Do speaker cables make any difference?

[planet10]

Originally posted by rdf
This can't be called science and gives little reason to take the wire tests at face value.

http://www.provide.net/~djcarlst/abx_data.htm

To ad to the other comments, this is an ABX test. These have been shown -- irregardless of the number of trials -- to be statistically invalid because of the high beta.

dave

[Dumbass]

Originally posted by planet10
high beta.

What is high beta?

[fokker]

Originally posted by planet10


To ad to the other comments, this is an ABX test. These have been shown -- irregardless of the number of trials -- to be statistically invalid because of the high beta.

dave

sure. When enough and sufficiently confusing terminology is used, anything will be stataistically invalid to anyone, including those who use them.

a true test of one's understanding of a subject matter is to see if that person can explain what they alledge in plain simple English.

More often than not, they are challenged by their own understanding of that subject matter.

[Dumbass]

Originally posted by macgyver10
the number of listeners on these tests is far too small to be statistically significant

To reject H_0: theta = .5 at alpha = .05 with 140 trials, you would need roughly 80 correct guesses.

There were seven listeners. Therefore, it would take just one listener out of the seven who could consistently guess correctly to reject the null hypothesis of no audible difference.

[macgyver10]

Originally posted by Dumbass
Therefore, it would take just one listener out of the seven who could consistently guess correctly to reject the null hypothesis of no audible difference.

I'd be content with one listener getting it right 80 of out 80....

However, if we're using the listeners essentially as the test equipment, then I'd like to see a large group in order to end up with an "average" human ear response. Anomalies like the listener above could be extracted from the data for further analysis.

Possibly to identify faults with the test procedure itself, even.

[macgyver10]

Originally posted by planet10


To ad to the other comments, this is an ABX test. These have been shown -- irregardless of the number of trials -- to be statistically invalid because of the high beta.

dave

Which is why I suggest a much larger sample of listeners. This beta error is controlled by increasing the number of listeners.

For a description of Beta:

http://wiki.hydrogenaudio.org/index.php?title=ABX

[macgyver10]

Originally posted by fokker



a true test of one's understanding of a subject matter is to see if that person can explain what they alledge in plain simple English.

Actually you'd be best to use the language most suited to the subject matter. That's not always going to be english. For most of electronics and physics it's going to be the language of mathematics that proves your understanding.

However, tests and results are going to actually prove your hypothesis, until then it's essentially an opinion.

[Dumbass]

Originally posted by macgyver10
Which is why I suggest a much larger sample of listeners. This beta error is controlled by increasing the number of listeners.

For a description of Beta:

http://wiki.hydrogenaudio.org/index.php?title=ABX

Statisticians generally refer to what they call "beta error" as a lack of power.

My previous post was an informal analysis of the power of the test. It would only have taken a single person out of the seven (who were presumably "audiophiles") who could tell the difference, to reject the null.

[rdf]

Originally posted by fokker
..a true test of one's understanding of a subject matter is to see if that person can explain what they alledge in plain simple English...

...ideally. Unfortunately perceptual phemomenon is notoriously hard to describe in independent terms. Try describing how 'red' looks without pointing to a red object. Sound is no different. The late Richard Heyser had a great series of editorials in the long-defunct Audio magazine about the need for a coherent language of sound. In the early days the tweak magazines made an honest (or honestly misguided depending on your perspective) effort at it but have since given it up for 'it rocks'.

[macgyver10]

Well, a million dollars is available to anyone who wants to try and prove that you can hear a difference between a $40 pair of cables and the most expensive ones available.

In fact, James Randi just emailed me directly to tell me so!

Anyone who's up for the challenge need only apply here:

http://randi.org/research/index.html

What have you got to lose? Read the instructions carefully, make an application, work out a protocol (that you are involved in designing), pass the test, and collect the cash.

Should be easy.