F*ck it, I wasn't going to bother but I'll do a charge of the Light Brigade and give a couple of reasons why this might not be the slam dunk everything think it is:
1. Voter registration: If you look at the 4 states that release the data (PA, MI, AZ and NC), the Republicans have outperformed the Democrats on registration numbers.
If you take PA (at least), and look at the regression analysis at a county level, there is a very strong correlation between changes in voter registration numbers and share of vote. That regression analysis suggests the PA polls showing Biden winning easily are wrong. Yes, it would mean a massive polling error if Trump wins. But there has to be an even more massive overturning of the rule that changes in voter registration numbers match voting trends pretty well.
2. The data is not marrying up. David mentions some of the state polls that maybe cast a slight doubt on his case for a landslide but. put bluntly, the two sets of data don't match. If Biden is winning +14/+16 at a national level, then he shouldn't be up +1 in AZ or +3 in NC or behind 8 in Maine (the last one is interesting because Mainers don't generally give a f*ck what anyone thinks of them). I see today Gideon's lead in the Senate is now down to +1. Given what we have been told about the decline in ticket splitting, she should be romping home.
3. The anecdotes on the ground are not marrying up. There have been two articles in the last 72 hours suggesting the Democrats are increasingly worried about NV. Now maybe NV is an unique case but there have also been articles suggesting local Democrats are also not happy with how things are being run in FL and, to a degree, in WI. Again, you can talk about "they are being cautious" but that is not sounding like a campaign that is optimistic.
4. The flaws in the polling data. We have plenty of people who use data day to day on this site and do models so they will understand the principle of "garbage in, garbage out". Polling companies are in a worst state now than they were four years ago in terms of margins, staff retention, morale, you name it. There are no accepted standards for American polling.
I accept 3 and 4 are more theoretical but 1 has a fair amount to back it and, for 2, the polls cannot be reconciled - some must be wrong
Of course you should bother, just think of the kudos you'll get if you turn out to be right.
Personally I just get overloaded by data with american elections so just go on my gut, which is unhelpful, but people seem to be able to use the data to make any argument anyway.
The one about anecdotes is interesting though, because I had noted that John Oliver made a crack about Biden's get out the vote operation on his show on Sunday. Might mean nothing, he certainly is not impartial and I get the impression no fan of Biden, but I don't discount the possibility things might still trip up.
As for your 2, I presume David feels the state polls are the ones which are more likely to be bogus.
I think the +14 / +16 are outliers, or at least are top-side MoE. The poll averages are roughly Biden +8 and that's what the state polls are pointing to. But +8 *is* a landslide in Electoral College terms. If Georgia is in the balance, then it's not close.
(FWIW, I agree that Nevada is doing it's own thing and that's one that Biden should be wary of but at the end of the day, if it is doing its own thing then by definition it's not indicative.
Thanks David and thanks for your header, very interesting as usual.
If there was one single piece of data I would be worried about if I was Biden it would be the voter registration number, where there seems to be a high correlation (at least in PA).
Re NV, yes it could be doing its own thing but a more worrying issue for the Dems is that these articles about dissatisfied Hispanic voters are popping up in multiple states. It doesn't seem to be limited to eg Florida or Nevada
F*ck it, I wasn't going to bother but I'll do a charge of the Light Brigade and give a couple of reasons why this might not be the slam dunk everything think it is:
1. Voter registration: If you look at the 4 states that release the data (PA, MI, AZ and NC), the Republicans have outperformed the Democrats on registration numbers.
If you take PA (at least), and look at the regression analysis at a county level, there is a very strong correlation between changes in voter registration numbers and share of vote. That regression analysis suggests the PA polls showing Biden winning easily are wrong. Yes, it would mean a massive polling error if Trump wins. But there has to be an even more massive overturning of the rule that changes in voter registration numbers match voting trends pretty well.
2. The data is not marrying up. David mentions some of the state polls that maybe cast a slight doubt on his case for a landslide but. put bluntly, the two sets of data don't match. If Biden is winning +14/+16 at a national level, then he shouldn't be up +1 in AZ or +3 in NC or behind 8 in Maine (the last one is interesting because Mainers don't generally give a f*ck what anyone thinks of them). I see today Gideon's lead in the Senate is now down to +1. Given what we have been told about the decline in ticket splitting, she should be romping home.
3. The anecdotes on the ground are not marrying up. There have been two articles in the last 72 hours suggesting the Democrats are increasingly worried about NV. Now maybe NV is an unique case but there have also been articles suggesting local Democrats are also not happy with how things are being run in FL and, to a degree, in WI. Again, you can talk about "they are being cautious" but that is not sounding like a campaign that is optimistic.
4. The flaws in the polling data. We have plenty of people who use data day to day on this site and do models so they will understand the principle of "garbage in, garbage out". Polling companies are in a worst state now than they were four years ago in terms of margins, staff retention, morale, you name it. There are no accepted standards for American polling.
I accept 3 and 4 are more theoretical but 1 has a fair amount to back it and, for 2, the polls cannot be reconciled - some must be wrong
Of course you should bother, just think of the kudos you'll get if you turn out to be right.
Personally I just get overloaded by data with american elections so just go on my gut, which is unhelpful, but people seem to be able to use the data to make any argument anyway.
The one about anecdotes is interesting though, because I had noted that John Oliver made a crack about Biden's get out the vote operation on his show on Sunday. Might mean nothing, he certainly is not impartial and I get the impression no fan of Biden, but I don't discount the possibility things might still trip up.
As for your 2, I presume David feels the state polls are the ones which are more likely to be bogus.
Thanks for that. My gut (biased, I know) says that Biden is going to trip up in three areas (1) white suburban voters who went it gets to the secrecy of the ballot box go with Trump but daren't advertise it to their neighbours for social harm (2) Hispanics who feel like they have been neglected by the Democrats and that BLM does not apply to them (3) ironically, Black voters, a small but enough chunk (mainly younger) who vote Republican combined with lower turnout in urban areas that have been impacted by the increased crime wave.
By gut is that those three issues are not sufficient to overcome a gap that has widened to about eight points.
Let's see. One of us will be wrong and the market is saying it will be me!!!!
The Trump +8 in ME CD2 is based on a sample size of just 234.
That poll has caused 538 to move ME CD2 back to Trump.
But surely the sample size is so small the poll should be almost completely ignored.
That’s poor from 538 to include such a rubbish poll. Though presumably they give it very little weight.
I remember raising this point re the small sample size of US polls and being told that it wasn't the size that was important (ahem) but the methodology. So why is this one different?
Genuine question.
Well, if you have a sample size of one, then no matter how good the methodology, you'll get a rubbish result.
However, so long as the ME2 is properly weighted that 234 person poll - while it will have a reasonably large margin of error - should not be summarily binned.
The Trump +8 in ME CD2 is based on a sample size of just 234.
That poll has caused 538 to move ME CD2 back to Trump.
But surely the sample size is so small the poll should be almost completely ignored.
That’s poor from 538 to include such a rubbish poll. Though presumably they give it very little weight.
I remember raising this point re the small sample size of US polls and being told that it wasn't the size that was important (ahem) but the methodology. So why is this one different?
Genuine question.
It's a question of degree. I've always understood that once you get above 500 the increase in accuracy is quite small. Above a 1000 and the increase is negligible. A sample of about 250 however really is a bit too small. I wouldn't dismiss it entirely, but it would have a very large margin of error.
Edit: For the avoidance of doubt, RCS did not collude. Honest.
I suspect being a branded a one-term loser will do Trump's health more harm than Covid-19 will.
What happens after a Biden landslide though? Will Trump's supporters simply refuse to believe the result and foster another "stabbed in the back" myth or will the spell finally be broken? Does the GOP establishment take back control of the party or does Trumpism simply regroup under a more competent leader?
Simply not being Trump will only get Biden so far. What happens when the realisation dawns that America's problems haven't gone away and the country is as divided as ever? Will his health hold up for four years? I expect once the honeymoon period ends AOC will turn on the Democratic establishment pretty quickly.
AOC isn't particularly influential: IIRC her and her PAC endorsed 17 candidates for Congress in the Primaries this year, and only three of them won. (And those three were in districts where they are unlikely to win in the general.)
When Biden hits the mid-term slump AOC et al will claim its because he isn't being radical enough. People claim her proposal of Sanders at the Democratic convention was a mere procedural formality but it still gave her enough distance from Biden to be able to turn fire on him at a later date.
So we're already looking towards Biden's mid-terms are we? I think this is the first speculation on here I've seen about that. Which may in itself be telling.
IIRC there have been a few mentions of 2020 on here but not many.
Re: the dreaded (or hoped for) mid-term slump for US presidents and their parties, note several factors:
> the size (or not) of the president's victory, or rather the extent of his coattails, and consequent size (or lack thereof) of seat margins in both houses of Congress.
> new OR ongoing trends that impact congressional-presidential relationship and public perceptions.
> in the case of 2020, impact of reapportionment and redistricting upon US House allocations and districting; note that in number of states (for example Texas) changing (or maintaining) party control of one or more legislative chambers can have MAJOR impact on 2020 congressional districts.
Redistricting is a feature of this election that has been notably overlooked. It may turn out to have the most impact over the medium term.
I’ve been thinking about that statistic and study that you quoted recently which appeared to show that people who die of coronavirus on average reduce their actuarially expected lifespan by 10 years. Does this actually tell us anything meaningful? Surely everyone who actually dies, dies earlier than their actuarial calculation would have predicted?
Say, before I catch an aeroplane, I am 59 years old, (which I happen to be), and I have an actuarial life expectancy of say 80, (which perhaps I do have; fingers crossed), and the plane crashes and I die. Then I have reduced my actuarial life expectancy by 21 years. Why? Because the plane I was on crashed. Everyone else who boarded a plane that day that didn’t crash has maintained their actuarial life expectancy – or increased it minimally. My life expectancy crashed because my plane crashed. The life expectancy of people who die of coronavirus crashes because they are the people who caught and died of coronavirus.
I suspect I’m missing something here. But it seems to me that most of the actuarial adjustment comes from the fact that the dead people being analysed were people that - died!
To me there seems to have been a misstep taken in this study somewhere between prior probability and known outcome. Bayes and his Bombs may or may not be relevant to my thinking? I’m not a statistician but I’m sure there are other on this blog can determine if I am onto something - or quite possibly talking bunkum!
Did you follow the link from the WSJ to the source papers? They have more detail.
Actuaries have their own mathematics, but my understanding would be that those who didn't die in a plane crash would actuarily gain life expectancy, albeit an infestessimal amount.
Foxy. Yes I did follow the link and read the paper. That's why I am questioning whether is is based on statistically flawed thinking. It seems to show that people who die, die sooner than people who live!
Re my plane crash example. I agree. Plane crash survivors and all survivors improve their life expectancy. That was part of my point.
I don't think you are correct. The survivors life expectancy is unchanged (at least it certainly isn't increased) as it is a calculation based upon everyone and everyone eventually dies. The person who dies however has reduced his personal life expectancy (zero) from what it was the day before.
I’ve been thinking about that statistic and study that you quoted recently which appeared to show that people who die of coronavirus on average reduce their actuarially expected lifespan by 10 years. Does this actually tell us anything meaningful? Surely everyone who actually dies, dies earlier than their actuarial calculation would have predicted?
Presumably it tells you that those who died were not solely the extremely old who were likely to die within the year of something else anyway -- in that extreme case the reduction in lifespan would be much smaller. Since there definitely were quite a lot of deaths among that group there must also be a lot of deaths among relatively healthy people who would otherwise have expected a couple more decades, to balance the average out.
There also needs to be understanding that life expectancy at any age is not the same as life expectancy at birth. For example a British male may have a life expectancy of 82 years at birth, but a life expectancy of 10 years at age 77 Hence the latter loses 10 Actuarial years if he dies at 77 from Covid.
Surely whenever anyone dies of anything, even at age 101, they lose some actuarial life?
I’ve been thinking about that statistic and study that you quoted recently which appeared to show that people who die of coronavirus on average reduce their actuarially expected lifespan by 10 years. Does this actually tell us anything meaningful? Surely everyone who actually dies, dies earlier than their actuarial calculation would have predicted?
Say, before I catch an aeroplane, I am 59 years old, (which I happen to be), and I have an actuarial life expectancy of say 80, (which perhaps I do have; fingers crossed), and the plane crashes and I die. Then I have reduced my actuarial life expectancy by 21 years. Why? Because the plane I was on crashed. Everyone else who boarded a plane that day that didn’t crash has maintained their actuarial life expectancy – or increased it minimally. My life expectancy crashed because my plane crashed. The life expectancy of people who die of coronavirus crashes because they are the people who caught and died of coronavirus.
I suspect I’m missing something here. But it seems to me that most of the actuarial adjustment comes from the fact that the dead people being analysed were people that - died!
To me there seems to have been a misstep taken in this study somewhere between prior probability and known outcome. Bayes and his Bombs may or may not be relevant to my thinking? I’m not a statistician but I’m sure there are other on this blog can determine if I am onto something - or quite possibly talking bunkum!
This is a smart observation. I don't know what the "average life expectancy at death" is for the UK, but it's clearly going to be a number significantly greater than zero!
The number is still meaningful, though, if we think of the road not taken. "But for COVID" those who died would have had an average of 10 years more, say. Just like your hypothetical plane journey cost you 21 years of average life. One could also say that catching COVID costs (making up a number) 0.1 years of average life expectancy, compared with not catching it. This is a reasonable way to quantify the (non-economic) cost of binary events, comparing the consequence of the event against the consequences of the opposite, but I agree with you that it's quite counterintuitive.
--AS
Surely one's "average life expectancy at death" is precisely zero?
Funnily enough, no! (I'm taking "at" to mean "the instant before".) It depends what the average takes into account -- an omniscient being would know that death was about to happen and know that your average life expectancy was zero -- but an actuary's calculation assigns everyone a positive life expectancy even the instant before death. What a positive attitude!
--AS
As a Buddhist the life expectancy at death is, of course infinity. Or until Enlightenment depending on your tradition.
F*ck it, I wasn't going to bother but I'll do a charge of the Light Brigade and give a couple of reasons why this might not be the slam dunk everything think it is:
1. Voter registration: If you look at the 4 states that release the data (PA, MI, AZ and NC), the Republicans have outperformed the Democrats on registration numbers.
If you take PA (at least), and look at the regression analysis at a county level, there is a very strong correlation between changes in voter registration numbers and share of vote. That regression analysis suggests the PA polls showing Biden winning easily are wrong. Yes, it would mean a massive polling error if Trump wins. But there has to be an even more massive overturning of the rule that changes in voter registration numbers match voting trends pretty well.
2. The data is not marrying up. David mentions some of the state polls that maybe cast a slight doubt on his case for a landslide but. put bluntly, the two sets of data don't match. If Biden is winning +14/+16 at a national level, then he shouldn't be up +1 in AZ or +3 in NC or behind 8 in Maine (the last one is interesting because Mainers don't generally give a f*ck what anyone thinks of them). I see today Gideon's lead in the Senate is now down to +1. Given what we have been told about the decline in ticket splitting, she should be romping home.
3. The anecdotes on the ground are not marrying up. There have been two articles in the last 72 hours suggesting the Democrats are increasingly worried about NV. Now maybe NV is an unique case but there have also been articles suggesting local Democrats are also not happy with how things are being run in FL and, to a degree, in WI. Again, you can talk about "they are being cautious" but that is not sounding like a campaign that is optimistic.
4. The flaws in the polling data. We have plenty of people who use data day to day on this site and do models so they will understand the principle of "garbage in, garbage out". Polling companies are in a worst state now than they were four years ago in terms of margins, staff retention, morale, you name it. There are no accepted standards for American polling.
I accept 3 and 4 are more theoretical but 1 has a fair amount to back it and, for 2, the polls cannot be reconciled - some must be wrong
You are aware that the Maine poll is just for ME-02?
I would be interested in seeing more about the link between registrations and vote share - is there a source for this data? One thing is for sure, if Trump wins (which I highly doubt) then you have PB tipster of the year nailed on.
Sorry, I didn't see the question re the registrations and vote share. This is what I was referring to, half way down the article.
The Trump +8 in ME CD2 is based on a sample size of just 234.
That poll has caused 538 to move ME CD2 back to Trump.
But surely the sample size is so small the poll should be almost completely ignored.
That’s poor from 538 to include such a rubbish poll. Though presumably they give it very little weight.
I remember raising this point re the small sample size of US polls and being told that it wasn't the size that was important (ahem) but the methodology. So why is this one different?
Genuine question.
It's a question of degree. I've always understood that once you get above 500 the increase in accuracy is quite small. Above a 1000 and the increase is negligible. A sample of about 250 however really is a bit too small. I wouldn't dismiss it entirely, but it would have a very large margin of error.
Edit: For the avoidance of doubt, RCS did not collude. Honest.
I’ve been thinking about that statistic and study that you quoted recently which appeared to show that people who die of coronavirus on average reduce their actuarially expected lifespan by 10 years. Does this actually tell us anything meaningful? Surely everyone who actually dies, dies earlier than their actuarial calculation would have predicted?
Say, before I catch an aeroplane, I am 59 years old, (which I happen to be), and I have an actuarial life expectancy of say 80, (which perhaps I do have; fingers crossed), and the plane crashes and I die. Then I have reduced my actuarial life expectancy by 21 years. Why? Because the plane I was on crashed. Everyone else who boarded a plane that day that didn’t crash has maintained their actuarial life expectancy – or increased it minimally. My life expectancy crashed because my plane crashed. The life expectancy of people who die of coronavirus crashes because they are the people who caught and died of coronavirus.
I suspect I’m missing something here. But it seems to me that most of the actuarial adjustment comes from the fact that the dead people being analysed were people that - died!
To me there seems to have been a misstep taken in this study somewhere between prior probability and known outcome. Bayes and his Bombs may or may not be relevant to my thinking? I’m not a statistician but I’m sure there are other on this blog can determine if I am onto something - or quite possibly talking bunkum!
An 80 year old in a first world country has a life expectancy of 10 years, so reduces her life expectancy by that amount, by dying of *anything*. Presumably the point is that covid victims are reducing their life expectancy by *only* that amount, unlike a 30 year old dying of cancer and losing 50 years. Which you would expect anyway, because covid victims tend to be old.
That's a great observation that I think supports my point. Thank you.
So how about this as a rough interpretation?
1. If people over 80 years old get coronavirus and it kills them, they die 10 years earlier than expected 2. One in ten 80 year olds who get coronavirus die from it.
3. So getting coronavirus reduces the life expectancy of people over 80 by one year on average.
Coronavirus reduces life expectancy by 1 year, (not 10 years) in the elderly who contract it and die of it.
The Trump +8 in ME CD2 is based on a sample size of just 234.
That poll has caused 538 to move ME CD2 back to Trump.
But surely the sample size is so small the poll should be almost completely ignored.
That’s poor from 538 to include such a rubbish poll. Though presumably they give it very little weight.
I remember raising this point re the small sample size of US polls and being told that it wasn't the size that was important (ahem) but the methodology. So why is this one different?
Genuine question.
Well, if you have a sample size of one, then no matter how good the methodology, you'll get a rubbish result.
However, so long as the ME2 is properly weighted that 234 person poll - while it will have a reasonably large margin of error - should not be summarily binned.
The Trump +8 in ME CD2 is based on a sample size of just 234.
That poll has caused 538 to move ME CD2 back to Trump.
But surely the sample size is so small the poll should be almost completely ignored.
That’s poor from 538 to include such a rubbish poll. Though presumably they give it very little weight.
I remember raising this point re the small sample size of US polls and being told that it wasn't the size that was important (ahem) but the methodology. So why is this one different?
Genuine question.
Not sure who said that but obviously both are important. A poll with 234 respondents has an MOE of 6% which is pretty bad. Really any poll with fewer than 1000 respondents should be viewed with scepticism.
I’ve been thinking about that statistic and study that you quoted recently which appeared to show that people who die of coronavirus on average reduce their actuarially expected lifespan by 10 years. Does this actually tell us anything meaningful? Surely everyone who actually dies, dies earlier than their actuarial calculation would have predicted?
Say, before I catch an aeroplane, I am 59 years old, (which I happen to be), and I have an actuarial life expectancy of say 80, (which perhaps I do have; fingers crossed), and the plane crashes and I die. Then I have reduced my actuarial life expectancy by 21 years. Why? Because the plane I was on crashed. Everyone else who boarded a plane that day that didn’t crash has maintained their actuarial life expectancy – or increased it minimally. My life expectancy crashed because my plane crashed. The life expectancy of people who die of coronavirus crashes because they are the people who caught and died of coronavirus.
I suspect I’m missing something here. But it seems to me that most of the actuarial adjustment comes from the fact that the dead people being analysed were people that - died!
To me there seems to have been a misstep taken in this study somewhere between prior probability and known outcome. Bayes and his Bombs may or may not be relevant to my thinking? I’m not a statistician but I’m sure there are other on this blog can determine if I am onto something - or quite possibly talking bunkum!
Did you follow the link from the WSJ to the source papers? They have more detail.
Actuaries have their own mathematics, but my understanding would be that those who didn't die in a plane crash would actuarily gain life expectancy, albeit an infestessimal amount.
Foxy. Yes I did follow the link and read the paper. That's why I am questioning whether is is based on statistically flawed thinking. It seems to show that people who die, die sooner than people who live!
Re my plane crash example. I agree. Plane crash survivors and all survivors improve their life expectancy. That was part of my point.
You are right. The flaw in the thinking is blindingly obvious now you've pointed it out. Of course if your sub-sample is restricted to those who die then you're going to come up with a statistic which is heavily biased towards a disagreeable outcome.
The correct measure should of course be the reduction in life expectancy amongst those who get infected. A rather different kettle of fish.
The original claim sounded wrong, and I'm kicking myself for not immediately seeing the flaw. Thanks for pointing it out. You are a genius!
FPT Philip_Thompson said: 'If the Tories increase a deficit during a period of growth then that would be bad. Can you name a year when that happened? '
How about 1973 /1974? The economy was racing along under Barber at well over 5% growth per annum yet the Budget Deficit was 4% - much higher than managed by Brown pre- 2007.
And the Barber boom was a notorious example of crap economic management for decades after. So in this case Philip is correct - it was bad.
I suspect being a branded a one-term loser will do Trump's health more harm than Covid-19 will.
What happens after a Biden landslide though? Will Trump's supporters simply refuse to believe the result and foster another "stabbed in the back" myth or will the spell finally be broken? Does the GOP establishment take back control of the party or does Trumpism simply regroup under a more competent leader?
Simply not being Trump will only get Biden so far. What happens when the realisation dawns that America's problems haven't gone away and the country is as divided as ever? Will his health hold up for four years? I expect once the honeymoon period ends AOC will turn on the Democratic establishment pretty quickly.
AOC isn't particularly influential: IIRC her and her PAC endorsed 17 candidates for Congress in the Primaries this year, and only three of them won. (And those three were in districts where they are unlikely to win in the general.)
What about AOC's endorsement of Jamaal Bowman, who defeated long-term incumbent US Rep. Elliot Engel for Democratic nomination in NY CD 16.
Agree she's NOT as powerful as many folks think, but it's a mistake to go overboard under-estimating her.
I’ve been thinking about that statistic and study that you quoted recently which appeared to show that people who die of coronavirus on average reduce their actuarially expected lifespan by 10 years. Does this actually tell us anything meaningful? Surely everyone who actually dies, dies earlier than their actuarial calculation would have predicted?
Say, before I catch an aeroplane, I am 59 years old, (which I happen to be), and I have an actuarial life expectancy of say 80, (which perhaps I do have; fingers crossed), and the plane crashes and I die. Then I have reduced my actuarial life expectancy by 21 years. Why? Because the plane I was on crashed. Everyone else who boarded a plane that day that didn’t crash has maintained their actuarial life expectancy – or increased it minimally. My life expectancy crashed because my plane crashed. The life expectancy of people who die of coronavirus crashes because they are the people who caught and died of coronavirus.
I suspect I’m missing something here. But it seems to me that most of the actuarial adjustment comes from the fact that the dead people being analysed were people that - died!
To me there seems to have been a misstep taken in this study somewhere between prior probability and known outcome. Bayes and his Bombs may or may not be relevant to my thinking? I’m not a statistician but I’m sure there are other on this blog can determine if I am onto something - or quite possibly talking bunkum!
This is a smart observation. I don't know what the "average life expectancy at death" is for the UK, but it's clearly going to be a number significantly greater than zero!
The number is still meaningful, though, if we think of the road not taken. "But for COVID" those who died would have had an average of 10 years more, say. Just like your hypothetical plane journey cost you 21 years of average life. One could also say that catching COVID costs (making up a number) 0.1 years of average life expectancy, compared with not catching it. This is a reasonable way to quantify the (non-economic) cost of binary events, comparing the consequence of the event against the consequences of the opposite, but I agree with you that it's quite counterintuitive.
--AS
Surely one's "average life expectancy at death" is precisely zero?
Funnily enough, no! (I'm taking "at" to mean "the instant before".) It depends what the average takes into account -- an omniscient being would know that death was about to happen and know that your average life expectancy was zero -- but an actuary's calculation assigns everyone a positive life expectancy even the instant before death. What a positive attitude!
--AS
As a Buddhist the life expectancy at death is, of course infinity. Or until Enlightenment depending on your tradition.
Or zero if you are a secular Buddhist.
If you are secular you ain't a Buddhist. You are something else. But this really isn't the time or place.
On Newsnight! Len defunds Labour and funds the Cobynistas in Labour. Internecine war erupts again.
Why doesn't Len stop dicking about, and just hand "his" money straight to Johnson. The outcome is the same. Five more years of Conservative Government's. What a fool!
FPT Philip_Thompson said: 'If the Tories increase a deficit during a period of growth then that would be bad. Can you name a year when that happened? '
How about 1973 /1974? The economy was racing along under Barber at well over 5% growth per annum yet the Budget Deficit was 4% - much higher than managed by Brown pre- 2007.
So pre-Thatcherism and nearly a decade before I was born?
Yes there may have been a mistake then. Pre-Thatcherism economics wasn't the best. But if that's the most recent you've got then it seems the Tories have learnt from that mistake and haven't repeated it even once in my lifetime.
Pre-Thatcherism economics may not have been the best - but the period 1945 - 1974 saw a far better performance than she managed in terms of macroeconomic parameters such as economic growth - balance of payments equilibrium - unemployment levels - and even inflation for most of that period. After 11.5 years of Thatcher, inflation declined by just 0.5% - from 10.2% in May 1979 to 9.7% in November 1990.
Nevada looked like a knife edge in 2016. They won comfortably.
Nevada is hard to poll but the voter registration numbers don't lie. Once we know what October's voter reg numbers are we know who wins the state.
Quite. But this is the point I made about PA. PA's voter numbers have trended more Republican since 2016. So why is everyone assuming PA is moving towards being safe for Biden when the voter registration data should be suggesting the opposite?
Nevada looked like a knife edge in 2016. They won comfortably.
Nevada is hard to poll but the voter registration numbers don't lie. Once we know what October's voter reg numbers are we know who wins the state.
Quite. But this is the point I made about PA. PA's voter numbers have trended more Republican since 2016. So why is everyone assuming PA is moving towards being safe for Biden when the voter registration data should be suggesting the opposite?
The article I pinged to @not_on_fire - check out the chart half way down - change in vote share by county vs party registrations
FPT Philip_Thompson said: 'If the Tories increase a deficit during a period of growth then that would be bad. Can you name a year when that happened? '
How about 1973 /1974? The economy was racing along under Barber at well over 5% growth per annum yet the Budget Deficit was 4% - much higher than managed by Brown pre- 2007.
And the Barber boom was a notorious example of crap economic management for decades after. So in this case Philip is correct - it was bad.
And Reggie Maudling in 1963/64 when he ran up a big balance of payments deficit.
I’ve been thinking about that statistic and study that you quoted recently which appeared to show that people who die of coronavirus on average reduce their actuarially expected lifespan by 10 years. Does this actually tell us anything meaningful? Surely everyone who actually dies, dies earlier than their actuarial calculation would have predicted?
Say, before I catch an aeroplane, I am 59 years old, (which I happen to be), and I have an actuarial life expectancy of say 80, (which perhaps I do have; fingers crossed), and the plane crashes and I die. Then I have reduced my actuarial life expectancy by 21 years. Why? Because the plane I was on crashed. Everyone else who boarded a plane that day that didn’t crash has maintained their actuarial life expectancy – or increased it minimally. My life expectancy crashed because my plane crashed. The life expectancy of people who die of coronavirus crashes because they are the people who caught and died of coronavirus.
I suspect I’m missing something here. But it seems to me that most of the actuarial adjustment comes from the fact that the dead people being analysed were people that - died!
To me there seems to have been a misstep taken in this study somewhere between prior probability and known outcome. Bayes and his Bombs may or may not be relevant to my thinking? I’m not a statistician but I’m sure there are other on this blog can determine if I am onto something - or quite possibly talking bunkum!
Did you follow the link from the WSJ to the source papers? They have more detail.
Actuaries have their own mathematics, but my understanding would be that those who didn't die in a plane crash would actuarily gain life expectancy, albeit an infestessimal amount.
Foxy. Yes I did follow the link and read the paper. That's why I am questioning whether is is based on statistically flawed thinking. It seems to show that people who die, die sooner than people who live!
Re my plane crash example. I agree. Plane crash survivors and all survivors improve their life expectancy. That was part of my point.
You are right. The flaw in the thinking is blindingly obvious now you've pointed it out. Of course if your sub-sample is restricted to those who die then you're going to come up with a statistic which is heavily biased towards a disagreeable outcome.
The correct measure should of course be the reduction in life expectancy amongst those who get infected. A rather different kettle of fish.
The original claim sounded wrong, and I'm kicking myself for not immediately seeing the flaw. Thanks for pointing it out. You are a genius!
FPT Philip_Thompson said: 'If the Tories increase a deficit during a period of growth then that would be bad. Can you name a year when that happened? '
How about 1973 /1974? The economy was racing along under Barber at well over 5% growth per annum yet the Budget Deficit was 4% - much higher than managed by Brown pre- 2007.
And the Barber boom was a notorious example of crap economic management for decades after. So in this case Philip is correct - it was bad.
Thank you.
I was wondering if I was missing something by standing by saying that was bad. It sounds as if Justin admires what Barber did rather than it being regarded as a mistake for the past half century.
I’ve been thinking about that statistic and study that you quoted recently which appeared to show that people who die of coronavirus on average reduce their actuarially expected lifespan by 10 years. Does this actually tell us anything meaningful? Surely everyone who actually dies, dies earlier than their actuarial calculation would have predicted?
Say, before I catch an aeroplane, I am 59 years old, (which I happen to be), and I have an actuarial life expectancy of say 80, (which perhaps I do have; fingers crossed), and the plane crashes and I die. Then I have reduced my actuarial life expectancy by 21 years. Why? Because the plane I was on crashed. Everyone else who boarded a plane that day that didn’t crash has maintained their actuarial life expectancy – or increased it minimally. My life expectancy crashed because my plane crashed. The life expectancy of people who die of coronavirus crashes because they are the people who caught and died of coronavirus.
I suspect I’m missing something here. But it seems to me that most of the actuarial adjustment comes from the fact that the dead people being analysed were people that - died!
To me there seems to have been a misstep taken in this study somewhere between prior probability and known outcome. Bayes and his Bombs may or may not be relevant to my thinking? I’m not a statistician but I’m sure there are other on this blog can determine if I am onto something - or quite possibly talking bunkum!
An 80 year old in a first world country has a life expectancy of 10 years, so reduces her life expectancy by that amount, by dying of *anything*. Presumably the point is that covid victims are reducing their life expectancy by *only* that amount, unlike a 30 year old dying of cancer and losing 50 years. Which you would expect anyway, because covid victims tend to be old.
That's a great observation that I think supports my point. Thank you.
So how about this as a rough interpretation?
1. If people over 80 years old get coronavirus and it kills them, they die 10 years earlier than expected 2. One in ten 80 year olds who get coronavirus die from it.
3. So getting coronavirus reduces the life expectancy of people over 80 by one year on average.
Coronavirus reduces life expectancy by 1 year, (not 10 years) in the elderly who contract it and die of it.
You were right to start with. If people over 80 years old get coronavirus and it kills them, they die 10 years earlier than expected. So Coronavirus reduces life expectancy by 10 years, (not 1 year) in the elderly who contract it and die of it.
I’ve been thinking about that statistic and study that you quoted recently which appeared to show that people who die of coronavirus on average reduce their actuarially expected lifespan by 10 years. Does this actually tell us anything meaningful? Surely everyone who actually dies, dies earlier than their actuarial calculation would have predicted?
Say, before I catch an aeroplane, I am 59 years old, (which I happen to be), and I have an actuarial life expectancy of say 80, (which perhaps I do have; fingers crossed), and the plane crashes and I die. Then I have reduced my actuarial life expectancy by 21 years. Why? Because the plane I was on crashed. Everyone else who boarded a plane that day that didn’t crash has maintained their actuarial life expectancy – or increased it minimally. My life expectancy crashed because my plane crashed. The life expectancy of people who die of coronavirus crashes because they are the people who caught and died of coronavirus.
I suspect I’m missing something here. But it seems to me that most of the actuarial adjustment comes from the fact that the dead people being analysed were people that - died!
To me there seems to have been a misstep taken in this study somewhere between prior probability and known outcome. Bayes and his Bombs may or may not be relevant to my thinking? I’m not a statistician but I’m sure there are other on this blog can determine if I am onto something - or quite possibly talking bunkum!
Did you follow the link from the WSJ to the source papers? They have more detail.
Actuaries have their own mathematics, but my understanding would be that those who didn't die in a plane crash would actuarily gain life expectancy, albeit an infestessimal amount.
Foxy. Yes I did follow the link and read the paper. That's why I am questioning whether is is based on statistically flawed thinking. It seems to show that people who die, die sooner than people who live!
Re my plane crash example. I agree. Plane crash survivors and all survivors improve their life expectancy. That was part of my point.
You are right. The flaw in the thinking is blindingly obvious now you've pointed it out. Of course if your sub-sample is restricted to those who die then you're going to come up with a statistic which is heavily biased towards a disagreeable outcome.
The correct measure should of course be the reduction in life expectancy amongst those who get infected. A rather different kettle of fish.
The original claim sounded wrong, and I'm kicking myself for not immediately seeing the flaw. Thanks for pointing it out. You are a genius!
F*ck it, I wasn't going to bother but I'll do a charge of the Light Brigade and give a couple of reasons why this might not be the slam dunk everything think it is:
1. Voter registration: If you look at the 4 states that release the data (PA, MI, AZ and NC), the Republicans have outperformed the Democrats on registration numbers.
If you take PA (at least), and look at the regression analysis at a county level, there is a very strong correlation between changes in voter registration numbers and share of vote. That regression analysis suggests the PA polls showing Biden winning easily are wrong. Yes, it would mean a massive polling error if Trump wins. But there has to be an even more massive overturning of the rule that changes in voter registration numbers match voting trends pretty well.
2. The data is not marrying up. David mentions some of the state polls that maybe cast a slight doubt on his case for a landslide but. put bluntly, the two sets of data don't match. If Biden is winning +14/+16 at a national level, then he shouldn't be up +1 in AZ or +3 in NC or behind 8 in Maine (the last one is interesting because Mainers don't generally give a f*ck what anyone thinks of them). I see today Gideon's lead in the Senate is now down to +1. Given what we have been told about the decline in ticket splitting, she should be romping home.
3. The anecdotes on the ground are not marrying up. There have been two articles in the last 72 hours suggesting the Democrats are increasingly worried about NV. Now maybe NV is an unique case but there have also been articles suggesting local Democrats are also not happy with how things are being run in FL and, to a degree, in WI. Again, you can talk about "they are being cautious" but that is not sounding like a campaign that is optimistic.
4. The flaws in the polling data. We have plenty of people who use data day to day on this site and do models so they will understand the principle of "garbage in, garbage out". Polling companies are in a worst state now than they were four years ago in terms of margins, staff retention, morale, you name it. There are no accepted standards for American polling.
I accept 3 and 4 are more theoretical but 1 has a fair amount to back it and, for 2, the polls cannot be reconciled - some must be wrong
You are aware that the Maine poll is just for ME-02?
I would be interested in seeing more about the link between registrations and vote share - is there a source for this data? One thing is for sure, if Trump wins (which I highly doubt) then you have PB tipster of the year nailed on.
Sorry, I didn't see the question re the registrations and vote share. This is what I was referring to, half way down the article.
FPT Philip_Thompson said: 'If the Tories increase a deficit during a period of growth then that would be bad. Can you name a year when that happened? '
How about 1973 /1974? The economy was racing along under Barber at well over 5% growth per annum yet the Budget Deficit was 4% - much higher than managed by Brown pre- 2007.
So pre-Thatcherism and nearly a decade before I was born?
Yes there may have been a mistake then. Pre-Thatcherism economics wasn't the best. But if that's the most recent you've got then it seems the Tories have learnt from that mistake and haven't repeated it even once in my lifetime.
Pre-Thatcherism economics may not have been the best - but the period 1945 - 1974 saw a far better performance than she managed in terms of macroeconomic parameters such as economic growth - balance of payments equilibrium - unemployment levels - and even inflation for most of that period. After 11.5 years of Thatcher, inflation declined by just 0.5% - from 10.2% in May 1979 to 9.7% in November 1990.
I'm sorry, but that is utter crap. You can't compare years or decades like that without considering the economic backdrop - such as oil prices.
It's very simple. Pre-1979, the UK was the worst-performing economy amongst major countries in Europe, and getting steadily worse ('the sick man or Europe'). After Thatcher, it was one of the best, probably the best. That's why countries all over the world subsequently copied her approach.
FPT Philip_Thompson said: 'If the Tories increase a deficit during a period of growth then that would be bad. Can you name a year when that happened? '
How about 1973 /1974? The economy was racing along under Barber at well over 5% growth per annum yet the Budget Deficit was 4% - much higher than managed by Brown pre- 2007.
And the Barber boom was a notorious example of crap economic management for decades after. So in this case Philip is correct - it was bad.
Thank you.
I was wondering if I was missing something by standing by saying that was bad. It sounds as if Justin admires what Barber did rather than it being regarded as a mistake for the past half century.
Not at all - the point is that it was far worse than anything Brown did.
The Wilson government of 1964 - 70 bequeathed to its Tory successor both a Budget Surplus and a Balance of Payments Surplus. No Tory Government has managed either!
F*ck it, I wasn't going to bother but I'll do a charge of the Light Brigade and give a couple of reasons why this might not be the slam dunk everything think it is:
1. Voter registration: If you look at the 4 states that release the data (PA, MI, AZ and NC), the Republicans have outperformed the Democrats on registration numbers.
If you take PA (at least), and look at the regression analysis at a county level, there is a very strong correlation between changes in voter registration numbers and share of vote. That regression analysis suggests the PA polls showing Biden winning easily are wrong. Yes, it would mean a massive polling error if Trump wins. But there has to be an even more massive overturning of the rule that changes in voter registration numbers match voting trends pretty well.
2. The data is not marrying up. David mentions some of the state polls that maybe cast a slight doubt on his case for a landslide but. put bluntly, the two sets of data don't match. If Biden is winning +14/+16 at a national level, then he shouldn't be up +1 in AZ or +3 in NC or behind 8 in Maine (the last one is interesting because Mainers don't generally give a f*ck what anyone thinks of them). I see today Gideon's lead in the Senate is now down to +1. Given what we have been told about the decline in ticket splitting, she should be romping home.
3. The anecdotes on the ground are not marrying up. There have been two articles in the last 72 hours suggesting the Democrats are increasingly worried about NV. Now maybe NV is an unique case but there have also been articles suggesting local Democrats are also not happy with how things are being run in FL and, to a degree, in WI. Again, you can talk about "they are being cautious" but that is not sounding like a campaign that is optimistic.
4. The flaws in the polling data. We have plenty of people who use data day to day on this site and do models so they will understand the principle of "garbage in, garbage out". Polling companies are in a worst state now than they were four years ago in terms of margins, staff retention, morale, you name it. There are no accepted standards for American polling.
I accept 3 and 4 are more theoretical but 1 has a fair amount to back it and, for 2, the polls cannot be reconciled - some must be wrong
Just to say I really appreciate having these thorough, well-reasoned, data-based posts that @MrEd is doing against the consensus (although in this case I think I agree with the consensus).
F*ck it, I wasn't going to bother but I'll do a charge of the Light Brigade and give a couple of reasons why this might not be the slam dunk everything think it is:
1. Voter registration: If you look at the 4 states that release the data (PA, MI, AZ and NC), the Republicans have outperformed the Democrats on registration numbers.
If you take PA (at least), and look at the regression analysis at a county level, there is a very strong correlation between changes in voter registration numbers and share of vote. That regression analysis suggests the PA polls showing Biden winning easily are wrong. Yes, it would mean a massive polling error if Trump wins. But there has to be an even more massive overturning of the rule that changes in voter registration numbers match voting trends pretty well.
2. The data is not marrying up. David mentions some of the state polls that maybe cast a slight doubt on his case for a landslide but. put bluntly, the two sets of data don't match. If Biden is winning +14/+16 at a national level, then he shouldn't be up +1 in AZ or +3 in NC or behind 8 in Maine (the last one is interesting because Mainers don't generally give a f*ck what anyone thinks of them). I see today Gideon's lead in the Senate is now down to +1. Given what we have been told about the decline in ticket splitting, she should be romping home.
3. The anecdotes on the ground are not marrying up. There have been two articles in the last 72 hours suggesting the Democrats are increasingly worried about NV. Now maybe NV is an unique case but there have also been articles suggesting local Democrats are also not happy with how things are being run in FL and, to a degree, in WI. Again, you can talk about "they are being cautious" but that is not sounding like a campaign that is optimistic.
4. The flaws in the polling data. We have plenty of people who use data day to day on this site and do models so they will understand the principle of "garbage in, garbage out". Polling companies are in a worst state now than they were four years ago in terms of margins, staff retention, morale, you name it. There are no accepted standards for American polling.
I accept 3 and 4 are more theoretical but 1 has a fair amount to back it and, for 2, the polls cannot be reconciled - some must be wrong
You are aware that the Maine poll is just for ME-02?
I would be interested in seeing more about the link between registrations and vote share - is there a source for this data? One thing is for sure, if Trump wins (which I highly doubt) then you have PB tipster of the year nailed on.
Sorry, I didn't see the question re the registrations and vote share. This is what I was referring to, half way down the article.
FPT Philip_Thompson said: 'If the Tories increase a deficit during a period of growth then that would be bad. Can you name a year when that happened? '
How about 1973 /1974? The economy was racing along under Barber at well over 5% growth per annum yet the Budget Deficit was 4% - much higher than managed by Brown pre- 2007.
And the Barber boom was a notorious example of crap economic management for decades after. So in this case Philip is correct - it was bad.
Thank you.
I was wondering if I was missing something by standing by saying that was bad. It sounds as if Justin admires what Barber did rather than it being regarded as a mistake for the past half century.
Not at all - the point is that it was far worse than anything Brown did.
Fair point. But Brown had all that history to learn from.
FPT Philip_Thompson said: 'If the Tories increase a deficit during a period of growth then that would be bad. Can you name a year when that happened? '
How about 1973 /1974? The economy was racing along under Barber at well over 5% growth per annum yet the Budget Deficit was 4% - much higher than managed by Brown pre- 2007.
So pre-Thatcherism and nearly a decade before I was born?
Yes there may have been a mistake then. Pre-Thatcherism economics wasn't the best. But if that's the most recent you've got then it seems the Tories have learnt from that mistake and haven't repeated it even once in my lifetime.
Pre-Thatcherism economics may not have been the best - but the period 1945 - 1974 saw a far better performance than she managed in terms of macroeconomic parameters such as economic growth - balance of payments equilibrium - unemployment levels - and even inflation for most of that period. After 11.5 years of Thatcher, inflation declined by just 0.5% - from 10.2% in May 1979 to 9.7% in November 1990.
I'm sorry, but that is utter crap. You can't compare years or decades like that without considering the economic backdrop - such as oil prices.
It's very simple. Pre-1979, the UK was the worst-performing economy amongst major countries in Europe, and getting steadily worse ('the sick man or Europe'). After Thatcher, it was one of the best, probably the best. That's why countries all over the world subsequently copied her approach.
I'm sorry too - because the 1974 Labour Government was hit by a more severe oil shock than the 1979 Thatcher government. Sadomasochistic monetarist policies which pushed up Base Rate to 17% in November 1979 and the raising of VAT from 8% to 15% in Howe's first Budget in June 1979 were self inflicted. By Spring 1980 the 10.2% inflation rate inherited in May 1979 had more than doubled to over 22%.
F*ck it, I wasn't going to bother but I'll do a charge of the Light Brigade and give a couple of reasons why this might not be the slam dunk everything think it is:
1. Voter registration: If you look at the 4 states that release the data (PA, MI, AZ and NC), the Republicans have outperformed the Democrats on registration numbers.
If you take PA (at least), and look at the regression analysis at a county level, there is a very strong correlation between changes in voter registration numbers and share of vote. That regression analysis suggests the PA polls showing Biden winning easily are wrong. Yes, it would mean a massive polling error if Trump wins. But there has to be an even more massive overturning of the rule that changes in voter registration numbers match voting trends pretty well.
2. The data is not marrying up. David mentions some of the state polls that maybe cast a slight doubt on his case for a landslide but. put bluntly, the two sets of data don't match. If Biden is winning +14/+16 at a national level, then he shouldn't be up +1 in AZ or +3 in NC or behind 8 in Maine (the last one is interesting because Mainers don't generally give a f*ck what anyone thinks of them). I see today Gideon's lead in the Senate is now down to +1. Given what we have been told about the decline in ticket splitting, she should be romping home.
3. The anecdotes on the ground are not marrying up. There have been two articles in the last 72 hours suggesting the Democrats are increasingly worried about NV. Now maybe NV is an unique case but there have also been articles suggesting local Democrats are also not happy with how things are being run in FL and, to a degree, in WI. Again, you can talk about "they are being cautious" but that is not sounding like a campaign that is optimistic.
4. The flaws in the polling data. We have plenty of people who use data day to day on this site and do models so they will understand the principle of "garbage in, garbage out". Polling companies are in a worst state now than they were four years ago in terms of margins, staff retention, morale, you name it. There are no accepted standards for American polling.
I accept 3 and 4 are more theoretical but 1 has a fair amount to back it and, for 2, the polls cannot be reconciled - some must be wrong
Just to say I really appreciate having these thorough, well-reasoned, data-based posts that @MrEd is doing against the consensus (although in this case I think I agree with the consensus).
May I second that. @MrEd is providing a valuable counter narrative which is never a bad thing. I don't buy it either, but am aware that I don't want to buy it.
Probably the only way Trump can win now is with a very large differential turnout between his best areas outside the big urban areas (in states like Pennsylvania and Wisconsin) and the Democrats' best areas in the cities.
White House Reception for Bereaved Military Families May Also Have Been COVID Spreader Event https://slate.com/news-and-politics/2020/10/white-house-covid-outbreak-may-have-spread-at-gold-star-family-event.html ... Photos from the event show Donald and Melania Trump, who have both since tested positive for COVID-19, standing close to guests. No one in the photos is wearing masks. The New York Times reported Monday that the White House is not doing any contract tracing for events prior to Sept. 28, which would mean that the families present have not been told they were potentially exposed and there has not been any effort made to quarantine any White House officials who were present....
FPT Philip_Thompson said: 'If the Tories increase a deficit during a period of growth then that would be bad. Can you name a year when that happened? '
How about 1973 /1974? The economy was racing along under Barber at well over 5% growth per annum yet the Budget Deficit was 4% - much higher than managed by Brown pre- 2007.
So pre-Thatcherism and nearly a decade before I was born?
Yes there may have been a mistake then. Pre-Thatcherism economics wasn't the best. But if that's the most recent you've got then it seems the Tories have learnt from that mistake and haven't repeated it even once in my lifetime.
Pre-Thatcherism economics may not have been the best - but the period 1945 - 1974 saw a far better performance than she managed in terms of macroeconomic parameters such as economic growth - balance of payments equilibrium - unemployment levels - and even inflation for most of that period. After 11.5 years of Thatcher, inflation declined by just 0.5% - from 10.2% in May 1979 to 9.7% in November 1990.
I'm sorry, but that is utter crap. You can't compare years or decades like that without considering the economic backdrop - such as oil prices.
It's very simple. Pre-1979, the UK was the worst-performing economy amongst major countries in Europe, and getting steadily worse ('the sick man or Europe'). After Thatcher, it was one of the best, probably the best. That's why countries all over the world subsequently copied her approach.
I'm sorry too - because the 1974 Labour Government was hit by a more severe oil shock than the 1979 Thatcher government. Sadomasochistic monetarist policies which pushed up Base Rate to 17% in November 1979 and the raising of VAT from 8% to 15% in Howe's first Budget in June 1979 were self inflicted. By Spring 1980 the 10.2% inflation rate inherited in May 1979 had more than doubled to over 22%.
I see that you have totally ignored my point.
Go off and find a graph of inflation-adjusted GDP 1960 to 2000 for the UK against other major European countries. You might find it eye-opening.
FPT Philip_Thompson said: 'If the Tories increase a deficit during a period of growth then that would be bad. Can you name a year when that happened? '
How about 1973 /1974? The economy was racing along under Barber at well over 5% growth per annum yet the Budget Deficit was 4% - much higher than managed by Brown pre- 2007.
And the Barber boom was a notorious example of crap economic management for decades after. So in this case Philip is correct - it was bad.
Thank you.
I was wondering if I was missing something by standing by saying that was bad. It sounds as if Justin admires what Barber did rather than it being regarded as a mistake for the past half century.
Not at all - the point is that it was far worse than anything Brown did.
Nearly half a century ago by someone long dead and the lesson was meant to have been learnt before I was even born. Why didn't Brown learn the lesson and why did he make the mistake?
If your argument is that Brown was simply repeating history and failed to learn from it then how is that an improvement?
I’ve been thinking about that statistic and study that you quoted recently which appeared to show that people who die of coronavirus on average reduce their actuarially expected lifespan by 10 years. Does this actually tell us anything meaningful? Surely everyone who actually dies, dies earlier than their actuarial calculation would have predicted?
Say, before I catch an aeroplane, I am 59 years old, (which I happen to be), and I have an actuarial life expectancy of say 80, (which perhaps I do have; fingers crossed), and the plane crashes and I die. Then I have reduced my actuarial life expectancy by 21 years. Why? Because the plane I was on crashed. Everyone else who boarded a plane that day that didn’t crash has maintained their actuarial life expectancy – or increased it minimally. My life expectancy crashed because my plane crashed. The life expectancy of people who die of coronavirus crashes because they are the people who caught and died of coronavirus.
I suspect I’m missing something here. But it seems to me that most of the actuarial adjustment comes from the fact that the dead people being analysed were people that - died!
To me there seems to have been a misstep taken in this study somewhere between prior probability and known outcome. Bayes and his Bombs may or may not be relevant to my thinking? I’m not a statistician but I’m sure there are other on this blog can determine if I am onto something - or quite possibly talking bunkum!
An 80 year old in a first world country has a life expectancy of 10 years, so reduces her life expectancy by that amount, by dying of *anything*. Presumably the point is that covid victims are reducing their life expectancy by *only* that amount, unlike a 30 year old dying of cancer and losing 50 years. Which you would expect anyway, because covid victims tend to be old.
That's a great observation that I think supports my point. Thank you.
So how about this as a rough interpretation?
1. If people over 80 years old get coronavirus and it kills them, they die 10 years earlier than expected 2. One in ten 80 year olds who get coronavirus die from it.
3. So getting coronavirus reduces the life expectancy of people over 80 by one year on average.
Coronavirus reduces life expectancy by 1 year, (not 10 years) in the elderly who contract it and die of it.
You were right to start with. If people over 80 years old get coronavirus and it kills them, they die 10 years earlier than expected. So Coronavirus reduces life expectancy by 10 years, (not 1 year) in the elderly who contract it and die of it.
I think you are over-thinking this stjohn!
Barnesian. Yes. My last sentence was wrong. Thanks. I think points 1-3 still work.
Expert John Edmunds on Newsnight was basically calling for another national lockdown.
Without being funny, his initial position back in Feb / March was more relaxed, however since then he has become a bit of a lockdown absolutist. It is totally unsurprising he is again calling for one.
FPT Philip_Thompson said: 'If the Tories increase a deficit during a period of growth then that would be bad. Can you name a year when that happened? '
How about 1973 /1974? The economy was racing along under Barber at well over 5% growth per annum yet the Budget Deficit was 4% - much higher than managed by Brown pre- 2007.
And the Barber boom was a notorious example of crap economic management for decades after. So in this case Philip is correct - it was bad.
Thank you.
I was wondering if I was missing something by standing by saying that was bad. It sounds as if Justin admires what Barber did rather than it being regarded as a mistake for the past half century.
Not at all - the point is that it was far worse than anything Brown did.
Nearly half a century ago by someone long dead and the lesson was meant to have been learnt before I was even born. Why didn't Brown learn the lesson and why did he make the mistake?
If your argument is that Brown was simply repeating history and failed to learn from it then how is that an improvement?
The deficits under Brown were not out of line for the entire period 1952 - 1967 when the average Budget Deficit was 2.5% of GDP - despite the absence of a recession over those years.
FPT Philip_Thompson said: 'If the Tories increase a deficit during a period of growth then that would be bad. Can you name a year when that happened? '
How about 1973 /1974? The economy was racing along under Barber at well over 5% growth per annum yet the Budget Deficit was 4% - much higher than managed by Brown pre- 2007.
So pre-Thatcherism and nearly a decade before I was born?
Yes there may have been a mistake then. Pre-Thatcherism economics wasn't the best. But if that's the most recent you've got then it seems the Tories have learnt from that mistake and haven't repeated it even once in my lifetime.
Pre-Thatcherism economics may not have been the best - but the period 1945 - 1974 saw a far better performance than she managed in terms of macroeconomic parameters such as economic growth - balance of payments equilibrium - unemployment levels - and even inflation for most of that period. After 11.5 years of Thatcher, inflation declined by just 0.5% - from 10.2% in May 1979 to 9.7% in November 1990.
I'm sorry, but that is utter crap. You can't compare years or decades like that without considering the economic backdrop - such as oil prices.
It's very simple. Pre-1979, the UK was the worst-performing economy amongst major countries in Europe, and getting steadily worse ('the sick man or Europe'). After Thatcher, it was one of the best, probably the best. That's why countries all over the world subsequently copied her approach.
I'm sorry too - because the 1974 Labour Government was hit by a more severe oil shock than the 1979 Thatcher government. Sadomasochistic monetarist policies which pushed up Base Rate to 17% in November 1979 and the raising of VAT from 8% to 15% in Howe's first Budget in June 1979 were self inflicted. By Spring 1980 the 10.2% inflation rate inherited in May 1979 had more than doubled to over 22%.
I see that you have totally ignored my point.
Go off and find a graph of inflation-adjusted GDP 1960 to 2000 for the UK against other major European countries. You might find it eye-opening.
For heavens sake - the policies pusued by Thatcher in 1979/80 were abandoned by Lawson anyway from the mid- 1980s.
FPT Philip_Thompson said: 'If the Tories increase a deficit during a period of growth then that would be bad. Can you name a year when that happened? '
How about 1973 /1974? The economy was racing along under Barber at well over 5% growth per annum yet the Budget Deficit was 4% - much higher than managed by Brown pre- 2007.
And the Barber boom was a notorious example of crap economic management for decades after. So in this case Philip is correct - it was bad.
Thank you.
I was wondering if I was missing something by standing by saying that was bad. It sounds as if Justin admires what Barber did rather than it being regarded as a mistake for the past half century.
Not at all - the point is that it was far worse than anything Brown did.
Nearly half a century ago by someone long dead and the lesson was meant to have been learnt before I was even born. Why didn't Brown learn the lesson and why did he make the mistake?
If your argument is that Brown was simply repeating history and failed to learn from it then how is that an improvement?
The deficits under Brown were not out of line for the entire period 1952 - 1967 when the average Budget Deficit was 2.5% of GDP - despite the absence of a recession over those years.
Yes they are. You might want to double check but yes Brown's were worse.
I'm old enough to remember the Barber boom. There's (not surprisingly) rather a lot of history re-writing on it. Whilst I wouldn't go so far as to defend him, it is worth pointing out that:
- At the time, almost no-one was criticising him for being too expansionist. Rather the reverse: the Labour Party was arguing that he wasn't being expansionist enough.
- Any discussion of the subject which doesn't mention the 1973 oil price crisis - which was comparable in impact to the 2008/9 world economic crisis - can safely be dismissed.
FPT Philip_Thompson said: 'If the Tories increase a deficit during a period of growth then that would be bad. Can you name a year when that happened? '
How about 1973 /1974? The economy was racing along under Barber at well over 5% growth per annum yet the Budget Deficit was 4% - much higher than managed by Brown pre- 2007.
So pre-Thatcherism and nearly a decade before I was born?
Yes there may have been a mistake then. Pre-Thatcherism economics wasn't the best. But if that's the most recent you've got then it seems the Tories have learnt from that mistake and haven't repeated it even once in my lifetime.
Pre-Thatcherism economics may not have been the best - but the period 1945 - 1974 saw a far better performance than she managed in terms of macroeconomic parameters such as economic growth - balance of payments equilibrium - unemployment levels - and even inflation for most of that period. After 11.5 years of Thatcher, inflation declined by just 0.5% - from 10.2% in May 1979 to 9.7% in November 1990.
I'm sorry, but that is utter crap. You can't compare years or decades like that without considering the economic backdrop - such as oil prices.
It's very simple. Pre-1979, the UK was the worst-performing economy amongst major countries in Europe, and getting steadily worse ('the sick man or Europe'). After Thatcher, it was one of the best, probably the best. That's why countries all over the world subsequently copied her approach.
I'm sorry too - because the 1974 Labour Government was hit by a more severe oil shock than the 1979 Thatcher government. Sadomasochistic monetarist policies which pushed up Base Rate to 17% in November 1979 and the raising of VAT from 8% to 15% in Howe's first Budget in June 1979 were self inflicted. By Spring 1980 the 10.2% inflation rate inherited in May 1979 had more than doubled to over 22%.
I see that you have totally ignored my point.
Go off and find a graph of inflation-adjusted GDP 1960 to 2000 for the UK against other major European countries. You might find it eye-opening.
For heavens sake - the policies pusued by Thatcher in 1979/80 were abandoned by Lawson anyway from the mid- 1980s.
So lefties like to claim, with no evidence. But even if it were true, so what? Lawson was Chancellor under which PM, remind me?
According to one fresher at the University of Manchester, the “Covid Positive” party in the university’s Fallowfield campus halls of residence was broken up by security on Saturday. It is just one of the increasing instances of students’ risky behaviour during lockdown restrictions.
“There was a flat party a few days ago which had a policy that you could only get in if you were positive. It was like their health-and-safety measure,” the 18-year-old physics student said.
The Wilson government of 1964 - 70 bequeathed to its Tory successor both a Budget Surplus and a Balance of Payments Surplus. No Tory Government has managed either!
The economy passed to Blair and Brown wasn't so bad. It had a rapidly shrinking budget deficit, and its current account was in balance.
Given national polls range from IBID/TIPP with Biden just 3% ahead and only 1% further ahead than Hillary to Biden 16% ahead with CNN ie Reagan landslide 1984 territory, a Biden landslide is in the range of possibilities at present but so is a narrow Trump re election in the Electoral College
FPT Philip_Thompson said: 'If the Tories increase a deficit during a period of growth then that would be bad. Can you name a year when that happened? '
How about 1973 /1974? The economy was racing along under Barber at well over 5% growth per annum yet the Budget Deficit was 4% - much higher than managed by Brown pre- 2007.
So pre-Thatcherism and nearly a decade before I was born?
Yes there may have been a mistake then. Pre-Thatcherism economics wasn't the best. But if that's the most recent you've got then it seems the Tories have learnt from that mistake and haven't repeated it even once in my lifetime.
Pre-Thatcherism economics may not have been the best - but the period 1945 - 1974 saw a far better performance than she managed in terms of macroeconomic parameters such as economic growth - balance of payments equilibrium - unemployment levels - and even inflation for most of that period. After 11.5 years of Thatcher, inflation declined by just 0.5% - from 10.2% in May 1979 to 9.7% in November 1990.
The balance of payments was a consequence of Bretton Woods: basically countries weren't allowd to run persistent deficits (or surpluses).
Trump seems to be present at these super spreading events....I am more shocked to hear Trump actually had a debate prep team, because it certainly didn't show.
FPT Philip_Thompson said: 'If the Tories increase a deficit during a period of growth then that would be bad. Can you name a year when that happened? '
How about 1973 /1974? The economy was racing along under Barber at well over 5% growth per annum yet the Budget Deficit was 4% - much higher than managed by Brown pre- 2007.
So pre-Thatcherism and nearly a decade before I was born?
Yes there may have been a mistake then. Pre-Thatcherism economics wasn't the best. But if that's the most recent you've got then it seems the Tories have learnt from that mistake and haven't repeated it even once in my lifetime.
Pre-Thatcherism economics may not have been the best - but the period 1945 - 1974 saw a far better performance than she managed in terms of macroeconomic parameters such as economic growth - balance of payments equilibrium - unemployment levels - and even inflation for most of that period. After 11.5 years of Thatcher, inflation declined by just 0.5% - from 10.2% in May 1979 to 9.7% in November 1990.
I'm sorry, but that is utter crap. You can't compare years or decades like that without considering the economic backdrop - such as oil prices.
It's very simple. Pre-1979, the UK was the worst-performing economy amongst major countries in Europe, and getting steadily worse ('the sick man or Europe'). After Thatcher, it was one of the best, probably the best. That's why countries all over the world subsequently copied her approach.
I'm sorry too - because the 1974 Labour Government was hit by a more severe oil shock than the 1979 Thatcher government. Sadomasochistic monetarist policies which pushed up Base Rate to 17% in November 1979 and the raising of VAT from 8% to 15% in Howe's first Budget in June 1979 were self inflicted. By Spring 1980 the 10.2% inflation rate inherited in May 1979 had more than doubled to over 22%.
I see that you have totally ignored my point.
Go off and find a graph of inflation-adjusted GDP 1960 to 2000 for the UK against other major European countries. You might find it eye-opening.
For heavens sake - the policies pusued by Thatcher in 1979/80 were abandoned by Lawson anyway from the mid- 1980s.
So lefties like to claim, with no evidence. But even if it were true, so what? Lawson was Chancellor under which PM, remind me?
And I see you are still ignoring my point.
Lawson was the second Chancellor to be appointed by the Anti-Christ.
FPT Philip_Thompson said: 'If the Tories increase a deficit during a period of growth then that would be bad. Can you name a year when that happened? '
How about 1973 /1974? The economy was racing along under Barber at well over 5% growth per annum yet the Budget Deficit was 4% - much higher than managed by Brown pre- 2007.
So pre-Thatcherism and nearly a decade before I was born?
Yes there may have been a mistake then. Pre-Thatcherism economics wasn't the best. But if that's the most recent you've got then it seems the Tories have learnt from that mistake and haven't repeated it even once in my lifetime.
Pre-Thatcherism economics may not have been the best - but the period 1945 - 1974 saw a far better performance than she managed in terms of macroeconomic parameters such as economic growth - balance of payments equilibrium - unemployment levels - and even inflation for most of that period. After 11.5 years of Thatcher, inflation declined by just 0.5% - from 10.2% in May 1979 to 9.7% in November 1990.
I'm sorry, but that is utter crap. You can't compare years or decades like that without considering the economic backdrop - such as oil prices.
It's very simple. Pre-1979, the UK was the worst-performing economy amongst major countries in Europe, and getting steadily worse ('the sick man or Europe'). After Thatcher, it was one of the best, probably the best. That's why countries all over the world subsequently copied her approach.
I'm sorry too - because the 1974 Labour Government was hit by a more severe oil shock than the 1979 Thatcher government. Sadomasochistic monetarist policies which pushed up Base Rate to 17% in November 1979 and the raising of VAT from 8% to 15% in Howe's first Budget in June 1979 were self inflicted. By Spring 1980 the 10.2% inflation rate inherited in May 1979 had more than doubled to over 22%.
I see that you have totally ignored my point.
Go off and find a graph of inflation-adjusted GDP 1960 to 2000 for the UK against other major European countries. You might find it eye-opening.
For heavens sake - the policies pusued by Thatcher in 1979/80 were abandoned by Lawson anyway from the mid- 1980s.
So lefties like to claim, with no evidence. But even if it were true, so what? Lawson was Chancellor under which PM, remind me?
And I see you are still ignoring my point.
I was unaware that any oil crisis was responsible for inflation in 1989 /90 returning - indeed at times exceeding - the level inherited by Thatcher in May 1979.
The Wilson government of 1964 - 70 bequeathed to its Tory successor both a Budget Surplus and a Balance of Payments Surplus. No Tory Government has managed either!
The economy passed to Blair and Brown wasn't so bad. It had a rapidly shrinking budget deficit, and its current account was in balance.
Is Trump playing a game here, or is he just bad at politics?
I think clearly the second one: He's being driven by stuff he sees on right-wing TV, rather than his own rational political self-interest.
This should also inform our take on his attack on postal voting, which another case of the classic dilemma where you're not sure whether he has some cunning, evil plan to get the Dem votes in a different box to the GOP votes and disqualify them in court, or whether he's an immense ignorant doofus who's left his unlucky party way behind on GOTV, and having to persuade voters to needlessly put their lives at risk by voting in person.
The Wilson government of 1964 - 70 bequeathed to its Tory successor both a Budget Surplus and a Balance of Payments Surplus. No Tory Government has managed either!
The economy passed to Blair and Brown wasn't so bad. It had a rapidly shrinking budget deficit, and its current account was in balance.
Worth remembering that the methodology has been adjusted since last time to take into account HS education - plus of course there are much fewer undecideds this time around. Last time undecideds broke to Trump late on, but if Biden is over 50% in PA then it doesn't matter how many undecideds break to Trump there this time around.
Worth remembering that the methodology has been adjusted since last time to take into account HS education - plus of course there are much fewer undecideds this time around. Last time undecideds broke to Trump late on, but if Biden is over 50% in PA then it doesn't matter how many undecideds break to Trump there this time around.
Biden's average lead in these states is 3.2% compared to 1.6% for Hillary Clinton.
Worth remembering that the methodology has been adjusted since last time to take into account HS education - plus of course there are much fewer undecideds this time around. Last time undecideds broke to Trump late on, but if Biden is over 50% in PA then it doesn't matter how many undecideds break to Trump there this time around.
Not really, apart from Nevada and Georgia Biden is within 5% of where Hillary was 4 years ago in all of them and in the must win states of Michigan, Pennsylvania and Wisconsin for Biden his lead is actually smaller than Hillary's, though he has a slightly bigger lead than she did in Florida and is closer than where she was in Texas.
In Michigan, Florida and Arizona without at least 1 of which Biden cannot win the EC he is also still under 50% on average
Is Trump playing a game here, or is he just bad at politics?
I think clearly the second one: He's being driven by stuff he sees on right-wing TV, rather than his own rational political self-interest.
This should also inform our take on his attack on postal voting, which another case of the classic dilemma where you're not sure whether he has some cunning, evil plan to get the Dem votes in a different box to the GOP votes and disqualify them in court, or whether he's an immense ignorant doofus who's left his unlucky party way behind on GOTV, and having to persuade voters to needlessly put their lives at risk by voting in person.
Worth remembering that the methodology has been adjusted since last time to take into account HS education - plus of course there are much fewer undecideds this time around. Last time undecideds broke to Trump late on, but if Biden is over 50% in PA then it doesn't matter how many undecideds break to Trump there this time around.
Biden's average lead in these states is 3.2% compared to 1.6% for Hillary Clinton.
Indeed. Double the lead, after the methodology change - and with fewer undecideds for Trump to turn around.
Nevada looked like a knife edge in 2016. They won comfortably.
Nevada is hard to poll but the voter registration numbers don't lie. Once we know what October's voter reg numbers are we know who wins the state.
Quite. But this is the point I made about PA. PA's voter numbers have trended more Republican since 2016. So why is everyone assuming PA is moving towards being safe for Biden when the voter registration data should be suggesting the opposite?
Well, the voter registration trends should affect the weightings on polls.
Imagine there are 100 voters in 2016, 30 Registered Dems, 40 Registered Independents, and 30 Registered Republicans. And imagine in 2016, independents broke 55% R / 45% D in the General.
In this case, the election would be 52/48.
Now imagine that it moved to 30/30/40, but that the remaining independents broke 2-1 for the Democrats. In which case, despite the change in registration numbers, it'd be 50%-50% in the General.
The campaign has four weeks to go, I have no doubt that Trump may turn things around - not sure how but Biden's team need to be very careful, he's not a great candidate and a dose of corona for JB could be even more disastrous for his campaign. It makes for an interesting run in to the final weeks, one term Presidents dont come round very often and I am not convinced JB is home and and dry
Worth remembering that the methodology has been adjusted since last time to take into account HS education - plus of course there are much fewer undecideds this time around. Last time undecideds broke to Trump late on, but if Biden is over 50% in PA then it doesn't matter how many undecideds break to Trump there this time around.
Not really, apart from Nevada and Georgia Biden is within 5% of where Hillary was 4 years ago in all of them and in the must win states of Michigan, Pennsylvania and Wisconsin for Biden his lead is actually smaller than Hillary's, though he has a slightly bigger lead than she did in Florida and is closer than where she was in Texas.
In Michigan, Florida and Arizona without at least 1 of which Biden cannot win the EC he is also still under 50% on average
Point of order: Biden can with the EC without any of Michigan, Florida or Arizona. Wisconsin plus Pennsylvania plus almost any other state with eight or more electoral college votes gets him there...
Worth remembering that the methodology has been adjusted since last time to take into account HS education - plus of course there are much fewer undecideds this time around. Last time undecideds broke to Trump late on, but if Biden is over 50% in PA then it doesn't matter how many undecideds break to Trump there this time around.
Not really, apart from Nevada and Georgia Biden is within 5% of where Hillary was 4 years ago in all of them and in the must win states of Michigan, Pennsylvania and Wisconsin for Biden his lead is actually smaller than Hillary's, though he has a slightly bigger lead than she did in Florida and is closer than where she was in Texas.
In Michigan, Florida and Arizona without at least 1 of which Biden cannot win the EC he is also still under 50% on average
In Florida 4 years ago today Clinton was polling 46.6% and on the election she got 46.6% - but she lost not due to losing share but Trump gaining it.
Biden is polling higher which not just means he's polling more than Trump won with last time, not just has he got a bigger gap in the polls, but there's a smaller pond for Trump to fish in to attract voters.
The lack of undecideds is one of the big stories of this election.
Worth remembering that the methodology has been adjusted since last time to take into account HS education - plus of course there are much fewer undecideds this time around. Last time undecideds broke to Trump late on, but if Biden is over 50% in PA then it doesn't matter how many undecideds break to Trump there this time around.
Not really, apart from Nevada and Georgia Biden is within 5% of where Hillary was 4 years ago in all of them and in the must win states of Michigan, Pennsylvania and Wisconsin for Biden his lead is actually smaller than Hillary's, though he has a slightly bigger lead than she did in Florida and is closer than where she was in Texas.
In Michigan, Florida and Arizona without at least 1 of which Biden cannot win the EC he is also still under 50% on average
Point of order: Biden can with the EC without any of Michigan, Florida or Arizona. Wisconsin plus Pennsylvania plus almost any other state with eight or more electoral college votes gets him there...
Trump is polling better in all his other 2016 states than in Wisconsin, Pennsylvania, Arizona, Michigan and Florida so if Biden does not win at least 3 of those states it is highly likely Trump will be re elected
Worth remembering that the methodology has been adjusted since last time to take into account HS education - plus of course there are much fewer undecideds this time around. Last time undecideds broke to Trump late on, but if Biden is over 50% in PA then it doesn't matter how many undecideds break to Trump there this time around.
Not really, apart from Nevada and Georgia Biden is within 5% of where Hillary was 4 years ago in all of them and in the must win states of Michigan, Pennsylvania and Wisconsin for Biden his lead is actually smaller than Hillary's, though he has a slightly bigger lead than she did in Florida and is closer than where she was in Texas.
In Michigan, Florida and Arizona without at least 1 of which Biden cannot win the EC he is also still under 50% on average
Point of order: Biden can with the EC without any of Michigan, Florida or Arizona. Wisconsin plus Pennsylvania plus almost any other state with eight or more electoral college votes gets him there...
Trump is polling better in all his other 2016 states than in Wisconsin, Pennsylvania, Arizona, Michigan and Florida so if Biden does not win at least 3 of those states it is highly likely Trump will be re elected
Well, this whole thing is premised on "what if the result isn't what the polls show now", so it's weird to take that as the premise but assume that none of the other states can show a divergence in the other direction.
Is there a single example anywhere of "focused protection" working in practice? ie has any country or area anywhere managed to have high levels of contagion everywhere, and yet keep it away from old people?
Is it me or has Trump's "release" raised a whole load more questions than if he had stayed in the hospital. The discharge has almost been more damaging than the original diagnosis.
I’ve been thinking about that statistic and study that you quoted recently which appeared to show that people who die of coronavirus on average reduce their actuarially expected lifespan by 10 years. Does this actually tell us anything meaningful? Surely everyone who actually dies, dies earlier than their actuarial calculation would have predicted?
Say, before I catch an aeroplane, I am 59 years old, (which I happen to be), and I have an actuarial life expectancy of say 80, (which perhaps I do have; fingers crossed), and the plane crashes and I die. Then I have reduced my actuarial life expectancy by 21 years. Why? Because the plane I was on crashed. Everyone else who boarded a plane that day that didn’t crash has maintained their actuarial life expectancy – or increased it minimally. My life expectancy crashed because my plane crashed. The life expectancy of people who die of coronavirus crashes because they are the people who caught and died of coronavirus.
I suspect I’m missing something here. But it seems to me that most of the actuarial adjustment comes from the fact that the dead people being analysed were people that - died!
To me there seems to have been a misstep taken in this study somewhere between prior probability and known outcome. Bayes and his Bombs may or may not be relevant to my thinking? I’m not a statistician but I’m sure there are other on this blog can determine if I am onto something - or quite possibly talking bunkum!
An 80 year old in a first world country has a life expectancy of 10 years, so reduces her life expectancy by that amount, by dying of *anything*. Presumably the point is that covid victims are reducing their life expectancy by *only* that amount, unlike a 30 year old dying of cancer and losing 50 years. Which you would expect anyway, because covid victims tend to be old.
That's a great observation that I think supports my point. Thank you.
So how about this as a rough interpretation?
1. If people over 80 years old get coronavirus and it kills them, they die 10 years earlier than expected 2. One in ten 80 year olds who get coronavirus die from it.
3. So getting coronavirus reduces the life expectancy of people over 80 by one year on average.
Coronavirus reduces life expectancy by 1 year, (not 10 years) in the elderly who contract it and die of it.
Comments
They won comfortably
Nevada looked like a knife edge in 2016. They won comfortably.
Nevada is hard to poll but the voter registration numbers don't lie. Once we know what October's voter reg numbers are we know who wins the state.
If there was one single piece of data I would be worried about if I was Biden it would be the voter registration number, where there seems to be a high correlation (at least in PA).
Re NV, yes it could be doing its own thing but a more worrying issue for the Dems is that these articles about dissatisfied Hispanic voters are popping up in multiple states. It doesn't seem to be limited to eg Florida or Nevada
However, so long as the ME2 is properly weighted that 234 person poll - while it will have a reasonably large margin of error - should not be summarily binned.
Edit: For the avoidance of doubt, RCS did not collude. Honest.
https://www.inquirer.com/politics/pennsylvania/pennsylvania-party-voter-registration-trump-biden-20200729.html
Yes, I do think the landslide is on. No certainty, obviously, but definitely on.
So how about this as a rough interpretation?
1. If people over 80 years old get coronavirus and it kills them, they die 10 years earlier than expected
2. One in ten 80 year olds who get coronavirus die from it.
3. So getting coronavirus reduces the life expectancy of people over 80 by one year on average.
Coronavirus reduces life expectancy by 1 year, (not 10 years) in the elderly who contract it and die of it.
https://twitter.com/freebeacon/status/1313579157826265088?s=21
The correct measure should of course be the reduction in life expectancy amongst those who get infected. A rather different kettle of fish.
The original claim sounded wrong, and I'm kicking myself for not immediately seeing the flaw. Thanks for pointing it out. You are a genius!
So in this case Philip is correct - it was bad.
Agree she's NOT as powerful as many folks think, but it's a mistake to go overboard under-estimating her.
You are something else. But this really isn't the time or place.
On Newsnight! Len defunds Labour and funds the Cobynistas in Labour. Internecine war erupts again.
Why doesn't Len stop dicking about, and just hand "his" money straight to Johnson. The outcome is the same. Five more years of Conservative Government's. What a fool!
https://www.inquirer.com/politics/pennsylvania/pennsylvania-party-voter-registration-trump-biden-20200729.html
I was wondering if I was missing something by standing by saying that was bad. It sounds as if Justin admires what Barber did rather than it being regarded as a mistake for the past half century.
If people over 80 years old get coronavirus and it kills them, they die 10 years earlier than expected.
So Coronavirus reduces life expectancy by 10 years, (not 1 year) in the elderly who contract it and die of it.
I think you are over-thinking this stjohn!
It's very simple. Pre-1979, the UK was the worst-performing economy amongst major countries in Europe, and getting steadily worse ('the sick man or Europe'). After Thatcher, it was one of the best, probably the best. That's why countries all over the world subsequently copied her approach.
Is Trump playing a game here, or is he just bad at politics?
https://medium.com/@tombonier/the-gops-claims-about-a-voter-registration-advantage-in-pa-aren-t-true-here-are-the-facts-471010a2cd74
But Brown had all that history to learn from.
https://slate.com/news-and-politics/2020/10/white-house-covid-outbreak-may-have-spread-at-gold-star-family-event.html
... Photos from the event show Donald and Melania Trump, who have both since tested positive for COVID-19, standing close to guests. No one in the photos is wearing masks. The New York Times reported Monday that the White House is not doing any contract tracing for events prior to Sept. 28, which would mean that the families present have not been told they were potentially exposed and there has not been any effort made to quarantine any White House officials who were present....
This was the day after the Barrett event.
Go off and find a graph of inflation-adjusted GDP 1960 to 2000 for the UK against other major European countries. You might find it eye-opening.
If your argument is that Brown was simply repeating history and failed to learn from it then how is that an improvement?
:-)
- At the time, almost no-one was criticising him for being too expansionist. Rather the reverse: the Labour Party was arguing that he wasn't being expansionist enough.
- Any discussion of the subject which doesn't mention the 1973 oil price crisis - which was comparable in impact to the 2008/9 world economic crisis - can safely be dismissed.
https://twitter.com/dreamweasel/status/1198972438442389504
And I see you are still ignoring my point.
https://www.dailymail.co.uk/news/article-8811631/Rich-world-close-normal-late-2021-vaccine-works-Bill-Gates.html
If we are lucky next Christmas won't be cancelled.
There has to be a number of industries who can't withstand 12+ months more of this.
“There was a flat party a few days ago which had a policy that you could only get in if you were positive. It was like their health-and-safety measure,” the 18-year-old physics student said.
https://www.theguardian.com/education/2020/oct/06/manchester-students-organising-covid-positive-parties
https://www.theguardian.com/world/2020/oct/06/packed-funeral-for-ultra-orthdox-rabbi-inflames-covid-tensions-in-israel
Nobody is going to be surprised if Rudy Giuliani tests positive in the next few days.
https://twitter.com/DanaBashCNN/status/1313614960900141057?s=20
This should also inform our take on his attack on postal voting, which another case of the classic dilemma where you're not sure whether he has some cunning, evil plan to get the Dem votes in a different box to the GOP votes and disqualify them in court, or whether he's an immense ignorant doofus who's left his unlucky party way behind on GOTV, and having to persuade voters to needlessly put their lives at risk by voting in person.
(Extend the graph to 25 years)
Worth remembering that the methodology has been adjusted since last time to take into account HS education - plus of course there are much fewer undecideds this time around. Last time undecideds broke to Trump late on, but if Biden is over 50% in PA then it doesn't matter how many undecideds break to Trump there this time around.
In Michigan, Florida and Arizona without at least 1 of which Biden cannot win the EC he is also still under 50% on average
Very impressive.
Imagine there are 100 voters in 2016, 30 Registered Dems, 40 Registered Independents, and 30 Registered Republicans. And imagine in 2016, independents broke 55% R / 45% D in the General.
In this case, the election would be 52/48.
Now imagine that it moved to 30/30/40, but that the remaining independents broke 2-1 for the Democrats. In which case, despite the change in registration numbers, it'd be 50%-50% in the General.
Biden is polling higher which not just means he's polling more than Trump won with last time, not just has he got a bigger gap in the polls, but there's a smaller pond for Trump to fish in to attract voters.
The lack of undecideds is one of the big stories of this election.
https://twitter.com/PatinkinMandy/status/1313441423962890242
https://twitter.com/kylegriffin1/status/1313625101833650176?s=21
https://twitter.com/realdonaldtrump/status/1313640512025513984?s=21
Returned ballots:
R 267,021 (28.5% of all)
D 491,562 (52.5% of all)
All 936,729
https://docs.google.com/spreadsheets/d/17YKazYaUMZ33vmd4YHDKcVdlTkP5DmnHQQlAttwOzL0/edit#gid=1837208868
2016 turnout was about 9.5 million.