Abstract

Regression analysis is used to model relationships between variables and to detennine the magnitude of the relationships.Then the models can be used to make predictions. Franc Anscombe, statistician gave case examples of regression analysis, which hyphothetics data were in the forms of some independent variables and dependent variables. Interesting phenomenon from Anscombe 's data were the result of data processing would produce the same statistical values although the data were different one from the other. Pursuant from regression resuls in the study of Frank Anscombe's data, this research compared the correct regression models between life expectancy with Gross National Income (GNI) among some countries in Asia. Results of scatter plots showed that the data pattern was not be obvious, if tending to be linear or quadratic. Therefore, bothapproaches were used in regresision analysis to be compared. Analysis results showed that the linear regression model for Live Expectancy is= 63.46 + 0.00069 GNI with R2 = 68.27% and the quadratic regression model for Live Expectancyis= 60.62 + 0.0015 GNI- 2.94E-08 GNI2 with R2 = 73.51%. If R2 as determinant to choose the best model, so the model of quadratic regression is choosen because it has bigger R2 than the R2 in the linear regression. Key words: Regression analysis, Frank Anscombe, life expectancy, Gross National Income