Stepwise nonparametric multiple comparisons in randomized complete block designs.
Groggel-D; Bena-J; Skillings-J
Joint Statistical Meetings, August 5, 2001, Atlanta, Georgia. Alexandria, VA: American Statistical Association, 2000 Aug; :300761
Many commonly used nonparametric multiple comparison procedures have poor power for pairwise comparisons involving treatments whose effects are near the middle of an increasing ordered set. Pairwise all-subset procedures are compared to procedures using joint ranks and pairwise ranks in a Monte Carlo study. Tha pairwise all-subset procedures often show higher power while still controlling the experimentwise error rate at the desired level.
Control-methods; Statistical-analysis; Mathematical-models
Abstract; Conference/Symposia Proceedings
Joint Statistical Meetings, August 5, 2001, Atlanta, Georgia