The conventional .05 significance level used to analyze typical listening tests can produce a much larger risk of concluding that audible differences are inaudible than concluding that inaudible differences are audible than concluding that inaudible differences are audible, resulting in strong systematic bias against those who believe differences are clearly audible between well designed components that are spectrally equated and not overdriven. This paper discusses ways to equalize error risks, introduces a quantitative measure of a listening test's fairness, discusses implications for literature reviewers, and presents a statistical table enabling readers to conduct equal-error analyses without calculations.
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.