I don't believe cables make a difference, any input?

[janneman]

You asked a reasonable question, and so far nobody was able to answer it. The significance of the null hypothesis means that it would need just ONE case where a difference was reliable identified under blind conditions, to put the thing to rest. The fact that with your setup it didn't happen is not your problem. Up to them to come up with a setup that DOES demonstrate it, under blind conditions. And really it is not so difficult to make a setup where the tested listener gets no clue to which cable actually is connected. The argument that the test setup (relays, switches, etc) maskes the audibilities actually works against them: that necessarily would mean that source selectors, preamps, connectors, power amp output relays etc make it impossible to hear differences. You can't have your cake and eat it too.

These people make extraordinary claims, but cannnot come up with a simple demonstration to back it up. Until this happens, the logical conclusion is, that their claim is false.

And as far as those mega-posts are concerned, I have always operated on the principle that if you need more than halve a page to explain something, there are two possibilities:

- it isn't true, or
- you don't really understand it yourself.

(I know, I'm getting close).

Jan Didden

[MBK]

Just for the record, I sympathize with Paulinator and I think he has a point.

Originally posted by janneman
The significance of the null hypothesis means that it would need just ONE case where a difference was reliable identified under blind conditions, to put the thing to rest. The fact that with your setup it didn't happen is not your problem. Up to them to come up with a setup that DOES demonstrate it, under blind conditions.

It's not that simple. No one "has to" come up with anything.

Some people claim they hear differences, some people claim they don't. If the null hypothesis is not rejected then we simply know nothing - we must remain undecided.

If a difference can be reliably identified just once, as you said, then yes we know that... in that specific case it made a difference. So, I would also like to see one well documented case where differences were audible.

But in this forum, or elsewhere, I haven't seen any actual data either way. So we don't know.

These people make extraordinary claims, but cannnot come up with a simple demonstration to back it up. Until this happens, the logical conclusion is, that their claim is false.

Nah. Logic has nothing to do with it. Common sense, yes. To me it just means that we have two contradicting hypothesis with no corroborating data either way.

if you need more than halve a page to explain something, there are two possibilities:

- it isn't true, or
- you don't really understand it yourself.

Well. There is a point here.

If you meant my post: it was so long because this forum is so darn sensitive that you have to reword everything 3 times with 5 disclaimers and 7 caveats. As if it even mattered.

[ThorstenL]

Konnichiwa,

Originally posted by janneman
You asked a reasonable question, and so far nobody was able to answer it. The significance of the null hypothesis means that it would need just ONE case where a difference was reliable identified under blind conditions, to put the thing to rest.

This brings me back to the basic problem, namely the statistics.

Many people (especially the ABX Mafia but not only them) have insisted in using the .05 level of significance in small smaple size double blind tests.

I recommend STRONGLY to consider the two following publications:

Les Leventhal: "How Conventional Statistical Analyses Can Prevent Finding Audible Differences In Listening Tests,"
Preprint 2275 (C-9), which had been presented at the 79th AES Convention in New York, October 1985

and

Les Leventhal: "Type 1 and Type 2 Errors in the Statistical Analysis of Listening Tests." JAES June 1986

The upshot is that if you wish to apply a .05 level of significance to your study, if you are studying fairly small audible differences and you wish to make equally certain that you do not incorrectly accept or reject the null hypothesis you need a very large sample size, commonly not provided in Audio ABX testing. The sample size would have to be >> 100 to allow the use of a .05 level of significance.

Thus, if you use such a level of significance with much fewer datapoints all you have done to make the acceptance of the null hypothesis near certain, COMPLETELY DISREGARDLESS OF THE ACTUAL AUDIBILITY.

The ABX test data previously published by the "Southeastern Michigan Woofer and Tweeter Marching Society" has sadly been taken off line (maybe because the published conclusions where open to severe criticism because of bad statistics and because once adjusted to not abuse the statistical methode failed to show the null results desired by the group publishing the results?).

http://www.pcavtech.com/abx/smwtms.htm

This is especially sad as these tests actually provided the raw data and allowed hence the assement of just how likely it was that their null result acceptance was in fact in error, if the difference evaluated was small.

Not surprisingly, equalising the "error budget" for both types of statistical errors leads more often than not to an indication that in their tests in fact a difference was heard in the test, yet due to the small sample size we cannot be particulary certain that it was hear, but such is statistics. At the same time it was clear that their "null" result was not defensible beyond a reasonable doubt. And that covered ONLY the statistics.

Please understand that the staistics issue presents a FUNDAMENTAL challenge and in most cases of published tests an immidiate invalidation of the claim that a "null" result was obtained, as the statistics where (deliberatly or by ignorance) rigged to reject anything short of day/night differences.

So, I will for one last time request Paulinator to provide his DATA instead of his conclusion so that the data cab evaluated using statistics that will not automatically return a "null" result due to ill application of said methode. The same applies of course for other tests.

I will repeat however that NON of the published ABX/DB Tests on the subject of speaker cables (and indeed many other features with debated audibility) provided and used enough data to allow an analysis to a .05 significance level. Hence, any analysis to this level (invariably found) and then claiming to have obtained a "null" result must be considered null and void, as in the most simple terms the risk of not picking up small differences was high (> .5) to very high (>.7 ... >.9) and indeed often high enough to make it an almsot certainty that small differences where missed.

I have raised the same self point repeatedly in this thread. Not once has anyone provided a reasonable rejection, here is the last chance. Unless you can suggests why such misapplied statistics would still give results that should be considered speak now, or I shall consider the fundamental criticism level at the statistics as carried without opposition.

Sayonara

[SY]

an immidiate invalidation of the claim that a "null" result was obtained, as the statistics where (deliberatly or by ignorance) rigged to reject anything short of day/night differences.

The fundamental fallacy lies here. Anyone who understands Jan's comment will appreciate how error budget should be apportioned.

[maynardg]

Some comments.

The domain of A/B testing for small audible differences in cables is one where individual differences likely rule - and many in this thread seem to think it appropriate that experimantal method be (mis)applied.

The true variance within condition (individual differences) is apparently very large relative to the true variance between conditions (any actual audible difference). Add measurement error (within-person judgement errors) - and then huge sample sizes are needed to have confidence that an observed mean difference was not observed by chance.

BTW, one cannot affirm the null, only reject it.

Because of individual differences in true aural acuity, and becuse each person is a measurement instrument, frought with error - A/B testing is probably a futile excersize.

Like nearly all human characteristics, the ability to detect differences is probably somewhat normally distributed.

Some people can detect differences of some magnitude and kind better than others. Why is that so difficult to accept?

In short - who cares about the mean?

I urge everyone not to view one side or the other of this argument as an article of faith. Truth is crystal clear in matters of faith.

Truth in reality is quite fuzzy - esp. in science.

MG

[janneman]

KYW,

I have two comments:

- As noted above, as long as the null is not proven beyond a reasonble doubt I think the phrase is, we are, in a strict scientific formalistic sense "undecided". But what do I do when offered these fantastic 250 $/meter cables (mono, without termination, that's extra) that really change my sound experience? Do I say, hey, I'm undecided, what the heck, I'll buy them? No sir, no me.

- I would be interested in Leventhal's papers if they are available electronic or on hard copy. But common sense tells me that if the results is so sensitive to these statistic manipulations, the case for audibility cannot be very strong. Again, that means that the chances that these great cables DO completely change my sound experience are probably too low to justify the layout. But I am now more practical than scientific, and I don't want to hijack the thread off the scientific track.

Jan Didden

[Christopher]

It is true that you cannot make any sweeping conclusions based on the various informal tests made by any of the contributors. I also see the issue from a practical point of view. How much of my available resources (time, money) should I devote to cables? The answer? From what I've seen here, not much.

I take great exception to statements that make claims of audibility, backed up by scientific principles which do not and cannot apply. Mathematical and theoretical models have a purpose, and in the face of such a question, can be used to narrow your choices so that you can finish the question off with an informal empirical test or two. Instead, distressingly, when the mathematical and theoretical models fail to show any significant differences, they are discarded and discredited as being inadequate, by people with no alternative explanations, and no reasonable data to properly support their objections. The biggest tool in the box of these people is the idea that it is far more complex than simple LCR parameters and impedances. Yet when pressed, not one of these people have ever been able to produce any of these alleged additional complexities. Most often skin effect and Maxwell effect are claimed, both of which are well understood, and known not to apply in the case of audio cables. There are people on this forum that could knock each and every one of these claims off by simply calculating the possible effect of any of these principles. All of them. A to Z. Silver vs copper, crystalline structures , eddy currents, and, as Ian showed, magnetostriciton. As boring as it may seem, the only reasonable conclusion for me is, it really is that simple. I completely accept that certain cables paired with certain systems can have an audible effect where they would not when paired with another system. I also think a quick investigation of this will reveal unreasonable LCR and Z parameters of the offending cable. This is not in any way an indication that things are too complex to be described by theoretical models as some would have you believe. It is an affirmation of the effectivity of these models.

I also take great exception to the way in which cables are portrayed by the Hifi media. Their continuous poetic descriptions of incredible tonality ,rhythm , transparency etc. has been so effective as to convince the masses that cables are of great import in the assembly of a hi end system. Yet on this forum, a group of experienced, and well educated audiohpiles cannot even agree that any audible difference exists, let alone which cable is better. In no other discipline of electrical engineering that I am aware of, would you find someone going beyond TEM to design a cable, yet there are cables with far more critical jobs on this earth than carrying a simple AC signal from amp to speaker.

Chris

Funny thing, in audio if you can hear things others cannot, you are revered, in society, you are sent for treatment.

[MBK]

Originally posted by Christopher
[B]I also see the issue from a practical point of view. How much of my available resources (time, money) should I devote to cables? The answer? From what I've seen here, not much.

I agree on that one

I also think a quick investigation of this will reveal unreasonable LCR and Z parameters of the offending cable. This is not in any way an indication that things are too complex to be described by theoretical models as some would have you believe. It is an affirmation of the effectivity of these models.

I guess picky amplifiers with picky loads, combined with a range of reasonable LCR characteristics, can probably still give different results (right below oscillation or right above)...

... the Hifi media. Their continuous poetic descriptions of incredible tonality ,rhythm , transparency etc. has been so effective as to convince the masses that cables are of great import in the assembly of a hi end system

I really get a good laugh out of most of that. But still - here we have a lot of earnest and passionate reviewers not all of which can be deluded or bought off. I don't really believe most of it, but I keep an open mind - I have been wrong before.

Besides that's the beauty of DIY no? If you DIY you can check out for yourself if solid core vs. stranded really does it for you. The problem lies not in the claims, but in the price people pay to verify them and in their credulity. KYW for once always recommends cheap and easy experiments. I made one of his cables, it didn't make a difference to me. But it cost nothing and I learnt something.

Funny thing, in audio if you can hear things others cannot, you are revered, in society, you are sent for treatment.

Again it's actually beautiful how so many earnest people can be so passionate about such a little thing as home audio. It reminds me of the (not always) gentlemanly fights up to the duel between rival scientists of the late 19th-early 20th century. The language was the same ... he claims ... dishonest ... falsify ... improper method ... not reproducible ... unsound mind ... whatnot. We can all only hope that our SO's never discuss what's going on late night on these threads.

Anyway, hearing seems to be surprisingly sensitive. At one point I read about an attempt to code unknown DNA data into sound because it was easier to hear possible patterns in seemingly random DNA than to attemt decoding it with algorithms.

[ThorstenL]

Konnichiwa,

Originally posted by SY

The fundamental fallacy lies here. Anyone who understands Jan's comment will appreciate how error budget should be apportioned.

How should it be apportioned, if not to give an even weighting to both types of errors? Unless of course, you are not interested in the truth, but simply wish to manipulate the statistical methode to provide something that superficially could be mistaken as proof for a untennable position?

Sayonara

[ThorstenL]

Konnichiwa,

Originally posted by janneman


- As noted above, as long as the null is not proven beyond a reasonble doubt I think the phrase is, we are, in a strict scientific formalistic sense "undecided". But what do I do when offered these fantastic 250 $/meter cables (mono, without termination, that's extra) that really change my sound experience? Do I say, hey, I'm undecided, what the heck, I'll buy them? No sir, no me.

If for you these cables make a material difference you need to ask - is the difference of a type and degree that makes the difference worth $ 500 to me?

I have not yet found any commercial cables I would ever consider owning (excluding all the really ultra-expensive stuff) which where capable to comprehensively outperform my "UBYTE" steries of homemade cables (note, the UBYTE Moinker stands for "Usually Beats Your Terrible Engineering" and was coined by the alter ego of Jocko Homo). Hence I have so far not found a reason to spend such money. But that is my own, personal counsel which I usually keep, in preference to limping proping up myself on other peoples opinions as crutches.

Originally posted by janneman

I would be interested in Leventhal's papers if they are available electronic or on hard copy.

As the AES Preprints and JAES articles are available from the AES - please pay their ransom and buy the copies.

Originally posted by janneman

But common sense tells me that if the results is so sensitive to these statistic manipulations, the case for audibility cannot be very strong.

Hmmm. May I ask again, have you ever actually "double blind" tested ABX testing?

Take a group of people with strong opinions on a given subject, tell them they will participate in an ABX test upon that subject, but instead test a different variable, one whose audibility is under no abiguity. Tabulate the results. You will find yourself very surprised. Even more so if you alter the significance levels (or indeed choose to use more sophisticated statistical methodes).

Originally posted by janneman

Again, that means that the chances that these great cables DO completely change my sound experience are probably too low to justify the layout.

I am perfectly in agreement. Most currently marketed "High End" audio cables actually are badly designed simply from a viewpoint of the fundamental parameters, the issues of interconnecting multiple pieces of mains powered gear etc.... Many cables actually DO change the sound, simply by being "tonecontrols".

However, I made the point before and will make it again. We have two COMPLETELY SEPERATE issues here and mixing one into the other is at the very best bad logic, at the worste a deliberate muddeling of the issues.

The Issues:

1) Do/Can speaker and interconnect cables make an audible difference?

2) Is a given cable marketed as "High End" cable good value for the money asked?

I notice that whenever the "Anti Cable League" runs out of arguments to defend the "null" position they something like:

"But what do I do when offered these fantastic 250 $/meter cables (mono, without termination, that's extra)"

You see, you are now DELIBERATLY mixing the issues. Given that you cannot, beyond a resonable doubt support the inaudibility of cables you switch to the "protect the dumb consumer form people after their money" line.

This has been a feature of the whole argument since the late 1970's. One would have thought the Anti Cable League could have come up with something else than rigged pseudo scientific tests and the endless argument about money, assuming they actually had a case, other than "sour grapes".

So, please once and for all either point out what bearing the cost of a given marketed cable has on the issue if certain differences in cable construction make an audible difference or not?

By muddeling the issues the way it always happens nobody benefits more than those people marketing overpriced and underengineered audio cables at a high premium. As long as there is a wholesale denial of any possible differences when enough people hear them at least to their own satisfaction you only leave a wideopen market without checks and balances whithin which ANYTHING can be claimed and anything sold.

I suspect any cable maker reading this thread is just sitting there and chuckeling to himself. Your side of the argument has prevented any accountability in the market for > 20 Years and made it what it is, in my view anyway. Without so much extremely noisy and long drawn out argument cables had never gotten the press they have and a more level view on the subject would have possibly prevailed....

Sayonara