The article featured today is from Statistics in Medicine and is now available to read in full here.
Two-sample test with $$ g $$-modeling and its applications. Statistics in Medicine. 2022; 1– 16. doi:10.1002/sim.9603, .
Two-sample comparisons occur frequently in statistical analysis, for example, testing the drug effect between the control and the treatment groups in clinical trials. Most existing methods focus on problems where observations are independently and identically distributed in each group. However, in some applications, the observed data are not identically distributed but associated with some unobserved parameters which are identically distributed. For instance, the observations are event counts following binomial distributions with a varying number of trials and a varying success probability across individuals, and the goal is to compare the distributions of success probability between two conditions. To address this problem, this paper proposes a novel two-sample testing procedure combining the g-modeling density estimation and the two-sample Kolmogorov-Smirnov test. Two versions of bootstrap procedures are introduced for the p-value estimation to cover the wide range of sample sizes and balance the needs for accuracy and speed. The proposed two-sample test can be applied to a wide range of applications, as it is capable of handling various types of data including count or continuous outcomes. The utility of the proposed approach is demonstrated with two biostatistical applications: the analysis of surgical node data with binomial models and the differential expression analysis of single-cell RNA sequencing data with zero-inflated Poisson models.More Details