Minitab ANOVA Gage R&R Error (vs. EXCEL ANOVA GRR Calculation)

I

Ingo1966

MiniTab includes the insignificant interaction Term Operator x Part in the final calculation of %GAGE R&R.

Attached I have an EXCEL ANOVA GRR Calculation for the same data's which I put into Minitab.

If you stay with alpha 0.25 to remove the insignificant interaction term Operator x Part the Results differ from Minitab. (even though MiniTAb Session Window stated to remove the Interaction Term also at alpha 0.25)

If you manipulate alpha to 1 in my Excel Sheet to INCLUDE the insignificant OPxPart Interaction --> everything is fine.

But this is wrong ! :mad: Ok it differs not much but unfortunately these Values are used to callibrate the Outcome from other Statistical Software.

And therefore some People who trust MiniTab would say my EXCEL Calculation is somehow wrong.
 

Attachments

  • GAGE_R_R_VERIFICATION TEST_ CALLIBRIATION DATA.xls
    144 KB · Views: 611
  • GRR DATAS for MINITAB INPUT.xls
    23.5 KB · Views: 437

Stijloor

Leader
Super Moderator
MiniTab includes the insignificant interaction Term Operator x Part in the final calculation of %GAGE R&R.

Attached I have an EXCEL ANOVA GRR Calculation for the same data's which I put into Minitab.

If you stay with alpha 0.25 to remove the insignificant interaction term Operator x Part the Results differ from Minitab. (even though MiniTAb Session Window stated to remove the Interaction Term also at alpha 0.25)

If you manipulate alpha to 1 in my Excel Sheet to INCLUDE the insignificant OPxPart Interaction --> everything is fine.

But this is wrong ! :mad: Ok it differs not much but unfortunately these Values are used to callibrate the Outcome from other Statistical Software.

And therefore some People who trust MiniTab would say my EXCEL Calculation is somehow wrong.

Comments anyone?

Thank you!!

Stijloor.
 
A

Allattar

It looks like you are using the EV calculation from the full model with the interaction. Even though you remove the interaction.

Note if you remove the interaction, the Sum of squares for the interaction must go to the error term, must go into equipment variation.

You are dividing g22 by h22. G22 is the EV found with the interaction included.

As far as I can see including or removing the interaction in your excewl sheet does not change the SS in the EV.
 
I

Ingo1966

Thankyou for your comments, you are right.

My Excel sheet is not very clearly you will find the modified SS for EV including the ERROR TERM in Cell C22

My Concern is MiniTab's Calculation why it comes up with different results
 
A

Allattar

I admit it is difficult following what the excel sheet is doing, but I think I may have traced it back.

The reproducibility estimates in your worksheet are different to Minitab, and hence the different figures.

Your calculation is MS(operator)/ (Parts*Replicates)

Minitab calculation

MS(Operator) - MS(repeatability) / (Parts * Replicates)
 
I

Ingo1966

Thankyou Allatar,

that was the point !
I was wrong, I have corrected it and than it fits with MiniTab.

I have to consult AIAG MSA 4th Appendix A again.

Dear Elsmar Forum:
"I am sorry for confusion"
 

Attachments

  • GAGE_R_R_VERIFICATION TEST_ CORRECTED VERSION.xls
    143.5 KB · Views: 547
Last edited by a moderator:
Top Bottom