Updated: Aug 9
Excerpt from the VIM: Vocabulaire International de Metrologie
To develop the concept that, in metrology, accuracy is not a number. It is something that we work towards. It does not make sense to say that a particular measurement result "is accurate". We can take certain steps to encourage accuracy such as using the mean or average of many results. The proper approach is provide a measurement result that is as accurate as possible together with a statement of the uncertainty of the measurement.
Note that Hodgson, in the article referred to in St-Onge, defines "accuracy" as:
Excerpts from Brian T. Hodgson, "The Validity of Evidential Breath Testing" Can. Soc. Forensic Sci. J. Vol. 41 No. 2 (2008) pp. 83-96
Q. I see. Now, I just want to ask you, in science, can you help me with what’s meant by an average or a mean? A. That’s the same word, meaning that it’s an average of the numerical values divided by the number of measurements. Q. I see. So, sometimes as a scientist, if you’re running an experiment and you get a large number of values, possible values, you might take the average of those values in order to try and give as accurate of a measurement result as you can, as a scientist? A. Correct. So, if you made eight measurements on that one analysis, you would take all those analyses, the quantitative result, and divide it by the number of measurements to get the average of all those results. Q. So, if we’ve got the truck on the weigh scale at the side of Highway 400 and there are 50 different measurement results, of – they average out to 177,500 of whatever the units is, whether it’s paper clips or whether it’s grams, or whether it’s kilo-paper-clips, or whether it’s kilograms. A. Okay. Q. The 177,500 we might consider to be an average of those 50 possible results. In other words, the measurement result that we take as good scientists would be to take all of the various numbers that we get that might
surround 177,500 and we work out an average of them, a mean. A. Well, you would take all the numerical values, divided by the number of measurements and that would give you the mean or average result of 177,500. Q. And we do that to enhance accuracy. A. Yes. Q. Now.... A. There’s variability associated with any measurement that’s made. Q. All right. I understand that, and we’ll get to precision later. A. Okay. Q. But for now, I just want to get a concept. A. That’s what that concept is. Q. I want to get a concept of what a mean is, what an average is, and I think you’ve explained that.