Jan Vigne wrote:
There was no answer required as there was no question posed. The sentence reads, "If you require a DBT to decide whether two speakers sound dissimilar to one another, then knock yourself out." That's true, if you can't tell whether one speaker is different than the other, knock yourself out. I don't care. That statement has no direct relationship to the article. It is simply a statement of personal belief.
And as such, a pointless strawmanly interjection, since no one is claiming that different loudspeakers aren't likely to sound *different*. (No one is saying it's impossible for them to, either). The question is whether 'sighted' loudspeaker preference is based primarily on these differences, as often asserted by the person doing the comparing, or not.
In this specific situation the text of the article reads, "A total of 40 Harman employees participated in these tests, giving preference ratings to four loudspeakers that covered a wide range of size and price."
The test determined whether one speaker at a time sounded "different" than another speaker each time.
No, that's not what the test did. The experment tested whether what was ostensibly a listener preference for loudspeaker 'sound' formed from sighted comparison, reliably predicted the preference when sound was in fact the only criterion for formation of preference. The results indicate that factors other than 'sound' play a big role in the formation of loudspeaker 'sound' preference, during sighted comparison. In other words, when listeners believe they are picking a loudspeaker based *just* on its sound, they're probably wrong.
The paricipants were asked for a "preference" which implies a they were searching for a "difference" between "A" and "B".
It 'implies' that difference was assumed to exist. Determining whether A and B were audibly *different* would have been a different test, and used a different, though still blind, method.
It would be improbable in the extreme for two devices to get significantly different *preference* ratings in a blind test, when they are audibly the same in a difference test. But it's quite possible for
there to be no statistically clear *preference* even when two devices pass a sonic difference test.
The listeners were not apparently asked which was the superior sound or the more accurate sound (something this tester might have found interesting, don't you think?), only the "preferred" sound. "Preferred" between "A" and "B" which are "different" from each other. Both "A" and "B" could have been horrible, the listeners needed only to express a preference for which was the preferred degree of horrible - which was different enough to make it acceptably horrible against the truly horrible. If "A" and "B" weren't "different enough", then the model was made into one version. That is stated in the article and I have quoted it above. Why are you having such difficulty understanding what is in print?
One would expect report of 'preferred' to correlate to report of 'superior', unless some specific criteria for 'superiority' were mutually established. Ditto a word like 'accurate', which would necessarily require either some reference to be 'accurate' too, or some training to hear same. If you actually familiarized yourself with Olive's work, you'd understand this and you'd understand that many of his experiments *did* involve listener training -- a he himself mentions on that page, where he refers to subsequent work.
DBT of loudspeakers is for bias-controlled study of *preference* of their different sounds, not simple detection of difference.
But there was acknowledged bias in the testing procedure.
....in the *sighted* tests. Such biases are *controlled for* by blind comparison...and as a result the supposed sound-based preference for the big, pricey loudspeakers over the others ones went away -- suggesting it wasn't due to the sound in the first place. THAT IS THE POINT.
The author does not state whether the "negligible perceptual differences" were found during the sighted or unsighted tests.
He doesn't need to. Can you not read the graphs? There's a wealth of information in them, including the answer to your question. The statement refers to Figure 1 (which is the line graph on the RIGHT of the page, btw), which shows that during sighted comparisons there is no significant difference (at a 95% confidence level) between *preference ratings* for G&D , but a huge difference between them for G&D versus the other two; there's also a significant difference between T and S, with S coming in 'worst'. When the same loudspeakers were compared blind, G&D were *still* insignificantly different from each other in terms of relative sound quality -- there's your answer, quality difference between them was 'perceptually negligable' in both conditions -- but the real kicker is that when sighted bias factors are removed, the smaller, cheaper-looking, last-place S is now 'negligably different' in sound quality from former top performers G & D, and also from T. The only significant difference remaining was between T and D, and it appears small.
The issue here is one of missing information. A good salesperson knows not to tell lies. However, a good salesperson, attorney, priest, physician or poll taker also knows when to leave out certain truths which would make a less compelling case. This brief article is such a statement IMO. What is said and implied is meant to lead to a predetermined conclusion. That conclusion is predicted in the title of the article, one really need not go any further to know the outcome that was sought. The full truth of the matter might be very "different" than what has been told. Or are you that unfamiliar with how "honesty" actually works?
This is another case of; if you don't understand this concept, take a print out of this thread with you when you do your next jury duty call. Show it to the a
attorneys, tell them you don't get this "leaving out details" stuff and I suspect you will be home early that day.
Silly blusterer. You confuse your own incomprehension with omission of details on Olive's part. The funny thing (besides your holding up 'good' salespersons, attorneys and priests as models for truth-tellers) is that, the more details Olive would add to this from his published work, the more your belief structure would be challenged, not supported. I can say with confidence this because I've read the work (including the paper cited on the page) -- have you?