Data and analysis thread, started 12 November

[NoGoodPunsLeft]

Previous thread here:

Data and Analysis Thread, started Oct 29 www.mumsnet.com/Talk/coronavirus/4064113-Data-and-Analysis-Thread-started-Oct-29

Regular lurker but I frequent poster, didn't want to lose the threads.

There’s actually a spreadsheet that I found linked to in the “about the data” section of the dashboard showing every single correction made on a day by day basis broken down by council
www.gov.uk/government/publications/covid-19-comparison-of-geographic-allocation-of-cases-in-england-by-lower-tier-local-authority

Richmond - the correction for October was about 10% of their cases I think?

I'm not sure they've updated the regions correctly yet. According to the daily summary, in my area there were no cases today, yet I know someone who got a positive result this morning.
There is no way they could put them into any other region-they've lived here all their life!

@GetAMoveOnTroodon

There’s Richmond though, right at the top of the table. Top marks to all those on this thread who called it a very long time ago

Love this thread.

I think they have put a zero for any local authority where the net movement is a reduction. For example I can see that in Trafford (which I track closely) there are approx 90 cases that have been added to the figures for the last 3 days, but the cumulative total has gone down by over 200 and so today’s cases are reported as zero.
Another reason to only ever look at the cases by specimen date, not reported date.

So since September 1st
Elmbridge lost 126, kept 1055 cases ~80,000
Richmond lost 212, kept 1859 cases ~100,000

Wiltshire lost 260 kept 2656 cases ~250,000

So 12% cases disappeared from those, big enough to make a difference? Certainly the grounds for the Tier 2 restrictions in Elmbridge but not the adjacent surrey areas is no longer there - Richmond was always the "all London" justification of course.

Another big blow to the trustability of the statistics.

@ceeveebee

There’s actually a spreadsheet that I found linked to in the “about the data” section of the dashboard showing every single correction made on a day by day basis broken down by council www.gov.uk/government/publications/covid-19-comparison-of-geographic-allocation-of-cases-in-england-by-lower-tier-local-authority

Richmond - the correction for October was about 10% of their cases I think?

@boys3 I did a quick calc based on this spreadsheet and my numbers are different from yours - this one is a direct comparison between the two methodologies. My LA now has fewer cases but you have it as having more...?

Richmond shows -240 if you use the 'official' comparison spreadsheet.

For the few LAs I've looked at, any changes only start on 1/9/20.

Just noticed that the official methodologies comparison spreadsheet here
www.gov.uk/government/publications/covid-19-comparison-of-geographic-allocation-of-cases-in-england-by-lower-tier-local-authority

has approx 18,000 cases 'unallocated' to a Local Authority at the bottom

The overall change for 'unallocated' is -724 with the new methodology.

I’m really sorry if I’ve missed anything up thread. I’ve been monitoring the Richmond numbers the past few wks via Covid Messenger. I saw them rocket with the uni debacle and then over the last couple of weeks, from half term, they’ve been decreasing, however over the last 3 days or so they’ve rapidly started rising again, along with places nearby, such as Kingston and Hounslow. Why do you think this would be? Was it people going out on the eve of lockdown? Halloween? Is the lockdown not working? Or are we still working through the lag? Thank you!

@Quarantino

Richmond shows -240 if you use the 'official' comparison spreadsheet.

For the few LAs I've looked at, any changes only start on 1/9/20.

Is the difference just todays “normal” case numbers? Trafford has 292 cases being removed per the spreadsheet, but then has about 90 cases being added (which is the figure that would have been reported today if there hasn’t been this correction) therefore a net reduction in the cumulative cases of 200 (compared to yesterday’s cumulative number)

For the few LAs I've looked at, any changes only start on 1/9/20

The announcement says that's only as far as they adjusted. Should cover all the mass uni testing though I'm sure there will be similar previous issues but I doubt anything like the magnitude or systemic bias as the uni testing. Prison outbreaks possibly though?

.

@ceeveebee - no, the difference is over the whole time period (changes since 1/9/20.
I created a third column of 'difference' (ie for each daily case number) and summed it for each LA.

I meant the difference between the 240 on the official spreadsheet, and the 212 that boys3 shows on the charts above - I think the difference will be that there should have been 28 new cases reported today.
So the movement between yesterday’s cumulative and today’s cumulative total is 240-28 = 212

@Quarantino well just looking the official spreadsheet it has a "before" number of 286 for cases in Richmond for 5-11 Oct; whereas yesterday, based on the published file on the dashboard I had 287, then for the next 7 days numbers match. I have the download file from yesterday so I'll compare it with the adjustment file now published. If I do have a miscalculation in somewhere not beyond the bounds of possibility, excel and sloe gin can have unintended results that should pick it up.

On a wholly separate data note it has been a spectacular sloe season in my part of the country.

@ceeveebee

I meant the difference between the 240 on the official spreadsheet, and the 212 that boys3 shows on the charts above - I think the difference will be that there should have been 28 new cases reported today. So the movement between yesterday’s cumulative and today’s cumulative total is 240-28 = 212

I'll double check in case I was a tad hasty in calculating columns earlier.

Ah, the official comparison spreadsheet only goes up to 14/11.

@Quarantino - I can think of a few reasons why cases might not be readily allocated, depending on the detail of the method:

• Tourist who has no UK address except hotel
• Foreign lorry driver who has no UK address
• Military travelling around UK
• Traveller (gypsy) of no fixed location
• Prison with no fixed abode
• Address has no valid postcode
• Untraced test, original user cannot be identified
• Asylum seeker
• Person taken to hospital unconscious or died, or who won't reveal address, so no known address attached
• Double-counting in a soon to be revealed error
• Cases from the last day or two who haven't yet been contacted by T&T and whose details are incomplete
• Homeless if they want to assign to MSOA not just LA

(Or something random like tests we agreed to do for Name-a-country).

I think numbers dropped because of half term. And are now rising again. Preston dropped 2000 on Zoe last few weeks....this week up again by 1000.

Why do you say numbers are rising, when the relative increase in case numbers has reduced?

I’ve just done the comparison for Ealing, it looks like overall 99 cases have been reallocated out, but it’s quite odd in that Sept to c.24 Oct, cases are being allocated out, but after that date more are being allocated in. Does anyone have any insight into why this might be? I’m wondering if it’s students returning for Halloween/reading weeks? The drama college?

I think University of Surrey has a larger number than perhaps you'd expect of home students, or even if they live out, students who live near so may not change GP, which may explain Guildford

Maybe, not sure, but they do have a relatively high percentage, 20% of international students.

It would have been interesting if PHE could have produced some sort of reallocation map for the council areas with some of the biggest changes.

So if Nottingham had a net addition of 958 in a specific week where did all these cases get re-allocated from.

Whilst the Before and After file that PHE have made available through the dashboard is undoubtedly useful they could have considered

• a net change column, rather than everyone doing it for themselves, and,

• showing both inflow and also any outflow of cases

Musing a bit further a simple matrix could have shown the movement - overall, or by month or week - between LAs. Or at least by region as a starter for ten.

Another thread posts this article, which raises doubts about safety of the new vaccines, could someone knowledgable critique it?

www.google.co.uk/amp/s/m.jpost.com/health-science/could-an-mrna-vaccine-be-dangerous-in-the-long-term-649253/amp?espv=1

(I've no idea whether The Jerusalem Post has an agenda.)