ISBIS4 Abstract

Contact Author's Name: John Henstridge
Title of Abstract: Business Models for Statistics
Author(s): John Henstridge
Affiliation: Data Analysis Australia

Many authors talk of applying statistics to business, but it is equally valuable to consider the opposite - how business methods might be applied to doing statistics. The argument can be made that statisticians will only have credibility with business if they are seen to be business-like in there approach.

This theme naturally leads to a number of questions, many of which are confronting to statisticians whose comfort zone is lies in academia or government. Standard concepts of inference are not always applicable when the aim is to help make a decision. Opinion can be just as important as proof. Mathematical detail is often counter-productive. And the bounds of what is statistics cannot always be defined the way that statisticians might traditionally do so. And finally there are always issues of cost.

One key outcome of this is that substantial changes are still required in what is taught as industrial statistics. While statistical methodology remains critical, approach is probably more important. This is perhaps best expressed by Box´s comment that it is better to have an approximate answer to the right question than an exact answer to the wrong one.

This talk is based upon over twenty years of exclusively commercial consulting, of running a business as well as consulting to business.