Portal de Periódicos da CAPES

Calculation of critical values for Dunnett and Tamhane's step-up multiple test procedure

2000; Elsevier BV; Volume: 49; Issue: 4 Linguagem: Inglês

10.1016/s0167-7152(00)00076-6

1879-2103

K.S. Kwong, Wei Liu,

Optimal Experimental Design Methods

Dunnett and Tamhane (1992, J. Amer. Statist. Assoc. 87, 162–170; 1993, Statist. Probab. Lett. 16 (1), 55–58; 1995, Biometrics 51, 217–227) proposed a step-up testing procedure for comparing k treatments with a control and showed that it is often more powerful than its competitors, the single-step (Dunnett, 1955, J. Amer. Statist. Assoc. 50, 1096–1121) and the step-down (Miller, 1966, Simultaneous Statistical Inference, McGraw-Hill, New York; Dunnett and Tamhane, 1991, Statist Med. 10 (6), 939–947) procedures. However, the applications of the step-up procedure may be hindered by the lengthy time for computing the required critical values, especially when the design of experiment is unbalanced or the value of k is large. In this paper we propose some more efficient approaches to the computation of the critical values to make the step-up test more applicable.