"Race" specificity and the femur/stature ratio

• Published in: American journal of physical anthropology Vol. 100; no. 2; pp. 207 - 224
• Main Authors: Feldesman, Marc R., Fountain, Robert L.
• Format: Journal Article
• Language: English
• Published: New York Wiley Subscription Services, Inc., A Wiley Company 01-06-1996; Wiley-Liss
• Subjects: African Continental Ancestry Group; Analysis of Variance; Anthropological methods; Asian Continental Ancestry Group; Body Height; body size; European Continental Ancestry Group; Female; Femur - anatomy & histology; forensics; fossil hominids; Human paleontology; Humans; Male; Methodology and general studies; Osteology; Physical anthropology; Prehistory and protohistory; Race; Reproducibility of Results; statistical validation; stature estimation; Human palaeontology; Anthropology; Method; Statistics; Morphology; Race
• ISSN: 0002-9483; 1096-8644
• DOI: 10.1002/(SICI)1096-8644(199606)100:2<207::AID-AJPA4>3.0.CO;2-U

This inquiry explores a series of problems related to the femur/stature ratio first raised by Feldesman et al. (1990). In particular, we used a revised data set and a more elaborate research protocol to address questions pertaining to: (1) whether the femur/stature ratios of three quasi‐geographic “races” (“Blacks,” “Whites,” “Asians”) are statistically significantly different; (2) whether these are statistically (as opposed to biologically) coherent groups; and (3) whether the “race”‐specific ratios are more accurate than the simple generic femur/stature ratio.
We used ANOVA, ANOCOVA, post hoc analysis, k‐means cluster analysis, linear discriminant functions, and approximate randomization to determine whether the group differences in the ratio were significant, and to assess the coherence of the “racial” groups themselves. We used validation procedures including mean absolute deviation, mean squared error, and Pitman's measure of closeness of a known sample of 798 femur/stature pairs to compare the accuracy of the generic ratio and the group‐specific ratios.
The results confirmed that the “Black” femur/stature ratio is significantly different from those of “Whites” and “Asians”; however, group coherence was poor, with results barely better than chance.
We found that “race”‐specific ratios slightly outperform the generic ratio when “race” is certain, but the gains are small for the assumptions required. More significantly, however, we found that when “race” attribution is uncertain or unknown, as in paleoanthropology, the wrong ratio (or the wrong regression equation) performs poorer than the generic femur/stature ratio. As a result, we recommend that researchers continue using the generic femur/stature ratio to estimate stature in pre‐ and protohistoric populations. An alternative equation, a generic regression, yields even better stature estimates; however, we urge further study before recommending that researchers use this instead of the more thoroughly tested generic femur/stature ratio. © 1996 Wiley‐Liss, Inc.