Decision Tree-Based Analysis Suggests Structural Gender Differences in Visual Inspection

Among manufacturing companies there is a widespread consensus that women are better suited to perform visual quality inspection, having higher endurance and making decisions with better reproducibility. Up to now genderdifferences in visual inspection decision making have not been thoroughly investigated. We propose a machine learning approach to model male and female decisions with classification trees and base the analysis of gender-differences on the identified model structure. A study with 50 male and 50 female subjects on a visual inspection task of stylized die-cast parts revealed highly significant structural genderdifferences (p = 0:00005), in spite of non-significant differences in overall accuracy (p = 0:34). Going beyond asking which sex is better at a given ability, our results suggest that classifier-based modeling is a promising approach to analyze differences in the structure of cognitive abilities.