3 ms·

If the rate were 1 in 2.3 million, then with 130,000 people vaccinated: p = 1/2,300,000 N = 130,000 Prob(0 cases) = (1 - p)*N = 0.9450... Prob(1 case) ~= p

If the rate were 1 in 2.3 million, then with 130,000 people vaccinated: p = 1/2,300,000 N = 130,000 Prob(0 cases) = (1 - p)*N = 0.9450... Prob(1 case) ~= p * N * (1 - p)*N = 0.0534... Prob(2 cases) ~= p*2 * (N*2 / 2) * (1 - p)*N = 0.0015... Prob(3 cases or more) = 1 - P(0) - P(1) - P(2) ~= 2.886e-05 Which is small enough to refute you comment, yes?

GavinMcG 6y ago
No. The evidence against the rate being 1 in 2.3 million is that the total number of cases divided by the total number of people vaccinated is a higher rate. You can't take a subset of the population, find that that subset is out of step with the rate in the total population, and then conclude that the rate in the total population is wrong. The number of cases isn't necessarily evenly distributed, and your model is one that applies to independent random events. But the fact that a case occurred in Norway rather than in another country doesn't dramatically change the overall rate.
- marvin 6y ago
  This probably doesn't convince you, but as additional data, Norwegian authorities have now confirmed five hospitalizations due to this presumed side effect, two of which resulted in death. https://www.vg.no/nyheter/innenriks/i/86OmPr/drama-i-kulissene-paa-astrazeneca-moete-norge-tok-dissens https://www.vg.no/nyheter/innenriks/i/86OmPr/drama-i-kulisse... Of course, Norwegian health authorities point out in the interview that the ethical consideration will be different between countries, depending on expected vaccine delivery and local incidence of covid.
  - GavinMcG 6y ago
    More data is obviously convincing. Why would you off-handedly suggest that I'm fundamentally irrational?
    - marvin 6y ago
      I figured you put more weight on «Norway and Europe are different populations and have different probabilities for this reaction» rather than the specific numbers from Norway. I wasn’t implying that you are irrational or stupid.
bainsfather 6y ago
I just realised that using * twice to represent a power shows up as just a single * - here is the correct version (the numbers are anyway correct): Prob(0 cases) = (1 - p)^N = 0.9450... Prob(1 case) ~= p * N * (1 - p)^N = 0.0534... Prob(2 cases) ~= (p^2) * (N^2 / 2) * (1 - p)^N = 0.0015... My apologies for that.

GavinMcG 6y ago

No. The evidence against the rate being 1 in 2.3 million is that the total number of cases divided by the total number of people vaccinated is a higher rate. You can't take a subset of the population, find that that subset is out of step with the rate in the total population, and then conclude that the rate in the total population is wrong. The number of cases isn't necessarily evenly distributed, and your model is one that applies to independent random events. But the fact that a case occurred in Norway rather than in another country doesn't dramatically change the overall rate.

marvin 6y ago
This probably doesn't convince you, but as additional data, Norwegian authorities have now confirmed five hospitalizations due to this presumed side effect, two of which resulted in death. https://www.vg.no/nyheter/innenriks/i/86OmPr/drama-i-kulissene-paa-astrazeneca-moete-norge-tok-dissens https://www.vg.no/nyheter/innenriks/i/86OmPr/drama-i-kulisse... Of course, Norwegian health authorities point out in the interview that the ethical consideration will be different between countries, depending on expected vaccine delivery and local incidence of covid.
- GavinMcG 6y ago
  More data is obviously convincing. Why would you off-handedly suggest that I'm fundamentally irrational?
  - marvin 6y ago
    I figured you put more weight on «Norway and Europe are different populations and have different probabilities for this reaction» rather than the specific numbers from Norway. I wasn’t implying that you are irrational or stupid.

bainsfather 6y ago

I just realised that using * twice to represent a power shows up as just a single * - here is the correct version (the numbers are anyway correct): Prob(0 cases) = (1 - p)^N = 0.9450... Prob(1 case) ~= p * N * (1 - p)^N = 0.0534... Prob(2 cases) ~= (p^2) * (N^2 / 2) * (1 - p)^N = 0.0015... My apologies for that.