Each recording therefore is of the same singer, same performance, and through identical preamplifiers. So this should make the test very accurate then?
Before considering the accuracy of this test, it is worth considering the alternative. One microphone could have been set up at a time and three takes recorded. The problem with this is that even with a very consistent signer, the three takes will be different. It is harder to draw conclusions, but even so this method can be illuminating.
Going back to the simultaneous test... the main problem with this is how do you point three microphones at a singer? For this test, the microphones were lined up horizontally, as close together as possible without touching.
So one microphone was pointing head-on at the singer's mouth, another from slightly to the left and the third from slightly to the right. Although the microphones are close to each other, even slight differences in position can make a difference to the sound quality.
In this instance though there was an additional factor that negated the effect of microphone position. The singer jiggled! I'm sure that every sound engineer prefers to work with musicians who stay still as they perform. But many musicians can't stop themselves from moving about.
The question would be whether you can stop them doing this without affecting the quality of their performance. Usually the answer is that you can't. You have to put up with it and do what you can to prevent it affecting the sound. Usually a greater microphone distance is the answer, perhaps coupled with using an omnidirectional pattern. Certainly there should be no reflecting surfaces around or this will cause comb filtering with shifting null frequencies. And that is bad!
One more factor is that the microphones will reflect sound off each other, so they don't pick up quite the same sound field as they would have individually.
But there is another important point...
Should microphones be tested using the same position for each mic, or should the optimum position for each individual mic be found? Now that is a big question. Using the same position each time is consistent and easy to do; finding the optimum position generates an inconsistency and is much more difficult.
Your comments are welcome. Suggested questions are...