Distribution Type

Researchers mostly use Gaussian and lognormal distributions to model biological traits. When considering infant birth weight and gestational age cohorts, we looked at a general two-component hybrid-lognormal mixture model. We found that all of the two-component mixture models provide reasonable approximations to birth weight and gestational age distributions. Specifically, birth weight is best fit with a two-component hybrid-lognormal and gestational age is best fit by a mixture of two Gaussian components. In the case of particular populations, the sum of two-Gaussian components provides the best fit except in the case of African American females where a Gaussian secondary component and a lognormal primary component fits birth weight better than the mixture of two Gaussians.