Sarius said:
I found
this link, which I think would satisfy most criteria of valid expermentation.
Thanks for providing that link. I had seen it before, but had forgotten it. I guess I wasn't very interested in whether power cords have an audible effect. The article is more interesting as a demonstration of how difficult it is to run a blind listening test. It is time consuming and difficult for the listeners to remain attentive, especially when comparing items like power cords which may, at best, have only quite subtle differences. For all their effort, they had results from only 9 listeners, which is a small enough sample to severly limit broad conclusions.
Despite its flaws, it may be among the better published blind listening tests of any audio gear. The Toole and Olive tests of speakers remain the best example I know of.
I especially liked the questionaire shown at the end of the article. It addressed the question of whether a listener's prior experience, attitudes, beliefs and expectations had any effect of the results of the test. The answer for those 9 listeners was a resounding NO.
What would be nice to see in the future, if anyone cares to stage his own blind listening test, are the addition of both positive and negative controls in the test. A negative control is easy to imagine. Use the same power cord for an A-B comparison. How many listeners report hearing a difference for a negative control would be the false-positive rate or background noise of the listening test. By positive controls, I mean listening trials that answer the question how many listeners report hearing a difference when one really does exist. Suitable positive controls could measure how many false-negative responses were in the test. I have seen listening tests that included a negative control, but I have never seen one with any positive controls.
I don't know what kinds of comparisons would be good positive controls, but imagine a series of veils draped over a speaker. Remove each veil, one by one, to see how many listeners can hear the difference. Obviously the veils are only my poor attempt here at describing a good positive control. Any one got some suggestions?
Positive controls (some obviously different sounding and some more subtle) might also provide welcome relief to the fatigue that listeners experience when subjected to a series of A-B comparisons that all sound very similar if not the same. Listeners need that mental reward for actually hearing a difference during the test. If they got to hear some A-B tests that were clearly different, they might fatigue more slowly, and be more confident of their responses to A-B tests where they heard no difference.
Positive controls could help argue against the criticism of short-term blind listening tests that subtle differences in sound due to different power cords can only be heard after long-term listening. It could provide a direct measure of just how many listeners could actually hear a series of genuine differences that spanned a range in difficulties.
A suitable positive control test would also be an important measure of the effectiveness of the audio gear used in the test and of variability among individual listeners. A listener who answers yes to the positive controls and no to the negative control would be considered as most reliable in the power cord test. The fraction of listeners who meet these conditions, regardless of their answer to the power cord test, might be considered as a measure of validity for the whole experiment. Ideally, all listeners would hear a difference in the positive control and none would hear a difference in the negative control test. However, it is possible to deviate from the ideal and still make useful conclusions, as long as suitable controls are included for each listener to determine the frequency of false negative and false positive responses.
