Does Outlier need to be removed from Regression Analysis? Case Study in Economics Research
Does Outlier need to be removed from Regression Analysis? Case Study in Economics Research

Summary/Abstract: The linear relationship between response and predictor variables in economics can be expressed in a regression model. The Ordinary Least Square in estimating the parameters in the regression will produce a less valid estimator whenever outlier appears in the data. The identification of the outlier needs to be done to improve the validity of the results. The objective of this study is to examine how is the effect of deleting outliers from the data sets will change the adjusted coefficient of determination R2adj as compared to the use of full data (with outlier). Using twenty five sets of data obtained from the final projects of undergraduate students in economics of University of Brawijaya, two methods of deleting outlier observation are examined, i.e. (1) by deleting observation with the largest residual without considering the number of outliers in the data sets and (2) by deleting all outliers at once. The results of the analysis show that deleting observation with the largest outlier from the data sets increase an average of R2adj by 6.33% and an average increase of R2adj by 7.78% after deleting all outliers at once. Therefore, it is required to have a very careful analysis so that it can be decided whether or not to include outlier in the process of the regression analysis.

• Details
• Contents