Cross-Validation of Methods to Predict Body Fat in African-American and Caucasian Collegiate Football Players

Publisher Website
Google Scholar
Abstract
Eight skinfold (SF) equations and bioelectrical impedance analysis (BIA) were cross-validated in 71 Division IA college football players. Criterion percent body fat (%FAT) was determined by hydrostatic weighing (HW) and computed from body density (BD) for African Americans (n = 28) and Caucasians (n = 43). Results were analyzed by method, with position and race effects examined simultaneously using a two-way ANOVA. Analysis included computation of mean difference (MD), correlation (r), standard error of estimate (SEE), and total error (TE). Skinfolds had MDs ranging from -1.8 to 2.9%, correlations from .85 to .93, SEEs from 2.3 to 3.2%, and TEs from 2.3 to 4.2%. Bioelectrical impedance analysis was significantly different from HW and yielded the highest MD (3.8%), lowest correlation (r = .78), highest SEE (3.9%), and highest TE (5.4%). The Jackson and Pollock (1978) SF equation provided the most valid prediction of HW-determined %FAT in a racially mixed sample of college football players, demonstrating the highest correlation, lowest SEE and TE, and an intercept and slope not significantly different than 0.0 and 1.0, respectively.