Type 1 and Type 2 Errors in the Statistical Analysis of Listening Tests
When the conventional 0.05 significance level is used to analyze listening test data, employing a small number of trials or listeners can produce an unexpectedly high risk of concluding that audible differences are inaudible (type 2 error). The risk can be both large absolutely and large relative to the risk of concluding that inaudible differences are audible (type 2 error). this constitutes systematic bias against those who believe that differences are audible between well-designed electronic components that are spectrally equated and not overdriven. A statistical table is introduced that enables readers to look up type 1 and type 2 error risks without calculation. Ways to manipulate the risks are discussed, a quantitative measure of a listening test's fairness is introduced, and implications for reviewers of the listening test literature are discussed.
Click to purchase paper 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, $5 for AES members and is free for E-Library subscribers.