A random effects model for ordinal responses from a crossover trial. by F. Ezzet and J. Whitehead, Statistics in Medicine, 10, 901–907 )1991(

I should like to comment on this interesting paper in which Ezzet and Whitehead developed a random effects model for ordinal responses from crossover trials. They presented a satisfying theoretical treatment of the problem and illustrated their approach with a genuine example. Nevertheless, it will be difficult for most data analysts to apply the Ezzet and Whitehead approach, and they themselves conceded that 'the fitting is rather complicated' (p. 905). Implementation of the Ezzet and Whitehead approach requires the programming of a numerical solution of a likelihood maximization in several parameters. I wish to propose an alternative.