ABSTRACT
This chapter considers modifications to the standard design, some of which result in complete separability of treatment and carryover effects. It provides modifications to the 2-treatment, 2-period, 2-sequence design that improve the estimability and efficiency of that design while retaining just two treatments. The use of baseline measurements alleviates the problem somewhat, leading to unique estimators of the parameters, but the correlation between the estimators of the treatment and carryover parameters is very high and the power of the test for carryover effects is so low that the design remains problematic. Incorporation of a wash-out period between the administration of the two active treatments does little to help solve the basic problem, although the use of a second wash-out period after the last treatment may be beneficial. The chapter explores various possibilities of extending the basic 2-treatment, 2-period, 2-sequence design to overcome the problem of over-parameterization that was inherent in that design.
