terminal digit preference

{{Short description|Statistical phenomenon}}

Terminal digit preference, terminal digit bias, or end-digit preference is a commonly-observed statistical phenomenon whereby humans recording numbers have a bias or preference for a specific final digit in a number. In medical science, this is often seen when recording measurements such as blood pressure by hand, where those taking measurements will round to the nearest 5 or 0.{{cite journal |last1=Thavarajah |date=1 December 2003 |title=Terminal digit bias in a specialty hypertension faculty practice |journal=Nature |volume=17 |issue=12 |pages=819–822 |doi=10.1038/sj.jhh.1001625 |pmid=14704725 |doi-access=free}} The phenomenon has been blamed for misdiagnoses.{{cite journal |last1=Nietert |first1=Paul J. |last2=Wessell |first2=Andrea |last3=Feifer |first3=Chris |last4=Ornstein |first4=Steven |title=Effect of Terminal Digit Preference on Blood Pressure Measurement and Treatment in Primary Care |journal=American Journal of Hypertension |year=2006 |volume=19 |issue=2 |pages=147–152 |doi=10.1016/j.amjhyper.2005.08.016 |pmid=16448884 |s2cid=25597886 |doi-access=free }} Terminal digit bias has been used to identify errors in research,{{cite journal |last1=Hla |first1=Khin |title=Observer Error in Systolic Blood Pressure Measurement in the Elderly |url=https://jamanetwork.com/journals/jamainternalmedicine/article-abstract/607571 |journal=Arch Intern Med |year=1986 |volume=146 |issue=12 |page=2373 |doi=10.1001/archinte.1986.00360240099017 |access-date=29 September 2021|url-access=subscription }}{{cite journal |last1=Hayes |date=2008 |title=Terminal digit preference occurs in pathology reporting irrespective of patient management implication |url=https://jcp.bmj.com/content/61/9/1071 |journal=Journal of Clinical Pathology |volume=61 |issue=9 |pages=1071–1072 |doi=10.1136/jcp.2008.059543 |pmid=18755731 |s2cid=10737432|access-date=29 September 2021|url-access=subscription }}{{cite journal |last1=Lusignan |date=23 March 2004 |title=End-digit preference in blood pressure recordings of patients with ischaemic heart disease in primary care |url=https://www.nature.com/articles/1001663 |journal=Nature |volume=18 |issue=4 |pages=261–265 |doi=10.1038/sj.jhh.1001663 |pmid=15037875 |s2cid=430764 |access-date=29 September 2021|url-access=subscription }} and is one method used in the identification of scientific fraud.{{cite journal |last1=Lawrence |first1=Jack |date=22 September 2021 |title=The lesson of ivermectin: meta-analyses based on summary data alone are inherently unreliable |journal=Nature |volume= 27|issue= 11|pages=1853–1854 |doi=10.1038/s41591-021-01535-y |pmid=34552263 |doi-access=free}} Severe terminal digit bias has been found in datasets for scientific papers that were later retracted {{cite web |last1=Brown |first1=Nick |title=Dr. |url=http://steamtraen.blogspot.com/2017/03/more-problematic-articles-from-food-and.html |website=More problematic articles from the Food and Brand Lab |publisher=Nick Brown |access-date=30 November 2021}}{{cite news |last1=Retraction Watch |title=JAMA journals retract six papers by food marketing researcher Brian Wansink |url=https://retractionwatch.com/2018/09/19/jama-journals-retract-six-papers-by-food-marketing-researcher-brian-wansink/ |access-date=30 November 2021 |publisher=Retraction Watch |date=19 September 2018}}