iSixSigma

Gage RR

Viewing 8 posts - 1 through 8 (of 8 total)
  • Author
    Posts
  • #47132

    Antonella
    Participant

    If the device of measure is improperly calibrated is most likely is that I obtain a errors due repetibility or bias?
    Thanks

    0
    #156795

    Craig
    Participant

    Bias

    0
    #156801

    Messy Problems
    Participant

    Antonella,
    Further to the other comment, it is really important to use ‘standards’ to calibrate the sensitivity of the equipment carefully. After all, we don’t want a robust measurement system in the sense that it is insensitive to changes we’re trying to measure. Although a low sensitive measurement system might give an acceptable P/T ratio, it is not going to be of much use to find sources of variation.
    We should remember there are two types of statistical usage. The first sumarises data; the second tears it apart.

    0
    #156806

    DJ
    Participant

    Wouldn’t the “insensitivity” show up in the analysis as an unacceptable number of distinct categories?

    0
    #156809

    Antonella
    Participant

    The results of the analysis is biased because the device was improperly calibrated, so obviosly the gage is unacceptable for the measure.

    0
    #156810

    CT
    Participant

    It is possible to have an R&R that is acceptable even if the calibration is off. Depends really on which type of R&R you are using. Obviously you want the calibration to be right, and cover the upper and lower specs of the product to be measured. When using equipment that requires Calibration, I like to use Known standards to compare the calibration to. That is the best way to determine R&R of a gage.
    CT

    0
    #156813

    Craig
    Participant

    GRR has nothing to do with accuracy (bias). It measures precision, which is repeatability and reproducibility. Your GRR study is probably OK unless the tool was so far out of calibration that it was completely unreliable.
    MSA has five parts:
    GRR: 1) Repeatability, 2) Reproducibility
    Bias: 3)Accuracy (how close to acutual value?)
    4)Linearity (Bias across the measurement range)
    5) Stability (How does the system measure over a long time frame?)
     

    0
    #156847

    Messy Problems
    Participant

    Measuring the variability of one category and multiplying the variability by six does not necessariliy descriminate six categories.
    The only way to guarantee the discrimination of six categories is to measure the repeatability of six standards.

    0
Viewing 8 posts - 1 through 8 (of 8 total)

The forum ‘General’ is closed to new topics and replies.