Micro results in a macro world

[Guest]

Re: Re: Re: Micro results in a macro world

There is no life after death. That subject has now been addressed.

Central Scroat,

I am a skeptic.

You made the extraordinary claim of: "There is no life after death."

Can you provide the proof of that claim?

Thanks.

 

[PixyMisa]

You have me on ignore? Uh, ok...

Yup. It's a pain responding to you this way, though.

Your 'analysis' amounts to name calling.

Hardly. It shows that the PEAR-people don't know where to begin writing a scientific paper.

The fact that you can't wrap your brain around the abstract is moot.

Wrong. A paper like this is intended to inform others. If it's incomprehensible, it's a bad paper.

The fact that a test wasn't blinded doesn't make the science apriori bad.

In this sort of research, yeah, it does.

The fact that the experimenters can dictate the experiment (encourage a playful mood) is not bad. That is what experimenters do, they set the parameters of the experiment.

"Playful" is not a parameter.

You said:

----
Not keeping records also being a popular tool of experimental science.
----


But they did NOT say they didn't keep records. They said that they did not "maintain systematic records on the relative effectiveness of the various personal strategies developed by the participants in their approach to the task, or on any of their psychological characteristics."

Yeah. But they chose to talk about those factors anyway. My point stands.

That is hardly the same thing as saying no records were kept, period.

Which I didn't say.

For example, you completely ignore, in your review, the sentences RIGHT AFTER all of that where they do talk about some of the things they recorded. You also ignore all the tables and graphs they had in the paper. Not to mention all their appendices. You think that these magically appeared without them keeping some type of records?

I do not ignore this at all. I revel in their craptacular data.

They also have several sections where their p-values were NOT significant. So so much for the 'only publishing the hits' theory.

Which I didn't say either. At all. Ever.

Can you explain the fact that the 653-trial formal database had a composite z-score of 5.418 (pvalue < 3x10^ -8 )? And I'll give you a hint, saying "meta-analysis" as your answer doesn't explain anything.

Yep. Their experimental methods are utterly hopeless, open to fraud and wishful thinking. When they tightened up the experimental controls, the positive results went away. And the PEAR-people are at a loss to explain this

Statistical analysis of junk data gives you junk statistics.

 

[Drooper]

Whodini,

A couple of points on the stats.

The type of tests you desribe rest on a "chi-square" distribution. This rests on the principle that the the sum of n independent squared standard normal distributed variables will have a chi-squared distribution with "n" degress of freedom. So the form is

X ~ (Z(1)^2 + Z(2)^2 + ..... + Z^2)

Summing standard normal distributions in the way you describe isn't really valid.

The key elements are

"independence", which could be breached if any number of the underlying experiments had a similar flaw, say methodology.

"standard normal distribution" could hold assymptotically for large samples (hundreds). But for experiments resulting un trials numberingnin the tens this assumtpion could be seriously compromised.

 

[Guest]

Re: Re: Re: Micro results in a macro world

There is no life after death. That subject has now been addressed.

Of course there is, you become worm fodder if you are buried and then you might in turn then fertilize the soil allowing new things to grow from you. or you get eaten and sustain another life that then makes anew life, there you go eternal life. Or your fried and scattered and the same cycle begins again.

Life after death. reborn into something else as food

 

[Guest]

PixyMisa,


----
It shows that the PEAR-people don't know where to begin writing a scientific paper.
----


It more shows that you don't know how to read and digest their information.

If you need help understanding something, you can post your questions here and we'll help you.


----
"Playful" is not a parameter.
----


The mood of the sessions most certainly is whether you like it or not.


----
Statistical analysis of junk data gives you junk statistics.
----


Hehehe, that still doesn't explain: the fact that the 653-trial formal database had a composite z-score of 5.418 (pvalue < 3x10^ -8 )?

 

[Guest]

Drooper,


----
Summing standard normal distributions in the way you describe isn't really valid.
----


Independence and asymptotic results are rarely met in practice, yet such things are done all the time.

 

[Brian]

The thing is, juryjone's original point was that even if the meta-analysis proved that psychics can reliably make a machine output 5% more 1's than it should, so what? If any one person were talented enough to do something more meaningful, then a meta-analysis wouldn't have been needed. It would require a slightly different way of thinking about things, but it would have no practical effect on people's lives.

I imagine you could get that same %5 more ones with a few lines of code. Reminds me of a Randi quote, "If Uri Gellar is using psychic powers to bend spoons, he's doing it the hard way."

Not quite verbatim of course.

 

[Guest]

PixyMisa,


----
It shows that the PEAR-people don't know where to begin writing a scientific paper.
----


It more shows that you don't know how to read and digest their information.

If you need help understanding something, you can post your questions here and we'll help you.

Woo-woo tactic #45: The smarter the person is that disagrees with you, the more you should treat them like a drooling idiot, regardless of your own level of intelligence.

 

[Guest]

Woo-woo tactic #45: The smarter the person is that disagrees with you, the more you should treat them like a drooling idiot, regardless of your own level of intelligence.

Militant skeptic tactic # 147: Throw in a bunch of logical fallacies and so-called 'woo-woo tactics' to try and discredit someone who has a valid point.

Pixy says that the PEAR people don't know even where to begin writing a scientific paper. That is simply false.

The paper made sense to me. If Pixy can't understand it, that says something about Pixy, not really about the authors.

 

[Guest]

Pixy says that the PEAR people don't know even where to begin writing a scientific paper. That is simply false.

The paper made sense to me. If Pixy can't understand it, that says something about Pixy, not really about the authors.

Pixy is undoubtedly one of the smartest people here. It is unworthy of you to talk baby talk to her. Chances are extremely remote that you are saying something that's over her head.

You speak as if you have knowledge of what should and shouldn't be in a scientific paper. What are your credentials? I'll bet Pixy could supply some.

 

[Drooper]

Drooper,


----
Summing standard normal distributions in the way you describe isn't really valid.
----


Independence and asymptotic results are rarely met in practice, yet such things are done all the time.

No, asymptotic results hold universally - a result known as the Central Limit Theory is a case in point and something used in studies of this type. At other times in asymptotic qualities are crucial in designing statistics - for instance a search for "consistency".

Independence in this context should apply readily enough if the experiments in question were conducted properly.

 

[PixyMisa]

You speak as if you have knowledge of what should and shouldn't be in a scientific paper. What are your credentials? I'll bet Pixy could supply some.

Just to make things clear: I don't have any scientific credentials, degrees, diplomas or doctorates, earned or honorary.

As to the PEAR-paper: I stand by what I said. There are plenty of badly-written scientific papers around, alas, but I have never seen one so full of waffle and nonsense as the PEAR meta-analysis paper. I have nothing against meta-analysis itself (in principle), but the PEAR-paper - if you take the trouble of reading the whole thing - comes to this conclusion:

They have no statistically significant results. This is clearly not because the effects they are searching for do not exist. This is due to the evil influence of (insert fringe-woo-woo theory here)

I'm not kidding. That's really what it says. They even provide a list of fringe-woo-woo theories to choose from.

 

[Guest]

Sundog,


----
Pixy is undoubtedly one of the smartest people here.
----


Yes, Pixy is smart and intellient, etc. But me saying that Pixy doesn't understand a paper doesn't mean that I am calling Pixy stupid.


----
You speak as if you have knowledge of what should and shouldn't be in a scientific paper. What are your credentials?
----


4 yr mathematics degree/econ minor BS, 2 yr statistics degree/minor applied statistics MS.

,which allows me to understand the statistics portion of a paper- which includes not only numbers, but experimental design and methodology.

I'm NOT saying the paper was flawless and that the paper is 100 correct, or that it proves the paranormal, etc. I'm asking if Pixy can explain the small p-value other than just "crappy data"-type of "explanations".

 

[Guest]

Drooper,


----
No, asymptotic results hold universally - a result known as the Central Limit Theory is a case in point and something used in studies of this type.
----


If n>= 30, one can rely on the Central Limit Theorem if one is doing inferernce on means (given simple random samples, independence, etc.)


----
Independence in this context should apply readily enough if the experiments in question were conducted properly.
----


So what is your evidence that an experiment done in 1980 is DEPENDENT with an experiment done in 1990?

 

[Guest]

----
They have no statistically significant results.
----


Really?

Did we read the same paper?

I saw a lot of small (some very small) p-values.

 

[juryjone]

Well, the thrust of this thread has certainly changed. But, that's what happens in a free discussion. We now seem to be discussing a PEAR report on remote viewing, which is not the report I referenced in my first post.

I must say that this remote viewing report seems way too subjective for me. Having to use judges is problematic at best; when you get results as quoted by PixyMisa,

It also is curious that these three judges agreed on the correct match of only one of the 31 trials.

then the results are meaningless. A properly structured test would try to eliminate as much subjectivity as possible. The fact that the judges don't agree points out even more strongly the fact that this experiment depends on too much subjectivity.

However, even in this case, the "results" are too small to be practical in everyday life.

 

[Guest]

Juryjone,


----
Having to use judges is problematic at best;
----


Did Pixy mention that mechanical ways of judging were implemented? No, of course that wasn't mentioned...

Maybe Pixy didn't understand that, or didn't find it relevant.


----
A properly structured test would try to eliminate as much subjectivity as possible.
----


Oh, I agree.


----
in this case, the "results" are too small to be practical in everyday life.
----


Which is itself subjective, and frankly irrelevant to the issue of 'are they statistically significant or not, and was this study done well or not'.