***Anyone that knows anything about trial design, knows that correctly designed and implemented DBTs are virtually impossible when comparing electronics. An incorrect or flawed DBT may introduce errors which can make the results of the DBT neither reliable nor valid. ***


Really..... see my post above. How hard could doing a single blind comparason without you knowing the identies of the units be? The volume difference issue could be easily solved by having the assistant reducing the volume to zero on each unit as it is put into the system, and you adjust it to taste from there. Personally, I wouldn't take the time to do such a comparason, since, being the outcast that I am, believe that the sound difference in most electronics is swamped by the differences in program material / how your hearing is on a particular day / if you just had an argument with the significant other / relative humidity / sound absorbtion from more people in the room etc.

Soundhound: Single Blind is NOT DBT(Double Blind Testing). And, Single blind testing introduces all its' own inherent biases and errors as a result of the investigator's knowledge of the unit being tested and the lack of testing randomness.

***Single Blind is NOT DBT(Double Blind Testing). And, Single blind testing introduces all its' own inherent biases and errors as a result of the investigator's knowledge of the unit being tested and the lack of testing randomness.***

We're not talking about rocket science or creative accounting practices here - simply which do you prefer!!!

Why would single blind inherently lack randomness?

brianca

Because someone is making an active decision as to which unit is being tested, when, and in what order.

Single blind can have as much randomness and double blind, it's just that the order of the 'subjects' is determined by a tester, who knows the identites of the units.

In that case, the tester can create a random number generator, using sophisticated scientific methods - like, uh, flipping a coin or something.....

EXACTLY my point. I prefer careful crossover listening without all the rocket science. However, if you intend on going the science route, either do it right or you're doing it wrong.

Having an investigator come up with and manually select the unit for testing, opens the protocol to suspicion and questioning, and probably introduces testing error variables which can confound the results thereby increasing the total trials needed to reach a significant p value.

If your study can reach a significant p value in ten tests or fewer, then the difference is so significant and obvious as to make the blind testing un-necessary in the first place.

Why not carefully listen to each unit a few times and then if you cannot discern a difference or make a decision, either buy the cheaper unit or then do your blinded testing?

I couldn't agree more - and no, I'm not in the drug business.



Um, because you're biased, like it or not? If the differences are signifcant as is often claimed, a double blind test (ABX or not) will show it.

Charlie

***Why not carefully listen to each unit a few times and then if you cannot discern a difference or make a decision, either buy the cheaper unit or then do your blinded testing? ***

Well, that was exactly my point in the first place. Going the scientific route is pretty pointless, I think, for something so elemental as home theatre, but you had previously asked me how I would propose to compare a 1066 vs a 950 in a blind situation, and I told you. I did not propose to make the test any more sophisticated than it deserves to be. You are absolutely right in that if you do only a few trials and can hear a difference, then the difference must be so gross that ANY comparasion (dare I say an ABX test...sorry...) would pick it up.