Direct and indirect interrelations between anthropometric and physiological variables in athletic and non-athletic adolescent girls: a path-analytic study.

Our previous analysis of anthropometric and exercise test data of 62 athletic and 56 non-athletic girls (age range 10.5-15.5 years) showed that the intensity of habitual exercise failed to discriminate between the group means of the studied variables. However, the patterns of intervariable correlations differed between the subgroups categorized by physical activity. The present paper studies the problem of this difference further by using exploratory multivariate regression of aerobic power (VO2max) on 10 anthropometric variables and age. The VO2(max) regression was significant (Y(nonathletic) = 0.0194x(1) + 0.004x(2) - 0.371x(3) + 0.045x(4) - 0.177x(5) + 0.070x(6) - 0.271x(7) - 0.170x(8) + 0.015x(9) - 0.0005x(10) + 0.185x(11), SEE = 0.37, R2 = 0.71, F(11.44) = 9.63; Y(athletic) = 0.029x(1) + 0.063x(2) + 0.277x(3) - 0.030x(4) - 0.069x(5) + 0.151x(6) - 0.148x(7) + 0.001x(8) + 0.018x(9) - 0.019x(10) - 0.065x(11), SEE = 0.32, R2 = 0.71, F(11.50) = 11.30), but none or only one of the independent variables had a significant partial regression coefficient. The individual VO2(max) estimates were studied in both groups by using the other group's regression formula to rule out sample dependence. Both formulae gave good approximations of the observed values in spite of the dissimilar regression coefficients. The path analysis of the respective criterion-predictor correlation coefficients confirmed that the relationship of the predictor variables with VO2(max) involved quantitatively direct and indirect effects in the non-athletic and athletic groups.