hirsch;159062 Wrote: 
> 
> DBX is useful for helping to interpret a positive result.  Nothing
> more.  *The absence of a positive result does not imply a negative
> result.* This cannot be overemphasized. The problem with DBX in audio
> is that people don't know how to interpret the results.

I really don't see what you're getting at.  Why do we need any help
interpreting negative results?  If you can't hear a difference in a
non-blind test most people would have little difficulty in deciding
what that means - it means that there any difference is too small to be
perceptible to them, and therefore they're not going to blow $1000 on
that component.  The *only* difficulty, and the source of all the
arguments on this and many other forums, is in interpreting positive
results from non-blind tests.

Suppose you think there might be a perceptible difference between two
components.  So you do a blind ABX test and try to distinguish between
the two components.  You identify X as A or B correctly n out of 10
times, say.

Now, in statistics 101 you learned about "rejecting the null
hypothesis."  That means you start from the hypothesis that there is no
difference, and you see with what confidence you can reject it given the
results of your test.  So you simply need to ask, what is the
probability p that you would randomly happen to be correct n out of 10
times if there is, in fact, no difference.  Then with confidence 1-p
you can reject the null hypothesis.  Usually we require 1-p > .95 (p <
.05) to say the result is significant, although of course that's a
convention.

In the example above you'd need to identify correctly 8/10 times to
reject the null hypothesis with 94.5% confidence, so that would be
almost but not quite at the conventional level of significance 
(I'm assuming a result of 8 or more *wrong* answers would not be
regarded as a positive result - otherwise it's only 89% confidence). 

Now, if instead you are correct, say, 5/10 times, does that mean you
can conclude there is no difference at all?  Of course not - it might
simply mean the difference isn't audible to you...  which is exactly
what you wanted to know in the first place.


-- 
opaqueice
------------------------------------------------------------------------
opaqueice's Profile: http://forums.slimdevices.com/member.php?userid=4234
View this thread: http://forums.slimdevices.com/showthread.php?t=29912

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

Reply via email to