1Anthropology, University of Illinois at Urbana-Champaign, 2Institute of Dentistry, Barts and the London School of Medicine and Dentistry, Queen Mary University of London
April 18, 2020 2:15PM, Diamond 8-9
Previous studies have indicated that differences between reference samples for age estimation may be due in part to the small sample sizes within ordinal categories. The effect of reference sample size and composition on age estimation parameters are tested here both directly and holistically. The data consist of Moorrees, Fanning, and Hunt scores of the first permanent incisor through third molar of 2607 London children between 2 and 23 years of age. Scoring was performed on radiographs by the second author. Direct testing is performed by computing the standard error of the difference in mean age at transition between groups, varying group sizes and degrees of admixture. Admixture of ethnic groups for a single sex and admixture of sex for a single ethnic group does not affect standard error, when admixture ranges from 0-50%. Standard error increases with decreasing sample size (100% < 80% < 60%) in both designs. The Akaike information criterion of a multivariate cumulative probit model fit to the full dataset shows a significant effect from sex but not ethnic group. Altogether, these findings suggest that the effect of within-stage sample size on mean age at transition may outweigh the effect of other factors such as sex and ethnic group, even when these factors are significant in the model as a whole.