Gage R&R for Attribute Data - Pump Noise Test

[kedarg6500]

Dear Friends

I was asked to do G RR for attribute data (noise test for pump). Here very expert/senior operator was used & taken as reference data. Line operator was asked to here the noise.

10 pumps were taken & 3 operators were asked to hear the noise & given opinion as GOOD/BAD.

I am attaching the file of results. Will anybody help me to calculate GRR & share excel sheet?

Thank you in advance

Kedar

 

[Bev D]

I did the calculations (after a fashion) and you can see them in the attached spreadsheet.

We really can't do a true GR&R because there aren't repeat readings for the operators. We can assess the single reading vs the 'truth'. Reproducibility has no meaning without first assessing repeatability. (who's to say that the second time Operator A reads the 10 units they won't get completely different answers?)

An additional complication is that the sample of 10 is really pretty small for an attributes study...

Kappa scores are:

OP A = 0.8
OP B = 0.4
OP C = 0.4

TOTAL = 0.66 (only a comparison of the operators without reagrd to truth)