ABSTRACT
Researchers are likely to have first-hand knowledge of the of how imminent risk changes with time for the lifetimes being studied. For example, lightbulbs tend to break quite unexpectedly rather than because they are suffering from old age; people, on the other hand tend to wear out as they get older. The analysts would expect the shapes of the hazard functions for lightbulbs and people to be different. People experience increasing hazard, whereas lightbulbs tend to exhibit constant hazard. An important feature of the exponential distribution is the 'memoryless property': the future distribution of the lifetime of a component depends on the past only through the present. Since many lifetimes are measured on the logarithm scale, and such transformations often increase the symmetry in data, it is important to examine the lognormal distribution where the transformed data are normally distributed. The chapter discusses a simple forms for the hazard function, including constant, power and linear.
