It's useful to note what is excluded. But you exclude bad data from the analysis.
Remember that what you're interested in is not the data but the ground truth that the data represent. This means that the full transmission chain must be reliable and its integrity assured: phenomenon, generated signal, transmission channel, receiver, sensor, interpretation, and recording.
Noise may enter at any point. And that noise has ... exceedingly little value.
Deliberately inserted noise is one of the most effective ways to thwart an accurate assessment of ground truths.
Outlier exclusion is standard practice.
It's useful to note what is excluded. But you exclude bad data from the analysis.
Remember that what you're interested in is not the data but the ground truth that the data represent. This means that the full transmission chain must be reliable and its integrity assured: phenomenon, generated signal, transmission channel, receiver, sensor, interpretation, and recording.
Noise may enter at any point. And that noise has ... exceedingly little value.
Deliberately inserted noise is one of the most effective ways to thwart an accurate assessment of ground truths.