PDA

View Full Version : About Statistics and stuff...



darin
11-30-2007, 09:53 AM
Fantastic powerpoint presentation relating to, and sourced from, my link in this thread: http://www.debatepolicy.com/showthread.php?p=161831





It is an incontestable fact that most negative or neutral results are not published. Editors do not like them and readers find them boring. Thus the published results are mostly winners in the one in twenty lottery; i.e. purely accidental.
Worse, there is now political censorship. Editors of once great journals, such as Nature, jump through hoops in order to prevent the publication of critiques of establishment dogma.


http://www.numberwatch.co.uk/Stuff.ppt

gabosaurus
12-04-2007, 03:45 PM
You know what they say:
There are lies, damn lies and statistics.

Hobbit
12-04-2007, 09:31 PM
I've taken statistics classes before, and while unethical, it is PAINFULLY easy to take a statistical analysis and alter the level of significance (accuracy) to change the conclusion. In fact, there's a value called the P-value, which is the value of the accuracy at which your conclusion changes.