Bias Study of a Measurement System

[weisusu]

Hi There:

We did a bias study on a CMM machine using the reference value representing the expected range for the parts that are measured. The average bias is -0.0002" or 16% of the process variation (total variation row of the Study Var (6*SD) column from Gage R&R study). The P-value for the average bias is 0.0000. My questions are: What do I do with this information? Do I need to add 0.0002" to the measurements I obtain from the measurement system?

 

[MikeHalloran]

Why did you bother doing the study?

Adding a bias to a measurement as a matter of course can very quickly drive you into terminal confusion; no one will know for sure if a recorded data point is the 'corrected' version or the uncorrected version, and you may even find instances of corrections to the corrected values, etc. What's reported should be exactly what the instrument said.



Mike Halloran
Pembroke Pines, FL, USA