Bigras-Poulin M
Faculté de médecine vétérinaire, Université de Montréal, Saint-Hyacinthe, Québec, Canada.
Vet Res. 1994;25(2-3):110-5.
The quantification of measurement errors is rarely considered during the preparation of a study protocol in epidemiology. This is in contradiction with the impact of these errors on the assessment of degree of association between various measures, which is often the objective of these protocols. A few important definitions of measurement errors are first proposed. The impact of these errors on the assessment of measures of central tendency and association is studied through examples and simple simulations. These measures must be exact since they play such an important role in estimating risk factors or in evaluating diagnostic tests.
在制定流行病学研究方案时，很少会考虑测量误差的量化。这与这些误差对各种测量之间关联程度评估的影响相矛盾，而评估关联程度往往是这些方案的目标。首先提出了一些测量误差的重要定义。通过实例和简单模拟研究了这些误差对集中趋势测量和关联评估的影响。这些测量必须准确无误，因为它们在估计风险因素或评估诊断测试中起着如此重要的作用。
