General Asylum

RE: There are two cases you need to consider:

70.16.99.162

In Reply to: RE: There are two cases you need to consider: posted by andy19191 on March 30, 2008 at 06:42:22

It's like having a debate with Wikipedia--full of facts but apparently incapable of independent thought. (I have no dog in the cable fight, by the way. I'm just a fan of an open man and against the misapplication of scientific concepts.)

>>If some of the samples are from people that can hear a difference then the distribution will not be random and the hypothesis will not hold.<<

Not sure because I can't quite tell what point you're making, but it appears to be the point I've been making. Every population has individual variations, so every random sample will have individual variations as well, whether it's a drug trial or an audio test. A null result based on a study of a group says nothing about the characteristics of individual members of the group--if it did you wouldn't need statistics or a statistics-based trial.

>>But what is measured directly in typical audibility experiments is sound perception. Sensitivity is completely meaningless.<<

The first sentence is obviously true, the second is nonsense--and would seem to be the point you're misunderstanding. A statistical test is an instrument, a rigorous way of observing reality. There IS a reality, even if you can't see it perfectly. A blunt statistical instrument will fail to perceive small effects, just as a low-power telescope won't see really tiny things. It is well established that the standard ABX methodology is biased towards a null result in the case of small effects. Ask a statistician.

But that's more technical than you really need to get to understand the broader point. An ABX test can be administered to a lot of different people (a sample of a larger population), in which case the results you get are characteristic of the population--and NOT of any individual member of that population. Or you can administer an ABX test to a single individual via a series of trials. But to get a statistically rigorous result, you need to do a larger number of trials (for reasons given, vaguely, in the previous paragraph) than is typically appreciated; fatigue enters, and stress is a real factor. Failing such a test fails to establish perception; it doesn't rule out the possibility of perception. Don't confuse your measurements with reality.

I've got no more time for this. I'm out.

Jim

Follow Ups:

Post a Followup:

FAQ

Post a Message!

Forgot Password?

Moniker (Username):
Password (Optional):
	Remember my Moniker & Password (What's this?) Eat Me
E-Mail (Optional):
Subject:

Message:	(Posts are subject to Content Rules)

Optional Link URL:
Optional Link Title:
Optional Image URL:
Upload Image:
E-mail Replies:	Automagically notify you when someone responds.

To view your new posting or follow-up, click on the RELOAD or REFRESH button on your browser.