Number of samples needed?
Number of samples needed?
(OP)
I need to find out a good way to estimate the number of samples I need to take for an analysis. I will have some HLW and LAW simulant that I will have tested for certain markers. I need to make sure that there is only a +/- 1% difference from the control specimen to my samples. Statistically how do I determine how many samples I need to take to get a accurate result?
Thanks a lot,
Thanks a lot,





RE: Number of samples needed?
JMW
www.viscoanalyser.com
Eng-Tips: Pro bono publico, by engineers, for engineers.
Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.
RE: Number of samples needed?
The decision about the accepted quality level is thus the very crucial for your sampling plans and depends on your company quality policy.
The procedure to determine the number of sample units is described in ISO 2859-1,2 and 3(1989)and ISO 3951. Regrettfully I do not know the corresponding ASTM standard but nevertheless, numbers in tables are the same because they are calculated on the same statistics.For the concept of the AQL (ISO 2859) you decide about accepted quality level, you enter the numbers of units in a lot and you choose the sampling plan: you take samples once, twice or more times. With those data you find the number of units you have to controll.
The entire proceedure is available with some commercial software.Personally I like "Statistica" or at the price top end "SAS", but I am sure you can find many more.But with a standard in a paper print form you can do it as well.
If you want to controll all units, there is an ellegant way to employ video camera (or more cameras at the same time) and an image analysing system.I have seen a system with five cameras operating parallely in a real time recognizing among a few thousand objects newly defined in MS Windows.
But as I said, your company's quallity policy will define the number of sample units in a lot that should be contolled.
m777182