What's Wrong With Loudspeaker Preference Testing?

[microstrip]

Toole (...)

Just wanted to point that IMHO it is impossible to debate or summarize anything on the second part (creating listening experiences) of the great F. Toole book in WBF without a proper discussion of the first part (fundamentals of sound reproduction) and his objectives.

WBF is mostly an high-end audiophile forum, where people valuate essentially their individual or small group experiences and particular preferences. The aim of Floyd Toole is essentially creating a loudspeaker enjoyed by the common listener, discarding most of the particular preferences most of us enjoy so much and spend a lot of time debating. The first part of the book is particular interesting - one of the the real reasons behind the Circle of Confusion exposition is destroying the usual references of audiophiles, explaining the limitations of real stereo recordings and how fragile is our individual evaluation in his perspective. The way preference and "better" are evaluated in his work is purely statistical - something most audiophiles and the high-end do not accept!

Anyway an excellent book including most of F. Toole research and work, that unfortunately is known mostly by the small parts that are exploited by the Harman marketing. IMHO these sections are just the less interesting and are limited by his views about stereo - an inferior format as he says somewhere - a “spatially-deprived medium”. But IMHO knowing the limitations of the format only helps us to better understand and enjoy our hobby.

 

[cal3713]

Lacking the resources of B&O, I opted for a less ambitious approach: Fairly narrow-pattern main array plus a level-adjustable rear-firing array. The dispersion pattern itself may not be adjustable, but the direct-to-reverberant ratio is.

Duke, all you need to do is create a remote control for the rear tweeter, call it the AudKen Remote One, and add some zeros to your price. "With no less than 3 (or 4) drivers and instantaneous situational adjustment via the AudKen Remote One..."

 

[Gregadd]

In order for statistical analysis to be proper one needs to take a sizable sample at random form the community to be analyzed.Otherwise it is just a "straw Poll"

 

[Duke LeJeune]

Duke, all you need to do is create a remote control for the rear tweeter, call it the AudKen Remote One, and add some zeros to your price. "With no less than 3 (or 4) drivers and instantaneous situational adjustment via the AudKen Remote One..."

Okay, but my family has been feeling left out so they want me to spell it "OddKin"...

 

[cal3713]

In order for statistical analysis to be proper one needs to take a sizable sample at random form the community to be analyzed.Otherwise it is just a "straw Poll"

It depends what your goal is. A convenience sample works just fine if your only goal is to determine *what* factors matter. It does not work at all if your goal is to determine *how much* they matter.

If, for example, you look at vision or hearing research in cognitive science, a lot of the studies are based on ~5 people -- all the scientists and grad students in the lab. It works because they're documenting the existence of processes, not trying to state how big of an impact those processes have within the population.

 

[Gregadd]

That's fine, and an appropriate trade-off in many cases; just don't make the mistake of assuming that you can extrapolate from that convenience sample back to your target population.

 

[cal3713]

Agreed. The only extrapolation possible from the convenience sample is that the process "exists in humans" or "can have a perceivable impact."

Of course anyone claiming that they've found the true population value with their "large" sample is probably just as far off. Gonna take a lot of people to zero in on the true value for a population of 7.8 billion.

 

[Gregadd]

Agreed. The only extrapolation possible from the convenience sample is that the process "exists in humans" or "can have a perceivable impact."

Of course anyone claiming that they've found the true population value with their "large" sample is probably just as far off. Gonna take a lot of people to zero in on the true value for a population of 7.8 b

Yet some have tried to it for cultures and nations. Personally I thought it bordered on the stereotypical.

 

[cal3713]

Yet some have tried to it for cultures and nations. Personally I thought it bordered on the stereotypical.

The census does a good job in America. Imagining the effort required to do that for 7.8b individuals is something else. I suppose they'll try at some point.

 

[Gregadd]

Just to be clear there are ways to limit the size of the community under examination. There is no need to define the entire world population.