Statistical Techniques Procedure, how much is too much?

Mustapha

Starting to get Involved
Worked at a few fortune 500 companies ranging from class I to class III. The statistical technique procedure was always very detailed, encompassing a whole slew of techniques, sampling, and ways of looking at data. If I needed to understand how to do something or why the sampling rate was as such, I'd reference the work instruction.

Current company has a procedure that's quite literally 6 sentences that says we do GR&Rs, use Minitab, and we use C=0, we may or may not use graphs to make business decisions. As the owner of the document, I feel its lacking only because I'm not used to something so vague, short, and...just lacking detail. The procedure was put in place to meet the ISO13485 requirement and nothing more. I want to update it to include everything that is done at the site, but also don't want to update for the sake of updating because it hurts to look at.
 

Steve Prevette

Deming Disciple
Leader
Super Moderator
Well, having gone through the exercise of writing the "Hanford Trending Primer" Statistical Process Control Library I have some experience with this.

Starting with the bad news first - you don't want to write an "audit trap". If management doesn't want "no stinkin charts" then it does little good to write a procedure requiring use of charts. I was in the odd situation of supporting a facility when we took over the operating contract, and I was asked to review their trending procedure. Come to find out, they were under pressure from the regulator to have a trending procedure, so they lifted my "Hanford Trending Primer" and made it their procedure. I had a good laugh when I was doing the due diligenct and discovered that. I said - this transition is going to be really easy - BUT you are actually going to have to live to your trending procedure (they had not been in reality).

Generally I found that facilities that "were in trouble" and management knew it, were very receptive to help from trending and data analytics. Those that did not believe they had any problems hated being told by the chart maker - there may me an issue here, here, and here.

One compromise (at least in the environment I was in) if mangement did not want a formal procedure for trending, I brought forth the "Hanford Trending Procedure" and declared it to be a "desk instruction" for the use of the data analysis personnel. That seemed to work.

I do know that some facilities incorporated the Hanford Trending Primer, but I've been out of the game for the past two years, so don't know the current state. By the way, the Instituted of Nuclear Power Opertors (INPO) reprinted my work in INPO 07-007 which they copyrighted.

Bottom line - I would not recommend trying to write "stricter" requirements for trending if your management will not support it and it will just become an "audit trap". But nothing prevents you from having a handy "desk instruction".
 

Mustapha

Starting to get Involved
Thanks for the response Steve. My intent wasn't to mandate the use of everything I was going to put into the procedure, but rather create a more how to and why. I definitely do not want to fall into the audit trap, so I'm leaning towards creating a more how to work instruction rather than putting a bunch of items in the top level procedure. Appreciate the insight.
 
Top Bottom