Items reflecting the problem being investigated are collected and two independent, representative samples taken from each item for analysis. A single calculated test variable is used, in conjunction with a computer and the analyst's intuition, to interpret the data. Although optimum performance cannot be guaranteed, the method is flexible and allows assessment of two samples for similarity, calculation of discriminating power for any set of measurements, and estimation of the discrimination achieved from the analysis of a particular item. Since formal statistical tests and tables are not used at any stage, analysts are competent to investigate their own data after only a little practice. The system is equally suited to research and operational requirements. Some applications and potential developments of the system are discussed. Footnotes, references, and tables are included. (Author abstract modified)