Identifying Minimum Effective Dose Under Heteroscedasticity
Michael J. Adjabui *
Department of Mathematics, CK Tedam University of Technology and Applied Sciences, Ghana.
Jakperik Dioggban
Department of Statistics, CK Tedam University of Technology and Applied Sciences, Ghana.
Irene D. Angbing
Department of Statistics, CK Tedam University of Technology and Applied Sciences, Ghana.
*Author to whom correspondence should be addressed.
Abstract
We propose a stepwise confidence procedure for identifying minimum effective dose (MED) without multiplicity adjustment.Stepwise procedures strongly control the familywise error rate (FWER) which is a critical requirement for statistical methodologies in identification of MED. The partitioning principle is invoked to validate the control of the FWER. Our simulation study indicates that the FWER was properly controlled in the case with balanced design but failed in some cases of sample sizes for situations of unbalanced design. In addition, the power of the procedure increases with increasing mean of ratio differences and the sample sizes.
Keywords: Balanced design, Fieller, FWER, MED, stepwise procedure, power