When performing a device move in FARO® Measure 10, Measure Q, or Measure X, the only data that can be reported is the statistical results of the data used in the form of an RMS, Standard Deviation or a Max Error. However, the algorithm used during a device move is a least squares and can be affected by geometry, spacing, quantity of points, etc. A device move with very few points of poor geometry could yield better statistical results but worse results throughout the measurement volume than a fit with more points of better geometry.
For example, a fit of 4 points with 3 of them almost in a straight line could yield a statistical average of 0.025mm, but may not be ideal. The 4th point off to the side is the only feature that is available to constrain rotation about the line created by the 3 other points, therefore, the constraint of this particular rotation is affected more by random measurement error than if there was an additional point or two scattered around. This could be seen as poor reproducibility on other points that are not included in the device move. The additional of other points may result in an average error of 0.040 mm but actually give better reproducibility on points that are not included in the device move or outside of the envelope of the control points.
For this reason, it is best to include many points over the full volume of the desired measurement envelope. When doing so, the average error of the device move is a better representation of the error that would be present in side the envelope on points not included in the device move. Of course, this is assuming that the environment is stable and the control point network used for the device moves is not moving itself.