n recent years, there are more and more empirical papers using panel data in economic and management field.At the same time, theoreticalresultsonhowtoprocesspaneldataarealsoincreasing.Howev-e,roneproblemisthatmosttheoreticalmethodologiesonpaneldataarebasedoninfinitesample, but we can only get limited samples in reality; another problem is that existing papers on using panel data don't dospecial treatment for panel data sample or don' t explain why adopting their processing methodology. Therefore, the two contributions of this paper are as follows:1)this paper compares the applicability of existing theoretical method ologiesby simulating based on the three different data structures— that with individual effect, with time effect and with two effects simultaneously; 2)considering that estimation results have significant differences for various processing methods in small sampl,e this paper provide sacriterion for judging what effects exist in the actual data.