Statistics
Outlier
A data point that differs significantly from other observations in a dataset.
What is Outlier?
An outlier is an extreme value that lies abnormally far away from the central tendency of the rest of the data. Outliers can arise from measurement errors, data entry mistakes, or genuine natural variations.
Why Outlier Matters
Outliers can drastically distort the mean, inflate variance, and pull regression lines away from the true relationship, potentially leading to incorrect research conclusions.
Example
In a dataset of human ages ranging from 18 to 85, a recorded age of 250 is an obvious outlier, likely due to a data entry error.
Common Mistakes
- Automatically deleting outliers without investigating their cause; true biological or behavioral outliers might contain the most interesting information in the study.
- Using mean-based outlier detection methods when the dataset is highly skewed, which falsely identifies valid data as outliers.