Extreme Value Theory

Ordinary life-expectancy plots did not capture the strong expected maximum lifespan increases found by GEV analysis. Most fitted GEV distributions were reversed Weibull distributions because estimated shape parameters were consistently negative. The GEV shape parameter determines whether the fitted extreme-age distribution has a finite upper bound. The study fitted daily block maxima for each death cohort to a Generalized Extreme Value distribution using maximum likelihood. The article argues that maximum age at death should be studied with Extreme Value Theory because it is an extreme value statistic.