Optimization problems appear in various engineering fields, such as control, design, and modeling. Many evolutionary algorithms have been developed for solving optimization problems. In recent years, one type of evolutionary algorithm, Differential evolution (DE), has been successfully applied in many problem domains. DE is a population-based stochastic search technique for solving optimization problems in a continuous space. The main idea of DE is to create a child as a new candidate solution by combining the parent individual and several other individuals of the same population. In previous research, we have proposed the Roulette Selection Based on Evolutionary Advance Level (REAL) for reducing the number of function evaluations. REAL is a generation alternation model which achieves an efficient search by an uneven evolution of individuals. In REAL, reproduction and the number of generated children are determined by evolutionary advance level. Therefore, it is considered that search mechanism of REAL is different from conventional DE. In this paper, we analyze the search mechanism of REAL and examine suitable control parameters using several test functions. Finally, we compare REAL and DE through experiments on high dimensional test functions and show that the search performance of REAL is superior to conventional DE.