Outliers are values in a data set that are far from the mean. The presence of outliers can skew our data; as a result, we may choose to remove them, or we may avoid reporting the mean (we may report the median instead, which is less skewed by outliers). Sometimes outliers indicate an error (perhaps a survey respondent accidentally typed “2000” instead of “200” when asked how many miles they drive per week). In other cases, it may be an accurate but unhelpful distraction; do we want our mean and standard deviation skewed by an outlier? Probably not. However, survey researchers should look at outliers carefully as they may indicate a “bad” record or signal emerging trends or niche opportunities.
Tip: if someone gives data analysis results that include a very unexpected mean value, ask to see that variable’s frequency distribution to see if an outlier is skewing the result.
« Back to Glossary Index
