The OSA isn't something that is applied in certain circumstances, it is something that is. Every scrap of an officer's dealings on every matter both high and low is subject to OSA the minute they sign on the dotted line. Whether or not actions are taken and becomes public knowledge is a different beast. In my line of work it was made very clear that I wasn't to discuss any aspect with anybody at any time, and this is how officers within the ONS will be operating. This is also why they have a sanctioned spokesperson.
I have a question and I am not sure I will say it right so bear with me. The last chart showing the missing vaxxed and unvaxxed is in absolute terms and it clearly shows the missing people. But, given this, shouldn't it be possible to go back and "adjust" the vaccine effectiveness calculations that the ONS did because now you know how to adjust the population numbers. What I am trying to get at is the vaccine effectiveness is % of people with success, and now you have produced data showing how to adjust the numbers so the %s are "correct". Thus, shouldn't it be possible to at least say, this is how much their calculations are biased/wrong on a month to month basis?
Partly possible but given we don't know their age we can't recalculate ASMR. Then there's all the sudden deaths that were excluded from both registers, along with all those folk not registered with a GP.
That's true. Bear with me while I say something that might be non-sensical: The ONS used counts of vaxxed populations and unvaxxed populations to calculate VE. Without knowing anything but that they were using their own data, shouldn't it be possible to calculate, at the population level (all ages), how much their incorrect ratios would have skewed their results. I realize that some other people have done back of the envelope calculations for this, but it seems like you have a better estimate of how much they must be off in terms of population counts. I have no idea if there is something interesting to see here or not, but it seems to me it would be helpful to understand how the miscounts change their VE calculations.
In addition, it seems like one could approximate the age bands VE because you do know from actual data how many people are in the age band. The problem of course, is at this fine grain, you don't know who they were including in each age band.
Also, I saw your reply to Rob Kay about why they have a sanctioned spokesperson. How does one go about asking the sanctioned spokesperson about the obvious problems with the data?
Dr Clare Craig is digging into this along with those other data warriors I mentioned, so there is every chance one of them will produce corrected figures, but we must realise this is only in reference to NIMS, which in itself is likely biased and incomplete in some way. I have a couple of folk looking more closely at NIMS data capture.
There is something screwy about the ONS use of ASMR that I suspect Profs Fenton & Neil will get to the bottom of. There's no need for age standardisation when analysing discrete age bands and I suspect they're hiding something. To this end I may well start out by deriving MR so we may get to see what it is that they're trying to hide.
It is important to realise that the officer responsible for producing the datafile isn't necessarily the same officer responsible for its release, and the officer responsible for its release is not necessarily the same officer responsible for engaging with the public. That's how it worked back when I served. Neither should we assume that these officers share information at the same level. By that I mean a need-to-know often operates as a safety mechanism. That partitioned mechanism also applies further up in that Ministers never get told what they would perhaps need to confess later.
Sarah Caul is the spokesperson in this instance and several data warriors are already obtaining responses. Whilst this is handy we have to realise Sarah is not the big cheese, merely a slice, and there will be a great deal she is not privvy to. If she is privvy to the 'hot stuff' then she will be duty bound not to reveal anything or, indeed, that she even knows what she knows. Her selection for the post she holds will not have been accidental since she will have displayed qualities suitable at lower grades. In short she'll be good at keeping her mouth shut as well as smart.
The TV series 'Yes Minister' and 'Yes Prime Minister' were uncannily accurate and I'd meet with colleagues in Westminster for a beer and fish 'n' chip lunch in a nearby pub to all roar with laughter at the games we'd then go back to play!
I finally got around to watching the John Campbell -- Norman Fenton discussion of the new ONS dataset. There is good information there, just as you said at the beginning of this post.
If you treat the ONS data as a biased sample you can have fun providing you then don't attempt to use results to generalise. NIMS is derived from GP registrations so is more robust but still not perfect since not everyone is registered to a GP and GPs will vary in their fastidiousness to record keeping. Neither are what we might expect given the importance of this matter.
Many thanks for helping us understand this. Q. I’ve only just noticed this release and previous refer only to England, so what’s the situation in other parts of the country.
JD, what do you think of this US Medicare data sent by a whistle-blower to Steve Kirsch? Kirsch says Clare Craig and Norman Fenton can't find fault with his analysis and it looks pretty damning for the vaccines to my non-statistician's eye.
Thanks for listing all of the other good analyses and discussions in one place!
Not so much a neat tidy traybake as a veritable pigs breakfast of official obfuscation and error. I nearly choked on it! Luckily I am not bound by the Official Secrets Act, but I think it is unlikely that it would ever be applied in this kind of context: Ministers know what happened when they started to bully people with it - it backfired. https://www.theguardian.com/world/2021/nov/27/inside-story-of-peter-wrights-spycatcher-blocked-by-cabinet-office-delay-and-deception
The OSA isn't something that is applied in certain circumstances, it is something that is. Every scrap of an officer's dealings on every matter both high and low is subject to OSA the minute they sign on the dotted line. Whether or not actions are taken and becomes public knowledge is a different beast. In my line of work it was made very clear that I wasn't to discuss any aspect with anybody at any time, and this is how officers within the ONS will be operating. This is also why they have a sanctioned spokesperson.
Super interesting!
I have a question and I am not sure I will say it right so bear with me. The last chart showing the missing vaxxed and unvaxxed is in absolute terms and it clearly shows the missing people. But, given this, shouldn't it be possible to go back and "adjust" the vaccine effectiveness calculations that the ONS did because now you know how to adjust the population numbers. What I am trying to get at is the vaccine effectiveness is % of people with success, and now you have produced data showing how to adjust the numbers so the %s are "correct". Thus, shouldn't it be possible to at least say, this is how much their calculations are biased/wrong on a month to month basis?
Partly possible but given we don't know their age we can't recalculate ASMR. Then there's all the sudden deaths that were excluded from both registers, along with all those folk not registered with a GP.
That's true. Bear with me while I say something that might be non-sensical: The ONS used counts of vaxxed populations and unvaxxed populations to calculate VE. Without knowing anything but that they were using their own data, shouldn't it be possible to calculate, at the population level (all ages), how much their incorrect ratios would have skewed their results. I realize that some other people have done back of the envelope calculations for this, but it seems like you have a better estimate of how much they must be off in terms of population counts. I have no idea if there is something interesting to see here or not, but it seems to me it would be helpful to understand how the miscounts change their VE calculations.
In addition, it seems like one could approximate the age bands VE because you do know from actual data how many people are in the age band. The problem of course, is at this fine grain, you don't know who they were including in each age band.
Also, I saw your reply to Rob Kay about why they have a sanctioned spokesperson. How does one go about asking the sanctioned spokesperson about the obvious problems with the data?
Dr Clare Craig is digging into this along with those other data warriors I mentioned, so there is every chance one of them will produce corrected figures, but we must realise this is only in reference to NIMS, which in itself is likely biased and incomplete in some way. I have a couple of folk looking more closely at NIMS data capture.
There is something screwy about the ONS use of ASMR that I suspect Profs Fenton & Neil will get to the bottom of. There's no need for age standardisation when analysing discrete age bands and I suspect they're hiding something. To this end I may well start out by deriving MR so we may get to see what it is that they're trying to hide.
It is important to realise that the officer responsible for producing the datafile isn't necessarily the same officer responsible for its release, and the officer responsible for its release is not necessarily the same officer responsible for engaging with the public. That's how it worked back when I served. Neither should we assume that these officers share information at the same level. By that I mean a need-to-know often operates as a safety mechanism. That partitioned mechanism also applies further up in that Ministers never get told what they would perhaps need to confess later.
Sarah Caul is the spokesperson in this instance and several data warriors are already obtaining responses. Whilst this is handy we have to realise Sarah is not the big cheese, merely a slice, and there will be a great deal she is not privvy to. If she is privvy to the 'hot stuff' then she will be duty bound not to reveal anything or, indeed, that she even knows what she knows. Her selection for the post she holds will not have been accidental since she will have displayed qualities suitable at lower grades. In short she'll be good at keeping her mouth shut as well as smart.
The TV series 'Yes Minister' and 'Yes Prime Minister' were uncannily accurate and I'd meet with colleagues in Westminster for a beer and fish 'n' chip lunch in a nearby pub to all roar with laughter at the games we'd then go back to play!
I finally got around to watching the John Campbell -- Norman Fenton discussion of the new ONS dataset. There is good information there, just as you said at the beginning of this post.
If the NIMS data is apparently better than ONS, why have people been doing tons of analyses on ONS data but not on NIMS the last three years?
If you treat the ONS data as a biased sample you can have fun providing you then don't attempt to use results to generalise. NIMS is derived from GP registrations so is more robust but still not perfect since not everyone is registered to a GP and GPs will vary in their fastidiousness to record keeping. Neither are what we might expect given the importance of this matter.
Many thanks for helping us understand this. Q. I’ve only just noticed this release and previous refer only to England, so what’s the situation in other parts of the country.
My pleasure! No idea what's happening elsewhere owing to difficulty in obtaining data but I can only presume it is the same story.
JD, what do you think of this US Medicare data sent by a whistle-blower to Steve Kirsch? Kirsch says Clare Craig and Norman Fenton can't find fault with his analysis and it looks pretty damning for the vaccines to my non-statistician's eye.
https://stevekirsch.substack.com/p/game-over-medicare-data-shows-the
Afternoon. Not had time to look at this yet - a bit manic here!
Great idea plotting missing individuals like that. :)
Thank you!