Google

Consider a dataset of four reaction times: $$200$$, $$250$$, $$280$$, and $$250$$ milliseconds (ms), which has a mean of $$245$$ ms. If an outlier score of $$5,000$$ ms is added—perhaps due to participant inattention—the mean drastically increases to $$1,445$$ ms. This new mean is larger than $$80\%$$ of the scores in the dataset, demonstrating how a single extreme outlier can pull the mean so far that it no longer accurately represents the typical behavior in the distribution.

Example of the Mean's Sensitivity to Outliers

Adding a single extreme outlier to a small dataset can pull the mean so far that it no longer accurately represents the typical behavior in the distribution.

Learn Before

Related