Interpretation of output for Minitab - P value & %linearity or % bias

[Anatta]

Hi all.....calling anyone that can help....

As discuss previous week, our organization is using Minitab to perform MSA study. And the output for Mintab is P value & %linearity or % bias, which is very different from the one results in AIAG reference manual.

For Bias study, how much of %Bias and how much of p value onwards from the Minitab output we need to consider the study fail and need to perform improvements. For AIAG reference guideline only spell out, if zero fall in between the 95% confidence interval of the bias, it is considered acceptable.

As for linearity, how much of % linearity, R-sq & P value from Minitab output is consider acceptable? In AIAG reference manual, we have to calculate the t-value.

There is a guideline for GR&R, where if the %R&R is less than 10%, it's acceptable. 10% ~ 30%, it's an indicator of rooms for improvement and >30% is unacceptable.

We just do not get the acceptable limit for Bias & Linearity.

Can someone please help me on these......

[Miner]

Originally Posted by Anatta

For Bias study, how much of %Bias and how much of p value onwards from the Minitab output we need to consider the study fail and need to perform improvements. For AIAG reference guideline only spell out, if zero fall in between the 95% confidence interval of the bias, it is considered acceptable.

If the p-value is greater than 0.05, there is not a significant bias effect. If the p-value is > 0.05, the %Bias is meaningless because there is no bias effect. If there is a bias effect, you should calibrate the gage until there is no longer a bias effect, so again the % bias is really meaningless.

Originally Posted by Anatta

As for linearity, how much of % linearity, R-sq & P value from Minitab output is consider acceptable? In AIAG reference manual, we have to calculate the t-value.

The R-sq percentage explains how much of the variation has been explained by the regression equation. In a Linearity study, this should be a lower the better number. A high value indicates that there is a linearity contribution from the gage. You also want a p-value > than 0.05, which accepts the null hypothesis that there is no linearity effect. You can also use the AIAG guidline that the blue center line falls within the red confidence interval lines. The same guidelines should be used here as with bias. If the p-value is > 0.05, there is no linearity effect, so no issue. If the p-value is < 0.05, there is a linearity effect and it should be corrected until it no longer exists (unless the gage will restricted to a portion of the range where the linearity is not significant).