winterkoninkje: shadowcrane (clean) (Default)

Hat tip to homasse:

Experimental evidence for precognitive ability

The paper, due to appear in the Journal of Personality and Social Psychology before the end of the year, is the culmination of eight years' work by Daryl Bem of Cornell University in Ithaca, New York. "I purposely waited until I thought there was a critical mass that wasn't a statistical fluke," he says.

It describes a series of experiments involving more than 1000 student volunteers. In most of the tests, Bem took well-studied psychological phenomena and simply reversed the sequence, so that the event generally interpreted as the cause happened after the tested behaviour rather than before it.

The alternative interpretation of this work is that the interpretation of previous studies as causation (e.g., typing a random selection of words causes improved recall of those words in a second task) is flawed. Technically all that has been found is that there's a typing-then-remembering correlation (in previous studies) and a remembering-then-typing correlation (in this study). Correlation does not imply causation; so even though the causative story is extremely plausible in the old studies and no non-precognitive causative story seems plausible in the current study, that doesn't necessarily mean that either causative story is correct.

For all the hard scientifical review the preprint is receiving, I'm surprised that this point hasn't been mentioned anywhere. IMO, refuting the standard causative story would be just as fascinating as supporting the precognitive story. It would imply that there are macro-scale quantum effects (e.g., where remembering words from a list and typing random words from that list are coupled), whereas it has long been believed that quantum effects only arise at scales too small to be detected by unaided human perception. If there really are macro-scale quantum effects, it seriously fucks with both physics and philosophy (to say nothing of psychology). Whichever way it turns out, it's still awesome.

Edit (2011.06.29): Here is an excellent examination of a lot of the debate this article caused.

Date: 2010-11-13 01:09 am (UTC)From: [personal profile] tim
tim: "System Status: Degraded" (degraded)
From this dude's home page:
"Are Self-identified Bisexuals Just Lying to Us—or to Themselves?"

0_o

(The contents of the link are just as 0_o as the title would imply.)

Also, I haven't read this carefully, but it looks like some people have already called shenanigans:
http://dl.dropbox.com/u/1018886/Bem6.pdf
Edited Date: 2010-11-13 01:14 am (UTC)

Date: 2010-11-13 05:15 pm (UTC)From: [personal profile] tim
tim: Tim with short hair, smiling, wearing a black jacket over a white T-shirt (Default)
I didn't find it reasonable. After all, he basically starts by agreeing with the premise that bisexuals can't be trusted to identify their own sexual orientation accurately. His assumptions were highly suspect: it only makes sense to draw his conclusion that "identifying oneself as bisexual is a straight or gay person's way of reconciling their attraction to one person who isn't their preferred gender" [I'm paraphrasing] if you start by assuming that "straight" and "gay" are the only real sexual orientations. His reasoning was baroque and it's very strange that he chose to link to this particular post on his home page. I've been in a lot of Internet arguments, and there were some where I even thought I wrote something clear and insightful, but I don't link to any of them on my professional home page.
Edited Date: 2010-11-13 05:15 pm (UTC)

Date: 2010-11-13 01:46 am (UTC)From: [personal profile] lederhosen
lederhosen: (Default)
I'm a bit dubious of the stats here:

"I purposely waited until I thought there was a critical mass that wasn't a statistical fluke," he says.

That bit there is a red flag to me. The decision about when to stop an experiment should NOT be based on the significance figures for results to date, otherwise you're at great risk of cherry-picking.

If I run a random number generator and keep rerunning the p-value, I will see that drift all over the place. Every so often, it'll drift into 'significant' territory; as the trial goes on that drift will slow, but wait long enough and you can meet any significance threshold you like.

If you have a few hundred colleagues running similar experiments, all holding off on publishing until they get a strong finding, you can get some *really* impressive-looking results.

Date: 2010-11-13 02:13 am (UTC)From: [personal profile] tim
tim: Tim with short hair, smiling, wearing a black jacket over a white T-shirt (Default)
...which would certainly explain why no one else has been able to reproduce his results.

Date: 2010-11-13 02:28 am (UTC)From: [personal profile] lederhosen
lederhosen: (Default)
Now I check the link you posted, it covers this and some other cherry-picking issues really well.

April 2019

S M T W T F S
 123456
78910111213
14151617181920
212223242526 27
282930    

Tags

Page generated 7 Jun 2025 01:43 am
Powered by Dreamwidth Studios