On Apr 8, 6:56 pm, Graham Cooper <grahamcoop...@gmail.com> wrote: > On Apr 9, 4:09 am, BruceS <bruce...@hotmail.com> wrote: > > > On Apr 7, 10:46 pm, Graham Cooper <grahamcoop...@gmail.com> wrote: > > > > > > how come this random noise doesn't affect Zenner Card tests guessing 1 > > > > > > card from 5 options? > > > > You beat me to it, Brad. That's pretty much it, Graham. The one in > > > > five is *very* subject to random noise, whether it's one guess from > > > WHAT ODDS PER TRIAL DO YOU ACCEPT EXACTLY? > > I've told you repeatedly that I'd be glad to support a test using the > > 1:50 odds you kept bragging you could do repeatedly. With those odds > > WOAH let's stop right there. > > Haven't I repeatedly told you that my paranormal bias is approx. a > CONSTANT +50% over the expected score. > > e.g. HALF RIGHT FROM 3 OPTIONS > > 1.5/50 from 50 options. > > A 3 options test would take 20 trials to be significant > A 50 option test would take 2000 trials to be significant > > I've told you 20 times that the consistency drops with more options. > > And I've told you 20 times the 100 number options was an illustrative > protocol example. > > Why do you keep ignoring this Bruce? > > There's going to be 20 or more trials to break 1000:1 preliminary odds > no matter how you do it. > > However, if you can MAXIMIZE THE CONSISTENCY then that MINIMISES the > number of trials needed. > > So basically we should be using a "comfortable" range for the number > of options. > i.e. the range in which the subjective positives are most easily > distinguished from the subjective negatives. > > So the 'correlation' of any random phrase to 'match' any random 'word' > seems to be about 1 in 5.