Difficulty of blind testing
Jun 7, 2009 at 2:40 AM Thread Starter Post #1 of 117

mike1127

Member of the Trade: Brilliant Zen Audio
Joined
Oct 16, 2005
Posts
1,114
Likes
25
I "believe" (so to speak---I'm not a "true believer") in small subjective differences between components and cables, the kind that science says shouldn't exist. Yet I'm somewhat embarrassed by the rarity of blind listening among audiophiles like me. If anyone out there is doing and passing blind tests of, say, cables, please let me know.

If you look at wine tasting, for example, it is common that testers drink a wine in an unmarked glass. We know that the placebo effect is real. We know that the preconception you bring to something will affect how you experience it. Why, then, do audiophiles not insist on doing most evaluative listening blind? (Again, if there are some who do, please let me know.)

One major reason, of course, is the difficulty. It's a lot harder than pouring wine into an unmarked glass.

Also, some of us who've done blind tests (I count myself here) feel confused during the test, like we "lost our way." I had a friend help me with a few cable tests, and I got the first few trials correct, but started to feel uncertain after that, and ended up doing no better than chance.

I had a thought about this recently, since upgrading to a much better source (added the Flat Cap to my Naim CD5x) and a much better headphone amplifier (the DNA Sonett).

I am finding myself much more involved with the music... really hearing and feeling at a gut level the nuances of the playing.

I noticed something interesting when, after a listening session, I took a drive and listened to some of the same music on my car stereo.

I was much more involved with it.

Having heard in my home system the exciting emotions, the expert nuances of playing, and so forth, I brought those experiences through my imagination to the low-resolution playback. If had never heard the high-resolution version, I wouldn't have been able to bring that much of a gut response to the low-resolution version.

Let me say more about getting "involved" with music. I mean we are not passive listeners, but respond to music with activity. That activity could include small body movements (even movements too small to observe), visual and auditory imagination, emotions, and probably a lot of things we don't have English words for.

I think this creates a major problem in doing comparisons. (And by the way, I'm not arguing for the validity of sighted testing here---I speak of comparison under any conditions.) If I listen to device A, followed by device B, who knows what new levels of activity carry over from A to B?

I do think small differences are real, though, and they make a difference in the long-term. If device A really fires up my imagination, then I am going to start bringing some of that to every other listening experience. But if I replace A by some inferior device B, even if I don't notice at first, eventually I'll have less of a gut response.

I suspect that initial impressions are often accurate as well.

The problem with statistically controlled testing is that it involves neither "initial impressions" nor "long-term listening." (Although it could perhaps be designed to do so.) For a test that doesn't take a ridiculous length of time, you have to switch between devices fairly rapidly, allowing for "imagination contamination."
 
Jun 8, 2009 at 3:36 AM Post #2 of 117
Quote:

Originally Posted by mike1127 /img/forum/go_quote.gif
I am finding myself much more involved with the music... really hearing and feeling at a gut level the nuances of the playing.

I noticed something interesting when, after a listening session, I took a drive and listened to some of the same music on my car stereo.

I was much more involved with it.

Having heard in my home system the exciting emotions, the expert nuances of playing, and so forth, I brought those experiences through my imagination to the low-resolution playback. If had never heard the high-resolution version, I wouldn't have been able to bring that much of a gut response to the low-resolution version.

Let me say more about getting "involved" with music. I mean we are not passive listeners, but respond to music with activity. That activity could include small body movements (even movements too small to observe), visual and auditory imagination, emotions, and probably a lot of things we don't have English words for.



I think you have a great point. But there are some fallacies that stood out to me:

1. The translation of hi-fi to lo-fi rigs and the placebo movement in between is something that I think is going to dramatically differ between individuals, something that I know you never denied, but will make a variable amount of difference none the less. To say that menial differences are going to make a real difference over time assumes that a fairly consistent level of active listening is going on. The hi to lo translation isn't going to come without focused, placebo influenced, attentive listening to details and tones and the differences between them. To call this a universally applicable transformation, even over a great period of time is not really assured, IMO.

2. Quote:

Originally Posted by mike1127 /img/forum/go_quote.gif
I had a thought about this recently, since upgrading to a much better source (added the Flat Cap to my Naim CD5x) and a much better headphone amplifier (the DNA Sonett).


This is a little bit more than a minor, potentially negligible upgrade. To debate the improvement of cable is one thing, a much better headphone amp, is going to have consequences on the sound that will maybe make a lot more easily picked out, making a double blind test rather unnecessary (because of the difference), and influencing your idea of our application of hi-fi to low.

Never hoping to troll in any of this, I thought your observations were really sort of profound, it's just important to really look at those observations and see how viable they are when applied on a level that's completely universal.
 
Jun 8, 2009 at 3:47 AM Post #3 of 117
There has been plenty of testing of cables and such. The results always come back the same - that no one can tell the difference.

Cable believers, and especially those who make money selling cables, then proceed attack test methodology.

The reason cable manufacturers refuse to test their cables is the same reason many prisoners refuse to submit to DNA tests that would free them. While a test could free a prisoner, it could also prove him guilty. Which is why prisons are crammed with the "innocent" who loudly protest the shortcomings of DNA testing.

Despite grunts and squeals otherwise, the real reason cable manufacturers refuse to test has nothing to do with methodology.
 
Jun 8, 2009 at 3:47 AM Post #4 of 117
OP -- interesting thoughts.

I have conducted very careful blind tests of digital coax S/PDIF cables, perfectly controlled, with pinpoint accurate protocols (some of which I designed, I have a PhD in Statistics).

I was the first subject, and I could not tell a $1000 digital cable from a $3 rat shack video cable (75 ohm RCA cable, so it works).

I have had the same "uncomfortable" thoughts you have, and for the time being have abandoned my blind testing research pending some new ideas on avoiding confusion and carry-over without giving in to the placebo effect.

Just like you, I started out correctly identifying cheap from high-end, but then fell in to "no better than random". That leads me to believe there is something real here, but I can't figure out how to measure it yet.

I do know that the big panels and group A/B tests (and A/B/X in general) as currently conducted and published in audio journals is NOT the right approach.

But I have given up posting on this topic because it gets unpleasant fast. Your thoughtful remarks drew me back in, however -- at least for one post!
 
Jun 8, 2009 at 6:31 AM Post #5 of 117
The only way to hear subtle differences is to spend lots of time with your system and to have a good memory for the music and then replace one part. Its not something you can do in one afternoon. Not unless you want to waste your time trying to find and acknowledge the differences. You need to get familiar with 10 or 20 tracks to the point were you don't have to listen for differences and then add the new cable. Then you can hear the difference and maybe even accurately describe what that difference is. All this flicking the switch between 2 cables is the biggest pile.

Is it everyone's opinion that cable differences are subtle or not??? To switch cables back and fourth on one afternoon the differences should be bigger as it would be when changing tubes or headphones or sources. Cables are great when you spend the time to find the right ones but if you think finding the right ones is scientific or a quick process then you are just fooling yourself.
 
Jun 8, 2009 at 6:48 AM Post #6 of 117
Quote:

Originally Posted by olblueyez /img/forum/go_quote.gif
All this flicking the switch between 2 cables is the biggest pile.

Is it everyone's opinion that cable differences are subtle or not??? To switch cables back and fourth on one afternoon the differences should be bigger as it would be when changing tubes or headphones or sources.



Well double blind ABX tests do not necessitate doing it in one afternoon. The test can be conducted across many days. As long as you are unable to visually identify cable X during testing, the protocol will be satisfied, no matter the length of time spent with AB or X.

Most proponents suggest quickly flipping between cables, as studies have shown that our aural memory is short. But if you have any reason to believe that you aural memory falls outside of the bell curve, you may wish to modify the ABX protocol as such.
 
Jun 8, 2009 at 7:06 AM Post #7 of 117
Quote:

Originally Posted by xolp /img/forum/go_quote.gif
Well double blind ABX tests do not necessitate doing it in one afternoon. The test can be conducted across many days. As long as you are unable to visually identify cable X during testing, the protocol will be satisfied, no matter the length of time spent with AB or X.

Most proponents suggest quickly flipping between cables, as studies have shown that our aural memory is short. But if you have any reason to believe that you aural memory falls outside of the bell curve, you may wish to modify the ABX protocol as such.



Readers Digest version: Take your time and forget science. Do you measure a girl before you take her home for some trim? Even if you did would it tell you if she was good in bed?

Loose the propeller beanies fellows. Its all bull, use your ears.
rolleyes.gif


200px-Casquette_a_helice.jpg
 
Jun 8, 2009 at 8:30 AM Post #8 of 117
Quote:

Originally Posted by olblueyez /img/forum/go_quote.gif
Do you measure a girl before you take her home for some trim? Even if you did would it tell you if she was good in bed?


You are comparing apples and oranges here. The sexual experience involves all senses, and is not dependent on need to blind. Besides, I am not suggesting any form of quantitative measurement. I am just suggesting that the double blind test is useful for determining if there are audible differences.

For issues that involve subjective inputs, we use a blind test to separate a specific aspect from the whole subjective experience. For instance, in testing audio cables, we would use the double blind test to isolate audible differences.

Of course, if we wanted to test if 2 ladies had any difference in the ability to give head, we could perform a double blind test to isolate other factors like visual beauty etc. Eg, between a granny and a barely-legal, they could both be equally good at giving head, but most people, if asked to comment if there was a difference without perform a blind test, would be unavoidably biased on their age rather than their true ability.

Quote:

Originally Posted by olblueyez /img/forum/go_quote.gif
use your ears


I'm not suggesting we do otherwise. I am also not putting any opinion forward of whether there is a audible difference in the whole cable debate. But as a part of a larger community who's based on logic, I fail to see how the double blind ABX test fails to satisfy as a credible test.
 
Jun 8, 2009 at 11:21 AM Post #9 of 117
xolp: A DBT is invalid if it a: Sets out only to prove a point, not find the truth, b: hasn't tested the listening ability of the subjects beforehand, c: cannot demonstrate that a measured positive result can be obtained during the test with the subjects, d: doesn't measure the differences between the things being tested (and compare those measurements to, say, the measured hearing ability of the subjects), and e: Cannot eliminate all points of failure (which is related to a, b, c and d). That doesn't mean to say that there haven't been credible ABX DBT tests done though.
 
Jun 8, 2009 at 12:20 PM Post #11 of 117
Quote:

Originally Posted by QQQ /img/forum/go_quote.gif
The human(human ears to be specific) is so innacurate "measurer" that it can't be used as a "tool" in any testing.


Indeed
 
Jun 8, 2009 at 1:16 PM Post #12 of 117
Quote:

Originally Posted by wavoman /img/forum/go_quote.gif
OP -- interesting thoughts.

I have conducted very careful blind tests of digital coax S/PDIF cables, perfectly controlled, with pinpoint accurate protocols (some of which I designed, I have a PhD in Statistics).

I was the first subject, and I could not tell a $1000 digital cable from a $3 rat shack video cable (75 ohm RCA cable, so it works).

I have had the same "uncomfortable" thoughts you have, and for the time being have abandoned my blind testing research pending some new ideas on avoiding confusion and carry-over without giving in to the placebo effect.

Just like you, I started out correctly identifying cheap from high-end, but then fell in to "no better than random". That leads me to believe there is something real here, but I can't figure out how to measure it yet.

I do know that the big panels and group A/B tests (and A/B/X in general) as currently conducted and published in audio journals is NOT the right approach.

But I have given up posting on this topic because it gets unpleasant fast. Your thoughtful remarks drew me back in, however -- at least for one post!



Wavoman, I'm glad there are still experts out there interested in audio that have such integrity as well as deep perception. I follow your posts closely.
 
Jun 8, 2009 at 1:44 PM Post #13 of 117
Does no one trust their own abilities of perception enough to just try a new peice of equipment once and awhile? If the answer is "No" then how the hell do you buy any audio equipment? Where is it stated that cables must carry the stigmata of placebo? It is the same as choosing a headphone but the only difference is it will serve you well to take longer periods of time to make a choice. Did the people who have the best equipment you have ever heard use blind testing iin order to assemble their systems? Since some of you must stigmatize cables then I think I will stigmatize people who persist with the blind testing. You guys will be called "Propeller Heads" by myself from now on.

Theard Title: Difficulty of blind testing
 
Jun 8, 2009 at 1:54 PM Post #14 of 117
Quote:

Originally Posted by mike1127 /img/forum/go_quote.gif
If anyone out there is doing and passing blind tests of, say, cables, please let me know.


Anyone besides Wavoman?
 
Jun 8, 2009 at 3:50 PM Post #15 of 117
Quote:

Originally Posted by QQQ /img/forum/go_quote.gif
The human(human ears to be specific) is so innacurate "measurer" that it can't be used as a "tool" in any testing.


I find this a very odd point of view for someone involved in audio. The human ear/brain is the reason audio exists, so it is the ultimate tool to be used in evaluating equipment. Measurements only mean something to the extent they can be related to what we hear.
 

Users who are viewing this thread

Back
Top