6.1. Choosing sample type, concentration levels, replicates and time range for precision

Most guidelines agree that the used matrices should correspond as well as possible to the matrices encountered in the routine analysis of real samples. If the precision of the whole method (not just the analytical step) (1) is determined, then each replicate determination should be made from a separate subsample and include all the steps of the method, including sample preparation. 

• The sample matrices should represent samples which are routinely analyzed in the laboratory. Sample(s) can in principle be of any origin: leftovers from large routine samples, leftovers from proficiency testing samples, etc. Accurate reference values are not necessary for precision determination and it is generally not advised to use certified reference materials (CRM) as samples for determining precision since these are often better homogenized than real samples, leading to too optimistic estimates of the precision.
• A sufficient amount of sample must be available so that the required number of replicate measurements with the subsamples can be carried out.
• A sample should be homogeneous, so that subsamples would have as similar matrix and analyte concentrations as possible.
• In the case of determining the intermediate precision, the sample has to be stable during a longer period of time (generally at least several months).  

Planning precision experiment 

http://www.uttv.ee/naita?id=23479

https://www.youtube.com/watch?v=L8gwuy41JD0

***
 

(1) It is almost always useful to determine the precision of the whole method, not just the analytical step. So, throughout this course we generally address the precision of the whole method, unless it is explicitly stated otherwise.