The reliability of human odontometric data.

The present analysis addressed the reliability of human dental mensuration, with special reference to intra- and inter-observer error, the effect of time and of tooth type. Two observers were asked to measure the same 35 incisors, premolars and molars, using the same mensurational technique and the same instrumentation. Standard mesiodistal (MD) and buccolingual (BL) measurements were repeated on the next day and after six months. Measurement error was evaluated by means of Pearsonian correlation coefficients, mean absolute error, the Bland-Altman procedure, percentage error and an analysis of variance. It was concluded that human dental mensuration is subject to a large error component which is more marked in the MD than in the BL dimension, higher in the anterior than the postcanine teeth and greater between than within observers. These error components were shown to be random rather than systematic.