De. Burmaster et Am. Wilson, Fitting second-order finite mixture models to data with many censored values using maximum likelihood estimation, RISK ANAL, 20(2), 2000, pp. 261-271
Finite mixture models, that is, weighted averages of parametric distributio
ns, provide a powerful way to extend parametric families of distributions t
o fit data sets not adequately fit by a single parametric distribution Firs
t-order finite mixture models have been widely used in the physical, chemic
al, biological, and social sciences for over 100 years. Using maximum likel
ihood estimation, we demonstrate how a first-order finite mixture model can
represent the large variability in data collected by the U.S. Environmenta
l Protection Agency for the concentration of Radon 222 in drinking water su
pplied from ground water, even when 25% of the data fall at or below the mi
nimum reporting level. Extending the use of maximum likelihood, we also ill
ustrate how a second-order finite mixture model can separate and represent
both the variability and the uncertainty in the data set.