My problem with the data and how it's given is that it's quite useless and the result, while not the null hypothesis, is the expected result. Namely, it can be

My problem with the data and how it's given is that it's quite useless and the result, while not the null hypothesis, is the expected result. Namely, it can be expected that no man would find a speaker dumber with superior audio, so let us assume that only 1% of the sample found the speaker more intelligent with superior audio, and for 99% it makes no difference, then the “average result” is still that “people” as a collective average find a speaker with better audio more intelligent, even though 99% of people don't. This is my problem with such research. — I am far more interested in the percentage of “people” onto whom this has an effect or how large the effect is in distribution. It goes without saying that this effect exists with some “people”, but I would like to know what percentage. It's entirely possible that the effect only exists with 20% of “people” and 80% are not biased, but the research is incapable of showing this. Such research, which does not, and cannot due to it's methodology make any claim as to the percentage of “people” it applies is often taken to apply to all “people”.