Fault mix and inspection performance

Publisher Website
Google Scholar
Abstract
Eight subjects performed a paced conveyor inspection task of small zinc-plated washers. There were two fault types, one considerably more difficult to detect than the other. Each fault type had a probability of occurrence of 0-0, 0-025 or 0 05 providing eight different batch conditions, four where only one fault type was present and four where both were present. All subjects were tested twice on each of the eight conditions. There were 1800 washers in each batch condition, which were inspected at a rate of 3 per second. The results demonstrated that the probability of detection of either fault type was significantly reduced when subjects were searching for the two fault types simultaneously, the decrease in probability being greater the larger the a priori probability of the other fault type. Signal detection theory analysis suggested that these results were due to changes in sensitivity rather than in response bias. The classical vigilance effect of signal probability was not observed.