Abstract
This is an attempt to improve the height-age style of weight prediction table currently used both for adults and for children. The measurements studied are: weight, net (W), age (A), stature (S), bi-cristal diameter of pelvic brim (Bc), bi-styloid diameter of wrist (Bs), bi-malleolar diameter of ankle (Bm). The material consists of private school boys aged 4 to 20 years, to the number of 810, except for the diameters of wrist and ankle with populations of 290 and 293 respectively. The linearity of some of the distribution plots is open to question. For instance, weight on stature looks curvilinear at the extremes and the criterion n2—r2 = .0229 ∓ .0072. Nevertheless, I am told by other people familiar with statistics that the plots are sufficiently rectilinear for the present tentative analysis.
The resulting correlation coefficients may be examined in Table I. If other things be equal, that trait most highly correlated with weight would be expected to be the most dependable for predicting weight. If we note especially the relationship of weight to various traits, it appears that this is highest for stature, then bi-cristal, then bi-styloid, then age, and finally bi-malleolar. Furthermore the difference between rwA and rwS proves to be statistically significant; .0379 ∓ .0036, yielding a ratio of the difference to its probable error of 10.4. Likewise the difference between rwA and rwBC is statistically significant: .0240 ∓ .0045, ratio 5.3. But r of weight with bi-styloid diameter of the wrist, although larger than rwA, is not significant: .0047 ∓ .0070. Hence one may conclude that weight is best referred to stature, next best to bi-cristal diameter, only third best to age, while wrist diameter is not worth further consideration.
Get full access to this article
View all access options for this article.