Main Coronavirus / Covid-19 Discussion Thread

[Kalimantan Gull]

A not very cheery read, but the data people behind this know their stuff ( predicted Johnson's landslide [ on another Twitter account ] when everyone else was predicting a hung parliament ).


So, believe or don't believe, and take your pinches of salt if you want.

As said, not cheery :

https://twitter.com/peoptog/status/1243123003636035584?s=19

I just read that and its pretty sub-standard analysis in my opinion. They're extrapolating off trendlines with R2 of less than 0.3 in several cases, very very weak trends. The sample size is 10 (the last 10 years), its not strong. In another thread on here we discussed how the 2019-20 seasonal influenza epidemic occurred very early this winter and was pretty much over by New Year, whereas last winter it occurred in January. So immediately we can see why respiratory illness deaths for January 2020 might seem low compared to previous Januarys. Moreover there has been very mild weather this January and February, not surprising that deaths are lower than usual. Finally, when you've got total weekly deaths in the 10-15,000 range, the 400 or so from coronavirus aren't going to make a big dent.

Now there is certainly a valid argument that people dying at home or in care-homes because of the virus might not be included in the total virus-death numbers, we know the French are only counting people who died in hospital. But this 'analysis' isn't based on potential mis-reporting of virus deaths as other causes, it is alleging that deaths are completely removed from the data, i.e. aren't even being reported as deaths. That seems ludicrously far-fetched to me.

Simple fact. Fewer people have died in 2020 so far compared to the same period in several previous years. Simple prediction. That fact won't be true in a month or so.

 

[Icy Gull]

Up to two year prison sentences for absolute wankers who “coronavirus” cough at officials to make them back off. Can we not just shoot them?

 

[Blue Valkyrie]

I just read that and its pretty sub-standard analysis in my opinion. They're extrapolating off trendlines with R2 of less than 0.3 in several cases, very very weak trends. The sample size is 10 (the last 10 years), its not strong. In another thread on here we discussed how the 2019-20 seasonal influenza epidemic occurred very early this winter and was pretty much over by New Year, whereas last winter it occurred in January. So immediately we can see why respiratory illness deaths for January 2020 might seem low compared to previous Januarys. Moreover there has been very mild weather this January and February, not surprising that deaths are lower than usual. Finally, when you've got total weekly deaths in the 10-15,000 range, the 400 or so from coronavirus aren't going to make a big dent.

Now there is certainly a valid argument that people dying at home or in care-homes because of the virus might not be included in the total virus-death numbers, we know the French are only counting people who died in hospital. But this 'analysis' isn't based on potential mis-reporting of virus deaths as other causes, it is alleging that deaths are completely removed from the data, i.e. aren't even being reported as deaths. That seems ludicrously far-fetched to me.

Simple fact. Fewer people have died in 2020 so far compared to the same period in several previous years. Simple prediction. That fact won't be true in a month or so.

Fair enough.

I know these guys were some of the few that correctly predicted Johnson's large majority from the available data, so they have a lot of trust in the bank from me.

 

[Kalimantan Gull]

We don't know that at all.

We don't know it for fact, sure... to say we don't 'know it at all' is wishful thinking I'm afraid. Official figures are round about 10,000 infected, with scientists estimating that maybe 10 times that are truly infected but not being tested or being asymptomatic. Thats 100,000 people, which is a long way from being anything other than 'a small proportion of our population'

Could they be out by a much greater factor? Highly highly unlikely based on the testing regimes, the growth curves and the proportions of sufferers who are mildly symptomatic / highly symptomatic / asymptomatic proportions from other countries.

I live in hope that a huge proportion of our population have been exposed to an earlier, fairly benign strain of this virus, and the antibodies against that are effective against this.

 

[beorhthelm]

along with the 21-year old girl yesterday, the 37 year-old deputy ambassador.... there's others....

get the point, but its perhaps not as rare as we were led to believe at the beginning, and we have to remember that only a small proportion of the population has so far come into contact with this virus

we were given a simple graph on the age distribution of mortality. if the trend follows when we have thousands of deaths, there will be many victims who are young, just many many fewer than are older.

 

[Guinness Boy]

Fair enough.

I know these guys were some of the few that correctly predicted Johnson's large majority from the available data, so they have a lot of trust in the bank from me.

There's always an element of hunch in that though. I made a few quid correctly guessing the spread of Johnson's seats margin of victory but it was based, not on extrapolating data, but how I thought the UK public would react to "Get Brexit Done" versus Corbyn and the fact that BoJo has always been good at elections.

 

[Kalimantan Gull]

Fair enough.

I know these guys were some of the few that correctly predicted Johnson's large majority from the available data, so they have a lot of trust in the bank from me.

I have no doubt they are able to run statistical analyses.... But there is a big difference between predicting a result that is 100% guaranteed to happen - i.e. there was always going to be an election result, one way or another - and looking for a result that may not be there at all. There are no data on 'missing deaths' for them to analyse, so they are trying to extrapolate the weakest of weak trendlines to imply that somewhere there are 2,000 bodies in a warehouse freezer that no-one is reporting. Poppycock.

Furthermore, the whole point of statistics is to prove or disprove hypotheses - and this is just as valid in assessing the performance of these people. I assume you have only become interested in these people because they predicted Johnson's majority. But that is just one datapoint and you cannot therefore prove they are the best because of it. What about all their other predictions? No, all you have is a hypothesis that these people are good at predicting things, but it will only be proved true if they keep on predicting things correctly.

There was a lot of fuss made about a statistic in the 1970's (I think) that Israeli fighter pilots children were 75% male. A lot of thought went into why that was. Was it because they had lots of testosterone or adrenaline, or something to do with the kind of person who chose to be pilots, or something to do with the women who fell for them or the frequency in which they had sex? But the statistician just used the information to form a hypothesis, and when they tested it over the next five years, exactly 50% of Israeli fighter pilots children were male. Hypothesis not proven.

The lesson there of course, is WHY was anyone looking at the sex of Israeli fighter pilot's children in the first place? The answer is that they were drawn to it precisely because the number seemed so unusual, it was such an outlier. In the same way that people are clearly drawn to the guys in your link because they predicted Johnson's majority so accurately, but it most definitely does not automatically follow that they have some innate ability that others do not have.

Sorry for the essay, home isolation gives one time to ramble.....

 

[Kalimantan Gull]

we were given a simple graph on the age distribution of mortality. if the trend follows when we have thousands of deaths, there will be many victims who are young, just many many fewer than are older.

Yeah, I know. Its just when its numbers on a page its easy to digest and rationalise. When its the picture of a pretty 21-year old girl or a 37 year old successful diplomat or a 38-yr old football fan, it all becomes far more real and close to home....

I'm sure all of us, at least those of us under 60, have at some point in this crisis said "I'm not worried about it as I'm young and healthy, just concerned for my parents and grandad" But that worry isn't zero anymore.

and really, assuming we are fundamentally healthy, are careful and don't take risks on the road, don't smoke or take drugs or drink too much, eat a healthy diet, live a sensible life - what is there out there that can kill us in our 30's or 40's? Very very little - almost nothing. And yet here is a virus which CAN infect us, which we will almost certainly come into contact with at some point in the next couple if years, and which CAN kill us at a non-negligible mortality rate.... chance may well be very low, but its also a lot higher than we normally have to face.....

sorry, don't know whats coming over me today.........

 

[Blue Valkyrie]

I have no doubt they are able to run statistical analyses.... But there is a big difference between predicting a result that is 100% guaranteed to happen - i.e. there was always going to be an election result, one way or another - and looking for a result that may not be there at all. There are no data on 'missing deaths' for them to analyse, so they are trying to extrapolate the weakest of weak trendlines to imply that somewhere there are 2,000 bodies in a warehouse freezer that no-one is reporting. Poppycock.

Furthermore, the whole point of statistics is to prove or disprove hypotheses - and this is just as valid in assessing the performance of these people. I assume you have only become interested in these people because they predicted Johnson's majority. But that is just one datapoint and you cannot therefore prove they are the best because of it. What about all their other predictions? No, all you have is a hypothesis that these people are good at predicting things, but it will only be proved true if they keep on predicting things correctly.

There was a lot of fuss made about a statistic in the 1970's (I think) that Israeli fighter pilots children were 75% male. A lot of thought went into why that was. Was it because they had lots of testosterone or adrenaline, or something to do with the kind of person who chose to be pilots, or something to do with the women who fell for them or the frequency in which they had sex? But the statistician just used the information to form a hypothesis, and when they tested it over the next five years, exactly 50% of Israeli fighter pilots children were male. Hypothesis not proven.

The lesson there of course, is WHY was anyone looking at the sex of Israeli fighter pilot's children in the first place? The answer is that they were drawn to it precisely because the number seemed so unusual, it was such an outlier. In the same way that people are clearly drawn to the guys in your link because they predicted Johnson's majority so accurately, but it most definitely does not automatically follow that they have some innate ability that others do not have.

Sorry for the essay, home isolation gives one time to ramble.....

No one is saying there are 2000 bodies somewhere !

Just that there may be misreporting, which includes delaying.

 

[Blue Valkyrie]

https://twitter.com/AndyNaylorBHAFC/status/1243178212290629633?s=19

 

[Springal]

https://twitter.com/AndyNaylorBHAFC/status/1243178212290629633?s=19

The 3 that went to Cheltenham then ?

 

[Kalimantan Gull]

No one is saying there are 2000 bodies somewhere !

Just that there may be misreporting, which includes delaying.

Well, no, they are suggesting just that, their conclusion is that there have been up to 2,300 additional deaths so far in 2020 which don't appear in the official mortality figures, and that most of these deaths were elderly coronavirus victims.

If true, the bodies have to be somewhere!

What earthly reason would there be for the release of these figures to be delayed? Isn't the government TRYING to scare us into isolation? Pretending fewer people have died would be the total opposite of that!

 

[Blue Valkyrie]

Well, no, they are suggesting just that, their conclusion is that there have been up to 2,300 additional deaths so far in 2020 which don't appear in the official mortality figures, and that most of these deaths were elderly coronavirus victims.

If true, the bodies have to be somewhere!

What earthly reason would there be for the release of these figures to be delayed? Isn't the government TRYING to scare us into isolation? Pretending fewer people have died would be the total opposite of that!

Can't be bothered to argue. Happy you are taking your pinch of salt.

 

[Green Cross Code Man]

What earthly reason would there be for the release of these figures to be delayed? Isn't the government TRYING to scare us into isolation? Pretending fewer people have died would be the total opposite of that!

This!

 

[Blue Valkyrie]

The 3 that went to Cheltenham then ?

Likely. They would have been at a higher infection risk.

 

[Mellotron]

along with the 21-year old girl yesterday, the 37 year-old deputy ambassador.... there's others....

get the point, but its perhaps not as rare as we were led to believe at the beginning, and we have to remember that only a small proportion of the population has so far come into contact with this virus

Again, unconfirmed but there are more and more suspicions that many more have been infected than first thought. I don’t think 50%, but could certainly believe 10-20% by now.

 

[Kalimantan Gull]

Can't be bothered to argue. Happy you are taking your pinch of salt.

OK, no worries. Surprised you still find validity in their analysis. I'm not arguing for the sake of it, I just hate dramatic conclusions presented on the flimsiest of data so wanted to point out the flaws. Their interpretations would get destroyed if peer-reviewed and yet they can happily post it on a web-page and share it through social media. Science is supposed to mean something but it struggles to keep up in this day and age.

 

[Kalimantan Gull]

Again, unconfirmed but there are more and more suspicions that many more have been infected than first thought. I don’t think 50%, but could certainly believe 10-20% by now.

10% of the population? 6.5 million people infected with 10,000 confirmed cases and 6,450,000 mild or asymptomatic cases? I don't think that chimes with the numbers from China or South Korea or elsewhere.

 

[Mellotron]

10% of the population? 6.5 million people infected with 10,000 confirmed cases and 6,450,000 mild or asymptomatic cases? I don't think that chimes with the numbers from China or South Korea or elsewhere.

10,000 confirmed cases is utterly, utterly meaningless though - given there will be plenty who aren’t being tested who may have, say, “mild-to-medium” symptoms. That stat doesn’t help us whatsoever until we get an idea just how many might be asymptomatic.

According to that study of the Italian village, as many as 75%.

Waiting on that Wuhan study by the Japanese to be peer reviewed, if that does pass muster and becomes accepted would it change your view?

 

[Thunder Bolt]

Yeah, I know. Its just when its numbers on a page its easy to digest and rationalise. When its the picture of a pretty 21-year old girl or a 37 year old successful diplomat or a 38-yr old football fan, it all becomes far more real and close to home....

I'm sure all of us, at least those of us under 60, have at some point in this crisis said "I'm not worried about it as I'm young and healthy, just concerned for my parents and grandad" But that worry isn't zero anymore.

and really, assuming we are fundamentally healthy, are careful and don't take risks on the road, don't smoke or take drugs or drink too much, eat a healthy diet, live a sensible life - what is there out there that can kill us in our 30's or 40's? Very very little - almost nothing. And yet here is a virus which CAN infect us, which we will almost certainly come into contact with at some point in the next couple if years, and which CAN kill us at a non-negligible mortality rate.... chance may well be very low, but its also a lot higher than we normally have to face.....

sorry, don't know whats coming over me today.........

This 38 year old football fan

[tweet]1243113217309761538[/tweet]