ABSTRACT
The exponential distribution, being the first distribution to be used in life-testing, has had some use in survival analysis. Unfortunately, since the constant hazard rate rarely holds in human and animal survival studies, its use has been rather limited. In most survival analysis studies involving either human beings or experimental animals who are placed on study, the variable of principal interest is the length of time until each responds in some manner. The vast majority of survival analysis studies is not analyzed by parametric methods, let alone a rather restrictive parametric survival distribution, the exponential. The idea, as with the Knapp studies, was to develop a test of the hypothesis of equality of the mean survival times for correlated exponential random variables according to the Block-Basu structure which also includes a parameter, actually a nuisance parameter, that describes the correlation.
