Modeling maxima of longitudinal contralateral observations

The paper gives the joint distribution of maxima of contralateral observations taken from the same individual at several occasions, when the data are normal with given constraints on the parameters. Different constraints lead to different well-known generalizations of the normal distribution. As an immediate consequence, evaluation of several features of these maxima is greatly simplified. The results of the paper are also useful in modeling other phenomena, such as measures of physical fitness based on the best of two trials. Further applications deal with statistical inference, for example, testing whether an identified treatment is best.