In this paper, we investigate the statistical characteristics of the three dimensional (3D) multiple- input multiple-output (MIMO) channels and their dependency on the dynamic range in channel impulse responses (CIRs). Our study is based on empirical results obtained from the 3D MIMO channel measurement campaigns at 6 GHz in a typical urban macrocell outdoor-to-indoor scenario. The space-alternating generalized expectation-maximization (SAGE) algorithm is employed to estimate parameters of multipath components from the measured data. The measured data is divided into groups based on the dynamic range in the measured CIRs. The second-order statistics in time and spatial domains are calculated for each group. It is found that the channel statistics shows strong dependency on the dynamic range in the measured CIRs.