Abstract
Although many data of cranial as well as other skeletal measurements have up to the present been accumulated in the field of anthropology, there can be seen few reports on their errors. In this article, the intra- and inter-observer errors of 22 measurement items, used frequently in Craniometry, were assessed and discussed.
Intra- and inter-observer measurement error variances of each item were firstly obtained through the analysis of variance for balanced hierarchal classification based on the data which were provided by six observers measuring the same five male skulls of Edo era independently twice. As a result, the greater intra-observer error variances were recognized, especially, in minimum frontal breadth, orbital breadth and facial profile angle (Tables 1 and 2). The significant systematic errors due to observers were found in eight of the 22 items, i, e., biauricular breadth, biasterionic breadth, orbital breadth and height, glabello-inion length, cranial base length, nasal height, and upper alveolar breadth (Tables 1 and 2).
Secondly, it was ascertained by the principal component analysis that there was no common error source among the 20 linear measurement items. Finally, the cluster analysis of the present sub-samples for assessing measurement errors and several samples from literature suggested that errors, especially inter-observer, might greatly affect the interpretation for the results of population studies (Fig, 1).
From the above, it was re-realized that any result based on measurements should carefully be interpreted taking their errors into account.
