Geez, I thought this would be an easy problem:)
Post# of 153771
(Total Views: 791)
Posted On: 06/11/2020 11:08:10 PM
Re: Rex Eupseiphos #37371
Geez, I thought this would be an easy problem
Thank you very much for your help, Rex. I don't think there is anything else you can do to educate me except for telling me to go study
Given the difficulty of this problem, the CYDY trial must have some predetermined mathematics that goes with it.
I was optimistic that I might have a clue with the Wolfram algorithm, but with a transition to Bayesian vocabulary, I have little hope of figuring this out.
On the other hand perhaps I can make an example that must demonstrate that the statistical algorithms are srbitrarily imputing features of the distribution which may not exist.
On the mavrilimumab data, its 13/13 and 19/26 for survival.
If I observe that the probability of 13 mavriliumab survivals given 19/26 placebo survivals is (19/26)^13=0.016, then I could assert that any p value other than 0.016 must necessarily impute some other value than 19/26. In other words, if you get any p-value besides 0.016 for the mavrilimumab results, then you must have done something funky to 19/26. And if there is one way to do something to 19/26, no doubt there is a second opinion as to what it is you should do to 19/26.
But at this point I am probably just wasting everyone's time, I think.
CYDY and the FDA must have some pre-specified math that they use for these trials. Good grief, it might not be the same for every trial!
Thank you very much for your help, Rex. I don't think there is anything else you can do to educate me except for telling me to go study
Given the difficulty of this problem, the CYDY trial must have some predetermined mathematics that goes with it.
I was optimistic that I might have a clue with the Wolfram algorithm, but with a transition to Bayesian vocabulary, I have little hope of figuring this out.
On the other hand perhaps I can make an example that must demonstrate that the statistical algorithms are srbitrarily imputing features of the distribution which may not exist.
On the mavrilimumab data, its 13/13 and 19/26 for survival.
If I observe that the probability of 13 mavriliumab survivals given 19/26 placebo survivals is (19/26)^13=0.016, then I could assert that any p value other than 0.016 must necessarily impute some other value than 19/26. In other words, if you get any p-value besides 0.016 for the mavrilimumab results, then you must have done something funky to 19/26. And if there is one way to do something to 19/26, no doubt there is a second opinion as to what it is you should do to 19/26.
But at this point I am probably just wasting everyone's time, I think.
CYDY and the FDA must have some pre-specified math that they use for these trials. Good grief, it might not be the same for every trial!
(1)
(0)