What is a common approach to handling outliers when using normative data in assessment?

What is a common approach to handling outliers when using normative data in assessment?

• A. Always exclude all outliers
• B. Re-test to verify results, check test protocol for errors, and consider excluding extreme outliers if justified, using robust statistics if needed
• C. Remove data points until only typical values remain
• D. Ignore outliers and proceed

Answer: (B)

When interpreting scores with normative data, you want the distribution that defines the norms to stay stable while still accurately reflecting real variability. The best approach is to re-test to verify the result and check the testing protocol for possible errors, because a measurement mistake or administration issue can create a spurious outlier. If, after verification, the outlier still seems not representative of the population and is justified to exclude (for example, due to a clear protocol violation or faulty data), then it may be excluded with documentation. If the outlier is valid data, you can reduce its influence without discarding the observation by using robust statistics, such as medians, trimmed means, or other robust estimators. This keeps the normative framework fair and reliable while acknowledging genuine variation. Excluding all outliers or removing data until only typical values remain biases the norms, and ignoring outliers can hide important issues or variability that need attention.

When interpreting scores with normative data, you want the distribution that defines the norms to stay stable while still accurately reflecting real variability. The best approach is to re-test to verify the result and check the testing protocol for possible errors, because a measurement mistake or administration issue can create a spurious outlier. If, after verification, the outlier still seems not representative of the population and is justified to exclude (for example, due to a clear protocol violation or faulty data), then it may be excluded with documentation. If the outlier is valid data, you can reduce its influence without discarding the observation by using robust statistics, such as medians, trimmed means, or other robust estimators. This keeps the normative framework fair and reliable while acknowledging genuine variation. Excluding all outliers or removing data until only typical values remain biases the norms, and ignoring outliers can hide important issues or variability that need attention.