Over-the-air (OTA) performance evaluations of wireless terminals need to be accurate and consistent. The radiated two-stage (RTS) method is a promising solution for the 5G MIMO OTA test due to its advantages in cost and efficiency. However, the accuracy of the RTS method hasn't been theoretically investigated. In this paper, an uncertainty analysis on the throughput performance using the RTS method regarding the reporting errors in the device under test (DUT) reporting system is presented. A mathematical model of the RTS method is constructed, and error factors are introduced. Based on the analysis, the reporting errors in the DUT antenna pattern measurement are eliminated using the proposed RTS implementation in the final throughput evaluations. In other words, the RTS method is free of the effects of the DUT reporting errors, contributing to the accuracy and consistency of the test. An example DUT was tested to demonstrate the impact of the reporting errors and validate that the reporting errors are eliminated in the proposed RTS implementation. The theoretical proof of the accuracy of the RTS method helps further develop the method for 5G MIMO OTA testing.