Ward R E, Jamison P L
Department of Anthropology, Indiana University, Indianapolis 46202.
J Craniofac Genet Dev Biol. 1991 Jul-Sep;11(3):156-64.
Craniofacial anthropometry has become an important tool used by both clinical geneticists and reconstructive surgeons. Yet little attention has been paid to the potentially serious problem of measurement error. This paper examines intra-observer measurement error and precision (also called repeatability or reliability) for 52 commonly used anthropometric variables of the head and face. Two factors proved critical to reliability: magnitude of the measurement in question and the degree to which its constituant landmarks could be readily identified. Thus, all of the measurement variables with means above 10 cm proved to have good or excellent reliability. In contrast measurement variables with means below 10 cm were more likely to have poor reliability. This trend was especially evident in variables with means of 6 cm or less where 18 of the 20 variables in this range had poor reliability. The least reliable variables were those like philtrum breadth, columella breadth, and nasal root breadth that combine small magnitude with difficult to define landmarks. While these results suggest that it may be prudent to avoid using craniofacial variables with small dimensions this may be neither practical nor desirable. In such cases repeat measurements may be the best means for optimizing reliability.
颅面人体测量学已成为临床遗传学家和整形外科医生使用的重要工具。然而，测量误差这一潜在的严重问题却很少受到关注。本文研究了头部和面部52个常用人体测量变量的观察者内部测量误差和精密度（也称为重复性或可靠性）。有两个因素被证明对可靠性至关重要：所讨论测量的大小以及其组成标志点能够被轻易识别的程度。因此，所有均值高于10厘米的测量变量都具有良好或优异的可靠性。相比之下，均值低于10厘米的测量变量更有可能具有较差的可靠性。这种趋势在均值为6厘米或更小的变量中尤为明显，在此范围内的20个变量中有18个具有较差的可靠性。最不可靠的变量是那些如人中宽度、鼻小柱宽度和鼻根宽度等，它们大小较小且标志点难以定义。虽然这些结果表明避免使用尺寸较小的颅面变量可能是谨慎的做法，但这可能既不实际也不可取。在这种情况下，重复测量可能是优化可靠性的最佳方法。
