3 ms·

I don’t think this is quite right of a reading “This would imply that they did not find a correlation between vaccination and severity of symptoms in their test

by mint2 3y ago

I don’t think this is quite right of a reading “This would imply that they did not find a correlation between vaccination and severity of symptoms in their testing.” Instead, I think a more accurate rough understanding of their process is they first backed out the effect of severity and then checked if vaccine was correlated with that adjusted number - they found it wasn’t. They didn’t make an assessment on how vaccines effect severity.

taeric 3y ago
My read wasn't that they made an assessment of vaccine on severity. Rather, I read that they found that it had no impact on transmission. They found that, on average, self reported severity correlated with transmission. However, they then state that transmission does not correlate with vaccination status. I could hope that they mean there is no extra relation between vaccination status and transmission above reduced symptoms. I don't know why they would state it in the way they do if that is what they mean, though.
- mint2 3y ago
  I agree it is poorly stated, but that is what they mean, no extra impact beyond reduced symptoms and the benefit thereof.
  - taeric 3y ago
    How do you know that?
    - mint2 3y ago
      Because their paper states it. That’s what is meant by “when accounting for etc” in: “We found that vaccinated and unvaccinated participants exhaled similar numbers of SARS-CoV-2 RNA copies (Fig 2C), when accounting for age, sex, presence of co-morbidities, days since symptom onset and symptom severity”
      - taeric 3y ago
        That... doesn't say anything about them exhaling less? It specifically says that they exhibit the same. You can make a logical jump to say that, combined with a high variance but statistically significant link to self reported symptoms, that maybe vaccinated would have people that self report lower levels. But, that is specifically not in the comparison that you quoted. It says only that vaccinated and unvaccinated are not distinguishable.
        mint2 3y ago
        It’s saying indistinguishable when adjusted for severity. It’s like saying night and day in Yosemite are equally bright when effects of the sun and moon are controlled for. Look, this doesn’t appear to be sticking, it’s a basic statistical modeling thing. At this point the options are accept it at face value, study stats, or be like the kid saying “why” 100 times after every explanation.
        taeric 3y ago
        But they specifically say that differently? As I said in another, you are jumping from severity correlating with vaccine to get to this idea. But, they actually call out self reported severity as correlating. Such that, if the correlations were strong enough it would have also showed up in their control for that variable. Stated differently, you have variables A = self-reported severity B = amount shed C = vaccinated status If A and B are correlated, and A and C are correlated, there will a floor to how correlated B and C can be. Their later sentence is that the correlation between B and C is below the significance floor. They never comment on how correlated A and C are. You could argue there is another variable D = Actual Severity, but that is not covered in the statements I remember.
        mint2 3y ago
        This is where the mistaken reasoning is, they aren’t measuring or reporting on that naive correlation between B and C that you’re focusing on. They are saying after backing out the correlation between A and B to get adjusted B, there is no correlation between C and adjusted B. That is an ELI5 explanation of their statement “controlled for” about the factors like severity. Yes their focus is not on how A and C relate, but we know that relation from other sources. That is what controlled for means. “Hello” means a greeting in English. It’s the definition. Same concept, it’s the meaning of controlled for. It’s what it was defined to mean.

by mint2 3y ago

taeric 3y ago

My read wasn't that they made an assessment of vaccine on severity. Rather, I read that they found that it had no impact on transmission. They found that, on average, self reported severity correlated with transmission. However, they then state that transmission does not correlate with vaccination status. I could hope that they mean there is no extra relation between vaccination status and transmission above reduced symptoms. I don't know why they would state it in the way they do if that is what they mean, though.

mint2 3y ago
I agree it is poorly stated, but that is what they mean, no extra impact beyond reduced symptoms and the benefit thereof.
- taeric 3y ago
  How do you know that?
  - mint2 3y ago
    Because their paper states it. That’s what is meant by “when accounting for etc” in: “We found that vaccinated and unvaccinated participants exhaled similar numbers of SARS-CoV-2 RNA copies (Fig 2C), when accounting for age, sex, presence of co-morbidities, days since symptom onset and symptom severity”
    - taeric 3y ago
      That... doesn't say anything about them exhaling less? It specifically says that they exhibit the same. You can make a logical jump to say that, combined with a high variance but statistically significant link to self reported symptoms, that maybe vaccinated would have people that self report lower levels. But, that is specifically not in the comparison that you quoted. It says only that vaccinated and unvaccinated are not distinguishable.
      - mint2 3y ago
        It’s saying indistinguishable when adjusted for severity. It’s like saying night and day in Yosemite are equally bright when effects of the sun and moon are controlled for. Look, this doesn’t appear to be sticking, it’s a basic statistical modeling thing. At this point the options are accept it at face value, study stats, or be like the kid saying “why” 100 times after every explanation.
        taeric 3y ago
        But they specifically say that differently? As I said in another, you are jumping from severity correlating with vaccine to get to this idea. But, they actually call out self reported severity as correlating. Such that, if the correlations were strong enough it would have also showed up in their control for that variable. Stated differently, you have variables A = self-reported severity B = amount shed C = vaccinated status If A and B are correlated, and A and C are correlated, there will a floor to how correlated B and C can be. Their later sentence is that the correlation between B and C is below the significance floor. They never comment on how correlated A and C are. You could argue there is another variable D = Actual Severity, but that is not covered in the statements I remember.
        mint2 3y ago
        This is where the mistaken reasoning is, they aren’t measuring or reporting on that naive correlation between B and C that you’re focusing on. They are saying after backing out the correlation between A and B to get adjusted B, there is no correlation between C and adjusted B. That is an ELI5 explanation of their statement “controlled for” about the factors like severity. Yes their focus is not on how A and C relate, but we know that relation from other sources. That is what controlled for means. “Hello” means a greeting in English. It’s the definition. Same concept, it’s the meaning of controlled for. It’s what it was defined to mean.