This paper examines the effect of dominance balance in many-objective optimization. The dominance balance can be defined by the ratio of the dominating space to the objective space. Here, CDAS, which is one of the most powerful evolutionary many-objective optimization algorithms, is known to be able to change the ratio of the dominating space by relaxing the definition of Pareto dominance with its user-specified parameter. However, the dominance balance is too difficult to control for the parameter in the higher-dimensional objective space. Therefore, we analyze the performance of CDAS by changing the ratio of the dominating space directly in even steps from the minimum to the maximum according to the number of objectives. The corresponding user-specified parameter in CDAS can be obtained from an equation which we assume in the paper. As benchmark test problems, we use DTLZ1, DTLZ2, DTLZ3, and DTLZ4 with two to ten objectives. From computational experiments, we can conclude that the optimal ratio of the dominating space differs depending on the problem at hand. It can be also said that the performance of CDAS is good especially when the ratio of the dominating space is small enough but not the minimum. Based on these observations, we propose a new version of CDAS called CDAS-D which controls the ratio of the dominating space dynamically during optimization.