Taves  states that “Minimization should be the method of choice in assigning subjects in all clinical trials”. Unfortunately, it shouldn’t. It is also stated that minimization can “reduce the risk of selection bias without adding randomization”. This is not true either. The idea behind minimization is brilliant, but its failure is the flip side of the same coin, and cannot be separated from its benefit. The claim to fame is that subjects are allocated not randomly but rather deterministically, so as to minimize an imbalance function (hence the name). It is this, the heart and soul of the method (and not some tangential aspect) that leads to its downfall.
Investigators using minimization can determine the group to which a prospective subject would be allocated, and then decide if this is a good thing or a bad thing, in terms of creating an imbalance with respect to some key predictor of outcome not considered in the imbalance function. In other words, there is no allocation concealment, and selection bias is there for the taking. To counter this, randomization may be added so that the treatment that minimizes the imbalance function for a given patient is not necessarily allocated to that patient, but rather there is a high probability of this being the case. It is unclear how selection bias can be “reduced” without this step, but even with it, are we really to believe that a betting man needs certainty, and not just good odds, to place a wager? If the odds of each treatment come close enough to 50% so as to truly take a bite out of selection bias, then the baby is lost with the bath water, and minimization no longer does what it purports to do. There is simply no way around this. Hence, what is needed is true randomization [2, page 23], and the best method among these for balancing the control of selection bias (by way of allocation concealment) with control of chronological bias is the maximal procedure . This, then, is what should be used, assuming that just one method is to be used. But the best solution of all seems to be a sensitivity design that employs a variety of techniques, including minimization .