Outlier Screening

Outliers are points that represent unusual, implausible, or impossible values. Typically an outlier is as many as three or more standard deviations from the mean. The presence of an outlier can adversely affect the estimates for means and correlations.

Multivariate techniques such as Factor Analysis assume that the data set to be analyzed is a joint Gaussian distribution. If the data is not joint Gaussian, then the model generated from the analysis may not accurately reproduce the measured density.

The elimination of outlier data can be done before or after data transformation. Outliers can be eliminated manually or through automatic data filtering. For more information refer to Transforming Distributions/Eliminating Outliers.