Modelling carryover for more complex designs.
Stephen Senn
University College, London
Abstract:
Some brilliant mathematical investigations have been caried out into
optimal cross-over designs. Unfortunately, however, the models employed
have been biologically innocent: it has ben assumed that carry-over
depends on the previous treatment only and lasts for one period. This
implies that steady state is reached by the
end of two periods. When biologically experienced modelling
is used it can be seen very easily that so-called 'optimal' designs are
not. This talk will give a very simple demonstration of the point.