A comparison of tests for the k-sample, non-decreasing alternative.

This paper compares, by means of Monte Carlo simulations, three non-parametric tests to detect a non-decreasing trend in location parameters. The three tests are the Jonckheere-Terpstra test, the Cuzick test, and the Le test. We estimated power for each test across a wide variety of non-decreasing location parameters under four different types of distributions. None of the tests produced an overwhelmingly higher power over their two competitors.