COVID-19

Covid-19 discussion, bring your own statistics
User avatar
shpalman
Light of Blast
Posts: 4912
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: COVID-19

Post by shpalman » Wed Jul 14, 2021 4:46 am

lpm wrote:
Tue Jul 13, 2021 9:23 pm
I really don't like the "cases/hospitalizations/deaths have reached their highest level since March" construct.

The correct comparison is with the upslope, not the downslope.
Especially when it ignores day-of-the-week effects i.e. Monday was a low number, Tuesday's number includes the catch-up. The 7-day average deaths-per-day (30) is similar to the ~23rd of September (on the way up). It's not unreasonable to expect ~100 deaths per day three weeks from now.
molto tricky

User avatar
lpm
Stummy Beige
Posts: 3263
Joined: Mon Nov 11, 2019 1:05 pm

Re: COVID-19

Post by lpm » Wed Jul 14, 2021 6:46 am

Herainestold wrote:
Tue Jul 13, 2021 11:33 pm
lpm wrote:
Tue Jul 13, 2021 9:23 pm
I really don't like the "cases/hospitalizations/deaths have reached their highest level since March" construct.

The correct comparison is with the upslope, not the downslope.

It's like saying "flood waters have risen to where they had fallen to a few days after the big floods of 2008".

Lazy journalists.
What they should say is its at the same level as Dec 20 in the second wave. It will double at least one more time again and may peak in August (or September). And ...
Across all scenarios explored, the team estimated that lifting restrictions on 19th July in the context of Delta could lead to a significant but highly uncertain third wave of hospitalisations and death with the total number of deaths ranging from 9,400 (estimate range: 4,600 to 19,800) in the most optimistic, to 115,800 (estimate range: 81,700 to 143,600) deaths in the most pessimistic scenario.
It will be nasty. Very nasty

https://medicalxpress.com/news/2021-07- ... lysis.html
That's Imperial.

They lost credibility long ago.

As far as I know they've never done a post mortem on their previous failures, they just merrily carry on.
What ever happened to that Trump guy, you know, the one who was president for a bit?

tom p
Dorkwood
Posts: 1090
Joined: Mon Nov 11, 2019 1:14 pm
Location: the low countries

Re: COVID-19

Post by tom p » Wed Jul 14, 2021 8:13 am

Millennie Al wrote:
Sat Jul 10, 2021 11:53 pm
tom p wrote:
Fri Jul 09, 2021 7:50 am
Herainestold wrote:
Fri Jul 09, 2021 1:20 am


So you are in favour of Big Tech censorship?
Everyone should be in the case of misinformation that can harm the public health.
In fact, I think big tech should be legally obliged to censor harmful content. Freedom of speech does not extend to the right to should "Fire!" in a crowded theatre. that's an old principle and one which should always be remembered.
Censorship does not mean that people are only allowed to speak the truth: it means they are only allowed to say what those in power want them to say. Censorship is a great friend of lies and an enemy of truth.
Hahahahaha.
The free for all of the last decade has been a *great* friend of the truth.
Ahahahahaha. tw.t.

hakwright
Bank Butt
Posts: 22
Joined: Mon Jan 20, 2020 12:58 pm

Re: COVID-19

Post by hakwright » Wed Jul 14, 2021 12:21 pm

shpalman wrote:
Wed Jul 14, 2021 4:46 am
lpm wrote:
Tue Jul 13, 2021 9:23 pm
I really don't like the "cases/hospitalizations/deaths have reached their highest level since March" construct.

The correct comparison is with the upslope, not the downslope.
Especially when it ignores day-of-the-week effects i.e. Monday was a low number, Tuesday's number includes the catch-up. The 7-day average deaths-per-day (30) is similar to the ~23rd of September (on the way up). It's not unreasonable to expect ~100 deaths per day three weeks from now.
It's really quite incredible that, after all this time, much of the reporting of covid stats in the media displays an appalling lack of understanding of what actually matters.

Comparing "today's figure" for any stat is absurd, as commented above. If you're not quoting some rolling average, you should consider resigning as a journalist.

Using a previous low-point as a reference to today's/this week's "high figure" is also a bit mad. If current values are high, compare with previous peaks.

Also, there's very little reference to the gradients, which are as important (if not more important) than the actual numbers. Many of the stats for new daily cases in Europe are still relatively low. But they are increasing at a rate much higher than all previous waves. This seems significant! A couple of weeks ago, you would have been correct to talk about new daily cases in Spain being at roughly their lowest for around 12 months. But because the rate of increase has been so scary, they just recorded their highest ever new daily figure, and the 7-day trend continues to shoot up at a near-vertical rate. It's this sharp gradient that's important.

In terms of the UK stats, we know new daily cases are rising fast. The most significant change that ought to be reported in the coming weeks is when this upwards gradient starts to reduce, i.e. at a point we are approaching the peak.

Howard

P.S. Strictly speaking, since the new daily cases values that are the basis of many stats are in themselves deltas, i.e. increases compared to the previous day, these are essentially already first derivatives. In other words, x new cases today is really giving you the current value of the *gradient* of total cases. These new daily cases values are, in themselves, a "rate of increase" or gradient value.

User avatar
bob sterman
Snowbonk
Posts: 580
Joined: Mon Nov 11, 2019 10:25 pm

Re: COVID-19

Post by bob sterman » Wed Jul 14, 2021 12:39 pm

hakwright wrote:
Wed Jul 14, 2021 12:21 pm
In terms of the UK stats, we know new daily cases are rising fast. The most significant change that ought to be reported in the coming weeks is when this upwards gradient starts to reduce, i.e. at a point we are approaching the peak.
I think James Annan's "weekly change in daily cases" figures are good for eyeballing this - just ratio of new cases reported on a day / new cases reported 7 days previously.

This is a few days ago - July 11th...

https://twitter.com/jamesannan/status/1 ... 5425837061
E6CXC9_WYAgHHzs.png
E6CXC9_WYAgHHzs.png (33.54 KiB) Viewed 472 times

User avatar
lpm
Stummy Beige
Posts: 3263
Joined: Mon Nov 11, 2019 1:05 pm

Re: COVID-19

Post by lpm » Wed Jul 14, 2021 1:00 pm

It's a good chart.

It's not clear yet whether the natural pace at our current unlockdown level is 1.7x a week, with some periods when this growth is suppressed for unclear reasons. Or whether it's 1.3x, with some periods when growth is very much higher for unclear reasons.

We're at another Weds figure. 1.3x would be 42,000. 1.7x would be 55,000.
What ever happened to that Trump guy, you know, the one who was president for a bit?

User avatar
Woodchopper
Stummy Beige
Posts: 3670
Joined: Sat Oct 12, 2019 9:05 am

Re: COVID-19

Post by Woodchopper » Wed Jul 14, 2021 1:57 pm

lpm wrote:
Wed Jul 14, 2021 1:00 pm
It's a good chart.

It's not clear yet whether the natural pace at our current unlockdown level is 1.7x a week, with some periods when this growth is suppressed for unclear reasons. Or whether it's 1.3x, with some periods when growth is very much higher for unclear reasons.

We're at another Weds figure. 1.3x would be 42,000. 1.7x would be 55,000.
Could be 1.3 plus football. Seriously, there is a marked gender gap whereby more young men have been infected than young women. Could be due to the former being more likely to gather to watch matches in pubs and private parties.

User avatar
lpm
Stummy Beige
Posts: 3263
Joined: Mon Nov 11, 2019 1:05 pm

Re: COVID-19

Post by lpm » Wed Jul 14, 2021 2:15 pm

I did wonder about football but do the timings fit? The first England match was 13 June - there was a cluster of 1.7x before that date.

I suppose the timing of matches doesn't help - catch it on 13 June from someone else in the pub for the Croatia game, infectious but asymptomatic at a party for the Scotland match on 18 June, the people you spread it to are then infectious for Germany on 29 June and they infect people on 7 July during the Denmark game.

The worst 1.7x zone is 23 June to 4 July in the James Annan chart above.

Not sure why the Germany match wouldn't have keep things hot - a week after that game in we're back in the 1.3x zone.
What ever happened to that Trump guy, you know, the one who was president for a bit?

User avatar
jimbob
Stummy Beige
Posts: 2719
Joined: Mon Nov 11, 2019 4:04 pm
Location: High Peak/Manchester

Re: COVID-19

Post by jimbob » Wed Jul 14, 2021 3:11 pm

lpm wrote:
Wed Jul 14, 2021 2:15 pm
I did wonder about football but do the timings fit? The first England match was 13 June - there was a cluster of 1.7x before that date.

I suppose the timing of matches doesn't help - catch it on 13 June from someone else in the pub for the Croatia game, infectious but asymptomatic at a party for the Scotland match on 18 June, the people you spread it to are then infectious for Germany on 29 June and they infect people on 7 July during the Denmark game.

The worst 1.7x zone is 23 June to 4 July in the James Annan chart above.

Not sure why the Germany match wouldn't have keep things hot - a week after that game in we're back in the 1.3x zone.
Good thread on it here:

https://twitter.com/VictimOfMaths/statu ... 39490?s=20

and this
https://twitter.com/VictimOfMaths/statu ... 9203309575
What's happened in Scotland, with a huge gender gap appearing and then closing again, aligns suspiciously well with Scotland's exit from the Euros.
Have you considered stupidity as an explanation

User avatar
shpalman
Light of Blast
Posts: 4912
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: COVID-19

Post by shpalman » Wed Jul 14, 2021 4:34 pm

lpm wrote:
Wed Jul 14, 2021 1:00 pm
It's a good chart.

It's not clear yet whether the natural pace at our current unlockdown level is 1.7x a week, with some periods when this growth is suppressed for unclear reasons. Or whether it's 1.3x, with some periods when growth is very much higher for unclear reasons.

We're at another Weds figure. 1.3x would be 42,000. 1.7x would be 55,000.
"UK figures for 14 July 2021: 42,302 new cases"
molto tricky

User avatar
lpm
Stummy Beige
Posts: 3263
Joined: Mon Nov 11, 2019 1:05 pm

Re: COVID-19

Post by lpm » Wed Jul 14, 2021 4:35 pm

jimbob wrote:
Wed Jul 14, 2021 3:11 pm
lpm wrote:
Wed Jul 14, 2021 2:15 pm
I did wonder about football but do the timings fit? The first England match was 13 June - there was a cluster of 1.7x before that date.

I suppose the timing of matches doesn't help - catch it on 13 June from someone else in the pub for the Croatia game, infectious but asymptomatic at a party for the Scotland match on 18 June, the people you spread it to are then infectious for Germany on 29 June and they infect people on 7 July during the Denmark game.

The worst 1.7x zone is 23 June to 4 July in the James Annan chart above.

Not sure why the Germany match wouldn't have keep things hot - a week after that game in we're back in the 1.3x zone.
Good thread on it here:

https://twitter.com/VictimOfMaths/statu ... 39490?s=20

and this
https://twitter.com/VictimOfMaths/statu ... 9203309575
What's happened in Scotland, with a huge gender gap appearing and then closing again, aligns suspiciously well with Scotland's exit from the Euros.
That is a good thread - but doesn't really answer why all the curves bend a bit or even stop while the tournament was still going on. For example 20-24 men peaks and actually falls back a bit. There's a slowdown in the exponential on all the little charts at the time when you'd expect the Germany match to fire them up a bit.
What ever happened to that Trump guy, you know, the one who was president for a bit?

User avatar
Woodchopper
Stummy Beige
Posts: 3670
Joined: Sat Oct 12, 2019 9:05 am

Re: COVID-19

Post by Woodchopper » Wed Jul 14, 2021 4:47 pm

lpm, everything is though very messy. Football might be an underlying factor which encourages more infections among men. However there are a lot of other things going on at the same time and testing is imperfect. So you may be overthinking if you're looking for curves in the graph to correlate with specific matches.

User avatar
lpm
Stummy Beige
Posts: 3263
Joined: Mon Nov 11, 2019 1:05 pm

Re: COVID-19

Post by lpm » Wed Jul 14, 2021 5:13 pm

Yes, but equally it is far too easy to come up with a good story and pick bits out of the mess to support it.

An easier story might simply be that men go to the pub slightly more than women do.

Or a slightly more complicated story might be that the age/social mix of pubs has changed from the 2019 norm - extra young men going to pub because other entertainment is not available, fewer women and older men going due to lingering desire to lockdown.
What ever happened to that Trump guy, you know, the one who was president for a bit?

User avatar
lpm
Stummy Beige
Posts: 3263
Joined: Mon Nov 11, 2019 1:05 pm

Re: COVID-19

Post by lpm » Wed Jul 14, 2021 7:22 pm

jimbob wrote:
Wed Jul 14, 2021 3:11 pm
lpm wrote:
Wed Jul 14, 2021 2:15 pm
I did wonder about football but do the timings fit? The first England match was 13 June - there was a cluster of 1.7x before that date.

I suppose the timing of matches doesn't help - catch it on 13 June from someone else in the pub for the Croatia game, infectious but asymptomatic at a party for the Scotland match on 18 June, the people you spread it to are then infectious for Germany on 29 June and they infect people on 7 July during the Denmark game.

The worst 1.7x zone is 23 June to 4 July in the James Annan chart above.

Not sure why the Germany match wouldn't have keep things hot - a week after that game in we're back in the 1.3x zone.
Good thread on it here:

https://twitter.com/VictimOfMaths/statu ... 39490?s=20

and this
https://twitter.com/VictimOfMaths/statu ... 9203309575
What's happened in Scotland, with a huge gender gap appearing and then closing again, aligns suspiciously well with Scotland's exit from the Euros.
Looking again at this twitter thread, on a big screen instead of a phone, and I'm not at all happy. I've been had.

There are times when you don't need to label the x or y axis properly because it looks pretty without. But at other times it's essential. If the VictimOfMaths guy had presented this to me at work I'd have immediately sent him or her off to label properly - plus, crucially, shove in some vertical lines to show the relevant football dates. I'm having to hold up a ruler against my screen which automatically gives this guy an F for fail.

The tournament began on 11 June. If men crowded into pubs for Italy vs Turkey then their cases would start to diverge vs women a week or two later. Covid takes a few days to kick in and be a positive test. But the 25-29 age group's divergence in cases between men and women on the little charts clearly begins too early - in fact around 11 June itself. The first England and Scotland matches are 13 and 14 June. By the time cases plausibly could have diverged a week later on 20 and 21 June there's already a long established difference between men and women. For example, the Scotland chart for 20-24 year olds shows the movement is far too early.

Likewise the story that Scotland crashed out and men stopped gathering to watch and so the gender gap disappears doesn't work. Scotland went out on 22 June. The peak divergence should therefore have been about 29 June, narrowing thereafter. And yet for 20-24 and 25-29 the gap narrows well before. It doesn't "align suspiciously well with Scotland's exit from the Euros", it aligns suspiciously badly with Scotland's exit from the Euros.

Plus all these charts aren't cases on the day, but 7 day rolling averages. The lags are even worse than my assumptions above.

I'm trying to be less cynical these days so let's say it was an accident that he didn't properly label and mark the x-axis, and hence failed to see his entire theory was wrong. But everyone enjoys a few hundred likes and retweets. To be fair, the guy has provided his R code and data.

Image
Image
What ever happened to that Trump guy, you know, the one who was president for a bit?

Millennie Al
Catbabel
Posts: 774
Joined: Mon Mar 16, 2020 4:02 am

Re: COVID-19

Post by Millennie Al » Thu Jul 15, 2021 12:12 am

tom p wrote:
Wed Jul 14, 2021 8:13 am
Millennie Al wrote:
Sat Jul 10, 2021 11:53 pm
Censorship does not mean that people are only allowed to speak the truth: it means they are only allowed to say what those in power want them to say. Censorship is a great friend of lies and an enemy of truth.
Hahahahaha.
The free for all of the last decade has been a *great* friend of the truth.
Ahahahahaha. tw.t.
You are falling for the fallacy that if something has flaws it must be worse than an alternative. In fact an alternative can be much worse. Censorship in favour of the "truth" of Lysenkoism helped cause famine that killed millions.
Covid-19 - Don't catch it: don't spread it.

tom p
Dorkwood
Posts: 1090
Joined: Mon Nov 11, 2019 1:14 pm
Location: the low countries

Re: COVID-19

Post by tom p » Thu Jul 15, 2021 8:15 am

Millennie Al wrote:
Thu Jul 15, 2021 12:12 am
tom p wrote:
Wed Jul 14, 2021 8:13 am
Millennie Al wrote:
Sat Jul 10, 2021 11:53 pm
Censorship does not mean that people are only allowed to speak the truth: it means they are only allowed to say what those in power want them to say. Censorship is a great friend of lies and an enemy of truth.
Hahahahaha.
The free for all of the last decade has been a *great* friend of the truth.
Ahahahahaha. tw.t.
You are falling for the fallacy that if something has flaws it must be worse than an alternative. In fact an alternative can be much worse. Censorship in favour of the "truth" of Lysenkoism helped cause famine that killed millions.
Yes, because there's definitely only one form of restriction on things people can say. FFS

nezumi
Catbabel
Posts: 774
Joined: Mon Sep 14, 2020 1:43 pm
Location: UK

Re: COVID-19

Post by nezumi » Thu Jul 15, 2021 8:20 am

One of the ladies on my team has caught it, she's double vaxxed so it has to be delta. She's still working through it (I'd be in bed!) but she freely admits it's hellish. Says the worst bit is her tastebuds can't decide what to like - one bite tastes of nothing, the next tastes OK, the next is vile. Poor lass.
Non fui. Fui. Non sum. Non curo.

User avatar
jimbob
Stummy Beige
Posts: 2719
Joined: Mon Nov 11, 2019 4:04 pm
Location: High Peak/Manchester

Re: COVID-19

Post by jimbob » Thu Jul 15, 2021 8:55 am

Bird on a Fire wrote:
Sun Feb 21, 2021 7:18 pm
So to come back to the question of the importance of data errors...

With "large" numbers of cases, the effect of natural variation is relatively small compared with the ongoing trend. Once the numbers are "small", that variation is a considerable proportion of expected change.

I suspect that measurement errors will show the opposite pattern. While there are lots of cases, systems and staff are overwhelmed, there's no time to check anything, and you'll get a comparatively large amount of mess. Once the case numbers are smaller and more manageable, there's time to double-check all the data for every covid patient in the hospital, to run samples twice, possibly even trace contacts and stuff.

I also suspect that as the numbers decline, the proportion of cases that actually get detected and end up in the data will increase for the same sorts of reasons. I have no clue how strong this effect would have to be for the decline to be detectably different from exponential, but it would be a fun thing to add to the model ;)
Really late to this. But the proxy measurement is probably the positivity rate.
Have you considered stupidity as an explanation

User avatar
Woodchopper
Stummy Beige
Posts: 3670
Joined: Sat Oct 12, 2019 9:05 am

Re: COVID-19

Post by Woodchopper » Thu Jul 15, 2021 12:28 pm

Short thread on how vaccination maybe isn’t going to stop transmission: https://twitter.com/adamjkucharski/stat ... 83713?s=21

User avatar
lpm
Stummy Beige
Posts: 3263
Joined: Mon Nov 11, 2019 1:05 pm

Re: COVID-19

Post by lpm » Thu Jul 15, 2021 12:43 pm

I find threads like that baffling.

Of course we can't stop transmission. Of course Covid is now endemic. Of course cases will come along every day for the rest of our lives. Of course cases will routinely build into waves. Of course there will be a permanent level of hospital beds needed for Covid patients and we need to devote more of our resources on health. Of course there will be thousands or tens of thousands of UK deaths a year on average. And - the big one - of course Covid is now a likely cause of our own deaths when we're very elderly assuming we dodge cancer etc.

It's like people still haven't realised global pandemics are really bad.
What ever happened to that Trump guy, you know, the one who was president for a bit?

User avatar
shpalman
Light of Blast
Posts: 4912
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: COVID-19

Post by shpalman » Thu Jul 15, 2021 6:30 pm

Of course we're going to have to keep locking down when it turns out that things got out of control again.
molto tricky

User avatar
Woodchopper
Stummy Beige
Posts: 3670
Joined: Sat Oct 12, 2019 9:05 am

Re: COVID-19

Post by Woodchopper » Thu Jul 15, 2021 8:17 pm

lpm wrote:
Thu Jul 15, 2021 12:43 pm
I find threads like that baffling.

Of course we can't stop transmission. Of course Covid is now endemic. Of course cases will come along every day for the rest of our lives. Of course cases will routinely build into waves. Of course there will be a permanent level of hospital beds needed for Covid patients and we need to devote more of our resources on health. Of course there will be thousands or tens of thousands of UK deaths a year on average. And - the big one - of course Covid is now a likely cause of our own deaths when we're very elderly assuming we dodge cancer etc.

It's like people still haven't realised global pandemics are really bad.
You know that, I know that and the writer of the tweet knows that.

However very large numbers of people still seem to assume that:

Vaccination + infection -> herd immunity -> everything goes back to how it was in 2019.

There’s been very little talk of what long term Covid will be like.

Herainestold
Dorkwood
Posts: 1159
Joined: Mon Nov 25, 2019 1:23 pm

Re: COVID-19

Post by Herainestold » Thu Jul 15, 2021 11:39 pm

lpm wrote:
Thu Jul 15, 2021 12:43 pm
I find threads like that baffling.

Of course we can't stop transmission. Of course Covid is now endemic. Of course cases will come along every day for the rest of our lives. Of course cases will routinely build into waves. Of course there will be a permanent level of hospital beds needed for Covid patients and we need to devote more of our resources on health. Of course there will be thousands or tens of thousands of UK deaths a year on average. And - the big one - of course Covid is now a likely cause of our own deaths when we're very elderly assuming we dodge cancer etc.

It's like people still haven't realised global pandemics are really bad.
Covid will not be endemic in countries like Australia, New Zealand and China, that have strong border controls and stringent lockdowns.
The rest of us are going to have to get used to mask wearing, indoors and out, distancing, cancellation of large events, much more limited travel, both internationally and domestically and recurrent seasonal lockdowns.
Delta changes everything.

User avatar
lpm
Stummy Beige
Posts: 3263
Joined: Mon Nov 11, 2019 1:05 pm

Re: COVID-19

Post by lpm » Fri Jul 16, 2021 8:01 am

Hi Herainestold, I should have put a note on this thread, but I started a new thread to talk about the long term implications of endemic Covid:
viewtopic.php?f=19&t=2628

I'll copy and paste your reply over.
What ever happened to that Trump guy, you know, the one who was president for a bit?

User avatar
lpm
Stummy Beige
Posts: 3263
Joined: Mon Nov 11, 2019 1:05 pm

Re: COVID-19

Post by lpm » Fri Jul 16, 2021 1:34 pm

There's someone called Andrew Lilico who is a Telegraph guy and has got himself loads of publicity recently by trolling on twitter.

Here's his latest. I can't make head or tail of it. He's saying R falls automatically if more vaccinated people catch Covid vs unvaccinated people.
https://twitter.com/andrew_lilico/statu ... 0310587400

He's assuming vaccinated people transmit the virus half as well as unvaccinated. And so he says we have "crossed over".

Image
What ever happened to that Trump guy, you know, the one who was president for a bit?

Post Reply