Wireless network simulation is used for research because of its simplicity and repeatability. While simple radio propagation models are evaluated quickly and are suitable for simple scenarios, sophisticated models can handle more complex environments and provide better accuracy. However, the cost of higher accuracy is slower execution speed. This paper describes experiments that validate ray-tracing and transmission line matrix models of the radio channel, within each approach considering versions that differ in their attention to detail and computational cost. We conducted the experiments under highly controlled conditions, within an anechoic chamber. Our main conclusion is that the errors due to lack of knowledge about beam forms and antennae shape significantly outweigh errors that might have been introduced if the experiments had not been within the anechoic chamber. While seemingly negative, the implication is that for our problem domain and level of information about the wireless environment, complex means of radio isolation aren't needed in validation studies.