There is an old joke which asks how many types of statistician are there, and the reply is “Three: those that can count and those that can’t” (this only properly works if the questioner is in fact a statistician). In fact, there are many more types than that and understanding what they do and how they think is crucial to navigating the world of statistics.
In order to maximize the impact of any piece of statistical work, it is important to tailor it to the right group. What kind of audience is your work aimed towards? For example, textbooks that are intended for students benefit from sections with problems and answers. For applied and medical statisticians it is essential to have relevant and modern examples. It is very frustrating for a statistician to find a new method applied to either invented data or very old data. It arouses suspicion in the reader that the method preceded the application, and so either the author has never come across a proper example, or couldn’t be bothered to look. Either way, this suggests that the method the work is describing is somewhat arcane and probably not worth studying in detail.
It is always beneficial to describe the types of software used to analyze the methods described in the text. The software package ‘R’, for example, is advantageous in that it is freely available and readers can therefore replicate exactly what is in the book or paper, without having to buy a particular package. However, ‘R’ is not suitable for the ‘non-statistician’ readers, who want something to which they can point and click. For non-statistical authors, it is always worth getting a statistician to review your work for howlers that may not be obvious (such as the proper definitions and meanings of p-values and confidence intervals).
Statistics plays a vast role in so many fields, but like creatures stranded on islands in the Galapagos, they rarely communicate, and have evolved into such distinct species that interbreeding is now very difficult.
So, who are the statisticians, and how do they vary from each other?
1) The mathematical statistician
Statistics arose from mathematics and can still involve some fiendish mathematics. This type usually occupies a post in a mathematics department, and is looked down upon by the pure mathematicians, and in turn looks down upon the applied mathematicians. They may also think of themselves as probabilists. In UK universities, they are increasingly concentrated in only a few departments.
2) The applied statistician
These statisticians work in a wide variety of areas, from universities, industry and public services. Excluding medicine, which I will discuss next, they can be found in departments of statistics and research groups looking at a whole variety of areas such as agriculture, veterinary medicine, opinion poll companies, and industry. There are a number who work on environmental statistics and areas such as climate change. Some of the best paid are the actuaries.
3) The medical statistician
This is a large genre and can be found in medical schools, drug companies and contact research organizations. It is often the case in UK universities that there are as many applied statisticians in the Medical School as there are in the Mathematics Department. They are in huge demand to service the medical research infrastructure, on review boards, editorial boards, ethics committees and trial monitoring committees. Recently, there has been a huge improvement in the conduct of clinical trials and Clinical Trials Units have been set up, and many employ large numbers of statisticians.
4) The Official Statistician
The word ‘statistics’ and the word ‘state’ derive from the same root. The UK civil service has its own statistician grade, and curates the vast array of government statistics. The Office of National Statistics in the UK also runs the decennial census. These are the people you turn to for the size of the population, the numbers of births and deaths, and migration. They also look at inflation, the retail price index, and even recently the ‘Happiness Index’. They often provide data for politicians to quote to journalists.
5) The ‘non-statistician statistician’
Statistics must be one of the few disciplines where it may be seen as an advantage not to be a statistician (but be good with numbers). I have seen a book on medical statistics where the blurb actually states that the book is written by non-statisticians and so is easy to understand! (A statistical reviewer of this book said ‘I think I’ll write a book on brain surgery- so much easier to understand than those written by brain surgeons, with all those complicated details’). From a statistical viewpoint they often appear to have a different emphasis, and seem more concerned with models than data.
For an alternative classification of statisticians as various forms of religious sects see Campbell MJ (2008). Dr Fisher: The doctor sees the light. Significance (5:4) 172.
For more on statistics and medical research, read Wiley books written by Mike Campbell.