Talk:Accuracy and precision

Definition of precision?

I'd thought the definition of precision was related to the smallest measurable unit of the measuring device, or the least significant digit reported in a measurement. For example, a balance which has a readout showing tenths of grams has a precision of 0.1g, whether or not it's accurate to that level. Similarly, a mass reported as 453.59g has a precision of 0.01g, even if the measuring equipment is not capable of that accuracy. That would seem to tie in to the "false precision" section better. Anthony Argyriou 15:34, 17 July 2008 (CDT)

I'd be delighted to find a definition that says that; it's pretty much what I always assumed, but those NIST and ISO publications I found disliked the term. My favorite practical definition came from one of my professors, who would tell her organic analytical chemistry students that there were between 2 and 4 unknowns in a sample. She knew that some of her students, already in the industry, just needed her course for graduation. She also knew the sensitivity of the instruments in the student lab, and strongly suspected that some of the graduation-requirement students were just taking the sample to their in-house lab and telling them to analyze it. Basically, she gave an unknown with 2 substances that could be detected with the equipment in the college lab, and two that could not. If a student handed in an an analysis with four correct identifications, she'd flunk them for cheating unless they could prove they did the analysis themselves at their corporate labs.

Alternate text is very welcome. Howard C. Berkowitz 16:17, 17 July 2008 (CDT)

So digging around on Wikipedia, there's a great image at Accuracy and precision which explains things pretty well. There's also Arithmetic precision, which is what I've understood precision to mean, partially incorrectly. Anthony Argyriou 17:02, 17 July 2008 (CDT)