Double Blind testing of audio components is the subject. Because of the subtlety of the differences that can be expected between these components it is important that the test be carried out in the environment that revealed the potential differences; in most cases this will be the subject's own listening room. What follows here is a method by which this can be carried out with some certainty of a successful outcome.
Obviously the subject and the proctor of the experiment. The proctor should be on the skeptical side of the argument. Additionally, there must be the person who makes the cable changes (we will use interconnects as the example here) who should also be skeptical, but he should be accompanied by one final person, preferably a friend of the subject, who also believes in the difference in sound who will act as an observer. So the situation is that there will be one believer and one skeptic involved in each aspect of the trial.
The subject should arrange his equipment so that it is easy to hide the interconnects (or whatever) that are being tested. The best way of doing this is probably to put the cd player and amp next to each other on the floor so the interconnect can be covered with a blanket.
The proctor should bring a CD with tones recorded on it at 100, 1000 and 15,000Hz. This will be used along with a meter at the speaker terminals to make sure that the necessary 0.1dB conditions of flatness and level are met by the two interconnects. If flatness is OK, but absolute level is not, this can be handled as will be described later.
The first thing that happens is that the reference and trial interconnects are checked by means of the test CD. Levels of tones at the three frequencies are measured with both cables, on both channels, and a record made of the levels.
Now the music chosen by the subject as the most revealing is put on the CD, and the level adjusted by him until he is comfortable with it. The subject can now swap cables a few times for himself to make sure he is familiar with the sound of both. The level should not be touched again for the duration of the test.
Both interconnects are now removed to another room. The subject and the proctor leave the room and the cable changer and observer also leave to a different room.
The interconnect for each trial will be chosen by a coin toss. The cable changer should at this point assign heads to one cable, and tails to another. The cable changer tosses the coin, takes the appropriate cable into the listening room and connects it, covering it carefully with the blanket so it is impossible to see which is connected. He is accompanied by the observer for this. He writes down the identity of the cable against the appropriate trial number.
The cable changer now makes some signal to the proctor to tell him that the first trial is ready. Both the cable changer and observer must leave the room before the subject returns.
Now the subject listens for as long as he wants, and when he is happy that he has identified the interconnect, he writes its identity against the appropriate trial number. The proctor makes a sign to the cable changer that the trial is done, and both the subject and the proctor leave the room.
The cable changer removes the interconnects, and returns to his room, where he tosses the coin for the next trial.
This is repeated until twenty trials have been completed.
At the end, proctor and subject will sign their form, as will cable changer and observer.
The two forms will be compared to see how many trials resulted in correct identification by the subject. If the number is fewer than 16, then the subject will be considered to have failed to identify a difference between the interconnects.
If the subject has scored 16 or more, then it must be considered a possibility that he has identified the interconnects by sound alone. But with a statistical process, there is a high probability that he has simply got lucky with his guesses. As this is a test of ability to hear differences, and not simply gambling to see if you get lucky, it is important to verify that it was not just a matter of chance.
So, if a score of 16 or more is achieved, the twenty trials must be rerun a second time. If the first score was not a matter of chance, it will be repeated. If it was simply an accidental statistical cluster, you can expect the score to revert to somewhere nearer the mean value of ten.
If this second run does not also yield a score of 16 or more, the subject will be considered to have failed to identify a difference between the interconnects.
Whatever the results, they will be published one one or more Usenet groups. The prime group will be the one which prompted the response to the challenge. The format will probably be a link to a web site, where a full report and scanned copies of the forms will be posted.
Suppose the interconnects would not yield sufficiently equal levels, but were otherwise OK for frequency response? In this case, at each trial, the cable changer should turn the volume control to zero before leaving the room, allowing the subject to turn it up to whatever level he finds comfortable for listening. This should only be done for relatively minor level variations as the position of the volume control might otherwise provide visual cues to the subject.
At any point during the trials, the subject may ask to re-aquaint himself with the sounds of the two interconnects. They will be returned to him, and the other three people will leave the room. Once he is content that he has re-familiarised himself, the trials can resume. The subject should complete the current trial before making his request.