Chaz McNulty
Footballguy
You are, but you get to decide which messages you bring to the board. I like that you argue from the other side, but I think a p-value of .005 is a load. I would have left that one on the other boards.
Fantasy Football - Footballguys Forums
Welcome to Our Forums. Once you've registered and logged in, you're primed to talk football, among other topics, with the sharpest and most experienced fantasy players on the internet.
***OFFICIAL CYDY/Leronlimab Thread*** (1 Viewer)
- Thread starter This_Guy
- Start date
Don Hutson
Footballguy
Multiple people have said it. It didn't come from Shorts. It came from Longs who know far more on the subject than you or I. You have absolutely zero idea of what p value is necessary.
Last edited by a moderator:
Chaz McNulty
Footballguy
Caesar
Footballguy
I have to agree with Don on this one as far as the source. Way better information there than YMB or IHUB.
While I think .005 is a load as well, I think the point they are trying to make is that best case would be .05 and worst case would be .005. They then gave the required numbers to reach both, and if it falls somewhere in the middle, then we have a pretty good idea where it needs to be.
Again, with the FDA, who the hell knows what they will require? If they show a 20-30% reduction in deaths, it seems like a no-brainer to save 20000+ people, give or take. But again, with the FDA, wtf knows?
The only time .005 was mentioned was that it was the penalty that comes if you unblind early. For statistical significance, which still remains the goal for the CD12 study, it's still 0.05.
Don Hutson
Footballguy
The .005 we are talking about came from the message boards, not from CytoDyn. Under normal circumstances, a p value of .005 might be what is needed to end a trial early in the interim analysis. But we aren't in normal circumstances so nobody knows. I think Caesar summarized it well with:
Last edited by a moderator:
Don Hutson
Footballguy
Last edited by a moderator:
FreeBaGeL
Footballguy
He was talking about the number to end the trial early, not the number required for statistical significance at the end.
For ending the trial early it's all purely 100% guesswork. It could be .05 (unlikely) or .005 or anything in between. I think the basic point is that to end a trial early the results at the half way mark have to be really over the moon good, not just statistically significant.
Caesar
Footballguy
you got my shares. I reduced. I'm willing to take a hit right now because if they get approval, the PPS will become somewhat irrelevant long term. I'm now down to a comfortable position I can afford to just wait it out.
I'm not looking to buy back in unless it trickles down under 2 while we wait. Even then, I am not sure I will make the move.
Dwayne Hoover
Footballguy
I understood this part but I thought that once he added the control group that the paper would be accepted.
If I understand correctly, the paper has now been rejected for a different reason? Im confused though because there has been talk about HIV things that I thought were unrelated.
Am I the only one having a problem following this?
Chaz McNulty
Footballguy
I think the HIV thing is to do with LL as a monotherapy and approval getting pushed back. This seems to be the reason.
Dwayne Hoover
Footballguy
How will they know this? My understanding is they aren't going to see any of the data (unless the trial is halted). Am I wrong?
Last edited by a moderator:
Chaz McNulty
Footballguy
Chaz McNulty
Footballguy
No idea. Aren't they the ones that have to go through the data and present it to the CDC?
Dwayne Hoover
Footballguy
My understanding is that they don't see any data unless trial is halted.
Other than that, they have to make the decision to unblind it, which they will be penalized for in the p value.
If that is the case, it would be extremely risky to unblind
Last edited by a moderator:
That's what I understand as well. Only 1 person on the review board sees the results.
Chaz McNulty
Footballguy
They must know what the penalty is beforehand. I have heard that they need .045 to reach statistical significance. Maybe that accounts for a .005 penalty? If this is the case, then they should definitely unblind. With 45 deaths, I don't think there is any combination of deaths between placebo and LL that put them around .045.
They really need to get information out to other countries.
Dwayne Hoover
Footballguy
Chaz McNulty
Footballguy
ILUVBEER99
Footballguy
I read somewhere that Nadar said he expects CytoDyn to be 10x's bigger than Gilead. The guy is absolutely delusional. 10x's Gilead would be almost 800 billion market cap (2x's bigger than the current #1 biotech/pharma Johnson and Johnson)
He's a legit moron.
He's a legit moron.
Charlie Harper
Footballguy
dkp993
Footballguy
Ummm. OK. Am I supposed to know I’m his schedule? Didn’t realize mentioning him on a FF forum was such a hassle....
Last edited by a moderator:
Don Hutson
Footballguy
If CytoDyn is planning on paying the .005 penalty to get the data, they aren't going to tell anyone beforehand. They are going to say that they aren't going to do it. They want the DMSB to think that they need to stop the trial for people to receive the drug. It puts the onus on the DMSB. But I hope that they do decide to pay the penalty. When looking at the p value charts, a change of .005 isn't a big deal. It is far less than 1 death even with 390 patients. Landing between .045 and .05 isn't likely. But maybe CytoDyn doesn't want the data. No data keeps the charade going for another 3-6 months.
A p-value at or below 1% (0.01) will definitely be seen as significant. The question may come if it falls between 1% and 5%. Non-life-threatening treatments can get away with the 5% threshold and their drug has a track record of safety so 0.05 or less should be sufficient. It wouldn’t shock me, though, if the FDA or other powers are more conservative and want to see the 1% threshold.
Look at me, I teach Statistics and I also own CYDY!
Look at me, I teach Statistics and I also own CYDY!
Don Hutson
Footballguy
Do the charts in the this link look accurate to you for a 2-tail p value?
https://investorshangout.com/post/view?id=5914267
ConstruxBoy
Kate's Daddy
I was told there would be no math.
ChiefD
Footballguy
fred_1_15301
Footballguy
I don’t know why he’s using a t-test, which is intended to compare the mean difference between 2 groups. The chart is not looking at means but rather a difference in proportion of deaths between the 2 groups. Chi-squared or Fishers exact would be more appropriate. And
at using Excel.Dwayne Hoover
Footballguy
Yep.
This is their dilemma with unblinding it. On top of having a penalty applied, they have no idea what they are going to find. Of course, they are hoping the data is good enough that it at least silences the critics and at best puts a lot of pressure on the FDA to do something if its out in the public.
However, there is a chance that the data is not going to be favorable. Are they going to release that to the public? Your scenario of no data is better than bad data and they would certainly prefer to keep the charade going I believe.
What a strange company to be invested in. I think the drug works though.
Last edited by a moderator:
Don Hutson
Footballguy
Don Hutson
Footballguy
Chaz McNulty
Footballguy
Bob Sacamano
Footballguy
Dave RL
Footballguy
@fred_1_15301 @Don HutsonI don’t know why he’s using a t-test, which is intended to compare the mean difference between 2 groups. The chart is not looking at means but rather a difference in proportion of deaths between the 2 groups. Chi-squared or Fishers exact would be more appropriate. And
I just used the tried and true method of the internet... post something quickly and you will get all the right answers handed to you. Seriously, I would have used R or SAS if I had more time but I was just trying get close to the right answer. DM me if you have any suggestions to get a better answer. I'd like to get what I have corrected if it's off.
What also I get from the numbers I put together is the small number it will take to move from significant to not ( 3 - 4 negative outcomes)
fred_1_15301
Footballguy
Sorry thought you were some random dude from the internet (not a footballguy). Didn’t mean to offend you but Excel is so ancient these days when it comes to stats.
I’ll look more closely later but a T test is not used to compare proportions (I.e categorical variable ). A T test is used to compare the mean difference (I.e. continuous variable) between 2 groups. I would need to look more closely at the data and unfortunately in my current role I no longer have access to SAS. @pecorino any thoughts?
Note - I completed my masters in biostats 18 years ago but haven’t really had to practice it much since. So I could be off here.
fred_1_15301
Footballguy
Because of the variation of tests, and because we don’t know if they have existing information on things like standard deviations of populations etc., I wouldn’t say one way or another whether those tables are correct. However it has been noted that, with such a small sample size, even a change of one or two deaths can have a big affect on the P value. That’s why these researchers would like to have larger samples.
ConstruxBoy
Kate's Daddy
Last edited by a moderator:
Don Hutson
Footballguy
This is all Greek to me. It's 195 patients with 45 deaths. Some have said it is supposed to be 2-tailed. Trying to figure out how many deaths in each arm would encompass the p values from .005 to .05.
ConstruxBoy
Kate's Daddy
Last edited by a moderator:
Don Hutson
Footballguy
CYDY hit $2.99. It is now $2.90.
Don Hutson
Footballguy
This would work out to:
Treatment arm
105 alive 25 deaths
Placebo arm
45 alive 20 deaths
If the placebo arm has more deaths:
Treatment arm
108 alive 22 deaths
Placebo arm
42 alive 23 deaths
Does the first scenario actually have a p value of .03? What is the p value of the second scenario?
Chaz McNulty
Footballguy
I was just quoting from Yahoo board. I don't know the formula for figuring this out. From what else I have read, I think the p-value would be higher and not SS in the first scenario. But it would be SS in the 2nd scenario.
