Extending statistics of extremes to distributions varying in position and scale and the implications for race models (Q701130)

The authors study extreme value models for the minimum, guided by the applicability of a model to psychological data sets. It is concluded that a model leading to the Weibull family of distributions is the most acceptable. In order to get better estimates for the parameters, the authors extend the model from independent and identically distributed variables to models in which the left end of the observations is allowed to be variable. In addition, it is shown that regular variation remains unchanged if the observations from two sets of data are pairwisely multiplied together. The independence is also relaxed by using a model first developed for reliability applications by the reviewer.