An Analysis of Errors in Craniometry

Twenty-six measurement items taken three times on 35 crania by 10 observers independently were analysed for the influence of careless mistakes and measurement errors. The median number of careless mistakes per trial was 7.5. Nested analysis of variance showed that 8 measurement items had greater than 10% total error variance (Table 1). The influence of careless mistakes and inter-observer error on population comparison was examined by cluster analyses of several distance coefficients. The results indicated that the influence of mistakes could be minimized by eliminating the outlying measurements outside the range of x±3S. D., and that the measurement items with greater total error variance should be excluded to minimize the influence of inter-observer measurement errrors.