Sam Heard wrote: > Dear Tom and all > > I must say that quantifying accuracy and uncertainty is very difficult > - and I do like the inclusion of ~ in the set of flags to mean > approximately - when there is no idea of accuracy from a mathematical > point of view. I think we may lose something if we try and get it all > in the notion we currently have of a measured accuracy. > > Sam > notwithstanding the fact that the '~' flag probably worked well in Vince's system, I cannot help but wonder what it really adds; if you don't know either the absolute limits (of the measuring device) or statistical confidence limits, how do you compute with it? How can I write a decision support program that takes notice of a '~' flag?
- thomas
