Validation of Two-sample Location-scale Model Using Empirical Likelihood Based Statistics
Leonora Pahirko, Jānis Valeinis, Janis Gredzens, Madara Krumina · Proceedings of the International Conference on Statistics, Theory and Applications (ICSTA ...) · 2023
One of the most popular semi-parametric relationships between two distribution functions is the locationscale model.Our goal is to develop a goodness-of-fit testing procedure for such location-scale models.For this purpose we use the plug-in approach by the estimation of the structural parameters and transformation of one of the samples.Further we propose to use the empirical likelihood test for the difference of two distribution functions and probabilityprobability plots.The effect of the estimation procedure on the limiting chi-square distribution is analysed by some simulation study.For practical applications we use the location-scale appropriate bootstrap resampling method to approximate the limiting distribution of the plug-in test statistic.