GRR a case study.

Hello there.

I'll try to keep the introduction as concise as possible.

A customer wanted a GRR study for his materials. The GRR analysis performed but was unsatisfactory.

I was trying to figure out what could be causing all this variability. So I ran the process again and tried to watch the process discreetly. It appeared that the operators were not following the procedure properly. That were, no one zeroed the micrometer before measuring, nor did they use the ratchet. The results were like as the first analysis (I will not post these results).

Learning at least this mistake I proceeded as follows:

1. I randomly chose 10 samples from the production.

2. I verified the instruments before use.

3. I mention to the operators again the work instruction (zeroing before measuring and using a ratchet)

4. I was present and when they forgot something I told them to repeat it.


As you rightly expect, there were reactions (I know how to count, I can't do it this way, etc.).

The participate 5 operators, 2 of them have experience more than 20 years and 3 have less than 5 years. I also noticed that the 3 operators were very similar between them and according to instructions. The 2 operators were not similar between them or with the other 3 or similar to instructions (at least they try).

The USL & LSL




First, I plot the results.







At the left side you will see the 5 operators and at the right the 3 operators

Not sure what your question is?

Instead of the individual plot by operator (the first graphs) try plotting the PAIRED values in a scatter diagram for each operator - this will show the repeatability directly and will provide the most insight. I provided you a link to a spreadsheet (MSA Tools) that will do this for you if you need.

Hello bev I don't have a particular question just an observation if i can say. I try to follow your suggestions about plot the data first. I think i try that plot but the diagramms were too much of information

Just plot the paired values for each operator on a scatter diagram with equal x and y axis…that is simple to do in minitab

Try doing the simple stuff first and stop ‘mintabbing’ your data to death. Read MIner’s blogs. Read my presentation on MSA.

you mean this?

NO I mean a real continuous data scatter plot. See pages 41, 42 and 43 of my MSA presentation. The scatter diagram is on page 43.

ONE plot per operator unless you color code each operator. Each sample has the first reading on the x axis and the second reading on the y axis to form the dot. In your case each plot will have 10 dots.

Scatter diagram--> Youden plot i believe that is correct (i make for ODmax only)


I am trying to make the continuous data (not progress)

Miner the minitab file