Joshua_G said:
Even more amusing : by numbers only, a peek at the posts of the last 4 days in this thread reveals the most prejudiced participator.
Some may sympathise with Mr Curl's views and experience, but feel that his cause is not served by Golden Earlobe groupy behaviour.
john curl said:
May your Bagel light up your Hannuka bush, but excuse me for preferring a different RAP
For 12 trials the minimum number of correct answers has to be 10 for sl<=0.05 .
@ Joshua_G.,
as pointed out before, a potential conflict could lie in the paralleling of two cable sets, as it forms very different current paths, could present a higher load on the output and therefore produce other results than with only one cable connected.
This has to be solved first (maybe it´s not a problem in reality, but it should be sorted out), then the other point was mentioned by SY, as there could be some influence if the switching person is still in the room while your´re listening.
If he/she would leave then it wouldn´t matter.
Despite that my main concern would be the describing terminology; you´ve said, after each trial you´ll describe the audio quality to your friend.
That leaves the possibility of misinterpretation, i´d propose to do an initial listening session to get used to the sonic nature and to choose a description for each DUT that would be counted as a correct answer in the blind test.
Of course that could be also a simple A better than B or vice versa.
Wishes
@ Joshua_G.,
as pointed out before, a potential conflict could lie in the paralleling of two cable sets, as it forms very different current paths, could present a higher load on the output and therefore produce other results than with only one cable connected.
This has to be solved first (maybe it´s not a problem in reality, but it should be sorted out), then the other point was mentioned by SY, as there could be some influence if the switching person is still in the room while your´re listening.
If he/she would leave then it wouldn´t matter.
Despite that my main concern would be the describing terminology; you´ve said, after each trial you´ll describe the audio quality to your friend.
That leaves the possibility of misinterpretation, i´d propose to do an initial listening session to get used to the sonic nature and to choose a description for each DUT that would be counted as a correct answer in the blind test.
Of course that could be also a simple A better than B or vice versa.
Wishes
You are correct- I slipped a line in the student test table.
Surely not for the differences he claims.
Regarding the description interpretation, you might wish to look at Randi's clinical disassembly of the SRI remote viewing trials. A simple choice of A or B removes all ambiguity.
I rechecked the number for n = 0.05 at 12 trials in David Clark's AES paper. It is 9, not 10. I'll read through more carefully to see why they allow a lower number of correct identifications than the student test.
OK, on rereading the paper, they are more lax in the confidence interval. Using a binomial table, a 12 trial 9 correct identification is more like n = 0.07, which is still pretty close. 10/12 correct is about n = 0.02.
I still think that triangle testing is a better way to do things (and what I used for most of my sensory research), but it does make the test itself more complicated to set up.
I still think that triangle testing is a better way to do things (and what I used for most of my sensory research), but it does make the test itself more complicated to set up.
SY said:
Of course it was meant to choose a unique descriptive word pair for the DUTs, maybe following the AES descriptive pairs (if suitable).
A and B could be a possible solution, but could otoh introduce a slight imbalance as there might be a bias to prefer answer "A" in general over answer "B".
It seems that you do have concerns regarding A/B-preference tests.
Could you elaborate about the reasons?
Wishes
Jakob: The choices need to be unambiguous; loosie-goosie "oh, can't you see what I meant?" ex post facto stuff invalidates any attempts at control and opens the opportunity for manipulation of the result to get the "desired" end. Preference is fine- the question is really more basic, can A and B even be distinguished? If one can show a consistent preference for (e.g.) A over B, then certainly one has established the ability to distinguish them. The trick is in instituting the controls, which are totally lacking in some of the suggestions I've seen here.
There is more statistical power in a triangle test (fewer trials needed for a given confidence level), but as I said, it's harder for an experimenter to set up. I think it's easier for the test subject, though.
There is more statistical power in a triangle test (fewer trials needed for a given confidence level), but as I said, it's harder for an experimenter to set up. I think it's easier for the test subject, though.
For 12 switches there are 24 correct answers – 12 for cable "A" and 12 for cable "B". Should it not be so, I'd extend the procedure.
When there are no marked differences between cables, I see no point in continuing the test.
As noted before, this is acknowledged and will be amended in all future tests.
There is no problem arranging that the switching person will leave the room in future tests, however, I don't see how it may be relevant, since my eyes are closed all along the session.
The switching person only writes down my comments, which at all 12 cases are very similar, almost identical, to cable "A" and the same applies to "B". The notes only serves to record my basic impression of each cable.
I don't see what can possibly be loose description when my description to "A" may be something like: "More high frequencies" and to "B": "Less high frequencies" – or similar descriptions.
Triangle test could be done, with double the control level, however, it looks like it's not necessary at all when my identification is 12 out of 12 for each cable.
SY said:
If the descriptive word pair (maybe following the AES examples) is choosen before the test and _has_ to be used in the test for a correct answer, then any "loosie-goosie" would be avoided. 🙂
Preference tests with forced choice answers seem to be more similar to the somewhat normal listening test routine of listeners.
That shorted in general the training time needed to get used to double blind testing.
It has been shown that discrimination tests do produce more likely wrong results, so it would be better to avoid these.
(AFAIAC i´d exspect that it is possible to get used to discrimination test routines too, but that it would take extended training time)
A 3-AFC (as a triangle) would require less trials compared to a 2-AFC, but there are good reasons to assume that this procedure is again quite unusual for untrained listeners.
Of course there is lack of controls (and control), but i think it is a realistic for the possiblities in any home diy test. I mean, what about the normally needed test of exhausted measurements, it would raise the barrs to high in a not scientific test setup.
Nevertheless i suppose, that most listener will find out, that even this quite simple test routine will take some training time to get used to it and to be able to produce reliable results.
Wishes
Joshua_G said:
If you compare cable A to cable B, then in every trial cable A would be randomly assigned to (for example) switch position 1 or 2.
Your task would be to identify the cable behind the switch position to complete the trial, so in every trial there is just one correct answer which sums up to a maximum of 12 correct answers in 12 trials.
Jakob: No-one said that establishing extraordinary fact claims was easy! It's certainly less fun than playing make-believe.
Jakob2 said:
Identifying cables may be others' task. You may be talking about another test, not about the tests I'm doing.
My task is to find the qualities of each cable, so that I may choose between the 2.
Also, switching is done sometimes randomly, that is, the same cable may be switched twice in a raw.
What you people don't get is that when a certain cable has certain qualities, I hear those qualities each and every time.
Qoute "What you people don't get is that when a certain cable has certain qualities, I hear those qualities each and every time."
Bollocks.
wire is wire is wire. A is A and E=mc^2 and Bush has been president for 2 terms.
What you hear are the interaction between the cable lumped elements (which do change depending on length, inductance and capacitance of the cables under test) and the amplifier out- put characteristics (output impedance which is a function of output inductance, capacitance and the moderating effect of feedback).
Bollocks.
wire is wire is wire. A is A and E=mc^2 and Bush has been president for 2 terms.
What you hear are the interaction between the cable lumped elements (which do change depending on length, inductance and capacitance of the cables under test) and the amplifier out- put characteristics (output impedance which is a function of output inductance, capacitance and the moderating effect of feedback).
OK, now I will tell my 'real' story about cables.
First, let me give you a little background about myself and my associates.
40 years ago, I was sharing whiskey shots with Paul Klipsch and Richard Heyser up in Paul's hotel room at the AES. I was also there when Klipsch formally protested Heyser implying (in public) that a 2 foot path difference between speaker drivers MIGHT be audible. (How far we have come)
Then I progressed to 1970, working the the Grateful Dead crew on a motion picture, and being cussed out by Larry Brilliant (now of the SEVA foundation) for perhaps exposing dozens of people to a communicable disease in a swimming pool.
In 1974, to actually (to me, no DB test involved) ascending to the Astral Plane with Ram Dass, until I could not even accept it, and broke away.
Being told by Oswley (the LSD manufacturer) that my design ideas came from God, not myself, and Oswley himself was the first person to recommend anything more than zip cord for loudspeaker wires. He recommended Belden 14 ga wire pairs.
This sort of loose association with a number of noted extra special people has loosened my position as what 'reality' is. I go with what works.
Now, let us skip to about 15 years ago, when I met BOB.
Bob is dead now, but he taught me something about wires.
Bob was an insurance salesman when I met him with a sincere interest in Audio. His degree was in psychology, and he knew what DB protocol was, with the best of them.
The problem with Bob was, is that HE trusted his ears.
He was not technical like many of us, but he could solder well, and he was willing to try things. One thing that he tried was to put high speed diodes into the Vendetta Research power supply in place of the 1.5A mini rectifier bridges (no small feat) and found that it made a REAL difference. He convinced me to try it, and I was sold. It is now the D mod for the Vendetta.
Then Bob started to make his own connecting wires and AC cables. I was not directly associated with him then, but when he shared a listening room at CES with a number of different audio manufacturers, that room would always sound 'right'. (Yes, another unprovable personal impression)
At this point Bob asked me to help him make the Blowtorch by building the electronics for it. He would supply the rest. (and what could that be, but everything that we debate around here?) Bob supplied the BOX, the SWITCHES, the SILVER WIRE, the CONNECTORS, etc, etc. I just supplied the circuit boards designed with Carl Thompson.
The preamp was a success at CES, and we began to win listening awards from reviewers, and we started to make a few.
We first used Bear's power amp, and it is a very good sounding design. Later we shifted to a modded HCA3500, called the Bar-B-Q, and then to the JC-1 power amp, because it had even more improvements than the Bar-B-Q. We were showing with $20,000 to $50,000 loudspeakers, and Demain's modified D-A, as well as records. We got lots of appreciation for the sound in our hotel room each year, and mostly because of Bob.
We didn't need any 'Stinkin ABX tests to do it, either!'
When people here DEMAND proof, it is because that are in a very tight intellectual 'straight jacket' and cannot think for, or trust themselves to listen openly. That's all it takes.
First, let me give you a little background about myself and my associates.
40 years ago, I was sharing whiskey shots with Paul Klipsch and Richard Heyser up in Paul's hotel room at the AES. I was also there when Klipsch formally protested Heyser implying (in public) that a 2 foot path difference between speaker drivers MIGHT be audible. (How far we have come)
Then I progressed to 1970, working the the Grateful Dead crew on a motion picture, and being cussed out by Larry Brilliant (now of the SEVA foundation) for perhaps exposing dozens of people to a communicable disease in a swimming pool.
In 1974, to actually (to me, no DB test involved) ascending to the Astral Plane with Ram Dass, until I could not even accept it, and broke away.
Being told by Oswley (the LSD manufacturer) that my design ideas came from God, not myself, and Oswley himself was the first person to recommend anything more than zip cord for loudspeaker wires. He recommended Belden 14 ga wire pairs.
This sort of loose association with a number of noted extra special people has loosened my position as what 'reality' is. I go with what works.
Now, let us skip to about 15 years ago, when I met BOB.
Bob is dead now, but he taught me something about wires.
Bob was an insurance salesman when I met him with a sincere interest in Audio. His degree was in psychology, and he knew what DB protocol was, with the best of them.
The problem with Bob was, is that HE trusted his ears.
He was not technical like many of us, but he could solder well, and he was willing to try things. One thing that he tried was to put high speed diodes into the Vendetta Research power supply in place of the 1.5A mini rectifier bridges (no small feat) and found that it made a REAL difference. He convinced me to try it, and I was sold. It is now the D mod for the Vendetta.
Then Bob started to make his own connecting wires and AC cables. I was not directly associated with him then, but when he shared a listening room at CES with a number of different audio manufacturers, that room would always sound 'right'. (Yes, another unprovable personal impression)
At this point Bob asked me to help him make the Blowtorch by building the electronics for it. He would supply the rest. (and what could that be, but everything that we debate around here?) Bob supplied the BOX, the SWITCHES, the SILVER WIRE, the CONNECTORS, etc, etc. I just supplied the circuit boards designed with Carl Thompson.
The preamp was a success at CES, and we began to win listening awards from reviewers, and we started to make a few.
We first used Bear's power amp, and it is a very good sounding design. Later we shifted to a modded HCA3500, called the Bar-B-Q, and then to the JC-1 power amp, because it had even more improvements than the Bar-B-Q. We were showing with $20,000 to $50,000 loudspeakers, and Demain's modified D-A, as well as records. We got lots of appreciation for the sound in our hotel room each year, and mostly because of Bob.
We didn't need any 'Stinkin ABX tests to do it, either!'
When people here DEMAND proof, it is because that are in a very tight intellectual 'straight jacket' and cannot think for, or trust themselves to listen openly. That's all it takes.
- Status
- Not open for further replies.