Linear Regression Model (LRM) is not only a time – hornored reserch method but a simple and essential analytical technique for social science researchers. However, it cannot reveal the meaning of the extrme value included in genuine data, which constitutes the major concern for researchers in social scineces. To solve this problem, most researhcers turned to quantile regression model. This model, hoever, is not only obscure because it divides data by quantile but difficult to perform because it needs special software. The study therefore proposed a new concept as well as method: equal part linear regression model (EPLRM) which divides the sample data into equal parts and builds LRM on each part so that the research can both observe the data distribution of sample data within each part and compare the results with that of LRM. As the case study shows, the poverty of fiscial expenditure on education would decrease Gross Nation Income (GNI) greatly and the promotion of educational input of private schools and social donation would boost the increase of GNI to a lage degree.