Computer packages as a substitute for statistical training?

Proposals are made for coping with the problems of teaching statistics to managers, and to students of management in higher education. The problems in question concern the fact that teaching statistics in these contexts is difficult and often ineffective: Numerous anecdotes indicate that these clients often neither like nor understand the statistics they are taught (see Wood & Preece, 1992, for a discussion of some of the problems this causes). For these clients, the discipline of statistics is a means to an end of being a better manager. Understanding statistics is not an end in its own right, as it might be for students on a degree course in statistics. It is for this reason that I have used the term "training" instead of "education" in the title. This is not to say that the purpose of learning statistics is simply to solve a particular problem or set of problems: Long-term utilitarians will learn ideas now because they think they may be useful in the years to come. However, if the subject is misunderstood, or avoided because it is disliked, it is of little practical use. This paper focuses on statistics and management, but I believe that very similar comments apply, for example, to statistics and medicine (Altman & Bland, 1991) and to mathematical modelling and management. In short, this paper addresses the problem of helping any novices use mathematical methods of any kind. The usual approaches to the difficulties of teaching statistics are based on advice such as practice more, think harder, get a better teacher, or even worse, become more intelligent. I think the problem is too difficult to be solved simply by these methods. Accordingly, this paper concentrates on a potentially more powerful strategy: that is, changing the methods and the technology used to make statistical analysis easier. The principles proposed here are not fully tested; however, I will illustrate them using specific examples, most of which I have used on an informal basis.