I have several problems with a test like that, ranging from quibbling to major problems.
The 4 different cables were of different gauge. A was 16 gauge, B was not reported, C was “heavy” gauge, and D was 18 gauge silver instead of copper.
Listeners reported their preferences after to listening to comparisons of various pairs. One of the stated conclusions of this test was “So cables do make a difference”. Although an optional answer of No Preference was available on the questionnaire, you cannot conclude that cables do make a difference when the questionnaire asked listeners what their preferences were without asking if listeners could indeed hear a difference. That’s circular logic.
For any test of the ability of a group of listeners to hear what might be subtle differences in sounds between cables, it is absolutely necessary to also ask just how many listeners could reliably tell the difference between two sounds that are known to be detectably different. It is unlikely to be 100%.
At the same time, it is also necessary to test how many listeners reported they heard differences when two cables were actually identical. Again, it is unlikely to be 0%. This was a test of human perception and it is extremely unlikely that people could always get it right. People who test food preferences or perform wine tasting tests are well aware of these problems with human perception. Why should audio be any different?
These tests, often called positive and negative controls, should not be considered as trick questions. They are essential to establish the real sensitivity of this test. If a negative control test (A vs. A) resulted in numbers such as 30% vs. 70%, 40% vs. 60%, or even 50% vs. 50%, the conclusions you could make of the results with paired tests of different cables would have been completely different.
Finally, I have a problem with statistics. 40 people turned in questionnaires. That may seem like a big enough number, but with numbers like 30, or 40 or 50 listeners are not large enough to make the conclusions that “Cables do make a difference”.
In medical clinical trials, where human testing and great expense are involved, blind trials of an experimental medicine involving roughly 40 people are often done. Even with apparently positive results similar to those in the article, you cannot say with any degree of statistical confidence that X% of patients responded positively to the test medication. Such numbers do allow you to say, YES or NO, that it is worth further testing. Usually it takes much larger numbers of people, at least 10-fold more people, before you can safely estimate a response percentage.
So I have several objections to claims about audible differences coming from speaker cables: 1) it assumed listeners could hear differences, 2) it failed to provide positive or negative control tests, and 3) it ignored standard statistics.
So my conclusion is that no convincing evidence was presented.