A1151
Title: The Lepage-type statistic for the one-sided location-scale alternative in the two-stage design
Authors: Kenji Nitta - Tokyo University of Science (Japan) [presenting]
Hidetoshi Murakami - Tokyo University of Science (Japan)
Abstract: In many practical applications, an increase in location is often accompanied by an increase in variance, suggesting that heteroscedasticity may signal a treatment effect. This motivates the use of location-scale tests, which jointly assess differences in both location and scale. A common approach involves combining separate statistics for location and scale into a unified test statistic. In the context of independent Stage 1 and Stage 2 data, two-stage testing procedures are frequently employed. The aim is to extend such procedures to account for potential correlation between the p-values derived from each stage by incorporating first-stage information into the design of the second stage. Through extensive Monte Carlo simulations under a variety of continuous distributions, it is demonstrated that the proposed test statistic achieves enhanced and stabilized power, making it a competitive alternative to existing location-scale test statistics.
