University of Washington - Department of Biostatistics & Environmental Health
A statistical rule of thumb is defined as a widely applicable guide to statistical practice with sound theoretical basis. Characteristics include intuitive appeal, elegance, and transparency. A rule states not only what is important but, by implication of what is not included, makes an assertion about what is less important. This talk is based on the recently published book, Statistical Rules of Thumb, Wiley and Sons, March 2002.
The rules are based on my experiences as a statistical consultant in environmental health, epidemiology and various biomedical areas such as Alzheimer's disease and Sudden Infant Death Syndrome.
The talk is essentially an abbreviated version of a workshop to be given at the 58th Deming Conference in Atlantic City, NJ next month. Three areas of rules have been selected: design of a study, its analyses, and reporting it. Rules covered under design include selection of study design; three principles of design: randomization, blocking, factorial treatment structure; and sample size calculations. Analysis deals with the following rules: analysis should follow design, parametric analyses are preferred, plan for multiple comparisons, and every analysis has a graphical analogue. Rules of thumb with respect to reporting of results deal with when to use words, when tables, when graphs.
I would be interested in statistical rules of thumb used by members of the audience particularly in the area of Bayesian statistics.
A flavor of the book can be found at my website that contains a copy of Chapter 2 of the book, and discussion of a rule of the month.