Rainbow Foil, Initial impressions - Page 9  

actually, you can ask Hirsch. he did do a test of cables among some peers last year or so. i never did see the results... only a mentioning that the results seemed to conclude that no one could clearly hear a difference. but i dunno... maybe he can speak up here--though it's a bit off subject. this was quite a while ago.

yes hirsch, if you wouldn't mind directing me to the results of that test!

We did the pilot run, but never followed through. I don't trust the pilot results, since there were some serious flaws in our methodology that didn't turn up until we were under way. The jury is still out. FWIW one person out of three could consistently identify interconnects, and none out of three could consistently identify power cords.

so, any plans to try the test again?--i think 3 people isn't a large enough pool though.

Three people indeed is not enough. The length of time we gave to listen to samples was not enough. We were able to perform 12 comparisons on interconnects and 12 on power cords, which wasn't enough. We were all exhausted at the end of the afternoon, which was too much. All of which goes to illustrate the difficulties in this type of thing. It's not a yes you hear it no you don't type of affair. Yes you hear it speaks for itself. One person reliably identifying interconnects is enough. The differences are there, and the fact that two others could not indicates that they were not as sensitive to what was happening as the other person, not that there are no differences. However, the presence of a total negative is not enough. You need to have some estimate of the variance present when a just noticeable difference (JND) is used (which you're going to need to experimentally test and calculate, so that responding is just above random) to generate a power analysis to determine exactly how many people you're going to need to have confidence in a negative result. You're then going to need to run a positive control, where there is a comparison at the JND to confirm that people are still sensitive to the JND. It's very easy for people to start guessing, which can hugely increase the variance and erase any chance of significant results in detection of differences. However, if it increases the variance at the JND, however small, you can discount it as experimental error. The gist of this little paragraph is that doing it right isn't easy. Something like ABX doesn't begin to approach some of these problems.

how exactly do you establish a JND in a cable test?

i'm familiar with it in other tests, but how would you do it with cables?

Audiophiles are stupid! Vol 999...

Can't blame them .

Wali, that title wasn't fair or accurate and was apologized for.

The amount of gullibility that has surfaced is surprising, though, I have to admit. The lack of knowledge of the scientific method (albeit with good intentions) has caught me off guard, too, but I don't think it's really the focus of this forum (which is certainly ok).

yeah, i think you're right. the other thing is, with only 12 evaluations each, it's very possible your one single person who demonstrated some ability might have been able to simply guess correctly.

but i know you have a job doing this kind of stuff... i also vaguely remember some formulas from my days in the stat classes that allows you to calculate when results are statistically significant. i dunno how many out of 12 that that person got right... but if he got all 12 right, well, then that sounds pretty overwhelming (though still possible to flip coins with the same result.) but if it's just like 9/12, well, that's only 3 above odds, which could be accountable by random chance, i would think.

anyway, i would like to discuss this with you further some time. you see, i do want to do such an experiment, but i think your results would be more credible than mine if you happen to find that cables don't make a difference. and of course, mine would be more credible if it was found that cables do. cause you know, i publicly don't believe too much, and you do.

but anyway, that's a subject for another thread in a non "dbt-free" forum. perhaps we can discuss this further in the Member's Lounge?

There would be at least one other problem I think you'd need to address. While I can hear differences in interconnects in my system very reliably, I'm not at all sure that I could do this in someone else's system. There are so many parameters to listen for it's incredible. When I change interconnects in my system, it is the differences that stick out to me but only because I am familiar with the sound that I have heard in the past.

Both simple and difficult at the same time. You see, there's no real need to do it with cables, even if that's what you're testing. Simplest would be a gain manipulation, but there are a host of other possible parameters, many of which are likely to have more face validity, if nothing else. Could be almost anything, and it's probably an interesting exercise to figure out what's best. Since it's not known what cables are actually doing, you're working in the dark, almost. You'd want a level that your sample performed above chance, but with some error so that it's not too easy. The idea behind a positive control is to insure that your test situation isn't altering sensory thresholds. The trick is picking the right threshold to use as your control.

then at least we can say that those who claim this or that cable makes a day or night difference in sound are exaggerating. if it takes so long to evaluate each cable, and if people doubt that they can hear the difference in other systems... well, i think it's fair to say, cables in general make small differences at most.
i don't think this is a good idea. it's also possible various cables might "sound" louder than others. it's better to stick to one single variable: the cable. same source, same amp, same headphones, and same volume... just change the cables.

Fine with me.

No, it's not. The differences may be small, but so what? I've been in the situation where in a controlled test I couldn't tell the difference between a pair of cables, and yet one of the cables in my setup literally saved it for me, when I thought getting the sound right was hopeless. The other was the one making it hopeless. You're a musician. How much difference does just "slightly" out of tune make in your music? Sometime small differences are important.

As gpalmer pointed out, we are much more sensitive to alterations in systems that we know and use a lot. There are also time factors to be considered. Sometimes I won't know what I like or dislike about what I'm hearing for days, or weeks. If I can't identify it, I'm not going to be able to tag it in an A/B situation.

yes, you're right. good point.

No, I can hear the differences in other systems, but I would bet that there are systems out there that would not be clear to me. This has always been one of the weakness in tests I have seen, researchers bring folks in to listen to systems they have never heard before which means they don't have a baseline for comparison. Seriously, think about how long it takes to learn the nuances of a new CD player or amplifer and be able to express them, then take that and extend it to the entire system. Generally I would say that cables are a finishing touch sort of difference and would agree that they are not a night and day change which is usually why I would compare them last after all other components are in place.