Practical Quality Engineering Resources Statistical Alchemy 2016-01-20

Hi Bev,

You have listed the following methods as amounting to Statistical Alchemy. I concur with you on most; haven't understood well about on a few.

The RPN value- You have explained very well in your note posted here as a resource. Very well taken.

Process Capability Indices- Again your note on Process Capability Indices explains it very well.

Measurement error as a percentage of tolerance (<10% is acceptable)- I do agree that the 10% number may have been conjured up from nowhere. What would be really wrong in comparing the % Error with the Tolerance? To me it would give an idea of the suitability of the measurement system to measure a tolerance I have in mind when I go out to procure one. Even if I decide to leave statistics out, the quantum of error would also help me compare Measurement methods I need to choose between.

p< 0.05- again the 5% can be argued as arbitrary, though statistics tables have been printed with alpha=0.05 from times before I was born , what would be wrong on banking upon p values? I have read Wheeler's texts that are skeptical about estimating how distributions behave under the tails, and have read a bit about Deming's views on Normality of distributions. I have heard too that Fisher didn't really subscribe to the concepts of alternate hypothesis to put it mildly. I would love to read your insights into this. I may have missed something you have already explained in other posts or resources about this.

Fishbone diagrams- After studying and practicing FMEA for a while, I do believe that cause-and-effect relationships do not exist as represented by the fishbone diagram. I believe that Many Causes singly or multiply lead to a Failure Mode, and once the Failure occurs all the consequent Effects of the failure mode are possible. So it's really a Many Causes-Failure Mode-Many Effects relationship; each effect differently serious than other.
I'm sure you have more than this in mind. Would be eager to read about it.

Thanks

Ravi

 

Just wanted to say "Thanks!" and give an anecdote if any find it amusing.

I love the term "Statistical Alchemy" and it reminds me of a book I found in a used book store in grad school.
Funds were short, but I had to get it and put it prominently on my shelf where my advisors could see it.

"Statistical Manipulation of Data" ... I never opened it, but its presence led to some interesting discussions...

 

But.... but .... my bosses all WANT fishbone diagrams. They insist that they are necessary ......

 

Old dog...new trick....we all know how it can go...
One man's "fact" is just an opinion in a crowd.

My mind is made up. Don’t confuse me with the facts - Attributed to Roy Durstine-

 

I am not known in the real world for my diplomacy. Perhaps that is why I cannot always convince people.

 

That said, please enjoy my poignant graphic. I don't know the source, unfortunately, but it is fantastic.

 

Nice graphic. You.must.assimilate.........