In this paper a short overview and a case study in a statistical comparison of stochastic optimization algorithms are presented. The algorithms are part of the Black-Box Optimization Benchmarking 2015 competition that was held at the 5th GECCO Workshop for Real-Parameter Optimization. The question about the difference between parametric and non-parametric tests for single-problem analysis and for multipleproblem analysis is addressed in this paper. The main contributions are the disadvantages that can appear by using multiple-problem analysis, in the case when the data of some algorithms includes outliers.