The relationship of body fat weight, determined densitometrically, to relative weight and triceps skinfold in American youths, 12-17 years of age

Abstract

The relationship between fat weight, measured densitometrically, and the triceps skinfold and Quetelet's Index has been analyzed by means of linear regression analysis of 235 American youths, 12–17 years of age; 48 males and 41 females were from Minneapolis and 121 males and 25 females from the Greater Philadelphia area. Validation was accomplished by using the results of 10 separate regression analysis, each on 90% of the sample, the remaining 10% serving in each instance as the validation group; this is known as the Jackknife Method. When the Minneapolis and Philadelphia components were pooled, suitable regression models were derived with standard deviations of the errors of prediction of 3.330 kg in males and 2.056 in females. When the sample was subdivided by city the SD's remained essentially the same, but the mean errors were larger. It is concluded that the linear regression models can be used on closely similar populations to estimate fat weight for groups, but estimates for individuals may be expected to have an error of ± 5–6 kg (i.e., 2 SD's) and are therefore not likely to be suitably accurate for use. The regression weights indicate a greater contribution of the triceps skinfold to fat weight than of Quetelet's Index in males. In females the reverse was found. This may affect sex differences in prevalence rates of obesity, if the diagnosis is based upon a single anthropometric indicator.