The present study attempted to determine regression equations which can be used to predict body density from anthropometric indices. Multiple regression and factor analysis were applied to variables including body density and other anthropometric indices such as height, weight, girth and skin-fold thickness (SFT), age, and sports experience. The data used were from 81 female college students. Body density was calculated by body weight and whole body volume, which was measured by an underwater weighing method in swimming pool. The main findings were as follows;
1. The body density of the above mentioned students estimated by this method was 1.0480±0.0126 (mean±S. D.: standard deviation). That of subjects with no sports experience (N=23) was 1.0433±0.0120. This approximately agrees with the figures for a normal group, measured and estimated by other researchers.
2. Variables for the factor analysis were age, sports experience, height, weight, body density, girth of chest, waist, upper arm and thigh, and SFT of triceps, subscapular, lateral abdomen and front of the thigh. Factors used for this analysis were Factor 1, which indicated body weight and girth, Factor 2, in which SFT was measured, Factor 3, indicating trunk SFT, and Factor 4, denoting body height. Body density was included in Factor 2. The eigenvalues of factors one to four were over 1.0, which accounted for 77.7% of the cumulative contribution rate (coefficient of determination).
Meanwhile, another factor analysis was done using variables frequently measured in the field of health care services: age, height, weight, body density, girth of chest, waist and upper arm, and SFT of triceps, subscapular and lateral abdomen. Factors used for this analysis were Factor 1, which indicated body weight and girth, Factor 2, SFT, and Factor 3, body height. Body density was included in Factor 2. The eigenvalues of factors one to three were over 1.0, which accounted for 67.3% of the cumulative contribution rate.
3. The multiple linear regression equation for screening those currently engaged in sports practices is Yc=1.062221-0.001166⋅X
1+0.005512⋅X
2, where X
1: triceps SFT, and X
2: sports experience score; 2=active, 1=active in the past, 0=no sports experience, and R=0.619(
p<0.001). For the screening of the obesity of general subjects, another equation is applicable: Yc=0.994044-0.000979⋅X
1-0.000755⋅X
2+0.000693⋅X
3, where X
1: triceps SFT, X
2: weight, and X
3: height, and R=0.590(
p<0.001). The results of these multiple linear regression equations showed no significant difference compared with thosee previously reported. Body density estimates using these variables and equations are simpler and more effective in the estimation of both aspects of body composition (lean body mass and fat mass), than those of other researchers.
View full abstract