Classification of hard core and petty criminals using anthropometric measurements

• Published in: Bioscience trends Vol. 4; no. 5; p. 239
• Main Authors: Agrawal, Girdhar G, Asthana, Akash, Maurya, Amit
• Format: Journal Article
• Language: English
• Published: Japan 01.10.2010
• Subjects: Analysis of Variance; Body Weights and Measures - methods; Criminals - classification; Discriminant Analysis; Face - anatomy & histology; Head - anatomy & histology; Humans; India; Models, Theoretical; Statistics, Nonparametric; India
• ISSN: 1881-7823; 1881-7823

The purpose of the present study was to compare the somatometric measurements among hard core criminals, petty criminals and community people. Using standard anthropometric procedures, somatometric dimensions were studied on 250 subjects each from the three groups: (i) experimental (hard core criminal) group, (ii) control-I (petty criminal) group, and (iii) control-II (community people) group. Univariate analysis of variance was used for making comparisons of somatometric measurements between these groups in the univariate case. Quadratic discriminant analysis (QDA) was used to develop a model based on measurements that classifies the cases into groups. The study revealed that the somatometric measurements such as morphological total facial height (p < 0.01), physiognomic total facial height (p = 0.015), nasal length (p = 0.001), height of lower face (p = 0.001), nasal depth (p = 0.002), sitting height vertex (p = 0.011), bigonial breadth (p < 0.001), maximum head breadth (p = 0.001), morphological upper facial height (p < 0.001), and physiognomic ear breadth (p = 0.039) were significantly different between the three groups. Morphological upper facial height, physiognomic total facial height, nasal length and height of lower face could be used as identifying factors for hard core criminals. Morphological total facial height, physiognomic upper facial height, physiognomic ear breadth, and sitting height were found different significantly in the experimental group as compared to the control-II group. The QDA provided an overall 72.4% correct classification of cases and 74.5%, 69.6%, and 72.9% correct classification for the experimental, control-I and control-II groups, respectively. The blurred distinction of the three groups could be explained by using the QDA model.