Measurement Parameters - How many can be considered all?

[jan.didden]

Jakob said:

"In addition the experimenters must get rid of the assumption that a negative test result is per se evidence for the fact that no detectable difference exists. "

Fully agree, yes it is a point often lost. A null result essentially means that a difference has not been detected. It does not mean it does not exist.
But it is seldom that a test series gives a null result. The usual outcome is that some trials come out correct, and some wrong. It is then a matter of statistics whether you conclude there was a difference of not.
For instance, if you do 20 trials, 13 came out correct, 7 wrong, what do you conclude? I guess you know that better than me Jakob; what is the probability that a difference existed?

Jan

 

[u301021]

Which is a common assumption, but a lot of counterevidence exists from good sensory experiments.
In fact, humans are still humans 'even' in a 'blind' test and therefore it is as easy to get incorrect results from 'blind' tests as it is from 'sighted' tests.

Experimenters need to consider the huge knowledge base existing about design of experiments especially in the sensory field, because we need a lot more of well planned and executed listening tests.

In addition the experimenters must get rid of the assumption that a negative test result is per se evidence for the fact that no detectable difference exists.

You can probably not call it a regular blind test, but if you are sent 2 music files and you cannot hear the difference between them and you subsequently put these files into eg audacity and can not distinguish the 2 files there either.
So what do you think the result is?
You know nothing about the files, the only thing you can do is listen to them and check them in eg audacity.

 

[Evenharmonics]

So what do you think the result is?

Unlike thor2, Jakob is already in audio business and he has been trying to badmouth DBT online for years, for his business sake, of course.

 

[jan.didden]

But Thomas C, that only proofs that no difference can be detected. It does not prove there isn't any.
One of the 'laws' of logic: you can't prove a negative.

Jan

 

[u301021]

But if no difference can be detected and it is not heard either, then it does not matter at all, or ?
It is, after all, improvements or the opposite we are all chasing

 

[Jakob2]

You can probably not call it a regular blind test, but if you are sent 2 music files and you cannot hear the difference between them and you subsequently put these files into eg audacity and can not distinguish the 2 files there either.
So what do you think the result is?
You know nothing about the files, the only thing you can do is listen to them and check them in eg audacity.

The crucial point is, which way to decide that one can't 'hear' a difference.

But my comment was related to the more formal tests tomchr probably had in mind.

 

[jan.didden]

But if no difference can be detected and it is not heard either, then it does not matter at all, or ?
It is, after all, improvements or the opposite we are all chasing

Sure. Except maybe a nagging feeling that perhaps there was something you missed ;-)
But it gets academic from here, and I agree with you.

Jan