LAUSR

More and more testing procedures with different principles are being developed covering both quantitative and qualitative analyses. To ensure the reliability of them and relevant testing results, performance evaluation and method validation are necessary. Following the investigation into a variety of testing procedures, this paper is intended to discuss the rational classification of them according to the nature and purpose of testing. Five kinds of testing procedures are summarized according to different combination of input and output properties of the testing model, which are properties on nominal, ordinal, interval and ratio scales. On this basis, the method validation and performance evaluation parameters, statistical approaches and tools available for each group of testing procedures are discussed. The classification of testing procedures seems helpful for inter-comparison of testing procedures with different purposes and application fields, and the development of some general rules for the performance evaluation and validation of them.