Outlier is a data in a set of observations which appears to be incoherent with the reminder of the set of observations. Outlier is problematic because its presence can distort the result of an analysis either by altering the mean performance or, by increasing variability etc. In this paper, six methods of detecting outliers are reviewed and assessed. These techniques are subdivided into two classes, one regarding parametric methods and the other one addressing non-parametric methods. Within these two classes, we are restricted to a case where the population distribution is known to be normal. A simple parametric method Z* has been proposed to modify the Z score to overcome its challenge of sensitivity in the presence of outliers.  Simulation studies were used to investigate the behavior of the different methods in terms of sample size, followed by ANOVA to assess efficiency of these methods, while the large and small sample size used, provides an insight of the strength and weakness of the different methods.

outliers, parametric methods, non-parametric methods, ANOVA.