Notes on Two Sample Tests for Partially Correlated (Paired) Data

Hani M. Samawi, Georgia Southern University
Robert L. Vogel, Georgia Southern University

Abstract

We provide several methods to compare two Gaussian distributed means in the two sample location problems under the assumption of partially dependent observations. Simulation studies indicate that our test procedure is frequently more powerful than other methods depending on the ratio of the unpaired data and the strength and direction of the correlation between the two variables. The tests used in our comparative study are illustrated with an example based on data from a small gynecological study.