Randomise or Randomize - that is the question
For 20 years of my professional practice, I have talked and written of “randomisation” and the process of “randomising”. My first published paper began with the words “Randomized controlled trials (RCTs) in surgery….”.(Cook, 2003). The first RCT I analysed was proudly stated to be “a pragmatic, randomised, double blind, placebo controlled trial”.(Avenell, 2004) Patients are “randomised” not “randomize”. Yet, over time it became apparent that many others think and write of “Randomizing”. Being a Brit, I tend to think of “ize” words on the whole as a rather American thing. It seems wrong, sounds wrong, and in my head it has a nasal Northern US accent (my apologies to Americans of different accentual persuasions). Similarly “minimisation” is to be preferred to “minimization”, clearly (to me). However, I have use “Randomized” and its variants in the past though “randomise” is my preference and most common and favoured practice.
A recent NEJM article was simply entitled ("A Placebo-Controlled Trial of Percutaneous Coronary Intervention for Stable Angina”) (Rajkumar, 2023) which is one way to get around this difference. Though in the text they clearly stated “We conducted a double-blind, randomized, placebo-controlled trial of PCI in patients with stable angina.” They also say “A total of 301 patient underwent randomization” which does not read that pleasantly aside from any disagreement in spelling; one cannot help but feel for the poor patients undergoing this experimental procedure. Let’s hope the surgeon knew what he or she was doing.
More happily for me, many others do as I do. For example, a recent publication in the BMJ clearly stated “we did a randomised controlled trial in which the primary aim was to compare attendance at BreastScreen Norway among immigrant groups who received an invitation letter and information leaflet in their language of origin and in Norwegian versus in Norwegian only”.(Hofvind, 2023) While I’m not a fan the “we did” language (I prefer de-personalising research as much as possible in terms of ownership to facilitate critique), I am subscribed to the “randomised controlled trial” though I might not feel the need for the word “controlled” as random allocation of treatment (or intervention if you prefer) is the only randomisation in town in clinical trials.
Probably the most enjoyable part of writing my recent book (Cook, 2023) was having a reason to read and engage with some of the earliest publication on clinical trials. However, it was a surprise to find that the publisher’s (Oxford University Press, OUP) copyeditor insisted on changing every “randomise” to “randomize”, and every “randomisation” to “randomization”. I was all ready to complain about this indecent Americanisation but thought as a good academic I better get my references in order. Given it was Oxford University Press, I thought the Oxford Dictionary would be the place to start.
WHat do the Dictionaries SAY?
The online Oxford University Press Dictionary give randomise and randomize (OUP, 2023a) one the same entry and it identifies a citation back to 1926 (Fisher, 1926) by a certain RA Fisher as the first recorded use (of “randomization”). While this was disappointing for me, the first reference to “randomisation” is given as shortly afterwards, in 1934. This paper was also authored by a certain RA Fisher in the Mathematical Gazette (Fisher, 1934). So not as early but pretty early still for “randomisation”. And so the dual practice goes back to possibly the originator of the word, and certainly one of the figures most synonymous with random allocation.
The OU dictionary definition for randomise/randomize is “To render random in order or arrangement; to employ random selection or sampling in (an experiment or procedure) in order to improve the statistical validity of the results.”(OUP 2023a)) That’s the kind of definition a statistician can get behind. OUP defines randomisation as the “The action, process, or result of randomizing”.(OUP, 2023b) Merriam-Webster (of the US) agrees 1926 is the first usage of Randomization (Merriam-Webster, 2023). Randomise, randomisation don’t get an entry in this dictionary though the “-ize” variants are offered as an alternative. The Cambridge Dictionary has “randomize” (CUP 2023a) as the relevant entry but notes the UK usually uses “randomise”.
What did key figures say?
Joan Fisher Box [a daughter of RA], when discussing the contribution of her father to the study design wrote of randomization.(Box, 1980) Being a Statistician I was well of Fisher’s key role in promotion of randomization (albeit not in medicine) so I checked further on his practice. For example in his key 1926 paper re wrote ‘“The distinction between errors eliminated in the field, and the errors which are to be carefully randomized in order to provide a valid estimate of the errors which cannot be eliminated, may be made most clear by one of the useful and flexible types of arrangements, namely the arrangement in “randomized blocks”.’ As an aside “error” here is not a mistake in the common sense but variation in the observed measurement (which can be caused by many things) and impact on what we want to measure. All in a publication of the “Journal of the Ministry of Agriculture of Great Britain” which suggests some authority. However, interesting his books reflect a mixed practice. In his both “Statistical Methods for Research Workers” and “The Design of Experiments”; he consistently wrote of “randomisation”, yet in “Statistical Methods and Scientific interference” it was “Randomization”.(Fisher, 1990) So Fisher it seems was both a “Randomizer” and a “Randomiser”.
Neyman may have conceptually conceived of random allocation in the 1920 separately from Fisher but without using the “randomize” as noted by Rubin : “If these statements from Neyman had been replaced by claims of priority based on his 1923 paper, it would have been difficult for me not to have accepted the position that he had independently envisioned physically randomized experiments even without a translation that used the word ‘randomization’."(Rubin,1990)
The word randomise or variants like randomisation irrespective of spelling was also not used by thought who were early users of random allocation in human experimentation such as Charles Pierce and Joseph Jastrow (Peirce, 1885) in the US, and Dora Colebrook (Colebrook, 1929) in the UK.
Bradford Hill, the strong promoter of clinical trials and the need for randomisation to achieve an adequate control referred to “random [number] sampling” in his classic papers on “Clinical trials” for example: “… recent British trials [which] have avoided the alternating method and adopted the use of random sampling numbers; in addition, the allocation of the patient to treatment or control is kept secret from the clinician….”.(Hill, 1951) Similarly he wrote of “…random allocation of patients…” (Hill, 1951,1952) but not of randomis(z)e or randomis(z)ation. By 1965 Hill certainly used “random order”, and “allocated randomly” and even “randomization”.(Hill, 1965) Usage around this time of others also tended to use randomisation or randomization.(Wrighton, 1953 and Mainland 1960). William Cochran and Gertrude Cox’s important book on Experimental Designs (Cochran, 1950) referred to “Randomization”.
The keen clinical trialist Thomas (Tom) Chalmers argued for Randomization from the 1st patient.( Chalmers 1975, 1977) though in 1968 he argued for “Randomisation” from the first patient (Chalmers, 1968). The late great Doug Altman who I was fortunate to work with on a few projects was a fairly consistent randomiser as far as I can tell.(Altman, 1984a, 1999, 2000, and Chan, 2005). His fine 1985 RSS Series A paper was on ”Comparability of randomised groups”.(Altman 1985) Though a related 1984 abstract is titled “Comparability of Randomized Groups randomized group”.(Altman 1984b) He also was I think a “minimiser” not a “minimizer.”((CUP2023b) see for example his BMJ stats note on minimisation(Altman, 2005)). The CONSORT statement of which he was a central figure, is a statement for reporting “randomized trials” in its original article (Begg, 1996) but when the update was published in 2001 in the Lancet (Moher, 2001a) it was “randomised trial” but also “Randomized trials” in JAMA (Moher, 2001b) and Annals of Internal Medicine (Moher 2001c). Similarly both spelling were used for the 2010 update.(Schulz, 2010a, 2010b)
What about the Journals?
What about some of the journals of relevance. Statistics in Medicine seems to allow both spelling (see for example Senn, 2004, and 2013) as does Trials and Clinical Trials (indeed a recent issue of the later had two articles sequentially next to each other one on a method for “randomised trial with time-to-event outcome”(Matteo, 2023) and another looking at methods of “futility monitoring for randomized clinical trials”(Wang, 2023). The BMJ seems to allows both and for a quick online journal website search [10 Jan 2024] has 23K citations form “randomization” on though slightly more “randomisation” just under 25K). 1939 seems to be the first “randomized” as in “randomized-blocks](Gardiner, 1939) in this journal with randomisation coming some time later. The New England Journal of Medicine has I think always been into “randomizing”.(MacMahon, 1955)
The Lancet similarly circa 1966 onwards seem to be a publisher of “randomisation” only: “We tested the randomness of these four groups; except for age-distribution, randomisation was satisfactory”(Vaishnava, 1966) and has been consistent publisher of “randomised trials” since.
Going forward
So what will I do going forward? I’ll maybe a bit nicer about those who like to randomize, but I think I’ll stick with randomise though a bit more humbly, and as before, if requested, I’ll become a “randomizer” for a project or paper. If Fisher could continue to smoke despite the evidence of harm, then I can keep being a randomiser even if randomization it seem is the original usage. In that I am may also be following Fisher. Ultimately being consistent within a specific work is the most important thing for clarity and pleasantness not for accuracy, and if you are a randomizer I promise I will not hold it against you.
Reference
Altman, DG. “Comparability of Randomized Groups.” Controlled clinical trials.1984. 5.3: 297–297.
Altman, DG. “Comparability of Randomised Groups.” Journal of the Royal Statistical Society. Series D (The Statistician). 1985. 34.1: 125–136.
Altman DG. A fair trial? BMJ. 1984;289(6441):336-7. doi: 10.1136/bmj.289.6441.336. [Altman 1984a]
Altman DG. Comparability of Randomized Groups. Controlled clinical trials. 1984;5(3): 297. [Altman 1984b]
Altman DG, Bland JM. How to randomise. BMJ. 1999;319(7211):703-4. doi: 10.1136/bmj.319.7211.703. PMID: 10480833; PMCID: PMC1116549.
Altman DG. Randomisation in the SYMPHONY trial. Lancet. 2000 Oct 28;356(9240):1521-2. doi: 10.1016/S0140-6736(05)73268-2. PMID: 11081555.
Altman DG, Bland JM. “Treatment Allocation by Minimisation.” BMJ. 2005.330.7495: 843–843.
Avenell A, Campbell MK, Cook JA, et al. Effect of multivitamin and multimineral supplements on morbidity from infections in older people (MAVIS trial): pragmatic, randomised, double blind, placebo controlled trial. BMJ. 2005;331(7512):324-9. doi: 10.1136/bmj.331.7512.324.
Begg C, Cho M, Eastwood S, et al. Improving the quality of reporting of randomized controlled trials. The CONSORT statement. JAMA1996;276:637-9.
Box JB. R.A. Fisher and the Design of Experiments, 1922–1926, The American Statistician, 34:1, 1-7, 1980. DOI: 10.1080/00031305.1980.10482701
Cambridge University Press [CUP]. Minimize, 2023 MINIMIZE | English meaning - Cambridge Dictionary. [CUP 2023a]
Cambridge University Press [CUP]. Randomize, 2023 RANDOMIZE | English meaning - Cambridge Dictionary. [CUP 2023b]
Chalmers TC. When should randomisation begin? Lancet. 1968;1(7547):858. doi: 10.1016/s0140-6736(68)90316-4. PMID: 4171344.
Chalmers TC. Randomize the first patient! N Engl J Med. 1977;296(2):107. doi: 10.1056/NEJM197701132960214. PMID: 318593.
Chan AW, Altman DG. Epidemiology and reporting of randomised trials published in PubMed journals. Lancet. 2005;365(9465):1159-62. doi: 10.1016/S0140-6736(05)71879-1.
Cochran, WG, Cox GM. Experimental Designs. New York: Wiley, 1950.
Colebrook DC Irradiation and Health. London: H. M. Stationery off. [printed by J. Johnson, at the University press, Oxford], 1929.
Cook JA, Ramsay CR, Fayers P. Statistical assessment of learning curve effects in a randomised controlled trial. Clinical Trials, 2004;1(2):209.
Cook JA. An Introduction to Clinical Trials. 1st Edition, OUP, 2023. Available from: An Introduction to Clinical Trials | Oxford Academic (oup.com).
Fisher RA. The arrangement of field experiments. Journal of the Ministry of Agriculture. 1926;33: 503-515.
Fisher, RA. Statistical Methods, Experimental Design, and Scientific Inference. Oxford: Oxford University Press, 1990.
Fisher RA. Randomisation, and an Old Enigma of Card Play Author(s): R. A. Fisher Source: The Mathematical Gazette. 1934;18(231):294-297
Gardiner A. M & B 693 for Gonorrhoea, BMJ. 1939;1:845.
Hill AB. The Clinical Trial. British Medical Bulletin. 1951;7(404):278-282. doi: 10.1093/oxfordjournals.bmb.a073919.
Hill AB. The Clinical Trial. N Engl J Med. 1952;247(4):113-9. doi: 10.1056/NEJM195207242470401.
Hill, AB. “Heberden Oration, 1965. Reflections on the Controlled Trial.” Annals of the rheumatic diseases. 1966;25:107-113.
Hofvind S, Iqbal N, Thy J E, Mangerud G, Bhargava S, Zackrisson S et al. Effect of invitation letter in language of origin on screening attendance: randomised controlled trial in BreastScreen Norway. BMJ 382:e075465,2023. doi:10.1136/bmj-2023-075465.
MacMahon, B. “Statistical Methods in Medicine.” The New England Journal of Medicine. 19955;253: 646–652.
Merriam-Webster. Randomize Merriam-Webster Dictionary online, 2023,Randomize Definition & Meaning - Merriam-Webster.
Mainland D. The clinical trial - some difficulties and suggestions. Journal of Chronic Diseases. 1960;11:484-496.
Moher D, Schulz KF, Altman DG. The CONSORT statement: revised recommendations for improving the quality of reports of parallel-group randomised trials. Lancet. 2001;357:1191-4. [Moher 2011a]
Moher D, Schulz KF, Altman D. The CONSORT statement: revised recommendations for improving the quality of reports of parallel-group randomized trials. JAMA. 2001;285:1987-91. [Moher 2011b]
Moher D, Schulz KF, Altman DG. The CONSORT statement: revised recommendations for improving the quality of reports of parallel-group randomized trials. Ann Intern Med. 2001;134:657-62. [Moher 2011c]
Rajkumar CA, Foley MJ, Ahmed-Jushuf F, et al; ORBITA-2 Investigators. A Placebo-Controlled Trial of Percutaneous Coronary Intervention for Stable Angina. N Engl J Med. 2023 Dec 21;389(25):2319-2330. doi: 10.1056/NEJMoa2310610. Epub 2023 Nov 11.
Oxford University Press. Oxford English Dictionary [online]. 2023. https://www.oed.com/search/dictionary/?scope=Entries&q=randomise [OUP 2003a]
Oxford University Press. Oxford English Dictionary [online]. 2023. randomization - Quick search results | Oxford English Dictionary (oed.com) [OUP 2023b]
Peirce CS, Jastrow J, "On Small Differences of Sensation," Memoirs of the National Academy of Science. 1885; 3:75-83.
Quartagno M, Morris TP, Gilbert DC, et al. A comparison of different population-level summary measures for randomised trials with time-to-event outcomes, with a focus on non-inferiority trials. Clin Trials. 2023;20(6):594-602. doi: 10.1177/17407745231181907.
Rubin DB. [On the Application of Probability Theory to Agricultural Experiments. Essay on Principles. Section 9.] Comment on: Neyman (1923) and Causal Inference in Experiments and Observational Studies. Statistical science. 1990;5.4: 472–480.
Schulz KF, Altman DG, Moher D; CONSORT Group. CONSORT 2010 statement: updated guidelines for reporting parallel group randomized trials. Ann Intern Med. 2010;152(11):726-32. doi: 10.7326/0003-4819-152-11-201006010-00232. Epub 2010 Mar 24. [Schulz 2010a]
Schulz K F, Altman D G, Moher D. CONSORT 2010 Statement: updated guidelines for reporting parallel group randomised trials BMJ 2010;340:c332 doi:10.1136/bmj.c332 [Schulz 2010b]
Senn S. Seven myths of randomisation in clinical trials. Stat Med. 2013;32(9):1439-50. doi: 10.1002/sim.5713. Epub 2012 Dec 17.
Senn S. Controversies concerning randomization and additivity in clinical trials. Stat Med. 2004;23(24):3729-53. doi: 10.1002/sim.2074.
Splawa-Neyman, J. “On the Application of Probability Theory to Agricultural Experiments. Essay on Principles. Section 9.” Statistical science. 1990;5.4: 465–472. Web.
Vaishnava H, Goyal RK, Neogy CN, Mathur GP. A controlled trial of antiserum in the treatment of tetanus. Lancet. 1966 Dec 24;2(7478):1371-4. doi: 10.1016/s0140-6736(66)90418-1.
Wang X, George SL. Futility monitoring for randomized clinical trials with non-proportional hazards: An optimal conditional power approach. Clin Trials. 2023;20(6):603-612. doi: 10.1177/17407745231181908. Epub 2023 Jun 27.
Wrighton RF. The theoretical basis of the therapeutic trial. Acta genet. 4, 312-343, 1953.