Stephen G. West

Arizona State University, Tempe, AZ.

SEEING YOUR DATA:  USING MODERN STATISTICAL GRAPHICS FOR MULTIPLE REGRESSION AND ANOVA

The use of standard statistical packages has sometimes had the ironic effect of distancing  researchers from their data.  In the pre-computer era, researchers often gained considerable insight into their data in the course of conducting laborious analyses by hand.  Familiarity with data can now be achieved rapidly and easily through the use of modern statistical graphical methods.  This talk will consider a number of methods of visualizing data within the context our most familiar statistical models--multiple regression and ANOVA.

I begin by presenting classic and modern plots of one dimensional distributions, showing their strengths and weaknesses in representing different aspects of the data. I then consider methods of representing two and three dimensional data.  A variety of methods of detecting relationships in continuous data are presented, with special attention given to curvilinear relationships and interactions.  These application of these methods will be illustrated with the analysis of a data set.  I then present several graphical  methods of comparing the distributions of groups, including several methods of displaying the results of complex ANOVA designs.  Finally, I will discuss some of the strengths and limitations of using graphical methods in conjunction with both hypothesis-testing analyses and exploratory data analyses.