Talk:Ganzfeld experiment
From Wikipedia, the free encyclopedia
 |
This article falls under the scope of WikiProject Paranormal, which aims to build a comprehensive and detailed guide to the paranormal on Wikipedia. If you would like to participate, you can edit the attached article, help with current tasks, or visit the project page, where you can join the project and discussions. |
| B | This article has been rated as B-Class on the quality scale. |
I've edited this to weaken the assertion to a claim. Can anyone give a cite for this claim? The Anome 08:01 Feb 28, 2003 (UTC)
- Um, what are the claims? --Brion 04:14 Mar 1, 2003 (UTC)
-
- The claims are that, over a very large number of individual tests (several thousand) the average "hit" rate is 33.2%, when 25% is expected (since the judge has to choose between one valid, and three dummy images). On the face of it, this claim actually seems to be correct, and certainly is statistically significant to a very high degree. However critics have noted quite a few problems which should be given more emphasis in the article, including:
- Most reported experiments seem to be uncontrolled. (Obviously if we repeat the experiment without the "sender" and get the same results, it is a clear sign something is wrong with the set-up);
- The design is overly complicated, making it harder to analyse and detect biases. In fact most of the changes introduced to eliminate "boredom" are changes that make errors easier. A better solution would be to simplify the experiment, but change subjects more often.
- The reliance on matching vague verbal descriptions to the "closest" image is particularly concerning.
- The so-called "auto-ganzfeld" experiments attempt to eliminate the possibility of conscious or unconscious biases on the behalf of the experimenters by computer automation of as much as possible of the process. However this in turn introduces the possibility of software errors - especially bearing in mind that less than a dozen labs have reported these experiments, and we don't know to what extent they shared software. Thus, the software should be audited.
- Securiger 04:39, 9 Dec 2003 (UTC)
- The claims are that, over a very large number of individual tests (several thousand) the average "hit" rate is 33.2%, when 25% is expected (since the judge has to choose between one valid, and three dummy images). On the face of it, this claim actually seems to be correct, and certainly is statistically significant to a very high degree. However critics have noted quite a few problems which should be given more emphasis in the article, including:
