Variability of the Absolute Auditory Threshold: A Psychophysical Study

Abstract

The purpose of this study was to explore the variability of the auditory threshold under experimental conditions, for comparison with the known variation in repeated clinical audiometric tests of the same subject. The literature on the reliability of audiometric methods is surveyed briefly. An experiment is reported in which the absolute threshold for pure tones of 1000 cycles per second was studied. Nine normal ears were studied in one to five sittings each. Each sitting involved ascending and descending presentation of the stimulus tones. Intensity was varied in one decibel steps. The subject was required to signal whether he heard each tone. The proportion of “yes” responses at each intensity level did not differ significantly between ascending and descending series. Psychometric functions were constructed by fitting the phi-gamma function to the data. Functions were computed for each sitting of each subject and for all sittings of each subject. No significance could be attributed to differences among the sittings of any subject. A psychometric function was computed which represents the total subject group. This function is the cumulation of a normal distribution with a standard deviation of 6.72 db. From this function a theoretical distribution of the average results of the type of series used in clinical audiometry is derived. The standard deviation of this distribution is 5.57 db for either ascending or descending audiometric series. The dispersion of audiometric tests of the same subject as reported in the literature is in reasonable agreement with the theoretical prediction, if allowance is made for the use of several series at each frequency in the clinical test. It is concluded that the fitting of the phi-gamma function is a satisfactory procedure for the study of the psychometric function for absolute auditory thresholds. This function (for a single subject) does not vary significantly over a period of several weeks. It is concluded also that the variation in clinical audiometric tests of the same subject is attributable to the variability of the subject's sensory behavior, rather than to the nature of the clinical test. The variability which has been observed over periods of several months is demonstrable in much shorter periods.