REVIEWS

conduct of “statistics ” in the medical literature has been found to be consistently poor,1 the transfer of innovative statistical techniques into the medical literature has been characterised by significant time-lags,2 and statistical input into medical research and publication, although “widely recommended … [is] … inconsistently obtained”.3 It may also reflect an undervaluation of statistical contributions to medicine, as articulated by the doyen of biostatistics, Norman Breslow.4 He observed that the work of econometricians Daniel McFadden and James Heckman on discrete choice models and selection bias received a Nobel Prize in 2000, but that similar contributions to medicine by statisticians and epidemiologists remain, as yet, unrecognised. Thus, we must “grapple ” with statistics in the same manner as Appleby urged with respect to health economics.5 To this extent, the now dominant evidence-based medicine movement has mandated “critical appraisal”, which incorporates, to varying degrees, statistical methods,6 and we reiterate that the discipline of statistics is increasingly engaged with “front-line science”.7 Statistics as we know and practise it today had its foundation in the first half of the 20th century and was established — more particularly, the “testing ” paradigm of P values and Type I and II errors8 — by two dominant figures, R A Fisher (1890–1962), who was born in England and died in Adelaide, South Australia (se