Of $90,000 turntables, Stradivarius violins, red wine, and blindfolds
On 2014-04-16 19:39, news wrote:
HOWEVER, it is not sufficient that they tell you whether they can tell a
difference, they must prove it in a statistically significant manner by
doing it at least 16 times, because if they only guessed one time it could
be luck. So, for example, if they can tell the white from the red, or the
fancy from the table wine, at least 12 out of 16 times (95% level of
confidence) then they would have proven that there is a difference, but not
a preference. If they can't even do THAT, then not only could they not have
a preference, but they cannot even tell under blind conditions whether there
is any difference between the two at all.
If I'm not misstaken five trials will suffice; the probability of
guessing correctly five times is 0.5^5 = 0.03125 which is less than five
percent.
-- August
|