Stature estimation from the hand length: an attempt to population non-specific method
|Year of publication
|Appeared in Conference without Proceedings
|The data were based on literature survey of studies where statistical parameters of body height and hand measurements were simultaneously available (right hand: 106 samples of both males and females). On these average values of all populations we computed regression equation by means of the method of reduced major axis (RMA). The newly developed cross-population equation was tested on samples of young adults from the Czech (sexes combined, 218 females, 214 males), Slovak (35 females, 30 males) and Bosnian (91 females, 100 males) whose hands were measured on 2D scans. As an alternative approach we computed simple linear regression equation (least squares, LS) based on the individual data of the Czech sample. When applied the simple LS equation derived from the Czech sample back to the Czech data, the differences (DIFFs) between the estimates and their true stature reached zero mean values and 95% of DIFFs ranged between -91.3 and 92.8 mm (50% between -29.6 and 29.8 mm). Similar results were obtained when tested the Czech LS equations on the Slovak sample. Estimates of the Bosnian sample were systematically higher on average of 10.69 mm. When used new cross-population RMA equations the results seriously underestimated stature of about 60 mm (95% of DIFF ranged between -179.4 and 36.8 mm). The performance of the cross-population equations, however, substantially improved when the data were limited to populations of European origins only, i.e. more affine to the testing samples. According to our results, the differences in hand vs. stature proportions between human populations are too large to allow development of a population-free method possessing an acceptable error of estimates.