ABSTRACT
The only way to measure reliability is to test completed products or components, under conditions that simulate real life, until failure occurs. Extensive testing, however, often results in undesirable expenditures of time and money. The life test is continuously monitored and a decision made as soon as there is sufficient supporting evidence for one of the two hypotheses. These tests take less time than non-sequential plans but estimation is complicated and not very 'robust'. A Weibull model is considered for data collected from five different material structures, N, O, P, Q, R. The data sets may be compared by considering the degree to which confidence regions for the model parameters do or do not overlap. Maximum likelihood methods are very important because they possess good properties and are extremely versatile. In particular, they are applicable to most types of censored data.
