Outlier Effect

Twenty-one surveys using the exact question asked adults nationwide in February and March of 2005 to rate the job performance of President Bush. In those 59 days, the results indicating approval of his performance ranged from a high of 57% to a low of 43%, a full fourteen percentage points different. Upon further analysis, it is revealed that the average was 49% and those two survey results were outliers among the other data points. Without the context of other data sources, an unsuspecting analyst looking only at the results of one of those two surveys might likely make erroneous conclusions or err in emphasis. Also, see outlier effect in the Glossary of Terminology.

Example 2 – The Whole Picture