Themis;373827 Wrote: > > When certain variables just can't be verified, you have to eliminate > them one after another. For instance, in this coffee problem, I would > start with measuring the audience's ability of choosing a coffee. Just > like Pat did with his magic switch : serve the same coffee and > eliminate all audience which finds a difference. Or (if you don't want > to eliminate them) at least you have to take into account that -for > instance- 30% of them give inconsistent answers. So, we have now one > more variable fixed. Because the assumptions of 0%, 50% or 100% of > consistency or inconsistency they are all wrong.
You're still completely missing the point. You managed to find an example in which the results were 50-50 - but it was an example in which the coffee machines' performances were identical on average. It's really very, very simple - in any case in which the machines' average performances differ, the experiment will yield a non-null result -no matter how inconsistent the subject is- (except in the extreme case that the subject has zero ability to distinguish - but that's precisely what we are trying to find out anyway). That's it. That's all that was ever claimed and all that is necessary. If there is a difference and the subjects can distinguish it at least sometimes, then with enough data there will be a significant result. If (with enough data) there is no significant result, then there is no difference distinguishable to the subjects. End of story. -- opaqueice ------------------------------------------------------------------------ opaqueice's Profile: http://forums.slimdevices.com/member.php?userid=4234 View this thread: http://forums.slimdevices.com/showthread.php?t=56712 _______________________________________________ audiophiles mailing list [email protected] http://lists.slimdevices.com/lists/listinfo/audiophiles
