Date of Award
Master of Arts
Masters Thesis-Campus Only
The present study addressed the problem of poor defect detection rates in industrial quality control inspection tasks, especially where the existing defect rate is low. The study conceptualized inspection as signal detection, with defects as signals and nondefective products as nonsignal stimuli. Subjects were assigned to find defects among 2,520 angle-shaped figures presented within a 20-min period in a computer-based batch inspection simulation. All subjects received 40 natural signals, analogous to unplanned defects, among the 2,520 stimuli. Subjects in two experimental groups received an additional 40 artificial signals, analogous to deliberate, tagged defects used to manipulate inspector performance. A control group received the natural signals only.
In one of the experimental groups, the artificial signals were identical to the natural signals. In the other experimental group, the artificial signals were in between the nonsignals and the natural signals on the dimension of angle width. It was expected that both of the experimental groups would detect greater numbers of the natural signals than the control group, and that the group with the non-identical artificial signals would perform better than the group with the identical signals by virtue of the phenomenon seen in postdiscrimination generalization peak shift.
The results indicated no differences among the groups in natural signal detection performance. Possible reasons why the expected effects were not seen were discussed.
Urbach, Michael Dennis, "The Use of Artificial Signals and Peak Shift in an Attempt to Improve Defect Dejection in an Inspection Task" (1995). Master's Theses. 3632.