Search the Elsmar Cove!
**Search ALL of** with DuckDuckGo Especially for content not in the forum
Such as files in the Cove "Members" Directory

Interesting Discussion Analysis of half normal distribution in minitab


Involved In Discussions
Johnny and Pau,
Thank you for your reply. I agree with how you look at this.

However it is an obligation for us to do so. We have to monitor the risk to the consumer in the market and compare it with our risk files and if needed update them or act on the risk in the market if needed if the ppm levels become to high compared to the accepted ppm levels in the risk file.
This way of PMS is obligated for medical devices and we do this also for non-medical devices.
Since I work for a big global company these metrics are also made mandatory by the business.
In our company the business groups need to develop this per business group and our business group is relatively young therefore we are looking into this how to calculate this with the challenges we face.

Therefore I'm not looking for different approaches on way of working (but thank you for the feedback) but I'm looking for the best way to calculate ppm levels with the difficulties we're facing, as explained.
Bev D, I wish it were so simple here. Unfortunately I have to proof that what they do with the numbers is not valid. And I can only do that when I proof that the numbers are meaningless due to the huge uncertainty involved. At this point here they have not taken the uncertainty of the production-complaint lag time into acount and go process the number placing them back to production period with a correction not taking in account that that correction has a distribution behind it. In short they just fit the number to an expected graph instead of making model that applies, is workable and realistic.
We will be obligated to practice risk (in market) monitoring this year. And only by giving proof I can make them think of different approach of monitoring.

Bev D

Heretical Statistician
Staff member
Super Moderator
Well if you can't calculate the proportion correctly because you don't have the requisite data then explain that you don't have the data to make the calculations that would have any 'meaning' and give them counts. In the end it's the number of complaints coupled with the severity that matters not the proportion THAT is what RISK is about. And the 'requirements' for Risk aren't about 'monitoring' it, it's about preventing it. Explain that.

not having the denominator makes counts more variable and can lead to people making false conclusions about count getting better or worse. but if you FIX the problems then those arguments don't occur. we must fight the right battles. No company ever won public or regulatory approval by playing games with math. Look at Boeing. does the proportion of crashed 737 max planes matter? no. the count did.

to put it in more common terms: if I see a mouse in my kitchen I am not going to try to count the actual number of mice in my house or how fast they are breeding or dying before I 1. set out traps to kill everyone of them and 2. go looking for every entry point I can find and plugging it up to prevent more mice from entering. Mice are bad. There are always more than 1. Customer complaints are like mice.
Top Bottom