3 ms·

how did this study end up being high voted on HN. I am very doubtful about mandatory vaccine for everyone, and know vaccines aren't the perfect solutions. Howev

by bsaul 5y ago

how did this study end up being high voted on HN. I am very doubtful about mandatory vaccine for everyone, and know vaccines aren't the perfect solutions. However, this study is probably one of the worst i've ever seen. It compares countries being in totally different situations regarding the infection spread. Some are done with delta, others are just starting..

necovek 5y ago
When numbers being looked at are large, they can tell us something even if we admit to them not being perfect in many ways. At worst, we can figure out trends or, at best, get a negative correlation to learn that something is not positively correlated or even causal. But everyone reading a statistical analysis, esp on HN, should know it for what it is: it does not account for many things (people have mentioned age, you bring up phase of the pandemic, my thought was how this doesn't account for probability of vaccinated people intermingling more which is related to someone else bringing up people in rural areas...). The entire study, based on the data it covered, does not proclaim that vaccines are bad (even though anti-vax people will take it as such), it's just that they are at an early phase and might need more development to increase their effectiveness, and that we'll need other solutions in addition to vaccines if we want to get back to any semblance of normalcy. I think it's of utmost importance to understand who is exactly at risk from a COVID infection (other than "old folk", which is not really saying much) so we can better prepare for dealing with them (us? :) in the future. While this study does not answer this, keeping a critical mind is important in science.
- ImaCake 5y ago
  While it is sometimes true that larger sample sizes can overcome many insufficiencies in the data I don’t think that is the case here. The data is still noisy and any attempt at statistics here would result in a non-significant result. The authors appear to be drawing their conclusions based on a lack of statistical significance. This is a basic fallacy, you can’t say something is not there because your data is too crap to see it. There are statistical tests that determine if something is not different from zero, the authors did not use those tests.
  - necovek 5y ago
    I have't looked that deep into results of this particular study (hopefully peer reviews will do that: it would be unmanageable to look at everything) — but that was not the GP's complaint, but rather that conditions of each region have not been taken into account, and I am saying that this is fine if numbers are sufficiently large and statistically relevant. You may be right that they are not, and hopefully that results in the paper being retracted.

by bsaul 5y ago

necovek 5y ago

When numbers being looked at are large, they can tell us something even if we admit to them not being perfect in many ways. At worst, we can figure out trends or, at best, get a negative correlation to learn that something is not positively correlated or even causal. But everyone reading a statistical analysis, esp on HN, should know it for what it is: it does not account for many things (people have mentioned age, you bring up phase of the pandemic, my thought was how this doesn't account for probability of vaccinated people intermingling more which is related to someone else bringing up people in rural areas...). The entire study, based on the data it covered, does not proclaim that vaccines are bad (even though anti-vax people will take it as such), it's just that they are at an early phase and might need more development to increase their effectiveness, and that we'll need other solutions in addition to vaccines if we want to get back to any semblance of normalcy. I think it's of utmost importance to understand who is exactly at risk from a COVID infection (other than "old folk", which is not really saying much) so we can better prepare for dealing with them (us? :) in the future. While this study does not answer this, keeping a critical mind is important in science.

ImaCake 5y ago
While it is sometimes true that larger sample sizes can overcome many insufficiencies in the data I don’t think that is the case here. The data is still noisy and any attempt at statistics here would result in a non-significant result. The authors appear to be drawing their conclusions based on a lack of statistical significance. This is a basic fallacy, you can’t say something is not there because your data is too crap to see it. There are statistical tests that determine if something is not different from zero, the authors did not use those tests.
- necovek 5y ago
  I have't looked that deep into results of this particular study (hopefully peer reviews will do that: it would be unmanageable to look at everything) — but that was not the GP's complaint, but rather that conditions of each region have not been taken into account, and I am saying that this is fine if numbers are sufficiently large and statistically relevant. You may be right that they are not, and hopefully that results in the paper being retracted.