A comparison of the effects of measured, predicted, estimated and constant residual volumes on the body density of female athletes.

The body density (BD), and hence the relative body fat (% BF) was measured for 182 female athletes. The residual volume (RV) was determined both before and after the underwater weighing by a multiple breath helium dilution technique with the subject immersed to neck level. The absolute mean difference (lXdl) and SEE between the two RV trials were 63 and 75 ml, respectively. These increased to values ranging 144-685 and 187-252 ml, respectively, when the mean of the two RV trials for each subject was compared with the RVs predicted via regression equations, estimated from the vital capacity (VC) and assumed to be a constant of 1000 ml. A similar trend resulted from variation of only the RV in the BD formula for each subject. The two RV trials resulted in an lXdl and SEE of .00121 (.5% BF) and .00141 g.cm-3 (.6% BF), respectively, but these increased to values ranging .00283 (1.3% BF) -.01291 (5.7% BF) and .00362 (1.6% BF) -.00527 g.cm-3 (2.5% BF), respectively, for predicted, estimated and assumed constant RVs. In all cases, the lowest lXdl and SEE were associated with the RVs predicted by a multiple regression equation (R = .725; SEE = 187 ml) which was generated on our sample while the largest lXdl values were registered by the other regression equations. These data emphasize that the use of predicted, estimated and constant RVs result in substantial errors in BD and % BF compared with those when the RV is measured.