Limitations of the rank transform statistic in tests for interactions
- 1 January 1987
- journal article
- research article
- Published by Taylor & Francis in Communications in Statistics - Simulation and Computation
- Vol. 16 (4) , 1133-1145
- https://doi.org/10.1080/03610918708812642
Abstract
Conover and Iman (1976), Iman (1974), and Iman and Conover (1976) have found the rank transform test to be robust and powerful when testing for interaction in experimental designs.The current study shows that, insofar as tests for interactions are concerned, the rank transform test is robust and powerful in some circumstances but is dramatically nonrobust and manifests power significantly below that of the usual F test in some cases. Therefore, this procedure should be used only withcaution when employed in designs suchas those examined here.Keywords
This publication has 0 references indexed in Scilit: