So, coronavirus...

[Biffer]

If there are false positive rates at 1%, how were we getting less than 1% of tests recorded as positive earlier in the summer?

[Bimbowomxn]

If there are false positive rates at 1%, how were we getting less than 1% of tests recorded as positive earlier in the summer?

https://www.oxfordreference.com/view/10 ... 3095449924

[Carter's Choice]

Doesn't sound like its been a great few days in the UK regarding Covid-19 infections. Good luck chaps, I hope your govt is able to make the right decisions, and lead you through this challenging period

[Ali Cadoo]

Doesn't sound like its been a great few days in the UK regarding Covid-19 infections. Good luck chaps, I hope your govt is able to make the right decisions, and lead you through this challenging period

This one is YOURS you jinxing cunt.

[Biffer]

If there are false positive rates at 1%, how were we getting less than 1% of tests recorded as positive earlier in the summer?

https://www.oxfordreference.com/view/10 ... 3095449924

Thought I would look at your post since you replied directly to me.

Unsurprisingly, you’re still talking shitebag. Base rate fallacy is of no relevance here. Different area of statistical analysis completely.

Expect no furthereplies, you you fascist mithering cunt.

[Jb1981]

Doesn't sound like its been a great few days in the UK regarding Covid-19 infections. Good luck chaps, I hope your govt is able to make the right decisions, and lead you through this challenging period

This one is YOURS you jinxing cunt.

This isn’t good. Hopefully there aren’t too many additional cases from close contacts not knowing to self-isolate

PHE said all of the cases "received their Covid-19 test result as normal".

But it means their close contacts were not approached to self-isolate, BBC Health Editor Hugh Pym says.

[Ted.]

I’m not looking for an exact number....

Just a rough idea.


More than or less than 1/2?

Is Bimbo really saying that half of all test results are false positives?

My post that he has taken exception to was in response to his implication that Homer's graph was explained by the results being "mostly false positives".

If most (ie. more than half) of the testing results are unreliable, what is the point of testing at all?

Well, that’s the question that’s being asked about the sheer scale and where testing is aimed in the UK. Test and trace is correct as a policy, testing for its own sake to say “we’ve made record tests” makes absolutely no sense at all.

The delay in reporting numbers seen this week show that folly even more clearly.

That this is dictating policy is much worse than folly.

To be balanced, Bimbo, an oxymoron if ever there was one, you need to also consider the mass of false negatives.

[Ted.]

If there are false positive rates at 1%, how were we getting less than 1% of tests recorded as positive earlier in the summer?

https://www.oxfordreference.com/view/10 ... 3095449924

Thought I would look at your post since you replied directly to me.

Unsurprisingly, you’re still talking shitebag. Base rate fallacy is of no relevance here. Different area of statistical analysis completely.

Expect no furthereplies, you you fascist mithering cunt.

[Rinkals]

Is Bimbo really saying that half of all test results are false positives?

My post that he has taken exception to was in response to his implication that Homer's graph was explained by the results being "mostly false positives".

If most (ie. more than half) of the testing results are unreliable, what is the point of testing at all?

Well, that’s the question that’s being asked about the sheer scale and where testing is aimed in the UK. Test and trace is correct as a policy, testing for its own sake to say “we’ve made record tests” makes absolutely no sense at all.

The delay in reporting numbers seen this week show that folly even more clearly.

That this is dictating policy is much worse than folly.

To be balanced, Bimbo, an oxymoron if ever there was one, you need to also consider the mass of false negatives.

Which would mean that only a tiny fraction of test results are reliable, surely?

I'm perfectly prepared to believe that those in charge are horribly incompetent, but implementing a testing regime where the results are more likely than not to be misleading seems counter productive and a waste of time and money.

[Rinkals]

https://www.oxfordreference.com/view/10 ... 3095449924

Thought I would look at your post since you replied directly to me.

Unsurprisingly, you’re still talking shitebag. Base rate fallacy is of no relevance here. Different area of statistical analysis completely.

Expect no furthereplies, you you fascist mithering cunt.

Is that not Bimbo's objective?

To get to the point where nobody confronts his misinformation because they either have him on ignore or they disdain from engaging with him because it inevitably descends into arguing about minor details, insults and questionable sources.

The problem with "peer-review" is that so many people feel that being educated at the University of YouTube and Facebook is enough for them to consider themselves fully qualified.

[Bimbowomxn]

If there are false positive rates at 1%, how were we getting less than 1% of tests recorded as positive earlier in the summer?

https://www.oxfordreference.com/view/10 ... 3095449924

Thought I would look at your post since you replied directly to me.

Unsurprisingly, you’re still talking shitebag. Base rate fallacy is of no relevance here. Different area of statistical analysis completely.

Expect no furthereplies, you you fascist mithering cunt.

Some one is a facist now for posting actual facts regarding the false positive rate.


Base fallacy rate calculations are central to false readings on this type of screening.

Read Carl Heneghans work in July and gain some understanding, or shout “maths is facist”instead.

[Bimbowomxn]

To be balanced, Bimbo, an oxymoron if ever there was one, you need to also consider the mass of false negatives.

Understanding something about the base fallacy rate would of course let you understand that this isn’t the case.

[Bimbowomxn]

The absolute determination of people on here to not understand the issue and the numbers of a false positive rate is stunning.

The blinkers people put in their lives is .


Do some reading .

[fishfoodie]

Doesn't sound like its been a great few days in the UK regarding Covid-19 infections. Good luck chaps, I hope your govt is able to make the right decisions, and lead you through this challenging period

This one is YOURS you jinxing cunt.

World Class !

[MungoMan]

Doesn't sound like its been a great few days in the UK regarding Covid-19 infections. Good luck chaps, I hope your govt is able to make the right decisions, and lead you through this challenging period

This one is YOURS you jinxing cunt.

The Mocker Gods smiled upon AC/CC, lightly tapped his shoulder and briefly lent him a portion of their powers.

[dpedin]

Doesn't sound like its been a great few days in the UK regarding Covid-19 infections. Good luck chaps, I hope your govt is able to make the right decisions, and lead you through this challenging period

This one is YOURS you jinxing cunt.

This isn’t good. Hopefully there aren’t too many additional cases from close contacts not knowing to self-isolate

PHE said all of the cases "received their Covid-19 test result as normal".

But it means their close contacts were not approached to self-isolate, BBC Health Editor Hugh Pym says.

Test and trace really is a clusterfuck that the Gov need to own!

Getting fed up with folk looking at bits of the whole system. We need to look at the whole process and the outcomes it delivers against what it is should be delivering. Timely access to reliable tests when required, quick turnaround of results, transfer of positive results into an efficient track and trace system, speedy tracing of contacts and contacts able and willing to then self isolate to avoid onward transmission of the virus is what is supposed to happen. In each and every step of this process this Gov has fecked up from lack of access to testing to insufficient lab capacity to poor messaging and support to folk who should be self isolating. Every measure has been missed and to keep talking about capacity as opposed to actual throughputs and performance of the whole system is just plain misleading - I have a bank account with enough capacity to hold a lottery win! Capacity is only as good as the worst performing element of the whole process and it aint anything like they are claiming. Getting one bit of this whole process wrong, such as 'losing' 16,000+ test results and not feeding them into the track and trace system is just plain criminal and reduces the performance of the whole system even more.

To be blunt test and trace is not working and it is the only thing we have that will allow us to open up society and the economy, unless we accept the consequences of letting the virus rip through the whole of society. To have failed in delivering a performing whole system test and trace system, despite the huge amounts of money spent on their private sector ex Uni/Bullington Club buddies, has been criminal and for this alone this Gov will go down as the worst in living memory.

[Bimbowomxn]

Dido Harding losing data .....


Either way it’s possibly designed to support larger numbers where required to justify the “circuit breaker” lock down that’s planned .

[Tichtheid]

Dido Harding losing data .....


Either way it’s possibly designed to support larger numbers where required to justify the “circuit breaker” lock down that’s planned .

Just out of interest, why would the government introduce lockdown if it wasn't necessary?

[fishfoodie]

Dido Harding losing data .....


Either way it’s possibly designed to support larger numbers where required to justify the “circuit breaker” lock down that’s planned .

Just out of interest, why would the government introduce lockdown if it wasn't necessary?

Please stop engaging with this fuckwit

[Mahoney]

Unless you are testing at random you cannot know.

This is not true, I realise. If we assume:
1) No false negatives
2) People are only tested once
3) We know for certain the false positive rate is 1%

then the maths to work out how many actual cases & false positives there are is fairly simple (see below).

Unfortunately it can't be applied to the data on https://coronavirus.data.gov.uk:
* there is a false negative rate, unaccounted for in my maths
* https://coronavirus.data.gov.uk/testing explicitly states there have been multiple tests for the same person
* the data on https://coronavirus.data.gov.uk/cases gives case by specimen date, which is what we want but the data at https://coronavirus.data.gov.uk/testing isn't by specimen date.

I guess if you knew the false negative rate, and you knew how many individuals were tested as well as the total number of tests done, and you had that data by specimen date, and you knew what they considered a case* then you could work it out, but the maths gets a lot more complicated & I haven't worked it out yet.

* presumably the same person testing positive twice is still one case; presumably if the same person tests positive and negative they are also a case?

let x = total people tested
let y = total people tested positive
let z = total people who actually have the lurgy

false positives = y - z
false positives = (x - z) / 100
(x - z) / 100 = y - z
x - z = 100y - 100z
x + 99z = 100y
z = (100y - x) / 99

And therefore the percentage chance that a positive is or is not false is:
(10,000y - 100x) / 99y

So for instance if you test 1,000 people and get 100 positives then 91 of them will be real and the odds of any one of the 100 actually having it will be 91%.

[Tichtheid]

The whole "inflated numbers of false positives" thing is based around a statistical hypothetical, Heneghan does this in his Spectator article, yeah with an infection rate running at about 1 in 1000 and a false positive rate at around 0.8% you'd expect to get 1 real positive and 8 false positive results from ANY GIVEN RANDOM 1000 PEOPLE.

That is not who is being tested, though, the vast majority of tests are conducted on people who are either symptomatic or have a high risk due to contact with people who are symptomatic. It's this heightened probability that leads to higher true positives, statistically as well as in real life.

I can't keep saying the same things, so that's it on false positives from me.

[Slick]

Very much looking forward to the replies to these last 2 posts.

[Mahoney]

The whole "inflated numbers of false positives" thing is based around a statistical hypothetical, Heneghan does this in his Spectator article, yeah with an infection rate running at about 1 in 1000 and a false positive rate at around 0.8% you'd expect to get 1 real positive and 8 false positive results from ANY GIVEN RANDOM 1000 PEOPLE.

That is not who is being tested, though, the vast majority of tests are conducted on people who are either symptomatic or have a high risk due to contact with people who are symptomatic. It's this heightened probability that leads to higher true positives, statistically as well as in real life.

I can't keep saying the same things, so that's it on false positives from me.

Yes; random testing is primarily about measuring the prevalence in the community, and importantly the change in prevalence over time, not about answering the question "does this individual have the virus".

Targeted testing is more about the individual's odds of having the virus.

Ideally you'd put the two types of testing in different buckets.

I think my maths applies in both cases, though obviously the inadequacies I mentioned render it relatively useless.

[Sandstorm]

Very much looking forward to the replies to these last 2 posts.

Please don't quote any replies, I'm blissfully unaware of any Russian Bots on this Bored if no-one quotes.

[SaintK]

Dido Harding losing data .....


Either way it’s possibly designed to support larger numbers where required to justify the “circuit breaker” lock down that’s planned .

Really?
It's just another shambolic, incompetent outcome of the test, track and trace programme that's costing £billions
PHE already being hung out to dry by the government as they try and shift blame.
Harding in hiding again and taking no responsibilty and Hancock putting his fingers in his hears and shouting lalalalala!

Matt Hancock just spoke at @Policy_Exchange Tory virtual fringe event. Chair said, "All questions welcome". He then ignored the many posted questions about the test and trace fiasco, including the one 'most recommended' by viewers. How odd. All felt very... cosy.
11:02 AM · Oct 5, 2020

[Bimbowomxn]

This is not true, I realise. If we assume:
1) No false negatives
2) People are only tested once
3) We know for certain the false positive rate is 1%

We can start with those assumptions.....


False negatives is a separate issue regarding the “whole” .

I wholly endorse retesting every “positive” that would solve the issue. There’s no evidence this is happening.

They made estimates of 0.5% - 1%..... in the work in July.

I’m glad you know better than Heneghan, I’m please.

[Ted.]

Well, that’s the question that’s being asked about the sheer scale and where testing is aimed in the UK. Test and trace is correct as a policy, testing for its own sake to say “we’ve made record tests” makes absolutely no sense at all.

The delay in reporting numbers seen this week show that folly even more clearly.

That this is dictating policy is much worse than folly.

To be balanced, Bimbo, an oxymoron if ever there was one, you need to also consider the mass of false negatives.

Which would mean that only a tiny fraction of test results are reliable, surely?

I'm perfectly prepared to believe that those in charge are horribly incompetent, but implementing a testing regime where the results are more likely than not to be misleading seems counter productive and a waste of time and money.

Oh, and Bimbo too.

[Lobby]

Dido Harding losing data .....


Either way it’s possibly designed to support larger numbers where required to justify the “circuit breaker” lock down that’s planned .

Really?
It's just another shambolic, incompetent outcome of the test, track and trace programme that's costing £billions
PHE already being hung out to dry by the government as they try and shift blame.
Harding in hiding again and taking no responsibilty and Hancock putting his fingers in his hears and shouting lalalalala!

Matt Hancock just spoke at @Policy_Exchange Tory virtual fringe event. Chair said, "All questions welcome". He then ignored the many posted questions about the test and trace fiasco, including the one 'most recommended' by viewers. How odd. All felt very... cosy.
11:02 AM · Oct 5, 2020

Despite the £billiions being spent on the 'world class' TTT system, apparently this latest problem is because the whole system is managed on a single Excel spreadsheet, which failed to update once its maximum column width had been reached.

[Bimbowomxn]

The whole "inflated numbers of false positives" thing is based around a statistical hypothetical, Heneghan does this in his Spectator article, yeah with an infection rate running at about 1 in 1000 and a false positive rate at around 0.8% you'd expect to get 1 real positive and 8 false positive results from ANY GIVEN RANDOM 1000 PEOPLE.

That is not who is being tested, though, the vast majority of tests are conducted on people who are either symptomatic or have a high risk due to contact with people who are symptomatic. It's this heightened probability that leads to higher true positives, statistically as well as in real life.

I can't keep saying the same things, so that's it on false positives from me.

Let’s be clear Heneghan did much more work than present an Article, his work in July actually changed official data from PHE on death measurement much of it was related to this maths.

Where’s the evidence that the “vast majority” of people being tested are symptomatic? But yes I agree the probability is less than random due to test and trace etc. However looking at the current trace scenarios I don’t hold out much hope.

Consider the simple facts that these “symptomatics” being tested are overwhelmingly presenting as negative.

[Bimbowomxn]

Dido Harding losing data .....


Either way it’s possibly designed to support larger numbers where required to justify the “circuit breaker” lock down that’s planned .

Really?
It's just another shambolic, incompetent outcome of the test, track and trace programme that's costing £billions
PHE already being hung out to dry by the government as they try and shift blame.
Harding in hiding again and taking no responsibilty and Hancock putting his fingers in his hears and shouting lalalalala!

Matt Hancock just spoke at @Policy_Exchange Tory virtual fringe event. Chair said, "All questions welcome". He then ignored the many posted questions about the test and trace fiasco, including the one 'most recommended' by viewers. How odd. All felt very... cosy.
11:02 AM · Oct 5, 2020

PHE wasn’t capable of any test and trace when the virus was at its peak, they refused help from the private sector for weeks.

Harding is a disgrace as pointed out.

Hancock has gone genuinely insane......

Let be clear though the “public “ health systems are f ucking crap and full of duds.

[Bimbowomxn]

Despite the £billiions being spent on the 'world class' TTT system, apparently this latest problem is because the whole system is managed on a single Excel spreadsheet, which failed to update once its maximum column width had been reached.

Please say that ain’t so.....

[Tichtheid]

The whole "inflated numbers of false positives" thing is based around a statistical hypothetical, Heneghan does this in his Spectator article, yeah with an infection rate running at about 1 in 1000 and a false positive rate at around 0.8% you'd expect to get 1 real positive and 8 false positive results from ANY GIVEN RANDOM 1000 PEOPLE.

That is not who is being tested, though, the vast majority of tests are conducted on people who are either symptomatic or have a high risk due to contact with people who are symptomatic. It's this heightened probability that leads to higher true positives, statistically as well as in real life.

I can't keep saying the same things, so that's it on false positives from me.

Let’s be clear Heneghan did much more work than present an Article, his work in July actually changed official data from PHE on death measurement much of it was related to this maths.

Where’s the evidence that the “vast majority” of people being tested are symptomatic? But yes I agree the probability is less than random due to test and trace etc. However looking at the current trace scenarios I don’t hold out much hope.

Consider the simple facts that these “symptomatics” being tested are overwhelmingly presenting as negative.

The majority of community testing is done at drive through centres, on people who go there because of symptoms. PHE said that 7% of tests were coming back positive in late September, so yes the vast majority are coming back negative but the important stat is the increase in numbers testing positive.

Taking 7% positive you'd have 70 for every 1000 tests and 8 expected as false positive, this is best-case scenario and ignores false negatives.

This was all in links yesterday.

[Bimbowomxn]

The whole "inflated numbers of false positives" thing is based around a statistical hypothetical, Heneghan does this in his Spectator article, yeah with an infection rate running at about 1 in 1000 and a false positive rate at around 0.8% you'd expect to get 1 real positive and 8 false positive results from ANY GIVEN RANDOM 1000 PEOPLE.

That is not who is being tested, though, the vast majority of tests are conducted on people who are either symptomatic or have a high risk due to contact with people who are symptomatic. It's this heightened probability that leads to higher true positives, statistically as well as in real life.

I can't keep saying the same things, so that's it on false positives from me.

Let’s be clear Heneghan did much more work than present an Article, his work in July actually changed official data from PHE on death measurement much of it was related to this maths.

Where’s the evidence that the “vast majority” of people being tested are symptomatic? But yes I agree the probability is less than random due to test and trace etc. However looking at the current trace scenarios I don’t hold out much hope.

Consider the simple facts that these “symptomatics” being tested are overwhelmingly presenting as negative.

The majority of community testing is done at drive through centres, on people who go there because of symptoms. PHE said that 7% of tests were coming back positive in late September, so yes the vast majority are coming back negative but the important stat is the increase in numbers testing positive.

Taking 7% positive you'd have 70 for every 1000 tests and 8 expected as false positive, this is best-case scenario and ignores false negatives.

This was all in links yesterday.

It’s disagreed with by Heneghan and others work. The Links yesterday were from a time before community testing.

[Tichtheid]

It’s disagreed with by Heneghan and others work. The Links yesterday were from a time before community testing.

The link explaining community testing and citing the 7% positive return on testing from PHE was from a BBC article two days ago, it quoted Dr Paul Birrell, "a statistician at the Medical Research Council's Biostatistics Unit at the University of Cambridge", it was him I paraphrased.

What seems to be the problem here is a misunderstanding, the tests are still the same, with the same degree of accuracy or inaccuracy. It doesn't matter that the BMJ paper was published in May, it is still relevant and is being used by researchers all over the world, indeed Heneghan used it for his article.

I'm not going to go over all of this again.

What you are doing is actually quite dangerous, someone might believe that a positive result is meaningless if they took what you are saying as even close to being accurate.

[Mahoney]

FWIW in the ONS data I'm looking at, between 3-16 July they tested 32,827 individuals and had 17 positive. Naively that suggests to me the false positive rate cannot be higher than 0.05%, because if the false positive rate were 0.05% and no-one had COVID then you'd expect 16-17 positives in 32,827. Sheet 1d, row 12:
https://www.ons.gov.uk/peoplepopulation ... surveydata

Sticking with naive calculations, and taking the false positive rate at 0.05%, sticking their most recent numbers of 245 positive out of 100,038 (11-24 Sept, 1d row 17) into my maths gives 195 actual positives, 50 false positives and therefore an 80% chance that anyone with a positive test actually has the disease. And that's with, as I understand it, random testing rather than targeted testing.

(This is just playing, thinking aloud really; I haven't fully understood the methodologies used by the ONS, and all the caveats in my original post apply. And I may have got the maths wrong.)

[dpedin]

Dido Harding losing data .....


Either way it’s possibly designed to support larger numbers where required to justify the “circuit breaker” lock down that’s planned .

Really?
It's just another shambolic, incompetent outcome of the test, track and trace programme that's costing £billions
PHE already being hung out to dry by the government as they try and shift blame.
Harding in hiding again and taking no responsibilty and Hancock putting his fingers in his hears and shouting lalalalala!

Matt Hancock just spoke at @Policy_Exchange Tory virtual fringe event. Chair said, "All questions welcome". He then ignored the many posted questions about the test and trace fiasco, including the one 'most recommended' by viewers. How odd. All felt very... cosy.
11:02 AM · Oct 5, 2020

PHE wasn’t capable of any test and trace when the virus was at its peak, they refused help from the private sector for weeks.

Harding is a disgrace as pointed out.

Hancock has gone genuinely insane......

Let be clear though the “public “ health systems are f ucking crap and full of duds.

PHE is directly accountable to Hancock, if he wanted them to get help from the private sector he would have told them to do so. They are not an independent body but Executive Agency of the Dept of Health and Social Care. Lets be clear that they do what they are told by Hancock and the Dept. They are not funded to deal with major pandemics. The PHE Framework agreement states:

The Secretary of State is accountable to Parliament for the health system (its “steward”), including PHE. This involves:

 setting national priorities and monitoring the whole system’s performance to ensure it delivers what patients, people who use services and the wider public need and value most;
 determining the level of resource allocations across the health system as a whole;
 setting objectives for PHE through the annual remit letter;
 supporting the integrity of the system by ensuring that funding, legislation and
accountability arrangements protect the best interests of patients, the public
and the taxpayer; and
 accounting to Parliament for PHE’s performance and the effectiveness of the
health and care system overall.

[MungoMan]

Unless you are testing at random you cannot know.

This is not true, I realise. If we assume:
1) No false negatives
2) People are only tested once
3) We know for certain the false positive rate is 1%

then the maths to work out how many actual cases & false positives there are is fairly simple (see below).

Unfortunately it can't be applied to the data on https://coronavirus.data.gov.uk:
* there is a false negative rate, unaccounted for in my maths
* https://coronavirus.data.gov.uk/testing explicitly states there have been multiple tests for the same person
* the data on https://coronavirus.data.gov.uk/cases gives case by specimen date, which is what we want but the data at https://coronavirus.data.gov.uk/testing isn't by specimen date.

I guess if you knew the false negative rate, and you knew how many individuals were tested as well as the total number of tests done, and you had that data by specimen date, and you knew what they considered a case* then you could work it out, but the maths gets a lot more complicated & I haven't worked it out yet.

* presumably the same person testing positive twice is still one case; presumably if the same person tests positive and negative they are also a case?

The website’s Cases page makes it clear that multiple tests on one person = one case, assuming a test is positive.

[Bimbowomxn]

You won’t find me defending Hancock.



I wonder who has the million pound spreadsheets.

[C69]

Doesn't sound like its been a great few days in the UK regarding Covid-19 infections. Good luck chaps, I hope your govt is able to make the right decisions, and lead you through this challenging period

This one is YOURS you jinxing cunt.

World Class !

Useless cnuts

[Tichtheid]

FWIW in the ONS data I'm looking at, between 3-16 July they tested 32,827 individuals and had 17 positive. Naively that suggests to me the false positive rate cannot be higher than 0.05%, because if the false positive rate were 0.05% and no-one had COVID then you'd expect 16-17 positives in 32,827. Sheet 1d, row 12:
https://www.ons.gov.uk/peoplepopulation ... surveydata

Sticking with naive calculations, and taking the false positive rate at 0.05%, sticking their most recent numbers of 245 positive out of 100,038 (11-24 Sept, 1d row 17) into my maths gives 195 actual positives, 50 false positives and therefore an 80% chance that anyone with a positive test actually has the disease. And that's with, as I understand it, random testing rather than targeted testing.

(This is just playing, thinking aloud really; I haven't fully understood the methodologies used by the ONS, and all the caveats in my original post apply. And I may have got the maths wrong.)

Even if all 245 in mid September were false positives it brings about a FPR of 0.2%, a variable rate of false positives on a range of 0.05 to 0.2% with the same tests?

...and that is taking every single positive result as false.

"Unlikely", is one word for it.