Robust Trend Tests Derived and Simulated: Analogs of the Welch and Brown-Forsythe Tests
将Welch和Brown-Forsythe稳健统计量扩展到有序均值限制下的趋势检验，通过蒙特卡洛模拟比较其与假设方差齐性的趋势统计量[Ebar]2的表现，发现稳健方法在多数情况下更优。
Abstract For the one-way layout, Brown and Forsythe (1974) have shown that the Welch and Brown-Forsythe statistics are robust against heterogeneous variances. In this article, these statistics are extended to the case in which order restrictions are imposed on the population means and compared via Monte Carlo simulations of [Ebar] 2, the trend analog of the ANOVA F statistic that assumes homogeneous variances. The conclusions are similar to those of Brown and Forsythe in the unordered case; that is, the robust procedures outperform [Ebar] 2 except when the sample sizes are exceedingly small and the variances are homogeneous.