Date of Award
Doctor of Philosophy
Dr. Joseph W. McKean
Dr. John A. Kapenga
Dr. Joshua Naranjo
Dr. Gerald L. Sievers
Statistical graphics combine statistics and graphics together to visually display analytical statistical results. They present information in a way that is very easy to understand and, furthermore, they frequently reveal information that is hidden from numerical summaries. Our purpose of this study is to explore the field of statistical graphics by investigating their application to statistical methods in the context of the R and GR methods in linear models. We consider the problems of score selection, R-diagnostics, and comparison of the R- and GR-fits.
We have developed a user friendly program which uses dynamic graphics and a user interface to perform the above problems. This program displays statistical results not only dynamically but also accumulatively. We have also developed a new graphical method called the QQBP which is an enhancement of two existing methods, the qq-plot (Wilk and Gnanadesikan, 1968) and the boxplot (Tukey, 1977). It is a very useful tool in differentiating between two distributions. Implementation of our applications is done by XLISP-STAT (Tierney, 1988) which is a statistical computer package equipped with many dynamic graphical tools. Our procedures can be easily extended to other situations in statistics.
Wang, Mei Huang, "Statistical Graphics: Applications to the R and GR Methods in Linear Models" (1996). Dissertations. 1727.