Analysis of the Errors in Somatological Measurements in Man

Abstract

Anthropometry plays a fundamental role in various fields of physical anthropology, but we have little knowledge about the error or the reliability of the measurement of the human body. The present study was made to assess the order and source of the error which were inevitable in execution of anthropometry and fluctuatingsubject to the effect of training. Subjects consisted of eight females aged 20 s (3 were slender, 2 were obese and 3 were medium ; according to Rohrer index). For each subject, 20 items (i. e., 7 heights, 2 breadths, 3 lengths, 5 circumferences, 2 subcutaneous fat thicknesses and body weight) were measured by 4 observers repeatedly at three successive times in a day and at the day after 2 weeks. The series of variances was analysed according to the analysis of variance method, and the following results were obtained :1) Standard deviations of errors (σE) varied from 1.0mm (subcutaneous fat thickness to 8.7mm (chest circumference). Errors rate was low in stature and cervical height (less than 0.5%), and high in bi-iliac breadth and waist girth (more than 0.5%).2) Measurement items more susceptible to the observer's bias presented larger standard deviaton of errors (σE). The deviation of the observers means (σB) was large especially in iliospinal height and biiliac breadth (7.07mm).3) Concerning the intea-observer and inter-observer stability, the stature was considered as the most stable and reliable measurement, neither influenced by the observer's technique nor by the subject's constitution.4) Classifying the measurement items according to the kinds of measuring tools, coincidence of observer's and subjects' rankings were examined. Some items of the lengths and the subcutaneous fat thick- nesses showed well coincidence in rankings. The mean value of the most reliable observer did not always coincide with the total mean. Errors of obese subjects were large in the items around the shoulder.5) In most items, errors became smaller and the accuracy of the measurement increased in the 2nd time examination than in the 1St. It was suggested, therefore, that the effect of training was appreciable.