Two sample comparisons with mixed discrete and continuous variants

The comparisons of means, dispersions or distributions are well understood and researched for univariate investigations. Much literature is available for the design and implementafion of studies involving muhivariate normal samples. Most of investigations into muhivariate, non-normal comparisons still remain open.

This dissertation presents what is available and provides two simple methods to transform the muhivariate two sample comparison into the well known chi-square test or logistic model. The methodologies and techniques will be examined by simulations and utilized to analyze the data from Texas Tech Health Science Center on the screening of Alzheimer's disease.