ABSTRACT
In the simple FPA (Forcing a Pre-fixed Allocation) rule in §1.2.1 the skewed randomised allocation of treatments was driven by the observed proportions of successes from the treatments. In this chapter we consider instead very different rules which are not response adaptive. The purpose is to allocate the treatments in a given proportion, often equally. Some randomness is however required and the rules vary depending on the relative importance of random allocation and treatment balance. The designs can therefore be described as “allocation adaptive” and viewed as randomised perturbations of sequential design constructions. A further important distinction from Chapter 1, where properties of the designs were analysed only when n = 100, is that the designs may be stopped for any number of patients. We require good properties whatever the value of n.
