Why does Dips occasionally calculate a dip average that is outside the range of dip directions?

The calculation of mean planes from set windows is based on vector averaging of the pole vectors.  It is not a simple numerical average, as this does not give you the correct results.

For a wrapped set window, it is possible to obtain a mean dip (for example) which is outside the actual range of the individual readings.  This is because the poles on the opposite side of the stereonet are actually assigned an upper hemisphere orientation so that the correct mean vector is obtained.  See the note about mean vector calculation for Wrapped Set Windows.

In general, straight numerical averaging of orientation data is not recommended and should be used with caution.  For example, with trend values on either side of 0/360, numerical averaging will give wrong answers.

Details of the mean vector calculation in Dips can be found in the Verification Manual.