View Complete Reference

Beer, TW (2009)

Terminal digit preference: beware of Benford's law

Journal of Clinical Pathology 62(2), p. 192.

ISSN/ISBN: Not available at this time. DOI: 10.1136/jcp.2008.061721



Abstract: INTRODUCTION: Recording numerical data in pathology reports is routine and in some cases may provide valuable prognostic data (eg, tumour size for cancer staging). Hayes has recently observed that there is a tendency for reporters to favour 0 and 5 as the last digits in measurements "terminal digit preference". This is perhaps not surprising as gross measurements are often approximations taken in a relatively imprecise fashion (eg, holding a ruler to an irregularly shaped and flexible tissue sample), and the observer may informally round to the nearest half unit. While microscopic measurements could be claimed to be more accurate, they typically only measure a three-dimensional object in two dimensions, and so are again, an approximation. Hayes states that "one would expect an equal frequency of values with each of the possible 10 terminal digits (i.e. a uniform distribution)". However, this is not necessarily the case.


Bibtex:
@article {, AUTHOR = {Beer, T. W.}, TITLE = {Terminal digit preference: beware of Benford's law}, JOURNAL = {Journal of Clinical Pathology}, YEAR = {2009}, VOLUME = {62}, NUMBER = {2}, PAGES = {192}, DOI = {10.1136/jcp.2008.061721}, URL = {http://jcp.bmj.com/content/62/2/192.1}, }


Reference Type: Letter to Editor

Subject Area(s): General Interest, Medical Sciences