchy data and it is a snapshot, not total cases...., so take these numbers with a grain of salt...maybe the whole shaker....
Try to stay healthy. Do the Distancing Thing. Be smart about this.
Wash your goddamned hands. Often.
One of the most basic tenets of survey research is that you can't assume answers for your non-respondents. This is especially true when you've got an estimated non-response rate of ~99%. At that point your assumption about non-respondents is essentially your entire dataset. Actual reader comments have so little weight that they might as well not exist.
ReplyDeleteHere you've assumed that every visitor who didn't leave a comment also didn't know anyone infected. It's not hard to think of a great many objections to that assumption.
Let's re-run the analysis solely based upon comments:
13 comments (plus one reply from you)
16 known cases (counting Kevin's response as 3, not counting Knucklehead's wife's knowledge)
13 * 50 potential cases (using your guess as to who each respondent knows)
16 / (13 * 50) = 2.5%
7% of 2.5% = 0.17% (using your death rate here)
375M * 0.0017 = 637,500
As you might expect from the numbers it's only higher by a factor of 2.5 (i.e. the difference between the case rate from your respondent and the 1% you assumed). That puts us well higher than Swine Flu and more than 10X seasonal flu. However, that's right now. Both your original calculation and this revised one assume that no more spread will occur, an assumption that's obviously incorrect.
To quote Knucklehead, "yes, it's real".
Your points are valid.
ReplyDeleteAs I said, it was a shitty sample and a crap methodology.
It is likely valid your way as mine, neither uses good statistical sampling nor decent analysis.
It does give us an idea as to what the final picture will look like, however.
The numbers so far do look closer to the 1% than not, but we shall have to see how things progress.