Dental nonmetric trait intraobserver precision: three observations of a large sample.

The utility of nonmetric dental traits in population phenetic studies is well documented. However, consistency in scoring trait expression is a concern due to the subjectivity in scoring quasicontinuous variables with graded scales. The aims of this study are to 1) analyze intraobserver results from an independent scorer and 2) interpret the least precise results. Six hundred individuals (300 of each sex) were scored for 48 dental nonmetric traits using ASUDAS, in one preliminary (100 individuals) and two full observations. Intraobserver precision between scoring sessions yielded adequate results (85% to 100%), with two notable exceptions (UI2 and LC distal accessory ridge) and three somewhat problematic cases (UM2 hypocone, UM1 Carabelli's trait, and LM1 anterior fovea). Kendall's τ-b rank correlation coefficients are usually moderate to strong (> 0.3). The results are similar to previously published studies of observer error. Error rates are likely caused by inexperience in observable surface selection and trait/grade definition. ASUDAS requires restraint from scoring teeth that are too worn or modified by caries or chipping. A careful study of trait and grade definitions is strongly encouraged. The key element in achieving high precision rates is experience. Intraobserver (and, if needed, interobserver) error should always be calculated. Overall, ASUDAS is a reliable and useful method, especially if the author's recommendations are followed.