[radrd]

I'm not sure I understand what all the controversy is about, but here's what I would do:

The hardware: amp with three inputs, source with dual outputs. My source's outputs are wired in parallel, so they are constantly outputting identical signals. My amp only requires the turn of a dial to move to the second and third inputs. You hook interconnect A up from output 1 to input 1 and interconnect B from output 3 up to input 3. Input 2 is silent (this will be important).

The administrator: you get someone who doesn't have the slightest idea what is going on to rotate the dial between the three inputs. A mom would be good at this because they are generally happy to help their sons/daughters with stuff, no matter how ridiculous it seems.

The listener: You *could* get someone who does not have any idea what is going on to listen. That person would sit down and listen to a CD being played, one that he or she is not familiar with. The listener would face away from the administrator. The listener could listen as long as he or she wanted and then tell the administrator to turn the dial. The administrator would have to rotate to the silent input, then could either rotate back to the first input or rotate to the third input. The listener would report whether they detected a difference after the change, and they would know the change had been completed because of that silent space, so there would be zero communication from the administrator to the listener.

If it were me, I would let the listener choose a CD that he or she were familiar with, and I would let the listener know exactly what is being tested.

The administration assistant: This would be yet another person who does not know what is going on. They would be instructed to be silent and record the listener's answers and which position the dial was at when the listener gave their response.

So, Rodbac, does my proposal work? Any questions?

 

[Orpheus]

come on guys...

we all know i ain't a believer. but i'm willing to conduct a fair and as unbiased experiment as i can muster.

rodbac, i think Hirsch knows what he's talking about. and you probably have intelligence too. so, instead of going after each other... try to limit your responses to actual facts. if you can quote directly from your statistics book, do so. it'll impress people much more.

ctn, do you think all these "believers" are incapable of accepting results from a well-designed experiment? we shouldn't be hypocrites--believers shouldn't be any more logical than "non-believers" just like non-Christians ain't any better at science than Christians. am i right?

i believe we are all capable of understanding the results, when they come.

let's continue designing this experiment.

 

[Edwood]

Iron_Dreamer and I tried "Single" Blind Tests. Where the listener is blind, but the tester is not.

The best way to test is to be able to switch back and forth between cables at the flip of a switch. And have both volumes the same. Getting a true A/B test is really difficult.

A nicely made custom switch box would probably do the trick. I'm sure you have something you could put together with all of your pro audio equipment.

This way the source, amp, and headphones are identical, and you can switch back and forth rapidly.

Also, using the same passage of songs, perhaps even custom edited together snippets of songs put back to back.

The key is to have control in the experiment. The more time that passes between cable changes, and the longer you rely on your memory, rather than direct comparison.

Also a randomizing of switching back and forth, maybe the switching back and forth can be done by a computer? That way the tester does not know and it would be true double blind. Better yet, if it is automated by a computer, then you wouldn't even need a tester, you could do it by yourself.

Just some thoughts. Probably a lot of up front work, but you did bring it up, Orpheus. But once this system is created it could be an invaluable Cable Test Rig.

-Ed

 

[radrd]

EDWOOD

If you read my previous post, you would see that I described exactly what you are talking about. Furthermore, I already own such a rig.

 

[rodbac]

Quote:

Do a mental experiment. Flip a coin ten times in sequence. It comes up heads all ten times. Compute the probability that this occurs by chance. Statistical analysis will show that you've got a weighted coin. Are you going to discount this probability because a coin flip is a two-choice event?

Statistical analysis will show precisely dick. Statistics 101 (which you surely took) will tell you that the chances of a coin flip are ALWAYS 50-50, so the chances of it happening 10 times in a row are no more impressive (and if you don't believe me, call a statistician and verify it).

However, let's use that example and expand it just a bit: people are claiming the can predict heads or tails. So my experiment involves them sitting down and trying to predict 10 coin flips. Would you really be that surprised when, out of 100 or 1000 trials, a few people got 10 in a row? Do you think that would reasonably prove people were capable of predicting heads or tails?

radrd- your method sounds acceptable at first glance, so I'm trying to think of pitfalls.

First, we'll need to verify that the signals coming out the other end of each channel of the amp are identical with the same IC.

Next, verify there is absolutely no feedback to the listener with the switch.

Also (IMPORTANT), make control runs with each listener first, telling them exactly which IC they're listening to. Only if they report a dif do you need to perform the double-blind portion.

Orpheus, thanks for the reality check, but I think I have stuck to facts. I don't mean to be attacking Hirsch, though, so I apologize to both you and him if I've been too aggressive in that regard. It doesn't change the preposterous nature of the proposed test as described, though, no matter how many trials were performed.

 

[Wodgy]

Quote:

Originally Posted by rodbac Statistical analysis will show precisely dick. Statistics 101 (which you surely took) will tell you that the chances of a coin flip are ALWAYS 50-50, so the chances of it happening 10 times in a row are no more impressive (and if you don't believe me, call a statistician and verify it).

Before everyone piles on you and points out that this statement is false, it's clear you actually learned something in Stats, which I applaud you for (most people have never acquired a basic understanding of statistics). You're not correct, though your error is subtle. It is true that any given sequence of flip results is equally likely (including 10 heads in a row). However, the chance of this specific sequence out of all possible sequences is 1 in 1024, not 1 in 2.

 

[Hirsch]

I've cleaned the thread of most of the dead wood.


Quote:

Originally Posted by rodbac Statistical analysis will show precisely dick. Statistics 101 (which you surely took) will tell you that the chances of a coin flip are ALWAYS 50-50, so the chances of it happening 10 times in a row are no more impressive (and if you don't believe me, call a statistician and verify it).

Statistical analysis will show that the odds of heads on one coin flip are 50% (a probability of .5). The probability of heads on two successive coin flips are .5 x .5, or 25%. The probability of heads on ten successive coin flips is .5 to the tenth power. Multiply it out, if the concept is tough. This is very small probability. You'll pass the .05 level, "statistical significance", around the fifth coin flip, and still have five to go. If you get ten heads in a row, you've got a weighted coin by any statistical analysis. The odds of the sequence occurring by chance are infinitesimal.

If there are mixed trials, say seven heads and three tails, you'll have several sequences where this result is possible. The probability of this result is the total number of possible sequences with this combination of results divided by the total number of possible sequences. This applies to any number of trials you care to run. The larger the number of trials, the more possible sequences that will have ten in a row. No big deal in that kind of probability calculation whatsoever.

EDIT: As Wodgy has pointed out, the probability of any one specific sequence is the same as any other. However, in the ten trial example, only one of the possible sequences has ten heads. If there are nine heads and one tail, you'll have ten possible sequences where this occurs, so the probability is ten times greater of this result than the ten head sequence, although the odds of any one of the ten possible sequences with exactly one tail is the same as the odds for the ten head sequence.

If the above statements don't make sense to you, stop now. You've got to understand elementary probability before you can understand any form of statistical analysis.

I used to teach statistics and experimental design, long ago.

 

[Ctn]

Quote:

Originally Posted by Orpheus ctn, do you think all these "believers" are incapable of accepting results from a well-designed experiment? we shouldn't be hypocrites--believers shouldn't be any more logical than "non-believers" just like non-Christians ain't any better at science than Christians. am i right?

If you believed in God, and I showed you a proof(not that I have any proof :> or willing to say I do) that s/he didnt exist. Would you believe me? The minority is labeled insane/uneducated as such.

Quote:

Originally Posted by Orpheus i believe we are all capable of understanding the results, when they come.

Dont count on it...

Understand it sure...except it...HAH !
This goes both ways mind you.

 

[Orpheus]

people will be believe what they want. but that's the whole point of this thread--to for once do a well-designed experiment that at least provides some credible data for either side. i mean, of course some experiments have been tried before about this subject, but they've all been approached half-assed. i've not seen one single experiment i've been satisfied with. and it's not cause i disliked the results or anything... just that the experiments really weren't conducted well at all.

no one has to accept the results as the last word in cables. but i do hope that the results would be respected, and at least provide guidance for further studies.

anyway, you don't have to worry about this pissing off anyone. there are plenty of "believers" who are looking forward to our results. so let's not dissappoint!

 

[Ctn]

Orpheus,

Im standing by my stance that good quality cables will sound as good if not better than expensive cables.

If I am happy with the experiment setup and the results show that exotic cables are experimentally better, I will eat my words and give expensive cables another shot.

 

[Edwood]

Quote:

Originally Posted by radrd EDWOOD If you read my previous post, you would see that I described exactly what you are talking about. Furthermore, I already own such a rig.

Ah. So you are splitting from one source three ways to three different IC's to one amp with three inputs?

I guess as long as you have an equal splitter, so that any degradation would be the same. Same would go for any switch box. I mention a switch box because not all amps have three or more identical types of inputs.

But I was also referring to taking the human element out of the administration. Not everyone lives with their mother.

Not necessarily computer automated, but perhaps some kind of random statistical model. I guess the ghetto solution to the double blind would be to cover up the wires.

-Ed

 

[rodbac]

Quote:

However, the chance of this specific sequence out of all possible sequences is 1 in 1024, not 1 in 2.

I know what the odds were- 1 in a 1000 is hardly more statistically relevant for proof of something like what we're trying to prove than 50-50 (which is why I used the term "impressive"- it was too late to do the math).

Hirsch, using the word "infinitesimal" with something like that is wholly inaccurate, and you know it.

Quote:

I used to teach statistics and experimental design, long ago.

Sorry, but I call complete bullsht. If you had done anything of the sort, you would not have even suggested an "experiment" as ridiculous as what you did.

Quote:

If the above statements don't make sense to you, stop now. You've got to understand elementary probability before you can understand any form of statistical analysis.

I've never claimed to be an expert in stats, but I do know enough to tell you that correctly predicting 10 coin flips comes nowhere ****ing near the improbability to "wow" anyone who knows anything about it. Only explanation would be a weighted coin? Probability "infinitesimal"? Seriously, maybe if you're in 3rd grade.

 

[cpw]

Quote:

Originally Posted by rodbac Sorry, but I call complete bullsht. If you had done anything of the sort, you would not have even suggested an "experiment" as ridiculous as what you did.

How 'bout a return to the facts.
I know zip about statistics. Why not point out specific pitfalls in the design rather than resort to words like "preposterous" and "ridiculous".
CPW

 

[rodbac]

Quote:

I know zip about statistics. Why not point out specific pitfalls in the design rather than resort to words like "preposterous" and "ridiculous".

CPW, the original proposal was so badly designed that it didn't even warrant critique. I have expressed my shortcomings with it a few times since then, though, and am actually embarrassed for having been asked to do it.

Hirsch, you're getting your wish- I'm going to bail on this thread. One last piece of advice, though: DO THE EXPERIMENT PROPERLY. Please.

None of your "the scientific community is way too obsessed with doing things properly" or "I can cut corners and still get valid results". This isn't about you doing the minimum to get something that impresses you (those infinitesimal odds against predicting coin flips or something)- it's about putting to rest the claims for or against million dollar cables, and, since it's going to be tough road for either side regardless of the outcome, leaving holes in your methodology will only serve to allow immediate dismissal of your results.

Again, DO IT RIGHT, PLEASE.

 

[PinkFloyd]

Quote:

Originally Posted by rodbac CPW, the original proposal was so badly designed that it didn't even warrant critique. I have expressed my shortcomings with it a few times since then, though, and am actually embarrassed for having been asked to do it. Hirsch, you're getting your wish- I'm going to bail on this thread. One last piece of advice, though: DO THE EXPERIMENT PROPERLY. Please. None of your "the scientific community is way too obsessed with doing things properly" or "I can cut corners and still get valid results". This isn't about you doing the minimum to get something that impresses you (those infinitesimal odds against predicting coin flips or something)- it's about putting to rest the claims for or against million dollar cables, and, since it's going to be tough road for either side regardless of the outcome, leaving holes in your methodology will only serve to allow immediate dismissal of your results. Again, DO IT RIGHT, PLEASE.

Oh God, you're an expert on cables and statistics as well as an expert on Rainbow foil.. what a man you are Rodbac

Dunno if any of you have read about the ABX Comparator but it may be of interest to you: http://www.bostonaudiosociety.org/ba...bx_testing.htm