Estimating Simultaneous Confidence Intervals for Multiple Contrasts of Means of Normal Distribution with Known Coefficients of Variation

This study investigated the performance of simultaneous confidence intervals (SCIs) to differentiate the means of multiple normal population distributions with known coefficients of variation (CVs). The researchers aim to find the means of several normal distributions with known coefficients of variation, SCIMOVER, SCIs, and SCIk, which are extended to k populations. The authors constructed SCIs for the difference between multiple normal means with known coefficients of variation. There are three approaches: the method of variance estimates recovery approach (MOVER), and two central limit theorem approaches (CLT). A Monte Carlo simulation was used to evaluate the performance of the coverage probabilities and expected lengths of the methods. The simulation results indicate that the MOVER approach is more desirable than the CLT approaches in terms of the coverage probability. The performance of the proposed approaches is also compared using an example with real data. Moreover, the coverage probability results for SCIMOVER were over the nominal level of 0.95, indicating that it is more stable than SCIs and SCIkand was thus more appropriate for use in this scenario. Finally, the researchers suggest using the MOVER approach for constructing the SCIs to determine the variation to achieve the best solution in related fields in the near future.