3 ms·

I don't think either of your examples of stats shenanigans apply here. And also: " there's increasing amounts of evidence that they may have some sort of immun

I don't think either of your examples of stats shenanigans apply here. And also: " there's increasing amounts of evidence that they may have some sort of immunosuppressive effect" No, you don't get to say vaccines have an immunosuppressive effect without providing evidence for that. If you have evidence that vaccines have some sort of immunosuppressive effect then you have to provide it, peer reviewed, not 'some guy blogged it'. ... 1) In terms of 'Dead Time Bias' - well, that doesn't apply if we are looking at daily data points of cases from the rates of those who contract COVID among vaxxed/unvaxxed population. There is no 'dead time' artefact by that kind of measure. We're looking at a large group of people, and watching who gets covid and not on a daily basis. If we were looking over larger time periods, it would start to play a factor. But especially for case counts, then all persons who 'get covid' on a particular day have obviously 'survived' to be in the sampled cohort on that day. 2) The links to Prof Fenton's data is interesting, but flawed. a) Particularly for infection and not mortality - rates of the former are very low among the general population for both vaxxed and unvaxxed. The 'distortion effect' due to delayed reporting in infections, is not really going to be a big deal. b) The distortion fades as quickly as the reporting delay. So the effect is null 1 week after vaccination rates stabilise (by his example of 1 week delay). His own 'real world data' comparison show vaccine effectiveness waning over a much longer periods. 3) The 'Datacrime' substack is rubbish, right from the start. In his examples, he tries to show that, if you don't count positive covid cases during the two weeks after the booster as belonging to the 'booster cohort' and instead group those positive cases into the '2 dose' cohort, then you get a false positive vaccine effectiveness for the booster. The obvious failure in his reasoning, would be that people would use for the 'numerator' a group of people counting as 'boosted' after a 2 week waiting period ... but then using for the 'denominator', using time at which people receive the jab to count them as boosted. Obviously if you count people as 'boosted' for a time period (in the denominator, at the moment they get the jab) ... and then move 'positive covid cases' out of that group into the '2 dose' group ... data will be skewed. I think it's a ridiculous thing to assume. Why on planet earth would people use two different criteria for 'who is boosted' literally in the same calculation. That would be stupid. The implication that health authorities 'count as boosted' anyone who received a jab, but then only count COVID cases based on 'received a jab + 2 weeks' makes no sense at all. As far as I can tell, it seems that health authorities are using the 'jab + 2 weeks' as the definition of 'who is boosted' in all calculations. By that obvious criteria, then this 'saline solution data misrepresentation' goes away. At least as presented in this article. I seenno evidence that Health authorities are doing otherwise. ... I will trade you a 'Boriquagato' for a 'Brandollini': "Brandolini's law, also known as the bullshit asymmetry principle, is an internet adage that emphasizes the difficulty of debunking false, facetious, or otherwise misleading information:[1] "The amount of energy needed to refute bullshit is an order of magnitude larger than is needed to produce it."

native_samples 5y ago
I just did provide evidence for it, from a public health agency. You haven't given any justification for why you think that data was wrong. At least majewsky implies a specific problem (albeit I didn't understand what they think it is). That's as good as it's possible to get because academics don't have access to the raw data with the correct definitions, so there can be no peer reviewed evidence. Not that peer review is any good anyway. "that doesn't apply if we are looking at daily data points of cases from the rates of those who contract COVID among vaxxed/unvaxxed population. There is no 'dead time' artefact by that kind of measure. Especially for case counts, then all persons who 'get covid' on a particular day have obviously 'survived' to be in the sampled cohort on that day." It does apply. Those populations are still classified as vaxxed/unvaxxed in a way that creates immortal time bias. Perhaps the explanation I give above isn't clear enough, but any classification system in which someone is assigned to a group some time after the intervention for that group actually started will have this problem. This was true both for the trials but also all government released stats (except it seems, those Alberta graphs which were so rapidly pulled). Again with cases this time: if you take the vaccine then it's not possible for you to be considered infected within the first two weeks after the shot, because if you were, that infection would be allocated to the unvaccinated cohort. This creates an "infection free period" for the vaccinated which would create an appearance of effectiveness even if the vaccine were a placebo. Fenton talks about it in the form of 'delayed reporting', but the underlying issue is more likely to be immortal time. "The distortion fades as quickly as the reporting delay. So the effect is null 1 week after vaccination rates stabilise (by his example of 1 week delay). His own 'real world data' comparison show vaccine effectiveness waning over a much longer periods." They didn't stabilize yet, did they? The vaccination campaign proceeds in waves in which new people are constantly becoming eligible, for the initial shots and then the boosters. His point is that for as long as we're doing this we will see what looks like fast-fading efficacy and we would see this even if the vaccine did nothing. We would also see the impossible data artifacts found in the ONS data, like vaccination increasing the non-vaccinated risk of non-COVID deaths. It's extremely problematic that this problem exists and creates the exact patterns in the data that are being used to drive the campaign forward. "The implication that health authorities 'count as boosted' anyone who received a jab, but then only count COVID cases based on 'received a jab + 2 weeks' makes no sense at all." That's his point. It doesn't make sense but that's what they do. When calculating how many people have been boosted by a given date, they report the total number of booster shots actually administered up to that point, because that's what their total records show. They aren't subtracting the number of people who received it recently. But when classifying the status of someone specific who has been infected/hospitalized/died, they classify people who have received a booster dose within the past 2 weeks as double vaxxed (because, they claim, it doesn't have any effect until 2 weeks later). And then when they sum those specific reports it creates immortal time, just phrased differently. You can see the effects of it in the ONS stats where you see the artifacts Fenton describes.

native_samples 5y ago

I just did provide evidence for it, from a public health agency. You haven't given any justification for why you think that data was wrong. At least majewsky implies a specific problem (albeit I didn't understand what they think it is). That's as good as it's possible to get because academics don't have access to the raw data with the correct definitions, so there can be no peer reviewed evidence. Not that peer review is any good anyway. "that doesn't apply if we are looking at daily data points of cases from the rates of those who contract COVID among vaxxed/unvaxxed population. There is no 'dead time' artefact by that kind of measure. Especially for case counts, then all persons who 'get covid' on a particular day have obviously 'survived' to be in the sampled cohort on that day." It does apply. Those populations are still classified as vaxxed/unvaxxed in a way that creates immortal time bias. Perhaps the explanation I give above isn't clear enough, but any classification system in which someone is assigned to a group some time after the intervention for that group actually started will have this problem. This was true both for the trials but also all government released stats (except it seems, those Alberta graphs which were so rapidly pulled). Again with cases this time: if you take the vaccine then it's not possible for you to be considered infected within the first two weeks after the shot, because if you were, that infection would be allocated to the unvaccinated cohort. This creates an "infection free period" for the vaccinated which would create an appearance of effectiveness even if the vaccine were a placebo. Fenton talks about it in the form of 'delayed reporting', but the underlying issue is more likely to be immortal time. "The distortion fades as quickly as the reporting delay. So the effect is null 1 week after vaccination rates stabilise (by his example of 1 week delay). His own 'real world data' comparison show vaccine effectiveness waning over a much longer periods." They didn't stabilize yet, did they? The vaccination campaign proceeds in waves in which new people are constantly becoming eligible, for the initial shots and then the boosters. His point is that for as long as we're doing this we will see what looks like fast-fading efficacy and we would see this even if the vaccine did nothing. We would also see the impossible data artifacts found in the ONS data, like vaccination increasing the non-vaccinated risk of non-COVID deaths. It's extremely problematic that this problem exists and creates the exact patterns in the data that are being used to drive the campaign forward. "The implication that health authorities 'count as boosted' anyone who received a jab, but then only count COVID cases based on 'received a jab + 2 weeks' makes no sense at all." That's his point. It doesn't make sense but that's what they do. When calculating how many people have been boosted by a given date, they report the total number of booster shots actually administered up to that point, because that's what their total records show. They aren't subtracting the number of people who received it recently. But when classifying the status of someone specific who has been infected/hospitalized/died, they classify people who have received a booster dose within the past 2 weeks as double vaxxed (because, they claim, it doesn't have any effect until 2 weeks later). And then when they sum those specific reports it creates immortal time, just phrased differently. You can see the effects of it in the ONS stats where you see the artifacts Fenton describes.