Interaction In R%R
October 20, 2004 at 12:31 am #37272
How should we deal it when we find part*operator is significant when we do R%R study?
Tks!0October 20, 2004 at 11:53 pm #109451
Can somebody who has experience on it help me?
Tks!0October 21, 2004 at 12:32 am #109454
Just so we are clear, your part to part variability is low, your person to person variability is low, but your person-part variability is high?
0October 21, 2004 at 2:23 am #109459
Thanks for your feedback first!
Pls see belowing sample from Minitab help file, we can find Part and Operator*Part is significant factor for their P-value is less than 0.05. So, my question is what should do on it?
Two-Way ANOVA Table With Interaction
Source DF SS MS F P Part 9 2.05871 0.228745 39.7178 0.00000Operator 2 0.04800 0.024000 4.1672 0.03256Operator*Part 18 0.10367 0.005759 4.4588 0.00016Repeatability 30 0.03875 0.001292 Total 59 2.249120October 21, 2004 at 11:56 am #109479
Since you haven’t provided any info regarding %Contribution the P-value itself may be strong but that doesn’t necessarily mean you have to react on it.
By the Degrees of Freedom and MS-values I made a quick calculation and came up with the conclusion that the %Part was approx 0,903 meaning your measurement equipment (including interaction) is less than 10%. This is just about accepted unless you have seen some trouble with it.
To deal with interaction it all comes down to why part(s) is behaving differently when measured with a specific instrument (device). You have to find the part, and the device, that contributed to the high interaction value. Then examine all factors that distinguish the part/device from the others, and hopefully you find the answers to the odd behaviour.
What part and device that you’re looking for could be seen in MINITAB’s GageR&R interaction graph (bottom, right hand graph; in ver 14).
“To solely react on P-values is like putting up the umbrella when a rain drop falls on your head, but the sky’s almost blue.”
The drop is real, true (that’s what the P-value says).But there’s no rain as far as the eye can see; that’s what the %Contribution says.So, no umbrella (no reaction).
I know, a lousy analogy, but I want you to look further than just a few inches, which I think you may do when looking at single parameters.
The forum ‘General’ is closed to new topics and replies.