Effect of step size on clinical and adaptive 2IFC procedures in quiet and in a noise background.

Abstract

Audibility thresholds for a 1000-Hz sinusoid were measured with a standard clinical (CLIN) procedure and a two-interval, forced-choice (2IFC) adaptive procedure bracketing 79% correct. Both used 2- and 5-dB step sizes in quiet and in a continuous, broadband noise background. Clinical thresholds were from 2 to 4 dB higher than 2IFC thresholds, depending on the condition. Step size had a larger effect on the CLIN thresholds than the 2IFC thresholds. For the CLIN procedure, thresholds with a 2-dB step size were 1.4 dB lower than with a 5-dB step size. For the 2IFC procedure, thresholds with a 2-dB step size were 0.8 dB higher than with a 5-dB step size. Reliability, as measured by the intrasubject standard deviation, was better for the 2IFC than for the CLIN procedure and better in noise than in quiet. Reliability was unaffected by step size. Adding extra trials to the 2IFC adaptive track decreased the variability across threshold estimates, but more for the noise background than the quiet background. The efficiency of the 2IFC procedure was fairly constant across track length in noise, but decreased for longer track lengths in quiet. In both quiet and noise backgrounds, CLIN procedures were much more efficient than 2IFC procedures.