An investigation of test-house variability in the mechanical testing of dental materials and the statistical treatment of results

Abstract

The suitability of a test for standard specification testing depends inter alia on the ability to repeat the tests in a reproducible manner at a number of test centres. This work involved the investigation of three tests currently proposed as standard specification tests for dental materials. It was found that one test (compressive strength of cements) is inappropriate for inclusion in standards due to an unacceptable variation in test results between test centres. The treatment of results suggested in standards should place less emphasis on the mean value of a relatively small number of test specimens and the use of a simple form of probability theory in which, say, 80 per cent of specimens are required to achieve a certain pass level. In some standards there may be a need to increase the numbers of test specimens significantly in order to achieve a more meaningful and reliable result. The use of Weibull statistics can be adopted as a means of identifying tests which are suitable for inclusion in standard specifications, although it is doubtful that a test at this level of sophistication is required or desirable for inclusion in the standards themselves.