Mixed messages with Black Belts and Masters

Statistical methods only have value when they provide insight and aid in meetin the objective of the project.  If the tool does not do this, don’t use it.
Get away from the debate a spend your time with defining the practical question and collecting good data.
statman

 I suppose the short “cute” answer to your question would be -too many cooks spoil the broth.  In truth, for any given problem there will always be multiple approaches to its resolution.  The plethora of tools and methods available for statistical analysis guarantees different recommendations and, assuming that you are dealing with skilled practitioners, any one of the suggested approaches will probably work when attacking your problem.  As has been suggested in some of the other posts – pick a BB whom you trust and follow his/her advice.  If you have some doubts about an approach – pull out your books and do a little research on your own.

Document everytime this type of situation with SixSigma occurs.   Process delays occur quite often with Sixsigma but are dismissed or ignored.  Times like this will require you to use your own judgement and intuition. Yes, that’s what I said.  These Blackbelts (dumb title, I think) directives may be self serving.  No one should be “going nuts” over interpretations of a manmade system.

I have gotten alot of great feedback…let me give you a little background.I started off as a green belt, and before you know it I was working on 2 GB projects, 3 BrB projects and one BB project. This has all been in the last 1.5 years. I have received my certification, but have become quite versed in Brown Belt skills.
I am handling global projects now at a BrB level, but we have had changes in management, and BB staff moving in and out. It has been difficult to retain one BB for mentoring, and my management has not have the skillset in SS to support me. This is changing, but the difficulty arises from our Finance dept. not currently having a trained BB on staff. This means I have had to ‘borrow where I can’ to leverage against the BB skillset in other departments. They have all been outstanding, and the feedback I have received has been great. However, the confusion arises out of asking guidance in a given situation as this leads to multiple directions without a devoted BB avalable.Everyone’s time and energy in responding has helped me with my situation, and I appreciate all input.
And you’re right. I need to determine the data (which I do) and decide the tools to be used. If I should have data results that need to be clarified, then I should go to the BB for assistance. However, I need to narrow the field of focus, as doing this in the past has slowed my outcome.Best Regards,ecole

Ecole,
Have you tried both test approaches? With software, they are fast operations. Often differing approaches in statistics reach the same conclusions if the data speaks sufficiently. If the 2 approaches give oposing results, then you have to resolve the debate and achieve higher precision. Do not sweat the precision question unless you determine you need to.
 

We may be creating a tempest in a teacup here. When the data strongly support either the null or the alternate hypothesis, both tests often lead to the same conclusion. Ususally the two tests “disagree” only when P is close to the threshold value, which happens less than often than we might expect. That’s the only time you have to worry about selecting the “right” hypothesis test.