EGM covered this phenomenon about a year ago, though you each worked it a little differently. They refer to the t<=14 period as the "running to the bunker" period and illustrate how starkly the situation is misrepresented when deaths in this interval are lumped into the "foxhole" condition when it was the "ordered to run from the foxhole to the bunker" condition that made the deaths possible in the first place.
The numbers have made it very clear for a while now that the 14-day window is MUCH higher risk with a MUCH more vulnerable immune system than the states before or after.
This probably relates to the elimination of "old guard" antigens while antigenic fixation is getting established by the goo.
What no one smarter than me has the data to play with yet is whether each subsequent booster is causing the same roll of the dice or whether it's simply causing direct harm through myocarditis & friends.
Yes - I didn't mean to imply this was some 'new' insight - lots of people have commented on this before now. It's clear it's a problem. What hasn't been clear (to me) is just how big this problem might be, so I've tried in this article (and the previous ones I've done on the topic) to figure out a simple alternative analysis approach that would help me to characterize the size of the effect.
I read Joel's article back in January but, at the time, didn't twig that it could be used to help with this kind of analysis. It is only a simple re-parametrization using this g factor but it actually helps quite a bit - especially because we have some real-world data to use from Alberta.
My approach here is really about analysing *principles* rather than real-world covid data - the idea here being to show, in a simple way, that this 14 day vaccine 'limbo' period thing is not sensible and can have a very significant effect on apparent efficacy. I think I've done that - at least to my satisfaction.
What I didn't expect to find was that using "diddle 2" you get this dependency on percentage vaxxed. That's disturbing - but I do need to double check my working. It seems sound in that the final formula passes several checks - but that's not a guarantee I haven't slipped up somewhere.
As for 2 doses and subsequent boosters, I don't think my simple approach to figure out the 'principles' is going to be productive, it's a much more complicated time-series, but maybe there's a way to do it.
It's difficult because so much datacrime has been done with these numbers that even the raw stuff, if you can get it, will yield things that don't make sense.
In either case, the short answer is, there has probably always been data crime around vaccine efficacy and side effects. As regards the specific window of heightened immunodeficiency, that's unique to the mRNA vaccines.
EGM has written a bunch about this with good statistical analysis. I'd go there for more in-depth answers to your question.
I should add that I sometimes try to use a simplified, less mathy example to show people that the datacrime you're describing works even if both the control group and the Goo group get saline solution. Then suppose that half of all adverse events occur in the first two weeks (which is not unreasonable if I understand the data at openvaers correctly). Finally, suppose that the Goo group is considered un-Goo'd during those first two weeks. Then a quarter of all adverse events get moved to control group, so that they now have 3/4 of all adverse events, and the Goo'd group has 1/4, so the Goo looks super-safe.
EGM covered this phenomenon about a year ago, though you each worked it a little differently. They refer to the t<=14 period as the "running to the bunker" period and illustrate how starkly the situation is misrepresented when deaths in this interval are lumped into the "foxhole" condition when it was the "ordered to run from the foxhole to the bunker" condition that made the deaths possible in the first place.
The numbers have made it very clear for a while now that the 14-day window is MUCH higher risk with a MUCH more vulnerable immune system than the states before or after.
This probably relates to the elimination of "old guard" antigens while antigenic fixation is getting established by the goo.
What no one smarter than me has the data to play with yet is whether each subsequent booster is causing the same roll of the dice or whether it's simply causing direct harm through myocarditis & friends.
Yes - I didn't mean to imply this was some 'new' insight - lots of people have commented on this before now. It's clear it's a problem. What hasn't been clear (to me) is just how big this problem might be, so I've tried in this article (and the previous ones I've done on the topic) to figure out a simple alternative analysis approach that would help me to characterize the size of the effect.
I read Joel's article back in January but, at the time, didn't twig that it could be used to help with this kind of analysis. It is only a simple re-parametrization using this g factor but it actually helps quite a bit - especially because we have some real-world data to use from Alberta.
My approach here is really about analysing *principles* rather than real-world covid data - the idea here being to show, in a simple way, that this 14 day vaccine 'limbo' period thing is not sensible and can have a very significant effect on apparent efficacy. I think I've done that - at least to my satisfaction.
What I didn't expect to find was that using "diddle 2" you get this dependency on percentage vaxxed. That's disturbing - but I do need to double check my working. It seems sound in that the final formula passes several checks - but that's not a guarantee I haven't slipped up somewhere.
As for 2 doses and subsequent boosters, I don't think my simple approach to figure out the 'principles' is going to be productive, it's a much more complicated time-series, but maybe there's a way to do it.
It's difficult because so much datacrime has been done with these numbers that even the raw stuff, if you can get it, will yield things that don't make sense.
Genuine question: that 14 days, is it something that has always been there in regards to all vaxxines or it applies only to this one?
Which 14 days: the vulnerability window or the data shenanigans that count T0-14 as "unvaxxed"?
That's what I was after: "As in regards..." Thanks for that G.
I don't remember the whole thing with 14 days being of import as an element of vaxination process before, that's all. But then I don't think
I've paid it much attention as I do now, hence the question. Thanks!
In either case, the short answer is, there has probably always been data crime around vaccine efficacy and side effects. As regards the specific window of heightened immunodeficiency, that's unique to the mRNA vaccines.
EGM has written a bunch about this with good statistical analysis. I'd go there for more in-depth answers to your question.
I should add that I sometimes try to use a simplified, less mathy example to show people that the datacrime you're describing works even if both the control group and the Goo group get saline solution. Then suppose that half of all adverse events occur in the first two weeks (which is not unreasonable if I understand the data at openvaers correctly). Finally, suppose that the Goo group is considered un-Goo'd during those first two weeks. Then a quarter of all adverse events get moved to control group, so that they now have 3/4 of all adverse events, and the Goo'd group has 1/4, so the Goo looks super-safe.
Here is El Gato Malo's article about this:
https://boriquagato.substack.com/p/bayesian-datacrime-defining-vaccine