Posted by John Matthews on April 8, 2026

John Matthews is Professor of  Medical Statistics at Newcastle University, UK.  In this blog, he talks about how Vern Farewell’s latest JLL article made him reflect on Donald Mainland’s 1952 classic Elementary Medical Statistics.

When I arrived in Newcastle in 1987, I well remember David Newell telling me that Mainland was a book all medical statisticians should read.  I did as I was told, up to a point, although I found the methodological details frustratingly non-mathematical and rather wordy.

This was not the book where the mathematically trained medical statistician would look to learn the mechanics of the tools of the trade.  Only after having been involved in a few consultations did more of the real merit of the text become apparent.

All those things that you did not learn from your mathematical texts were there – what questions are being asked and why?  What samples are being used to address these questions and are they up to the job? Is the underlying logic sound?  Are the interpretations being sought the ones that are justified?

Only after cutting at least some of your consulting teeth did the insights of the book really hit home.  This was also accompanied by a tinge of disappointment that in 1987 we were still plagued by so many of the infelicities that were addressed in the first edition of Mainland’s book in 1952!

My naïvety has not been so completely supplanted by cynicism that I think things are no better in 2026, but we have not, by any means, eradicated such problems.  However, rereading the text after many years, one cannot help but be struck by the prescience and continued relevance of Mainland’s cautions and recommendations.

Vern Farewell’s article is a splendid piece of work which prompted me to reread large passages of Mainland; I hope it will do the same for others, as well as introducing it to those not yet aware of this book.  The historical background provided by the first two sections in Farewell’s article is both interesting in its own right and also very useful for a full appreciation of what follows.

It is remarkable that a book which is approaching 75 years-old still has lessons that are pertinent today, and which could usefully be read by those just starting out in biostatistics.

More modern books are available, most of which are written by mathematically trained biostatisticians.  While these often contain sections on, and illustrations of the mistakes alluded to by Mainland, these often seem to be additions and do not form the essence of the book in a way that they do in Mainland.  Perhaps the fact that few modern texts in this area are written by medically qualified authors is relevant.

Having said that, it is also the case, judging by Doug Altman’s biography, that by the time he wrote the first edition Mainland had probably seen more ferrets than patients.  Perhaps an extraordinary career is what is needed to write an extraordinary book.  It may be worth taking a moment to reflect on whether the lasting merit of the book is due to the unusual background of the author.

In his biography Altman noted that the second edition was such a different work that it could effectively be considered a new book.  While it would be out of place to extend Farewell’s article about the first edition, further work on the changes between the editions may pay dividends (not that I recall seeing a copy of the second edition).