Hide Ya Wives, Hide Ya Kids: Worldwide Coronavirus Pandemic!

[160jordansdeep]

Nurse says patients are attending ‘coronavirus parties’ to intentionally get sick

A North Carolina nurse said she’s treating patients who have attended “coronavirus parties” in an attempt to catch the deadly bug and hopefully develop immunity. “Over the last few days, we h…

nypost.com

more selfish people

 

[everysingletime]

I'll try my best, although I don't have the JPM paper itself, so I have to make some assumptions.

R0 is an estimate of the number of new infections that are caused by a single infected person. R0 of 1 means that the virus will neither increase or decrease over time. If 1000 are infected today, 1000 will be infected in a month (a different 1000). R0 over 1 means the number of infected people will grow. R0 below 1 means the number of infected people will decrease over time and eventually go away completely. You can think of it like blackjack. If you make on average $1 for every $1 you bet, you will (on average) have the same amount of money forever. If you only make $0.99 for every $1 you bet, you will eventually run out of money if you play long enough.

R0 is not set in stone, and can change dynamically depending on many factors, such as social distancing. More precisely, there is Rt, which is the instantaneous version of R0. Pretend there's no lockdown. Rt on a busy travel day, like the Wednesday before Thanksgiving, might be 10. On the Saturday after Thanksgiving, it might be 2 because everyone's staying home to put up lights.

On to the actual JPM data: we can try to estimate Rt (they should've called it Rt, not R0, I think) based on models that look at the number of cases and deaths reported each day. This can get complicated because it would incorporate many things, such as the incubation period of the virus, to try to recover an estimated Rt for each day. I have no clue how JPM actually did this for their data. Because there are lags (we can't really estimate today's Rt until a few weeks from now) it's unclear how JPM came up with these measures, given that a lot of these states only opened up recently.

So what they did is estimated Rt during the lockdown and then after the lockdown was lifted for each state. If Rt is smaller after the lockdown is lifted, which would indicate the virus is spreading more slowly (because the smaller the Rt, the less the virus is spreading), then the dot on that graph would fall to the right of the diagonal line. They argue that most of the dots fall to the right of the line, indicating that removing the lockdown has actually led to less transmission of the virus.

This is flawed because these estimates of Rt carry huge uncertainties. If I say Rt is 0.86, I'm really saying I think it's between 0.80 and 0.92, for example. So when South Dakota goes from 0.89 to 0.86, the real statement should be: South Dakota goes from somewhere in the range of 0.83-0.95 to somewhere in the range of 0.80-0.92. With uncertainties that large and a change that small, you can't really conclude anything. That's why the tweet said these estimates are "implausibly precise." The dots should really be big circles.

I followed you all the way up to the last paragraph. Meaning, the last paragraph is the only confusing part. Is there another way you can explain that? Lol

 

[160jordansdeep]

Demanding Answers: Flights From Countries Hard Hit By Coronavirus Still Coming, So Are Precautions In Place?

CBS2 has learned some of the same international flights that have long been blamed for causing coronavirus to explode in our area are still operating.

newyork.cbslocal.com

 

[whywesteppin]

I followed you all the way up to the last paragraph. Meaning, the last paragraph is the only confusing part. Is there another way you can explain that? Lol

Two parts --

First, these estimates are highly uncertain. It's like asking you to estimate someone's batting average for the 2020 MLB season (if it happens lol). You could try to pick a specific number, like 0.317, but really you would say, "I think it'll be between 0.29 and 0.33, and I'm 95% certain of that." To get that estimate, you are incorporating a lot of data in your head and spitting out your best guess of where you think the number will fall. Similarly, these estimates of Rt are taking a lot of data, making assumptions, and spitting out a range of values where they think the true number resides. So even though they plotted a point at 0.86, for example, no serious data scientist would plot a single point. You would plot a mean and a standard deviation for each measure. So it would really be 0.86 +/- 0.06.

That comes to the second point. If we have two points and they have a range that overlaps, you can't statistically say that one data point is larger than the other. Here, the two data points they are comparing are Rt during lockdown and Rt after lockdown for a given state. They don't tell us what the uncertainty or standard deviation in their estimate of Rt actually is, but it's probably fairly big. Let's say it's +/- 0.06. That means, if we really wanted to say that there is a difference during lockdown vs after lockdown, the mean value would need to change by something like double the uncertainty, which would be 0.12. But, from the plot, we can see that the largest changes are closer to 0.03, which is relatively small and insignificant, statistically speaking.

 

[everysingletime]

Two parts --

First, these estimates are highly uncertain. It's like asking you to estimate someone's batting average for the 2020 MLB season (if it happens lol). You could try to pick a specific number, like 0.317, but really you would say, "I think it'll be between 0.29 and 0.33, and I'm 95% certain of that." To get that estimate, you are incorporating a lot of data in your head and spitting out your best guess of where you think the number will fall. Similarly, these estimates of Rt are taking a lot of data, making assumptions, and spitting out a range of values where they think the true number resides. So even though they plotted a point at 0.86, for example, no serious data scientist would plot a single point. You would plot a mean and a standard deviation for each measure. So it would really be 0.86 +/- 0.06.

That comes to the second point. If we have two points and they have a range that overlaps, you can't statistically say that one data point is larger than the other. Here, the two data points they are comparing are Rt during lockdown and Rt after lockdown for a given state. They don't tell us what the uncertainty or standard deviation in their estimate of Rt actually is, but it's probably fairly big. Let's say it's +/- 0.06. That means, if we really wanted to say that there is a difference during lockdown vs after lockdown, the mean value would need to change by something like double the uncertainty, which would be 0.12. But, from the plot, we can see that the largest changes are closer to 0.03, which is relatively small and insignificant, statistically speaking.

Wow it just clicked when you compared it to baseball stats, even though I don't watch baseball at all.


Thanks man.

 

[whywesteppin]

Wow it just clicked when you compared it to baseball stats, even though I don't watch baseball at all.


Thanks man.

I don't watch baseball either lol.

 

[marikomorose]

man stopped sticking needles in his buttcheeks:

 

[everysingletime]

I don't watch baseball either lol.

So just to make sure I got it. Some stat ppl guessed at a number that we were going to be at post lockdown, but the guess was off. But just because the number was lower than the guess, you can't reasonably conclude anything from it because you're supposed to give yourself some leeway in the first place (like a range of numbers rather than one specific number). Plus the guess wasn't off by that much.

.... Stats is not my forte but this is just me trying to put what you said in simpler terms.

 

[IATT]

 

[whywesteppin]

So just to make sure I got it. Some stat ppl guessed at a number that we were going to be at post lockdown, but the guess was off. But just because the number was lower than the guess, you can't reasonably conclude anything from it because you're supposed to give yourself some leeway in the first place (like a range of numbers rather than one specific number). Plus the guess wasn't off by that much.

.... Stats is not my forte but this is just me trying to put what you said in simpler terms.

this is where my analogy wasn't great. in the baseball analogy, we were guessing a future outcome, but it's a number we will directly observe. in this case, we're guessing something that has already happened. the difference is we can directly observe batting average (it just hasn't happened yet), but here we're guessing something we can't directly observe (although it has already happened). in both cases, whether it's a future thing we can observe (like batting average) or a number we can't directly measure in a current situation (like Rt), we are trying to make a guess at the true number, so in both cases there's uncertainty.

in this case, the stats people took data from during the lockdown and, using some sort of model, guessed what the value of Rt best explains the data. then they took the data after the lockdown and, using the same model, guessed what the value of Rt was during this other period. so they're generating two guesses of Rt for two different sets of data (let's say, data from April 1 to April 15 compared to data from May 1 to May 15) and comparing those numbers.

otherwise, what you're saying is correct. in their guess of Rt for April 1-15 and again for May 1-15, they need to give themselves leeway on both guesses because there's uncertainty with both.

 

[fargin]

man stopped sticking needles in his buttcheeks:

Hellboy faded.

 

[ksteezy]

The wife and I tested positive for the Antibodies.

come at me, Rona.

 

[everysingletime]

this is where my analogy wasn't great. in the baseball analogy, we were guessing a future outcome, but it's a number we will directly observe. in this case, we're guessing something that has already happened. the difference is we can directly observe batting average (it just hasn't happened yet), but here we're guessing something we can't directly observe (although it has already happened). in both cases, whether it's a future thing we can observe (like batting average) or a number we can't directly measure in a current situation (like Rt), we are trying to make a guess at the true number, so in both cases there's uncertainty.

in this case, the stats people took data from during the lockdown and, using some sort of model, guessed what the value of Rt best explains the data. then they took the data after the lockdown and, using the same model, guessed what the value of Rt was during this other period. so they're generating two guesses of Rt for two different sets of data (let's say, data from April 1 to April 15 compared to data from May 1 to May 15) and comparing those numbers.

otherwise, what you're saying is correct. in their guess of Rt for April 1-15 and again for May 1-15, they need to give themselves leeway on both guesses because there's uncertainty with both.

Now I'm sort of lost again. At the part where you say 'we're guessing what's already happened.' If it's already happened, then, we wouldn't need to guess?

It's late at night. I might need to just sleep on the info.

 

[DB Cooper]

The wife and I tested positive for the Antibodies.

come at me, Rona.

Rona looking for a rematch...

 

[fargin]

Rona's looking sharp.

 

[whywesteppin]

Now I'm sort of lost again. At the part where you say 'we're guessing what's already happened.' If it's already happened, then, we wouldn't need to guess?

It's late at night. I might need to just sleep on the info.

more precisely -- we're guessing a number that generate something that's already happened. the something that's happened that we can measure is the number of cases (or deaths). Rt has in a sense already happened but, because we can't directly measure it, we have to infer it. that's where the uncertainty arises.

let me give a concrete example. let's say somehow we know that Rt is 2. if we start with 1 case, then the next time period it would be 2 cases (a time period could be 1 day, 2 weeks, or whatever. it doesn't matter for now). then it would be 4, then 8, then 16, etc.

so if the case numbers we observe look like 1 - 2 - 4 - 8 - 16, then we would guess with high certainty that Rt is 2.

but what if we observe numbers that go 1 - 2 -5 -7 - 22? Now we're more uncertain of what Rt is. It could be 2, or it could be 2.2 or 1.7.

we will never truly know Rt. it is just a parameter in a model that we are using to explain the numbers we observe. technically speaking, we're finding the range of Rt values for our model that generates data that looks reasonably similar to the actual data we observe.

 

[whywesteppin]

The wife and I tested positive for the Antibodies.

come at me, Rona.

that's cool, especially since we're getting more evidence that reinfection is unlikely.

i forget -- did you have symptoms? or do you have an idea of when you may have been infected?

 

[Blake P]

The wife and I tested positive for the Antibodies.

come at me, Rona.

Not my intention to rain on your parade, but I know I’ve read about some antibody tests that aren’t very accurate and can detect the presence of antibodies to OTHER coronaviruses. Do you know for sure that you have antibodies to COVID-19?

 

[everysingletime]

more precisely -- we're guessing a number that generate something that's already happened. the something that's happened that we can measure is the number of cases (or deaths). Rt has in a sense already happened but, because we can't directly measure it, we have to infer it. that's where the uncertainty arises.

let me give a concrete example. let's say somehow we know that Rt is 2. if we start with 1 case, then the next time period it would be 2 cases (a time period could be 1 day, 2 weeks, or whatever. it doesn't matter for now). then it would be 4, then 8, then 16, etc.

so if the case numbers we observe look like 1 - 2 - 4 - 8 - 16, then we would guess with high certainty that Rt is 2.

but what if we observe numbers that go 1 - 2 -5 -7 - 22? Now we're more uncertain of what Rt is. It could be 2, or it could be 2.2 or 1.7.

we will never truly know Rt. it is just a parameter in a model that we are using to explain the numbers we observe. technically speaking, we're finding the range of Rt values for our model that generates data that looks reasonably similar to the actual data we observe.

Hmm. I think, I get what you're saying now.

I know one thing for sure. It took you three posts for me to understand, cuz I'm slow.

It just means that anyone can take any chart and bs a statement based on the chart and no one but statistic ppl will know its bs.

 

[freeze]

'Come at me brahhhhh!!!'

 

[QU!K]

The wife and I tested positive for the Antibodies.

come at me, Rona.

Meaning you got it and recovered? Are you infectious?

 

[QU!K]

'How Can I Be Sick?' Woman Who Took Hydroxychloroquine For 19 Years To Treat Lupus Still Got COVID-19

A Wisconsin woman is warning people that the anti-malarial drug, hydroxychloroquine, wil not protect someone from COVID-19.

detroit.cbslocal.com

 

[Blake P]

Meaning you got it and recovered? Are you infectious?

You develop antibodies after you have recovered and your immune system has cleared the virus, so yes it should indicate that they are no longer infectious and should have some level of immunity to the virus for an unknown amount of time. Reportedly, early antibody tests were not being regulated and the results of many being produced were not accurate enough. I’m not sure if that has changed, but when I looked into having it done at Quest they had a disclaimer that it may falsely detect the presence of antibodies to other coronaviruses, so I declined.

 

[had pants]

 

[freeze]

'How Can I Be Sick?' Woman Who Took Hydroxychloroquine For 19 Years To Treat Lupus Still Got COVID-19

A Wisconsin woman is warning people that the anti-malarial drug, hydroxychloroquine, wil not protect someone from COVID-19.

detroit.cbslocal.com