The Automation of Regression Modeling in Osteometric Sorting: An Ordination Approach1. Issue 3 (21st July 2017)

• Record Type: Journal Article
• Authors: Lynch, Jeffrey James
• Is Part Of: Journal of forensic sciences. Volume 63:Issue 3(2018)
• Page Start: 798
• Page End: 804
• Publication Date: 2017-07-21
• Subjects: forensic science -- osteometric sorting -- commingling -- principal components analysis -- canonical correlation analysis -- regression; Medical jurisprudence -- Periodicals; Forensic sciences -- Periodicals; Forensic Medicine -- Periodicals; Gerechtelijke geneeskunde; Gerechtelijke chemie; Gerechtelijke psychiatrie; 363.2505
• DOI: 10.1111/1556-4029.13597
• Languages: English
• ISSNs: 0022-1198

Abstract

Abstract: This study compares the original osteometric sorting association method with an ordination approach across all combinations of the humerus, ulna, radius, femur, tibia, and fibula. This includes both the original prediction interval and t ‐statistic approaches. Standard measurements are utilized in the models with full measurements combined and without length measurements. The sample is the osteometric sorting reference from the Defense POW/MIA Accounting Agency. A full set of performance statistics is provided. Results indicate the ordination approach outperforms the original in the majority of bone combinations. Models with length measurements have more exclusion power than those without. It is recommended for the ordination approach to supersede the original when applied to large commingled assemblages.