The outliers in Turkish economic data and their effects on scientific inference

Outliers in a data set can be defined as observations, though smaller than halfnumber of the data, that prevent the detection of information contained in most of thedata and that cause misleading estimators. This thesis is prepared for investigating theexistence of outliers in Turkish economic data and their effects on scientific inference.For this aim in the second part, information is given about outlier detection methods. Inthis part, firstly initial and commonly used outlier detection methods are explained thenthe results of an extensive literature search about contemporary methods for outlierdetection are presented and the results are used to comment on the validity of methodsused in this thesis. In addition, the effect of outliers on outcomes are stressed with dataderived for this thesis. In the third part the existence of outliers in Turkish economic dataand their effects on scientific inference is investigated with panel data. In the forth andlast part the results are discussed. The results provide evidence for the existence ofoutliers and their negative effects on models, estimators and results.