Some parametric and non parametric inferences under non standard conditions

Abstract

In this thesis, multiple comparison procedures are proposed for the testing of the location and scale parameters of different location-scale family of distributions under non-standard conditions. The main concern in multiple comparison procedures (MCPs) is to control the familywise error rate (FWER). The MCPs are classified broadly into two categories, such as single-step and stepwise procedures. The stepwise procedures generally control the FWER strongly and are more efficient in terms of power than the single-step procedure. Therefore, stepwise test procedures have been proposed to address the different testing problems as follows: (i) ordered pairwise comparisons; (ii) successive pairwise comparisons; and (iii) comparison with the best. The required critical constants to implement these proposed procedures are computed using the different techniques and are tabulated in this thesis. To assess the performance of these proposed procedures, simulated studies have been performed to compute the powers of these procedures. Also, the simulated powers of these proposed procedures are compared with the existing procedures, if available. The finding of the thesis is that the proposed procedures have remarkable performance to declare the more significant results. The numerical examples are also taken to demonstrate the implementation of the proposed procedures. newline