How to choose the data model, namely to take the systematic effects into consideration

摘要

The paper suggests guidelines in the choice of data model. The most useful approach is to establish the relationships and a logical order between the following concepts, not listed in no particular order: Intra-laboratory vs. inter-laboratory measurements; Within-laboratory vs. between-laboratory knowledge; Repeated vs. non-repeated (replicated) measurements; Systematic effect, bias and corrections; Nonexistent, fixed- vs. random- vs. mixed-laboratory effect; Class 1 vs. Class 2 standards; Single vs. several setups; Intra- and inter-laboratory comparison types; Calibration vs. testing; Measurements in series. They should be ordered according to the hierarchical order in which the knowledge is acquired and incremented. The evolution of the knowledge in time can be ordered in three basic layers: within-laboratory (single setup), within-laboratory (intra-comparisons), between-laboratories (inter-comparisons).