Optimum threshold crossings and time-window validation in threshold pure-tone computerized audiometry.

Two versions of the "10 dB down-5 dB up" rule were computerized: (1) with a "lax" time-window response validation in which a subject had to press and release a button within the window of 15-3750 ms after onset of a 1-s tone in his earphone (this was designed to stimulate the condition often found in manual audiometry) and (2) with a "strict" condition in which he had to respond within 60-795 ms after onset of the tone and release the button within 15-750 ms following the tone. A threshold-seeking run was continued using 5-dB steps until the threshold had been crossed six times. With 24 subjects it was found that 13 of 288 runs did not yield the "ASHA" criterion (three ascending series ending on the same hearing level (HL), out of any six or fewer consecutive series), but of these 13 runs all but one run did meet the "ANSI" criterion (two ascending series ending on the same HL, out of any three or fewer consecutive series). The ANSI Hearinnd yielded essentially the same means. An ANSI HTL was established within four or fewer ascending series for 95% or more of the runs, while an average of five extra stimuli was needed to pursue the run until an ASHA HTL was achieved (a savings of nearly 5 min per person). It was recommended that the ANSI Criterion for HTL be adopted for many audiometric situations, that a maximum of four ascending series be completed in any single threshold-seeking run, and that rather strict time windows be set within which onset and offset responses may register.