Evaluation of Juvenile Stature and Body Mass Prediction Paul W. Sciulli* and Samantha H. Blatt Department of Anthropology, Ohio State University, Columbus, Ohio 43210 KEY WORDS juvenile; stature; body mass; accuracy; bias ABSTRACT This investigation evaluates the per- formance of juvenile stature (from tibia and radius lengths) and body mass (from breadth of the femoral distal metaphysis) prediction equations based on the Denver Growth Study sample (Ruff C. 2007. Am J Phys Anthropol 133 698–716). The sample used here for evaluation is an independent sample of juveniles brought to the Franklin County (Ohio) Coroner in 1990–1991. The Ohio sample differs somewhat from the Denver reference sample: it includes 25% African- Americans (rather than all European-Americans), a significant number of right limb bones were measured (rather than all left side), it includes a wider range of economic statuses and it includes individuals who died from disease and trauma. As such the composition and measures of the Ohio sample correspond more gener- ally to that seen in skeletal samples so that the accu- racy of the estimates from the present sample should approach those found in practical applications of these methods. Results indicate that both juvenile body mass and stature are estimated relatively accurately. Accu- racy of body mass estimates for 1-13-year-old juveniles is similar for African-American and European-Ameri- can males and females. The least accurate estimates are for individuals in the 8–13 years age class (excluding individuals with body mass indices greater than the age specific 95th percentile): n 5 9, 6 2.9 kg, 95% confidence interval 1.4–4.4 kg. Accuracy of stature estimates for 1-17-year-old juveniles is comparable for the tibia and radius and, as with body mass estimates, are similar for African-American and European- American males and females. For combined age, sex, and ancestry groups average accuracies are in the 63.5 to 66.5 cm range. Some limitations of the meth- ods are discussed. Am J Phys Anthropol 136:387–393, 2008. V V C 2008 Wiley-Liss, Inc. Ruff (2007) presented juvenile stature and body mass prediction equations based on skeletal measures for a subset of the Denver Growth Study sample (McCammon, 1970). The prediction equations followed 20 individuals longitudinally, 10 males and 10 females with the most complete radiographic information, and covered yearly intervals from 1 to 17 years of age. The results of the analyses indicated that stature and body mass can be estimated from juvenile skeletal measures with errors equal to or less than errors associated with stature and body mass estimates for adults. As noted by Ruff (2007), variation in body size is com- monly used to assess juvenile health status within and among populations. Comparisons among living and recent populations are facilitated by the data available for a large number of living and recent populations (Eve- leth and Tanner, 1990). The ability to estimate accu- rately juvenile stature and body mass in skeletal sam- ples would allow more direct comparisons between these populations and living populations. Furthermore, these estimates would expand information on growth and de- velopment to include a greater range of genetic composi- tions, environmental situations, and genotype-environ- ment interactions. Accurate estimates of juvenile stature and body mass would also provide valuable forensic in- formation for identification of recent skeletal remains. However, even though the results of the analyses of the Denver Growth Study sample indicated that juvenile stature and body mass could be estimated relatively accurately the investigation had a number of limitations which in practice could affect the accuracy of the esti- mates when the methods are applied to other samples. These limitations include the small sample sizes (n 5 18–20) on which the prediction equations are based, the composition of the sample which consists of modern juve- niles from the United States, mostly of northern Euro- pean ancestry, and primarily middle to upper class, pre- diction equations derived from skeletal measures from the left side only, and the necessity for an age determi- nation to the nearest year in order to apply the predic- tion equations (Ruff, 2007). In addition, errors in esti- mating juvenile body size differ for the skeletal meas- ures. For example, femoral head breadth has smaller associated errors in estimating body mass than the dis- tal femoral metaphyseal breadth in the 7–13 years age range and lower limb bone lengths provide better stature estimates than upper limb bone lengths (Ruff, 2007). Thus, it would be useful to know how these prediction equations perform for an independent sample of individ- uals that is known to differ in some respects from the composition of the original study sample, whose skeletal measures differ (right and left sides) from those in the original sample, and does not have measures available that have yielded body size estimates with the smallest associated errors. *Correspondence to: Paul W. Sciulli, Department of Anthropology, The Ohio State University, 219 Lord Hall, 124 W. 17th Ave., Colum- bus, Ohio 43210. E-mail: sciulli.1@osu.edu Received 9 November 2007; accepted 28 January 2008 DOI 10.1002/ajpa.20820 Published online 18 March 2008 in Wiley InterScience (www.interscience.wiley.com). V V C 2008 WILEY-LISS, INC. AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 136:387–393 (2008)