Comments by "N Marbletoe" (@nmarbletoe8210) on "Convalescent Plasma, Poor results" video.

  1. 5g can be ruled out due to the limited rollout
    4
  2. there are coronavirus vaccines for animals
    3
  3. certainly can be, but not always
    2
  4.  @gooner9772  I mean that we can read the results of a particular study and see if the sample size was large enough. Mathematically 40 samples is often enough, but more samples are done for measuring multiple groups or if there is a high variance. . If the study in question had p<0.00003 we can say it has a firm conclusion. If the p-value was 0.05 then it is considered marginal significance, and no firm conclusion. . For negative results, use power instead of significance. . Authors will often report the "power" of the test, which is the chance of detecting a difference, given that it exists and you take x samples. It is subjective but you can decide for yourself if, say, 90% or 99% is powerful enough. . Power essentially tells you the confidence of a negative result, while p-value the confidence in a positive result.
    2
  5. nope
    2
  6.  @gooner9772  The 40 number is a rule of thumb from stats class... with 40 samples from a uniform population, you get a confidence interval of a few percent, for a binomial choice (like Biden v Trump). . Real life is messy so most researchers get as much data as they can! . I agree it is not solid to extrapolate from one study, even if it's highly confident. If it matters it should be replicated. . I guess you have convinced me lol.
    1
  7.  @BatteryExhausted  That sounds like a strategy for when they don't know what to do. . i
    1