Veil of Fire

Someone pointed me to a book by Dean Radin, called "Conscious Universe: The Truth of Psychic Phenomena." Apparently he uses actual scientific research methods and comes to some odd conclusions on psychic stuff.

I asked around, and came up with a few articles about it...

<a href="http://mail.cruzio.com/~quanta/review.html" target="_blank">I.J. Good's Review in 'Nature'</a>

Various Replies to Such:
<a href="http://www.tcm.phy.cam.ac.uk/~bdj10/psi/doubtsregood.html" target="_blank">http://www.tcm.phy.cam.ac.uk/~bdj10/psi/doubtsregood.html</a>
<a href="http://members.cruzio.com/~quanta/badgood.html" target="_blank">http://members.cruzio.com/~quanta/badgood.html</a>
<a href="http://www.tcm.phy.cam.ac.uk/~bdj10/psi/rossman.html" target="_blank">http://www.tcm.phy.cam.ac.uk/~bdj10/psi/rossman.html</a>

Opinions?

Veil of Fire

No one has an opinion?

butswana

I am of the opinion that it is all bullshit.

doubtingt

As Good points out, Radins arguments rest heavily on the meta-analysis, and meta-analyses are frought with obstacles. First, many published studies in psychology do not provide enough information for you to actually examine the effect size. Unfortunately, psychology has gone down the misguided path of placing undue faith in the utility of Null hypothesis significance testing and its associated p-values. Therefore, these meta analyses are prohibited from doing what they should, which is to combine all of the actual observations (raw data) from all the studies into a single data set with a standardized metric and conduct a statistical analysis on this data.
A related issue is that many journal editors hold the unfounded belief that null findings are uninformative, thus even when a null result is published no information is given about the actual data other than "non-significant".

In addition, the file drawer problem is more than just and issue of journal refusing to publish null findings. Studies have shown that a null result is far less likely to even be submitted by a researcher to be reviewed for publication.
Also, it is not only null findings that get lost, it is every experiment where the subjects guesses were significantly worse than chance. When these "anti-esp" findings are taken into account, far fewer "file-drawer" studies would be needed to balance (and cancel out) the pro esp findings that make it into publication.

Meta-analsis has its place, but given these inherent almost insurmountable limitations, I do not see how meta-analysis techniques can ever be used to establish the existence of a phenomenon that has otherwise avoided empirical verification.

My argument is that a controlled study with sufficient statistical power that reveals positve results in a series of replications is the only way that we can rule out random and systematic errors in measurement and sampling as alternative explanations for our observed data.