ANALYZING DATA SETS FROM LOGNORMAL DISTRIBUTION WHEN UNDETECTED OBSERVATIONS PRESENT

There are several different ways in the literature to analyze data sets from normal or lognormal population distributions when the observations before the starting observation time are missing. One of the methods, was proposed by McBean and Rovers (1998). It is shown in this paper that the method proposed by McBean and Rovers is not scale-invariant. A modified method is presented in this paper. To evaluate the performance of the method presented in this paper, statistical simulation is conducted.