A methodological comparison of discriminant function analysis and binary logistic regression for estimating sex in...

Area of Science:

• Forensic Anthropology
• Biometrics
• Statistical Analysis

Background:

• Accurate sex estimation is crucial in forensic investigations.
• External ear anthropometry offers a potential source of biometric data.
• Multivariate statistical methods are increasingly used for biometric analysis.

Purpose of the Study:

• To evaluate the accuracy of discriminant function analysis (DFA) and binary logistic regression (BLR) for sex estimation using human external ear anthropometry.
• To compare the performance of DFA and BLR in discriminating sex based on anthropometric ear measurements.

Main Methods:

• A cross-sectional study involving 497 participants (233 males, 264 females) aged 18-35 years from North India.
• Collection of 12 independent anthropometric measurements from both left and right ears using sliding calipers.
• Application of DFA and BLR to discriminate sex based on the collected anthropometric data.

Main Results:

• Both DFA and BLR achieved comparable accuracy in sex estimation, with predictive percentages of 76.3% (DFA) and 76.2% (BLR).
• Sensitivity, specificity, positive predictive value, and negative predictive values were nearly identical between the two methods (difference ~0.02).
• Classification error, B-index, and Matthews correlation coefficient indicated equivalent performance for both statistical models.

Conclusions:

• Discriminant function analysis and binary logistic regression are equally capable of estimating sex from external ear anthropometry when statistical assumptions are met.
• The study recommends the interchangeable use of DFA and BLR in forensic research and casework for sex estimation and other forensic applications.
• External ear anthropometry provides a reliable basis for sex estimation using established multivariate statistical techniques.