Submitted to: Medicine and Science in Sports and Exercise
Publication Type: Abstract Only
Publication Acceptance Date: 3/19/2007
Publication Date: 5/7/2007
Citation: Turpin, I., Bray, M.S., McFarlin, B.K., Jackson, A.S. 2007. Percent body fat equations for diverse population, aged 17 to 30: the TIGER study [abstract]. Medicine and Science in Sports and Exercise. 39(5 Suppl):S34-S35. Interpretive Summary:
Technical Abstract: Research from our lab showed that the Jackson-Pollock generalized equations gave biased percent fat (%fat) estimates of men and women. The likely sources are increased %fat and the diversity of contemporary subjects. Our goal was to develop anthropometric %fat equations valid for a diverse population of young adults. The sample included 251 men age 18 to 30 years, with an ethnic composition of White 40%, Hispanic 27%, Black 19%, Asian-Indian 5%, and Asian 9%; and 465 women aged 17 to 30 years, with an ethnic composition of White 31%, Hispanic 25%, Black 32%, Asian-Indian 4%, and Asian 8%. Dual energy X-ray absorptiometry (DXA) was used to measure %fat. The mean DXA %fat was 21.7 +/- 8.4 for males and 33.2 +/- 7.4 for females. Multiple regression was used to develop the prediction model. The dependent variable was DXA %fat. Independent variables were the sum of chest, abdomen, and thigh skinfolds (SS) for men; the sum of triceps, suprailium, and thigh skinfolds (SS) for women; and ethnic group. General linear models showed that the quadratic form of SS and ethnic group were independently related with DXA %fat (R = 0.95, SEE = 2.54 for males; R = 0.88, SEE = 3.56 for females). Post-hoc analysis showed that the regression weights for Hispanic (1.1 males, 1.65 females), Asian-Indian (2.76 males, 3.75 females) and Asian (1.90 females only) groups were significantly (p<0.01) greater than 0. These prediction models provide valid anthropometric %fat prediction models for diverse populations of young men and women, aged 17 to 30 years. The multiple correlations found with these data for males of 0.95 is higher and the SEE of 2.54 is substantially lower than the correlation of 0.90 and the SEE of 3.40 reported by Jackson and Pollock (1978). Similarly, for females the multiple correlation and SEE of this model of 0.90 and 3.56 were more accurate than the correlation of 0.85 and SEE of 3.8 reported by Jackson and Pollock (1980).