News for hackers. Stuff that matters.
Re:Is Precognition Real? by tokenadult on Saturday November 06, 2010 @09:51PM (#1878366)
Earlier and more varied discussion here:

http://forums.randi.org/showthread.php?t=188366

On an email list of local psychologists, one psychologist pointed out that the showings of different stimuli were almost certainly not random in the way that Bernoulli trials would have to be for the calculation of statistical significance to be correct. I'll try to paraphrase some of his words below:

We have to find out whether or not Bem was really using Bernoulli trials. Did he allow for cases comparable to having an experimental subject guess heads versus tails in a sequence of twelve coin tosses, when heads might come up twelve times in a row? Did he also allow stimuli to alternate back and forth, like a coin toss sequence of H T H T H T H T H T H T? Many times when experimenters try to make sure sequence LOOKS random, it actually makes a sequence more predictable by excluding some possibilities. This always makes the sequence more predictable.

If an experimenter shows subjects sequences that are meant to look random, but for every subject there are just as many "off" states as "on" states, in twelve trials, then there are fewer possible sequences. With Bernoulli trials, there would be 2^12 = 4096 possible sequences of 12 trials, but with "balanced" choices of stimuli there would be only 12-choose-6 = 942. Experimenter-designed sequences can also create dependencies so that knowing any part of the sequence helps to predict the rest of it. A subject who guessed that over twelve trials in a sequence there are equal numbers of trials in each of two states would be right 64.3 percent of the time, on average, not 50 percent.

The psychologist went on to note that such studies are very hard to design properly, but so far information isn't available on how the study was done to that level of detail. He thinks it is more likely than not that the experimenter gave his subjects clues to the structure of the sequences of trials he set up, and that he set up conditions for which performance above the 50 percent accuracy level is expected and not remarkable.

Re:Is Precognition Real? by Fargren on Saturday November 06, 2010 @11:40PM (#1878578)
From the article : "The sequencing of the pictures on these trials was randomly determined by a randomizing algorithm … and their left/right target positions were determined by an Araneus Alea I hardware-based random number generator."

At the very least they were using Araneus Alea wich is a hardware random number generator, so the numbers were not predictable. It's possible that the "randomizing algorithm" did something dumb and made the sequence not random, but I doubt it. I'm downloading the replication package to see if it has the source code. If it doesn't, that's a pretty big red flag for me.

I think that it's more likely thet the sudy was done so many times that it eventually gave significant results than it is that the sequence was not random. Or maybe prescience is real to some degree, or the study is a statistical glitch.

EDIT: the replication package doesn't have the source code for the progrma that randomizes the sequence.

Re:Is Precognition Real? by tokenadult on Sunday November 07, 2010 @12:12AM (#1878639)
Thank you for pointing to that statement from the article. I'm still pondering what scope that statement leaves for imbalance in the number of pictures of one kind or another during the trails with each experimental subject.