Through a quasi-standard experiment,it is confirmed that the third level discipline catalogue should become the reference standard to distinguish peers from non-peers. Furthermore, the research application evaluation difference made by peers and non-peers who are defined with this reference standard between natural science and humanity, social science is analyzed comparatively. The test results about the hypothesis originated from "chess game experiment" indicate that the idea about evaluating system design quality dose not suitable for the research application evaluation, and further prove that the peers play the key role in the guarantee of evaluation effectiveness.