Minitab and Excel Giving Different Cpk Values for the same Data

Miner

Forum Moderator
Leader
Admin
Im not seeing that.

Run chart with subgroups across rows.
P-value clusters - .155
P-value mixtures- .845
P-Value Trends- .256
P-Value Oscillation- .744

On an x-bar chart using standard rules, test 5, 1, and 6 are broken at places. Not a huge effect but not really cyclical evidence.

You are right it isn't stable, but I cannot see two processes there.
Just a comment. Runs tests are very crude and will miss all but the most obvious examples. When I teach run charts, I show examples of all four categories that pass the runs test, but fail other tests.

For example, I create two statistically different samples, mix them and plot them on a run chart. They will pass the runs test for mixtures, yet fail on the IMR chart.

In my experience, by the time it is bad enough to fail a runs test, your naked eye has already told you the obvious.
 
P

Palt88

Could you make a screenshot of minitab to see what you are putting in the subgroups box, etc..

Because I have tried with different datas order vs subgroups in minitab i'm unable to get the same Cpk than you.

Could you confirm that your datas order are horizontal?
Subgroup1>s1>s2>s3>s4>s5>Subgroup2>s1>s2>s3>s4>s5...etc...
 
A

Allattar

Just a comment. Runs tests are very crude and will miss all but the most obvious examples. When I teach run charts, I show examples of all four categories that pass the runs test, but fail other tests.

For example, I create two statistically different samples, mix them and plot them on a run chart. They will pass the runs test for mixtures, yet fail on the IMR chart.

In my experience, by the time it is bad enough to fail a runs test, your naked eye has already told you the obvious.

I was just quoting those becuase I was too lazy to upload a picture.

Maybe I should have.

I can see the pattern, when plotting without the subgroups.
 
Last edited by a moderator:
Top Bottom