Proof of Immortality, the IV league

[jt512]

yep
and if you include biased data (anecdotes) then your results will tend tobe biased, whether bayesian or not.

You might want to try to actually understand what it means in statistics for something to be biased.

 

[Mojo]

Mojo,
- What if the report (anecdote) given was that his wife was in the waiting room, wearing red -- but she wasn't?

Jabba,
- I could post a series of videos of me throwing a die and it coming up as a six. Would you accept that as evidence that I throw a six every time I throw a die?

Mojo,
- I don't see how your example is analogous.

Jabba,
- As I pointed out yesterday, the last time you did it, you are quoting me in a misleading way. It isn't intended to be "analogous". As anyone can see by looking at the post you quoted, it was not a direct response to your post. If you want a response to your post, look at the post that quoted it.

Anyway, would you accept the videos as evidence that I throw a six every time?

 

[Mojo]

Argumemnon,
- I think that LL was wrong.

On what basis can you think that? By your own admission you didn't understand what he had posted.

 

[annnnoid]

The point is, annnnoid, that no one cares.

Edited by jsfisher: 

<snip> Edited for compliance with Rule 12 of the Membership Agreement.

 

[annnnoid]

Literally anywhere. Look up the scientific method. Then follow it.

Since you're a lawyer...we have a legal question that needs answering:

In the judicial system…if the testimony of a credible witness cannot be corroborated with other evidence (verified...IOW), is that testimony ‘thrown out’.

Second question: Can the exclusively anecdotal unverified testimony of a credible witness contribute to a conviction in a criminal or civil case?

 

[annnnoid]

The bit I quoted.

You have yet to identify where I made this claim. The bit you quoted does not include me making any such claim.

…and you have yet to demonstrate any understanding of what variety of information science deals in.

Now we're getting somewhere. Can you elucidate on what you've said here, please? Give a couple of examples of anecdotes, hypotheses and "whole pile of other things", please.

Why on earth should I waste my time going to all this trouble? Just explain what your point is. Otherwise don’t expect me to bother.

 

[Nonpareil]

Meanwhile, we're still waiting I, on the other hand, don’t care enough to bother with your incoherent garbage anymore. That’s what ‘not caring’ means.

No doubt you will continue to 'not care' enough to dribble your vitriol over my future posts. For someone who 'doesn't care'...you sure seem to get riled up.

The fact that you think I have been angry with you at any point is honestly hilarious.

You have never seen me angry. Your inability to differentiate between vitriol and blunt dismissal of obvious nonsense is your own problem, not mine.

Since you're a lawyer...we have a legal question that needs answering:

In the judicial system…if the testimony of a credible witness cannot be corroborated with other evidence (verified...IOW), is that testimony ‘thrown out’.

As already mentioned, if a witness' testimony cannot be supported, they are not credible.

You have yet to identify where I made this claim. The bit you quoted does not include me making any such claim.

It has you doing exactly that.

Like I said, it's very hard to take you seriously when you so obviously don't even understand what you yourself are saying. If you wish to retract or clarify your explicit response to the question, then do so. But don't sit here trying to say "I never said that" when you can be directly quoted as doing so. It's just wasting everyone's time.

…and you have yet to demonstrate any understanding of what variety of information science deals in.

As I said previously: by all means, annnnoid. Educate us. It should be entertaining.

 

[Jedi]

You might want to try to actually understand what it means in statistics for something to be biased.

for a statistic to be an unbiased estimate of a population parameter, the data values used in calculating the statistic are random variables, identically and independently distributed (iid), inter alia. so the sample mean is an unbiased estimate of the population mean only if the data used are random variables, iid, etc. the central limit thm also requires random variables. you know why?

because random error can be dealt with in statistical models but systematic error cannot, as systematic error is evidence of bias. and anecdotes are wrought with systematic error which you refuse to deal with. in your bayes examples, you're using anecdotes in both your prior and likelihood. again, if all the data you're looking at is biased, your result will be biased. if you include any biased data in the likelihood (prior is where it goes) then you are introducing bias in your posterior estimates.

well, you did offer a counter-example (anecdote) of how people discovered cures in the stone age. you claim they didn't use clinical trials as if this was a counter-arguemnt to whether clinical trials are the gold standard (they didn't use bayes either, btw). but your anecdote suffers from the same fatal flaw that you've been avoiding responding to, repeatedly. in fact they actually did do trials in those days. it was called 'trial and error.' someone with a cold may have accidentally walked by a burning eucalyptus grove and realized he felt better. someone else may have done the same near a burning poison oak bush and had a different outcome. others may have walked by something inert burning but believed they felt better for reasons not associated with what they were breathing. but we only hear the eucalyptus anecdote so you think there's no need for any controlled trial. no double-blinding. no control for a placebo effect. so if 1000 people got sick or died trying stuff to cure their cold they're not taking into account. it's that one that worked we all focus on to show clinical trials are not needed.

so now you're promoting stone-age methods like trial and error over clinical trials? you may as well drop bayes while your at it. and computers. and... all these things have added nothing to drug discovery, right?

and no answer for any of this except to self proclaim your statistical superiority, repeatedly. proclaiming and demonstrating are two different things. you're aware of what an ad homonym argument is? 2+2 is not 534 even if Einstein tells you so.

and by the way, what's your understanding of bias in statistics?

 

[jt512]

Unverified claims - that is, anecdotes - are face-down playing cards. They might be exactly what you need to build the winning poker hand, but they might also be completely worthless. In order to know, you have to look at them. Five cards face-down in front of you are not evidence that you have a royal flush.

How about five face-up cards, specifically 6H, 7H, 8H, 9H, 2C. The game is Texas Holdem, and those cards are on the board. Ten players each have two face-down cards. The rules of the game are that each player makes the best 5-card hand he can out of his 2 cards + the 5 board cards. Is there any evidence here that at least one player has a straight flush? If you were holding a pair of 9s and first to play, would you bet or pass?

 

[Jedi]

How about five face-up cards, specifically 6H, 7H, 8H, 9H, 2C. The game is Texas Holdem, and those cards are on the board. Ten players each have two face-down cards. The rules of the game are that each player makes the best 5-card hand he can out of his 2 cards + the 5 board cards. Is there any evidence here that at least one player has a straight flush? If you were holding a pair of 9s and first to play, would you bet or pass?

cards, in your analogy, are actual data we're looking at. verifiable. not made up unverifiable stories (anecdotes). bad analogy. better analogy is if you told me a story that you had two 9's and won. may be true, may not. unverifiable. and i most likely won't hear many stories about someone who had a pair of 9's in your scenario and lost.

 

[Nonpareil]

How about five face-up cards, specifically 6H, 7H, 8H, 9H, 2C. The game is Texas Holdem, and those cards are on the board. Ten players each have two face-down cards. The rules of the game are that each player makes the best 5-card hand he can out of his 2 cards + the 5 board cards. Is there any evidence here that at least one player has a straight flush? If you were holding a pair of 9s and first to play, would you bet or pass?

In short, and without calculations because I have neither the time nor the desire to run them: yes, there is evidence one way or the other. Because there are face-up cards, which influence your knowledge of what the face-down ones can or cannot be, and what hands would result if they were or were not a certain value and-slash-or suit.

Which is the whole point. On their own, face-down cards tell you nothing; with supporting evidence, you have a case for what those face-down cards could be, and what that would mean.

 

[Loss Leader]

LL,
- An NDE anecdote claims an event. If the other involved witness(es) disavow the claimed event, the claimed event is effectively falsified.

That has nothing to do with anything. The researcher attempts to show that a particular outcome could only have been possible if NDE's had to do with immateriality. If the outcome could be true under either materialism or immaterialism, then nothing has been evidenced either way.

If you want to use NDE's as evidence of a soul, then you need evidence that could have been obtained no other way than by an immaterial soul. Nothing less, nothing more.

 

[Loss Leader]

Tell me Loss Leader…in the judicial system…if the testimony of a credible witness cannot be corroborated with other evidence, is that testimony ‘thrown out’ as Nonpareil claims it is?

Second question: Can the exclusively anecdotal unverified testimony of a credible witness contribute to a conviction in a criminal or civil case?

Thank you for putting my name in red. I'd been skimming through your posts so quickly that I would otherwise have missed it.

I already know the answers to these questions (my brother is a lawyer…and a very very successful one at that)…but I think Nonpareil needs an education.

Well, that is amazing. I've neve seen such a demonstration of telepathy. Your brother went to law school and somehow you learned the rules of evidence.

Congratulations on whatever it is you wrote.

Thank you for putting my name in red

 

[jt512]

for a statistic to be an unbiased estimate of a population parameter, the data values used in calculating the statistic are random variables, identically and independently distributed (iid), inter alia. so the sample mean is an unbiased estimate of the population mean only if the data used are random variables, iid, etc. the central limit thm also requires random variables. you know why?

because random error can be dealt with in statistical models but systematic error cannot, as systematic error is evidence of bias. and anecdotes are wrought with systematic error which you refuse to deal with. in your bayes examples, you're using anecdotes in both your prior and likelihood. again, if all the data you're looking at is biased, your result will be biased. if you include any biased data in the likelihood (prior is where it goes) then you are introducing bias in your posterior estimates.

well, you did offer a counter-example (anecdote) of how people discovered cures in the stone age. you claim they didn't use clinical trials as if this was a counter-arguemnt to whether clinical trials are the gold standard (they didn't use bayes either, btw). but your anecdote suffers from the same fatal flaw that you've been avoiding responding to, repeatedly. in fact they actually did do trials in those days. it was called 'trial and error.' someone with a cold may have accidentally walked by a burning eucalyptus grove and realized he felt better. someone else may have done the same near a burning poison oak bush and had a different outcome. others may have walked by something inert burning but believed they felt better for reasons not associated with what they were breathing. but we only hear the eucalyptus anecdote so you think there's no need for any controlled trial. no double-blinding. no control for a placebo effect. so if 1000 people got sick or died trying stuff to cure their cold they're not taking into account. it's that one that worked we all focus on to show clinical trials are not needed.

so now you're promoting stone-age methods like trial and error over clinical trials? you may as well drop bayes while your at it. and computers. and... all these things have added nothing to drug discovery, right?

and no answer for any of this except to self proclaim your statistical superiority, repeatedly. proclaiming and demonstrating are two different things. you're aware of what an ad homonym argument is? 2+2 is not 534 even if Einstein tells you so.

and by the way, what's your understanding of bias in statistics?

Edited by jsfisher: 

<snip> Edited for compliance with Rules 0 and 12 of the Membership Agreement.

 

[Jedi]

Edited by jsfisher: 

<snip> Moderated content redacted.

Edited by jsfisher: 

Edited for compliance with Rules 0 and 12 of the Membership Agreement

 

[jt512]

<snip>

<snip>

Edited by jsfisher: 

Moderated content and response to same redacted.

 

[zooterkin]

I like the analogy with the face-down poker cards. They are not evidence of a royal flush until you turn them over and verify. However, they would seem to be evidence that you are playing poker.

Which makes me think that an anecdote may not qualify as evidence of one thing but might of another. And, by symmetry, evidence may have a high value in one dimension and none at all in another. Strange bedfellows.

I have often had five face-down playing cards in front of me. I have never played poker.

 

[marplots]

I have often had five face-down playing cards in front of me. I have never played poker.

So you say. I'm afraid I cannot accept your anecdote as evidence for your claim, but since it also supports my claim, and it's not my anecdote, it's strong evidence for the poker hypothesis.

I have learned all this by paying close attention in this thread. Or so my story goes...

 

[Jedi]

How old are you?

old enough to make cogent logical statistical arguments. you?

the only reason i responded to you is cause you wrote:

You might want to try to actually understand what it means in statistics for something to be biased.

if you want me not to respond to you, stop making irrelevant snarky comments and nit-picky spell check comments. if you want to have a discussion, then you need to back your assertions, like the one i quoted you on above. you asserted that i don't understand statistical bias. i explained it to you. your turn to refute. you can't, that's fine. i'm fine with dropping it.

 

[jt512]

old enough to make cogent logical statistical arguments.

I do not think those words mean what you think they mean.