The value of specialized tests in studies where ordered group means are expected

The performances of three tests for ordered means (Bartholomew's, Williams' and a modified method of least significant differences) are compared with the performances of more general multiple comparison procedures. The tests are assessed using simulated data representing various patterns of results from a toxicological study with one (zero dose) control group and three treatment groups. The specialized tests are considerably more powerful than the general tests and their application is recommended whenever ordered group means are expected.