Re: COVID-19
Posted: Tue Oct 06, 2020 3:35 pm
14,500
*amends shopping order to include wine*
*amends shopping order to include wine*
It is that bad- local authority numbers just for Nottingham from the gov website are 222, 342 and 342 for 30th Sep, 1st and 2nd October respectively and the other boroughs add even more to those.
You're right. The numbers in Nottingham are greater than those in Nottinghamshire.
Different councils. There's the City council and then the County council which is made up of several smaller local authorities.
You're right. Nottingham is not included in Nottinghamshire . wikipedia says:mediocrity511 wrote: ↑Tue Oct 06, 2020 5:27 pmDifferent councils. There's the City council and then the County council which is made up of several smaller local authorities.
The council does not have jurisdiction over Nottingham, which is a unitary authority governed by Nottingham City Council.
Same for Derbyshire and Derby. Although there seem to be some links - presumably as it made sense for the councils to coordinateKAJ wrote: ↑Tue Oct 06, 2020 5:43 pmYou're right. Nottingham is not included in Nottinghamshire . wikipedia says:mediocrity511 wrote: ↑Tue Oct 06, 2020 5:27 pmDifferent councils. There's the City council and then the County council which is made up of several smaller local authorities.The council does not have jurisdiction over Nottingham, which is a unitary authority governed by Nottingham City Council.
Ah those heady days of R=1.1, seems so long ago. There has been some very poor statistical analysis around thislpm wrote: ↑Thu Oct 01, 2020 8:27 amYep. Many of these things have been truly terrible. There was the one that claimed cases had halved when confirmed actuals had doubled, another had claimed for weeks that R=1 when cases were rising week after week.PeteB wrote: ↑Thu Oct 01, 2020 7:48 amhttps://www.bbc.co.uk/news/health-54366478
https://twitter.com/jamesannan/status/1 ... 1037154304
In this case it's perfectly obvious R was never 1.7, nationally.
R=1.7 corresponds to approx 10% growth per day, a doubling in about 7 days. R=1.3 corresponds to approx 5% per day, around 1.4x in a week, doubling in about 14 days.
James Annan's twitter has a very simple chart that plots each day's cases as a multiple of the number a week previously. If cases were doubling every week, then the plots would be clustering around the 2.0x mark. If cases were static at R=1 then they would be cluster at 1.0x.
A brief eyeballing of the chart shows each day tends to average at 1.4x the previous week's figure, with the usual statistical noise around this. Most noticeably, it never reached the 2.0x level except for a single outlier. The coloured lines show it's never diverged from the 1.4x per week, 5% per day, approx R=1.3.
I wish the media would stop wanging on about the R number. It's hard to calculate, and meaningless in real time because it describes how many people an infected person goes on to infect in total, regardless of the time period. Obviously other measures also have limitations, but over the span of a week or so the doubling rate of cases is easy to calculate and probably less unreliable than most other measures.
Exactly. What's important is what the rate of rise/fall is. With the current levels of infection, exponentials are still adequate for anything beyond complex models - where R itself is pretty poor as we know that superspreading events are important and most people *don't* infect others.sTeamTraen wrote: ↑Wed Oct 07, 2020 9:13 amI wish the media would stop wanging on about the R number. It's hard to calculate, and meaningless in real time because it describes how many people an infected person goes on to infect in total, regardless of the time period. Obviously other measures also have limitations, but over the span of a week or so the doubling rate of cases is easy to calculate and probably less unreliable than most other measures.
We're not comparing different viruses so R is good. There's no change to the infectious period of SARS-CoV-2.jimbob wrote: ↑Wed Oct 07, 2020 9:55 amExactly. What's important is what the rate of rise/fall is. With the current levels of infection, exponentials are still adequate for anything beyond complex models - where R itself is pretty poor as we know that superspreading events are important and most people *don't* infect others.sTeamTraen wrote: ↑Wed Oct 07, 2020 9:13 amI wish the media would stop wanging on about the R number. It's hard to calculate, and meaningless in real time because it describes how many people an infected person goes on to infect in total, regardless of the time period. Obviously other measures also have limitations, but over the span of a week or so the doubling rate of cases is easy to calculate and probably less unreliable than most other measures.
And an R of 1.5 but an infectious period of a week is worse than an R of 5 but an infectious period of 6-months.
There was a decent piece in the Atlantic about this recently. I've been more concerned about the fact that testing (in the UK) isn't really adequate to measure it accurately, but they add the point that R is essentially a mean being presented without any measure of dispersion. Annoyingly they kind of frame it around a different one-variable-to-rule-them-all instead of a more sensible nuanced approach.sTeamTraen wrote: ↑Wed Oct 07, 2020 9:13 amI wish the media would stop wanging on about the R number. It's hard to calculate, and meaningless in real time because it describes how many people an infected person goes on to infect in total, regardless of the time period. Obviously other measures also have limitations, but over the span of a week or so the doubling rate of cases is easy to calculate and probably less unreliable than most other measures.
I suppose it is easier to say "We want R to be 0.7" rather than "we want cases to be falling at 30% over 3-weeks" or whatever that equates tolpm wrote: ↑Wed Oct 07, 2020 10:04 amWe're not comparing different viruses so R is good. There's no change to the infectious period of SARS-CoV-2.jimbob wrote: ↑Wed Oct 07, 2020 9:55 amExactly. What's important is what the rate of rise/fall is. With the current levels of infection, exponentials are still adequate for anything beyond complex models - where R itself is pretty poor as we know that superspreading events are important and most people *don't* infect others.sTeamTraen wrote: ↑Wed Oct 07, 2020 9:13 am
I wish the media would stop wanging on about the R number. It's hard to calculate, and meaningless in real time because it describes how many people an infected person goes on to infect in total, regardless of the time period. Obviously other measures also have limitations, but over the span of a week or so the doubling rate of cases is easy to calculate and probably less unreliable than most other measures.
And an R of 1.5 but an infectious period of a week is worse than an R of 5 but an infectious period of 6-months.
R shows how badly we are adrift of R<1. All the details of complex calculations of R don't matter compared to the fact that current lockdown measures are nowhere near enough to get R<1. If the figure is 1.4, 1.5, who cares, we need it 0.7.
Personally I'm shocked that closing pubs at 10 pm didn't do the trick. Utterly shocked and bemused.
Over 3 weeks or so and with a decent percentage of infected people being tested, the logarithm of the case multiplication rate is going to be close to R anyway. I think they are only using R because that's the standard virology/epidemiology measure, which is useful for comparing between pathogens. It reminds me a bit of the use of BMI as a diagnostic criterion for individuals.
You basically know if R < 1 or R > 1 by whether cases or falling or rising; to get the R from the exponential time constant requires you make an assumption about how long a subject is infectious for. But talking in terms of doubling time or whatever is better because it avoids that assumption.sTeamTraen wrote: ↑Wed Oct 07, 2020 12:19 pmOver 3 weeks or so and with a decent percentage of infected people being tested, the logarithm of the case multiplication rate is going to be close to R anyway. I think they are only using R because that's the standard virology/epidemiology measure, which is useful for comparing between pathogens. It reminds me a bit of the use of BMI as a diagnostic criterion for individuals.
I will not have this divisive language, you do-gooder lawyer activist!
What about drakes?Bird on a Fire wrote: ↑Thu Oct 08, 2020 2:20 amI will not have this divisive language, you do-gooder lawyer activist!
As an ornithologist, I feel professionallly obliged to point out that, while many birds may be bally - for instance, the Hazel Grouse, Maribou Stork or Magnificent Frigatebird, focussing merely on the neck area - few if any birds can be accurately described as dicky (myself notwithstanding).
One more...Bird on a Fire wrote: ↑Thu Oct 08, 2020 2:20 amfew if any birds can be accurately described as dicky (myself notwithstanding).
Fun fact: Dickie Bird was actually trained as an elephant proctologist.bob sterman wrote: ↑Thu Oct 08, 2020 8:04 amOne more...Bird on a Fire wrote: ↑Thu Oct 08, 2020 2:20 amfew if any birds can be accurately described as dicky (myself notwithstanding).
Exactly - talking about R gives the illusion of knowing more than we do about this epidemic. And it's not going to be useful in a proper model as we know that superspreading events are highly significant.shpalman wrote: ↑Wed Oct 07, 2020 4:12 pmYou basically know if R < 1 or R > 1 by whether cases or falling or rising; to get the R from the exponential time constant requires you make an assumption about how long a subject is infectious for. But talking in terms of doubling time or whatever is better because it avoids that assumption.sTeamTraen wrote: ↑Wed Oct 07, 2020 12:19 pmOver 3 weeks or so and with a decent percentage of infected people being tested, the logarithm of the case multiplication rate is going to be close to R anyway. I think they are only using R because that's the standard virology/epidemiology measure, which is useful for comparing between pathogens. It reminds me a bit of the use of BMI as a diagnostic criterion for individuals.