“測量”(measurement)被廣為使用於一般的生活中,其主要目的為估計未知品(unknowns)的特徵值(characteristic)。然而,在測量的過程中通常會受到誤差(errors)的幹擾。因此,在實驗室中或工業產品的產製過程之測量,通常會利用標準品(standards)來估計誤差,進而來校準未知品的測量值。本文主要由實驗設計(experimental designs)的角度來介紹有關測量過程的研究。首先,介紹測量過程中常見的兩種統計模型:累加性模型(additive models)及線性校準模型(linear calibration models)。再針對不同的模型,介紹相關的實驗設計之研究。實驗設計的主要議題(issues)包含量測過程中標準品與未知品的配置(allocation)及排列方式(arrangement)。最後提出一些關於測量設計未來可能的研究方向。
Measurement making is an activity for most people. The purpose of the measurements is to estimate the degree to which some characteristic of a specimen is present. A measurement process is usually subject to errors which may be classified as random errors only or a combination of both random errors and systematic errors. An experimenter may monitor the errors by measuring standards at appropriate intervals during a measurement process in a laboratory or a product process in industry. This is because that the true values of standards are known, so the errors can be observed whenever standards are measured. Hence, one may use standard measurements to estimate the parameters associated with the errors, and obtain a precise estimate for the true value of unknown specimen by calibrating its observed values. This article discusses some literature concerning two important design issues of a measurement process, including the allocation and the arrangement of order of measurements for standards and unknown specimens. Some possible future research problems are also presented.