测量的不确定性 - 如何量化结果的歧义？

Ambiguity or uncertainty refers to a degree of doubt on the results of testing. It is an accepted fact that it is never possible to report the actual value as there is always a scope for error, however small it may be, that results in variation between the true value and reported value.

Measurement is truly valid if it is accompanied by definition of degree of uncertainty. Such uncertainties arise from deviations from stipulated environmental controls, non-homogeneity of sample, method or instrumental limitations and personal operator errors. Some of these errors are measurable and can be corrected through calibration whereas others are indeterminate. In other words every measurement or test should be defined in terms of two parameters. First is the bandwidth or interval defining the accuracy of the measurement and the second is the level of confidence.

许多因素导致化学实验室进行测量的不确定性。其中一些因素是：

• 样品同质性
• Sample degradation between collection and analysis
• Contamination during sampling, sample treatment or storage
• Incomplete extraction of analyte from sample matrix
• Interferences arising from other sample matrix components
• Variations in recommended environmental controls
• Personal bias in recording observations
• 重或volu中的错误metric transfer of solutions
• Degradation of standards and reference materials over storage

表达不确定性

A set of repeat observations should be made and both mean and standard deviation (SD) are reported. From these the value of uncertainty, U, is calculated as

\（u = \ frac {sd} \ sqrt {n} \）

其中n是观察的数量。

The combined uncertainty resulting from several factors is expressed as

Combined uncertainty,

\(U_C = \sqrt{a^2 + b^2 + c^2 + d^2+……}\)

where a, b, c, etc are individual uncertainties.

Expanded uncertainty, Ue, is arrived at by multiplying the combined uncertainty with the coverage factor, k

\(Ue = kU_c\)

Average factor k of 2 is used to report expanded uncertainty at a confidence level of 95%

Finally the observation is reported to 2 significant digits as

结果=平均值+/- UE

在任何分析中，重要的是要欣赏可能导致测量不确定性的因素的作用。有必要通过频繁地校准参考材料和实验室间的能力测试来识别，最小化和尽可能量化它们。