Not all measurement systems (or measurement processes) will only have the two (major) sources of variation gage & appraiser. In real life the outcome of a measurement process will depend on several sources for variation. To evaluate a measurement process appropriately, these sources should be gathered (e. g. with an
Ishikawa_diagram for the
measurement process), reviewed and prioritized previous to a measurement uncertainty study.
In the beginning you should also decide what kind of response you want to examine (pass/fail or measurements) and what kind of analysis should be done with the data collected:
- pass/fail data
- attribute gage R&R
- attribute agreement analysis
- measurements
- type I gage study
- gage linearity and bias study
- gage R&R
- gage R&R expanded
If the data could be both (attribute and continuous) and the attribute result is simply achieved by categorizing the continuous values (e.g. <10: pass, >=10: fail), you will always get more detailed information out of the values than out of the pass/fail results. (But if your customer insists on an attribute study, you should provide the attribute analysis as well provided the customer knows what he is doing and why he choose this kind of analysis.)
Hami812 said:
Who is the real appraiser the guy who hits the button and awaits the pass/fail, or is the PC who collects the parametric data?
Maybe it is easier to find the answer for the question "If some results were found to be wrong, where would you search for the root cause? Would you at first take a look into the PC and/or software (maybe some kind of bug) OR would you ask guy who hits the button what he did wrong (maybe hitting too hard)?"
Best regards,
Barbara
I have to perform a dry run on Minitab prior to a client a Gage R&R. I noted the client had previous performed what I thought was a Gage R&R< but as it turns out, it was an Attribute Analysis. Can the data used in Attribute Analysis be fitted for Gage R&R or does the testing have to be done differently? Also, when you consider Appraisers, can the appriaser be a PC? If you have 16 slots and your checking electronic blackboxes for some anomaly. Who is the real appraiser the guy who hits the button and awaits the pass/fail, or is the PC who collects the parametric data (unknown to the appraiser in a log file somewhere). Since the Auditor in this case will only want ot see the parametric data, can the operator be considered the appraiser or is it the PC with home grown software, and the parametric detection scheme (Data loggers, D/a COnverters, etc) are what is collecting data and deciding on Pass/Fail?. I have Minitab v16. This may be more a philosophy question, but i need to understand how to select the data rows and columns correctly for samples, appraisers, number of trials. Does the 16 slots testing these Blackboxes, count as number of times samples if they do the testing nearly simultaneously? Sorry.. I am getting hung up on details. I can provide sample data. Bit let me know.