Techie Corner – 2018 – Post #2 – Outliers

Techie Corner – 2018 – Post #2 – Outliers

When may a real scientist discard an outlier data point?

My two cents, but please share yours:

An outlier may be discarded, if there exists proven evidence it was caused by one of the following.

• Tool malfunction
• Operator recorded the data incorrectly
• Operator calculated the wrong number

Otherwise, the outlier should be kept. It is, however, perfectly fine to curve fit the data both with and without the outlier point to see how large of a shift occurs.