Multivariate Approaches to Neonatal Assessment of Newborn Babies

Abstract

This study examines gender-based differences in neonatal physical characteristics using multivariate statistical techniques. A total of 1,000 newborns (male and female) were sampled from the Federal University Wukari Teaching Hospital, Taraba State, Nigeria. Key anthropometric variables measured included occipito-frontal circumference (OFC), cranial circumference (CC), length of birth (LOB), abdominal circumference (AC), and weight (WT). Due to perfect correlation with other variables, AC was excluded from the multivariate analysis. The objective was to determine whether statistically significant physical differences exist between male and female neonates at birth. The study employed Hotelling’s T² test and profile analysis; however, the assumptions of homogeneity of covariance matrices (tested via Box’s M) and independence (assessed via scatter plots) were violated. To address these issues, a robust non-parametric permutation-based Hotelling’s T² test was conducted, yielding a statistically significant result (p < 0.001), indicating notable gender-based differences in multivariate mean vectors. While the main effect of Feature was highly significant (p < 0.001), revealing differences among OFC, CC, LOB, and WT, the Gender × Feature interaction was non-significant (p > 0.05), suggesting parallel measurement patterns across genders. The study concludes that gender significantly influences neonatal physical traits and that advanced multivariate methods, including Hotelling’s T² and profile analysis, are effective for analyzing high-dimensional neonatal data—even under violations of classical assumptions such as normality and homoscedasticity.