P Floding wrote:
> However, it is meaningless given that we do not have the perfect system
> to test components in.
> I.e your "null hypotheses" is incomplete.

If your null hypothesis is badly written, you are totally correct the 
whole expensive process proves nothing.

But you don't have to have perfection to start. Many (most?) drug tests 
give drugs to sick people, and just want to prove that the drug fixed 
one thing, not that it made them all as healthy as a 21 year old Marine.

And many tests are badly worded, there has been a lot of lawsuits 
because a drug claimed: "This will help you lose weight"
when the null hypothesis should have been "helps people lose weight at 
the cost of an increased chance of heart attacks"

Stereophile and others claim that AB testing just shows that some people 
can detect that something sounds different, which is not very 
interesting. What you want to be able to test is a null hypothesis of:
"a SB3 sounds as good as a  SB3 feeding a Larvy"

If you get 10 people to prove that by selecting the Larvy as better in 
ten out of 15 tests, you have something.

-- 
Pat
http://www.pfarrell.com/music/slimserver/slimsoftware.html

_______________________________________________
audiophiles mailing list
[email protected]
http://lists.slimdevices.com/lists/listinfo/audiophiles

Reply via email to