Measures of Body Fat and Related Factors in Normal Adults—II
1297
variation for all body measurements within each group. For example, the overall
range of percentage standard weight was 77-233 per cent. Correlation matrices*
were prepared for all anthropometric measurements for the 2053 subjects by
category.
Table 3 presents the means, standard deviations and ranges for anthropometric
measurements, total body fat and LBM by subject category for those subjects who
had laboratory measurements. The ranges of body measurements within categories
Pag
ee
ay
oh
£
total body weight, the correlation is much lower and more variable. Thus, by
subject categories this group of r values was: I, 0.782; II, 0.588; ILI, 0.584; IV,
0.842; and V, 0.613. Standard weight tables are derived from observations of
population groups and therefore require revisions from time to time dependentin
part on changes in the characteristics of the population and alterations of environmental factors. As shown above these tables provide an inconsisient indication of
the relative body composition and the body fatness among race-sex-age groups.
Iliac crest and thigh diameters were not included in Tables 2 and 4 because
they were not obtained on all subjects. However, trial computer runs revealed that
they added practically no predictive information that was not already implicit in
my
the other anthropometric measurements.
EQUATIONS
subject category and combinations of categories. First, the best single-variable
regression equation was calculated using that anthropometric variable which correlated highest with body fat. Second, a two-variable regression equation was calculated by adding that variable which correlated highest with the deviations between the
actual and predicted body fat using the first equation. The second variable, in other
words, was the one which best explained what was unaccounted for by the first
equation. In a similar way, other anthropometric variables were added one by one.
Anthropometric indexes were purposely omitted from the development of the
equations to avoid unnecessary mathematical steps in the practical use of the
equations.
For the most part, the variables entering the equations were ones for which excellent reproducibility between examiners was obtained (see Discussion). Subsequent
fd7
UY
Noa tat
*Copies are available from the authors.
cers
tre
Bllgabe ck
igh ta tte
Sinaies
REGRESSION
Using only those subjects on whom TBW-body density determinations had been
made, the following procedure was used to develop regression equations for each
niet
ix-nh
jh ge ibneny
L. Sahin
Ponderal index
% standard wt.
Saito
atts PEtes 2
apySiebc S ike Fe tate
4
bo
oe ate mabe
senate SAMANL? saad 2 Watistiaah,.VaricealLd
for these laboratory subjects did not differ appreciably from those for all subjects.
The overall range of percentage standard weight for the laboratory subjects was
77-203 per cent. Kilograms of body fat and of LBM showed similar widespread
ranges within the subject categories.
Correlation coefficients of anthropometric measurements and the three indexes
(percentage standard weight, Gubner and ponderal) with kilograms of body fat are
presented in Table 4. Of particular interest is the correlation of percentage standard
weight with body fat in kilograms. For the first five subject categories in the table
these correlation coefficients (* values) vary from 0.811 to 0.958, the older white
men and the Negro men having the lowest correlation coefficients. When the r
values are calculated between percentage standard weight and body fat as percentage