A tale of two methods: randomization versus matching trials in clinical research<FNR></FNR><FN>editorial note: the editors invite correspondence on the issues raised within this paper and the previous one by fox which will be published as ‘letters to the editor’. </FN>

This article comments on the research paper &quot;A hypothesis about Spiegel et at&#39;s 1989 paper on psychosocial intervention and breast cancer survival&quot; by Bernard H. Fox. Dr. Fox underestimates the importance of randomization done by present author. The reason that the authors have done randomized trials is because any attempt to match a study sample with some other comparison sample is open to the possibility of bias. The fundamental premise of randomized clinical trials is that the only thing that systematically distinguishes intervention from control patients is their randomization status. If this is the case, prognostic variables, such as site of metastasis, tumor type, and other potential confounds, are unlikely (but not unable) to account for the differences observed. Dr. Fox assumes improbable results as fact in his analysis of the study conducted by authors of the present article. There is a bit of a bias in Dr. Fox&#39;s analysis of sample differences.