Plotting capability study data - I have a bimodular data distribution

[GMan]

Plotting capability study data, and I have a bimodular distribution of the data which about six data points are .01 larger than the other points.
The dimension is the thickness of a gear molded from 15% glass filled polypropylene in a single cavity mold. The average of all 30 points is .652, and the range of the 6 points is .660 - .664 Wondering if these outlying data points could be parts produced from before the mold raised to its normal run temperature, therefore did not shrink as much? Any insight would be greatly appreciated.

 

[Marc]

bimodular

Do you mean bimodal?

 

[Miner]

You should have collected your data in time sequence and plotted it on a control chart to determine whether your process was stable prior to calculating the capability. If these data were as you speculate then you would see them as the initial points on the control chart and presumably out of control.

 

[GMan]

Yes, bimodal.

 

[Bev D]

Miner is correct. unfortunately no amount of statistical manipulation or scientific guessing or rationalization can save a poorly designed study.
If you are really interested in an informative capability study, let's talk about study design before you 'collect some data'...