Seminar

Statistician heal thyself - have we lost the plot?

Thursday, 25 October 2012
Time: 
9.30am - 10.30am
DEPM, Seminar Room 1, Level 5, The Alfred Centre, Monash University
Australia

In joint work with Sue Finch, I draw on the seminal work of Edward Tufte and Bill Cleveland on excellence in statistical graphics to develop five simple principles for producing quality graphs. Our focus is on static graphics showing data, data summaries and inferences. We review some of the resources available to teachers and learners of statistics to examine the consistency of visual displays provided with our five principles. We also review the default graphics generated by a range of different statistical software packages for a few standard graphical forms. Although major developments in statistical graphics have influenced the types of functionality provided by software packages, default graphics do not generally conform to all our principles.

We report on an empirical examination of the quality of statistical graphics for data and inference found in the statistical and scientific literature. We considered published papers from high ranking (A*) journals, according to the Australian Research Council’s 2010 Excellence in Research Australia evaluation, and compared a sample of graphs published in statistical journals with a sample of graphs published in applied disciplines. We draw inferences about the relative quality of graphics produced by statisticians and by other researchers. We suggest a simple checklist to be used when producing graphs for presentations and publications.

 

Prof. Ian Gordon

Department of Mathematics and Statistics
University of Melbourne

Professor Ian Gordon is a Professor of Statistics and the Director of the Statistical Consulting Centre at The University of Melbourne.

He is a founding member of the Australasian Epidemiological Association. He has provided statistical consulting to several hundred clients from business, industry and government over the last 25 years, and carried out research on many epidemiological projects, as well as work on statistical methodology in meta-analysis.