Outlier Lies: An Illustrative Example of Identifying Outliers and Applying Robust Models

Abstract

The presence of outliers can contribute to serious deviance in findings of statistical models. In this study, we illustrate how a minor, typographical error in the data could make a standard OLS model “lie” in the estimates and model fit. We propose robust techniques that are insensitive to extreme, outlying cases and provide better predictions. With implementation examples, we demonstrate how robust technique improves estimations over conventional models based on normality and outlier-free assumptions.