Schwartz's latest gem

[Paul C. Anagnostopoulos]

Have we discussed Gary Schwartz's latest mediumship extravaganza yet?

http://veritas.arizona.edu/papers/Beischel%20EXPLORE%202007%20vol%203.pdf

This study involves proxy sitters to get around the objection that the mediums interact with the sitters. The proxy sitters were "... experimenter blinded to the identity of the sitters ..." I wonder who those experimenters were?

Also, check out how the sitters were paired up. And check out the AWRRS numerical score #3.

~~ Paul

[CFLarsen]

How much is "enough" information to establish necromancy?

He has 7 scores, but only 1 without any evidence of mediumship. That means that if the sitter can in any way massage the babble to mean "communication", mediumship exists. It is only in the cases where the psychic completely fails to make any kind of hits that there is no "communication". And that is virtually impossible, given the incredible slack the mediums get.

The rating system is designed to prove mediumship.

Gary Schwartz is a fraud. And you can quote me on that, baby.

[skeptic of oz]

I can't believe that there are people susceptible to all this nonsense BUT:
Obviously there are otherwise there wouldn't be a need for organisations such as the JREF!:D

[skeptic of oz]

It's disheartening sometimes that no matter how much people like Uri Geller are debunked there are always a new generation of charlatans coming along...
Ahh well. Such is the job of a skeptic.

[Paul C. Anagnostopoulos]

Am I misreading the paper, or did each sitter get only two readings to rate, one that was his and the other one for his paired sitter? Note that if his deceased was a parent then his paired sitter's deceased was a friend, and vice versa.

~~ Paul

[Soapy Sam]

I read the "Methodology" section of the precis four times.
Am I missing something, or is this utterly silly?

If this protocol achieves statistically significant results, why do they need a sitter at all, "real" or "proxy" -?

[Paul C. Anagnostopoulos]

If this protocol achieves statistically significant results, why do they need a sitter at all, "real" or "proxy" -?
Presumably the medium somehow connects with the sitter's deceased person by knowing the deceased's first name. The proxy sitter is merely a formality to make the medium more comfortable.

The amusing thing is where they describe how unlikely it is that the results could be explained by super-psi because the medium would somehow have to locate the unidentified sitter is psi-space. They don't seem to mind that the medium has to locate the dead person or vice versa.

~~ Paul

[Paul C. Anagnostopoulos]

From the paper:

Each reading was transcribed and a corresponding numbered list of individual items (i.e., separate, stand-alone pieces of information) was created by an experimenter blind to details about the sitters or discarnates. Each sitter in a pair acted as a matched control for the other sitter in the pair ... In each pair, one sitter was given his/her intended reading to score first, and one was given the control reading to score first.
I presume that one job of the transcriber was to remove any mention of the discarnate's first name, since that was known by the medium and proxy sitter.

But who handed the readings to the sitters? Did that person know which reading was the sitter's and which was the control? I've asked Julie Beischel these questions.

~~ Paul

[fls]

What am I missing? What are you complaining about?

Yes, the scoring system encourages 'hits', but that would only explain why the scores are high overall, not why the scores are different between the intended reading and the control reading.

Linda

[Ersby]

I'm a little surprised that in the 0-6 scoring system for the readings, a score as low as 3 is specifically described as indicating that communication with the deceased has occurred.

[Paul C. Anagnostopoulos]

I think the protocol would be improved by selecting sitters from somewhere other than the researchers' university. Four of the sitters described a deceased "peer" and the proxy sitters, who are researchers, knew the peer's first name. I wonder if they recognized any of the names?

The whole scoring thing is silly. The only thing that matters is how many times the sitters selected the correct reading. That happened 13 out of 16 times.

~~ Paul

[Paul C. Anagnostopoulos]

Statisticians: The paper says that 13 out of 16 correctly chosen readings has a p=0.01 using a one-tailed exact binomial test. Is that correct?

~~ Paul

[CFLarsen]

I'm a little surprised that in the 0-6 scoring system for the readings, a score as low as 3 is specifically described as indicating that communication with the deceased has occurred.

No, no:

2: Some correct information, but not enough to suggest beyond chance that communication occurred.

Some correct information. Ergo: Communication with the deceased has occurred.

1: Little correct information or communication.

But still some correct information. Ergo: Communication with the deceased has occurred.

0: No correct information or communication.

Oops. No communication with the deceased.

In 6 out of 7 scores, communication with the deceased has occurred.

Schwartz has rigged the scoring so it is virtually impossible not to show that psychics talk to dead people.

[Ixion]

I am in Tucson, AZ, at the University of Arizona. I wonder if there is a way to be able to "participate" in the research, just to see it for myself firsthand. It seems silly enough on paper. I probably wouldn't be allowed to be a research subject though, since they probably prescreen individuals and weed out the skeptics.

[fls]

Schwartz has rigged the scoring so it is virtually impossible not to show that psychics talk to dead people.

Schwartz doesn't seem to claim the results are significant because of the absolute value of the numerical scores. He claims they are significant because the intended readings ended up with relatively higher scores, and because the sitters accurately picked out the intended readings from the control readings.

How you explain differences in scoring between the intended readings and the control readings, considering that the scorers didn't know in advance which was which?

Linda

[polomontana]

I just downloaded the study but I have a few preliminary questions based on the conversation taking place.

Was there a control group?

Did this control group score less than the psychics?

If so, how could he rig the test?

[polomontana]

I see now, and I think Linda's question stands.

How you explain differences in scoring between the intended readings and the control readings, considering that the scorers didn't know in advance which was which?

[CFLarsen]

How you explain differences in scoring between the intended readings and the control readings, considering that the scorers didn't know in advance which was which?

There aren't a lot of readings, so a few lucky ones will skew the result tremendously.

Read this closely:

It is noteworthy that three mediums produced dramatic findings in terms of the meaning of summary scores of 5.0 and 5.5 (see Methods section); two mediums produced moderate findings (summary scores of 3.5); and none of the mediums produced reversals (i.e., control ratings higher than intended ratings).

You see what happened? The mediums don't score overall better than average. You have three who got lucky, and the result of those make up for the very average performance of the rest of the mediums.

Remember that Schwartz said in the beginning:

Eight mediums who had previously demonstrated an ability to report accurate information in a laboratory setting performed the readings.

Yet, five of those mediums performed about or close to average (with a small number of trials).

[polomontana]

Three got lucky?

Give me a break, that's your bias on the underlying subject of the study.

Three mediums had dramatic findings and two mediums produced moderate findings.

You ASSUME that three got lucky because your pre-existing belief about mediums.

Three got lucky is not an answer. What is your proof that three got lucky?

[fls]

There aren't a lot of readings, so a few lucky ones will skew the result tremendously.

Isn't that the point of significance testing - figuring out how (un)likely you are to find a lucky few?

Linda

[Gord_in_Toronto]

Three got lucky?

Give me a break, that's your bias on the underlying subject of the study.

Three mediums had dramatic findings and two mediums produced moderate findings.

You ASSUME that three got lucky because your pre-existing belief about mediums.

Three got lucky is not an answer. What is your proof that three got lucky?

Har de har har! JB Rhine built pretty much his entire career on the same principle. He kept on finding the scores of those readers who gave the best results tended to the mean the more they were tested. Surprise. Surprise. :boggled:

[Paul C. Anagnostopoulos]

How you explain differences in scoring between the intended readings and the control readings, considering that the scorers didn't know in advance which was which?
Yes, that is the question. I have a problem with the fact that the proxy sitters where experimenters. I also have questioned the authors about who handed the readings to the sitters for evaluation, since they had to know which readings were the correct ones. As always, I wonder if the statistics were done correctly. Which measure of effect size did they use? I think an effect size of 0.5 is considered a medium effect (so to speak).

I've asked for the raw data, too. Why do they keep using the average of the readings scored by the sitters, rather than showing us all the data?

~~ Paul

[Beth]

Statisticians: The paper says that 13 out of 16 correctly chosen readings has a p=0.01 using a one-tailed exact binomial test. Is that correct?

~~ Paul

Assuming the probability of success by random chance alone, I get p=0.0106

P(X >= 13) = 0.0106 given n = 16 and p = .5,

[fls]

Yes, that is the question. I have a problem with the fact that the proxy sitters where experimenters. I also have questioned the authors about who handed the readings to the sitters for evaluation, since they had to know which readings were the correct ones. As always, I wonder if the statistics were done correctly. Which measure of effect size did they use? I think an effect size of 0.5 is considered a medium effect (so to speak).

I've asked for the raw data, too. Why do they keep using the average of the readings scored by the sitters, rather than showing us all the data?

~~ Paul

If you're asking for raw data, see if you can get the names of the discarnate, as well. I imagine a 'Kaylee' gets a different reading than a 'Gladys'. Same thing for a 'Kumar' and a 'Matthew'.

I can't tell how they got an effect size of 0.50. Maybe they based it on the population mean?

Linda