Article Text

Download PDFPDF

Random, chance, or hazard?
  1. P G Ince
  1. Department of Neuropathology, `E' Floor, Medical School, Beech Hill Road, Sheffield S10 2JR, UK

    Statistics from

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    The recent correspondence exchanged between Dr Batman1 and Dr Carter and colleagues (ibid) airs the issue of the use of the expression “random”, which is usually invoked to describe the approach to selecting fields or structures under the microscope for quantification. Having been initiated (from a research perspective) within the “post-Nick Wright” incarnation of the cell kinetics group in Newcastle during the 1980s, I was heavily influenced (you might say subdued) by the statistical austerity of Dr D Appleton. His view was that what all of us were doing, in choosing areas to quantify, was never really random and I doubt that much has changed across most pathological research. To be truly random the investigator would have to define and assign coordinates to all the suitable fields/structures within the area represented in the slide, number them, and then use a source of random numbers to make an unbiased selection. In practice, what most people do (including, I suspect, Dr Carter and colleagues) is to select by a more haphazard or chance strategy, which most of us never properly define. Dr Appleton (from the benefit of a good Scottish education) suggested that the most appropriate expression would be “selected at hazard”, which I did slip by various editors a few times. Having moved on, whenever I try to put this in a paper now my co-authors always strike it out and substitute “selected randomly”; presumably they think it sounds more scientific or less prone to bias. Even I would balk at admitting to selection “by chance”. I have no doubt the statistical literature in relation to biology must contain the answer to this issue but I have not found it. I admit to no great diligence in my search. Perhaps some of your readers could point us in the right direction. In practice, the issue probably illustrates the gut feeling that most investigators have, when undertaking quantitative morphology, that statistical rectitude is not always superior to common sense. Common sense suggests that I am unwise to take as an example work from my new colleagues in Sheffield to make these observations but I am sure they are of a forgiving nature.