Abstract
Inspection of printed circuit boards frequently requires a judgement of whether an electronic component, such as a resistor or diode, is properly centered. The present study investigated observers' ability performing this discrimination by modeling a Just Noticeable Difference (JND), δS, based upon Weber's fraction, k=δS/S. Using computer-generated figures of mis-centered components, twenty graduate students participated in the experiments of making ‘centered’ versus ‘not centered’ judgements. Both the lengths of the leads (varying from 0% to 16% different) and the separation between the leads (width of resistor) were manipulated, while both errors and RTs were measured. Individual differences and percentage differences between the two sides were quite significant on errors and RTs, regardless of absolute line lengths. A sigmoid threshold response function was calculated within individuals and linearized using the logit transformation (see equation (1)). The composite function was more linear due to cross-subject averaging and separately modeled for each of the 20 subjects. The logit transformation models for 17 out of 20 subjects did not show any significant differences. Weber's Law accurately and significantly captured individual observer decision making in this study. The 50% threshold JND difference judgement of 7% was 3.5 greater than Laming's (1986) 2% JND for line lengths. Clearly, the present task had an additional difficulty associated with their horizontal separation. While differences in separation did not impact accuracy, the mere presence of a separation made this a difficult judgement task. Task design implications were also proposed.
Get full access to this article
View all access options for this article.