[jsaliga]

I'm done with this thread and won't be back. I'll say that your personal gripes with SmellyGas have simply sucked the life out of this topic for me and it just isn't worth any more of my time.

I have asked a moderator to review this thread. If they think it's ok then I'm fine with that...but regardless: you two really need to take this offline. No one else cares about the pissing contest going on between you two.

--Jerome

 

[SmellyGas]

Quote:

Originally Posted by jsaliga /img/forum/go_quote.gif I'm done with this thread and won't be back. I'll say that your personal gripes with SmellyGas have simply sucked the life out of this topic for me and it just isn't worth any more of my time. I have asked a moderator to review this thread. If they think it's ok then I'm fine with that...but regardless: you two really need to take this offline. No one else cares about the pissing contest going on between you two. --Jerome

Hey, I don't even know who this guy is!! Regardless, at least a rough calculation of alpha and beta needs to be done for a given test methodology in order to determine the required sample size beforehand. This is very basic and essential component to experimental design.

Calculation of beta requires a prediction of the expected test result if there truly is a difference between the two groups (e.g. cables). For instance, in a simple blinded A/B or same/different test, you would need to predict what % correct your listeners can achieve, given the expected differences between cables. In this fashion, you can estimate that your particular test methodology is "powered to detect ABC difference with a small XYZ chance that you will fail to detect an actual difference." [where ABC is the "difference," or the minimum predicted % correct your listeners can achieve if given an unlimited # of trials; and XYZ chance is the probability of type II error].

Failure to consider alpha and beta represents a serious flaw. For a simplified explanation, look no further than an article in Stereophile:
Stereophile: The Highs & Lows of Double-Blind Testing

For a more mathematical explanation of alpha and beta and how it applies specifically to blind listening tests, I have two JAES articles I can provide you a reference for (if anyone is interested). PM me for this.

 

[b0dhi]

Well done troll, the thread is dead. Mission successful.