Data & analysis thread, started 1 December

[NoGoodPunsLeft]

New thread!

Link to previous:

Data and analysis thread, started 12 November www.mumsnet.com/Talk/coronavirus/4077794-data-and-analysis-thread-started-12-november

@boys3 thank you. That sounds better, the graph was freaking me out!

Is anyone still watching the Zoe app?

I have found that every time an area drops into the 1000-2000 zone, instead of changing it to a lighter shade of pink, it changes to grey/not enough data.

Today there are quite a few in East Anglia, southern England, and some others dotted around Cornwall, Northern England and Scotland.

[quote PrayingandHoping]@TheCountessofFitzdotterel yes it is just England. But remember to look at rolling average not day to day.... the rolling average is down. You can't make assessments based on just comparing 2 days.

Agreed it is too early for a levelling off.
Hope not

My council posted a massive number yesterday though... same as the day before lockdown 🤦🏼‍♀️🤦🏼‍♀️[/quote]
Comparing with that day the previous week, eg a Friday with the previous Friday, tells you how the 7 day rolling average has moved.
I wouldn’t bother commenting based on one day’s rolling average but a few days of the rolling average levelling off looks like something is happening.
That’s not the same as comparing from day to day which I think everyone knows by this stage is pointless as in addition to the usual noise there are clear weekly patterns, particularly in death numbers, due to reporting differences on different days of the week.

@TheCountessofFitzdotterel comparing one Friday to the following Friday is one view but not the most complete. The more reliable way is the 7 day rolling average.

The line on that graph is the 7 day rolling average. Not a day to day comparison

I don’t think you understand my point about the rolling average. Each day the total number that you divide by 7 to get the rolling average will lose one day’s data and gain a new lot. So today it will lose the number from last Saturday and gain the number from today. Thus the difference between those two numbers will tell you what the rolling average is doing.

I do understand about comparing the 7 day rolling average. But that is down to a decent degree so I'm not sure why you're concerned? That is exactly what the line in the graph I posted is showing

Whereas the number of cases reported from Friday last week to Friday this week is closer, but that is not a useful comparison

But comparing cases/deaths/hospitalisations on one day to the same day the previous week is useful - and gives a sense of where the 7 day rolling average may be going, for the reasons the Countess has outlined.

To a point... @IrenetheQuaint over a period of time rather than a one off. They have been quoted on here before, but a months worth. That is more useful.

But comparing the 7 day rolling is more reliable.

My husband often quotes off the Daily Fail who compare day to day figures and in isolation they don't have a lot of meaning.

Comparing Friday with the Friday before is simply another way of saying ‘ah, I see the 7 day rolling average is up/down today’.

Prayingandhoping, you keep telling me I should be looking at the rolling average. Talking about the rolling average is exactly what I am doing!

@TheCountessofFitzdotterel it was clear initially

I get u now are but from the graph I posted the line is the 7 day rolling average and it is going down not levelling off so I don't understand what you're seeing?

The actual numbers on one day week to the same day following week do get posted (not the rolling average) on here and described as such

Here’s the Worldometer graph massively expanded so you can see the last few days more clearly. The graph you posted just shows the little tick for one day of beginning to level off but because yours is specimen date and the data isn’t complete the rolling average line doesn’t go all the way to the end so you can’t see from it what the last few days have been doing.
Basically the rolling average was plummeting but over the last few days it has levelled off. This might be a blip or it might not. Fingers crossed it will go back to the nice falls that we were seeing a week ago.

I might be misunderstanding, but given we have just come out of lockdown and are going into Christmas (when there will be some sort socialising), wouldn't you expect falls in numbers, corresponding to lockdown then progressive increase until 23-27 December, Then there will be a massive increase as people get tested in early January, and lots of sad stories in the papers about ill relatives following family get togethers. Unless of course there is a press embargo on such stories, and only reporting of vaccine uptake is permitted.

@herecomesthsun that pretty much sums up the likely way things will play out. If Christmas was a once a century event I could understand the easing, but given it happens annually the plan as it stands will hinder rather than help. I genuinely hope I’m wrong on that, but the pattern of covid thus far suggests the contrary.

I agree @Countess - the numbers just in the last few days have looked more like levelling than continuing to fall fast, but it could yet be just a blip. You'd expect more downward trend as a result of lockdown for another week or so yet. Though when you look at different areas it does look at if the lockdown worked well in some and not at all in others (non compliance?) So I'm not sure we will suddenly start seeing drops in those areas.

There is also a bit of a hope that people who are seeing family at Christmas will be particularly careful to avoid catching Covid for a week or two beforehand, and of course schools will be closed for two weeks, and many workers taking a break. Though I doubt that will be sufficient to counterbalance the effects of family socialising.

dashboard seems to be in the process of updating.

43.000 lateral flow tests showing so those numbers really started to increase now - presumably Uni test numbers starting to feed through.

Based on today's reporting looking at England the rate of decline has slowed. 7 day moving average at 1st Dec based on specimen dates, 12001; down from a peak of 22176 on 13th November. 12001 is roughly the same as the 7 day average to 7th October. 12001 is still though c 5 times higher than than on 8th September.

Regional graphs would look different again.

We still have the effect of students going home to factor into the December figures. Even those being offered tests, which we know isn’t all, are not all taking them up.

Have we had today's figures?

15, 539 new cases
1444 admissions
397 deaths reported on the BBC website

@boys3

Is their a glitch on CM? Redcar & Cleveland has been on 198 7 day average for 3 /4 days now.

[quote AbstractDot]@boys3

Is their a glitch on CM? Redcar & Cleveland has been on 198 7 day average for 3 /4 days now.[/quote]
@AbstractDot Its correct just a quirk of the maths. This is cases and the moving seven day total. Still a big fall from the peak seven day figure on 12th November

If Christmas was a once a century event I could understand the easing, but given it happens annually the plan as it stands will hinder rather than help.

Yes, Christmas gatherings will add fuel to fire but the problem is people are very likely to go ahead with them irrespective of what the government says / does.

My nanny is the most law abiding and risk averse person I know (I guess she has to be given her profession - she is very good at what she does). She has told me several times she is planning to spend Christmas with her family "even if it means she will get arrested for that". I am pretty sure many people share her thoughts.

The behavioural science shows that if people break the rules once (e.g. meet their friends/family for Christmas when they are not allowed to), they are much less likely to take the rules seriously in the future. As a result, if cases shoot up and there is a need for the third lockdown/fire breaker in January - February, it may not be particularly effective if people "learn" to ignore rules before that.

It is a tricky situation and it is almost certain there will be another "wave/spike" in January. The question is how we minimise the impact. Unfortunately, I do not see how it can be avoided.

With regards to falls slowing down would that be inevitable as it gets lower? For example South hams was recently having falls of around 50% on the seven day rolling average, today it had 5 positive cases posted so won't it get harder to show big drops as the numbers get smaller? I'm not clear if that is how it works, just trying to figure it out.

@ancientgran If everywhere was falling at the same rate then the falls should go on being at the same proportional rate (eg 20% down per week) - obviously the falls would get less in numerical terms.

But I think you might in fact be right that the places that were causing the fall - mainly the North of England, but also some low rate areas like South Hams that went up while in tier 1 during October - have effectively run out of steam and even if they go on falling at the same rate this is now contributing little to the overall fall - increasingly outweighed by parts of Kent, Essex and London in particular which have appeared unaffected by the lockdown and have been rising throughout. I'm not sure what the answer is for these areas - if the 4 week lockdown didn't work, Tier 3 isn't going to either.

Firefliess thanks, that helped clarify it. I'm a bit parochial so tend to follow the South Devon rates which definitely showed great falls during lockdown, I just hope it continues. It is a bit of a mystery why it seems to work in some areas and not others.