# 1.5 Sigma

Viewing 21 posts - 1 through 21 (of 21 total)
• Author
Posts
• #50261

Azmat Khan
Participant

Hi, The shift of the mean in the long term is always 1.5 sigma, why it is not more than or less than 1.5

0
#172739

Anup
Participant

Hi Azmat, this is a very intricate question, motorola after a lot of experimentation found out a factor of 1.5 as a shift, for GE its 1.33 and so on. They introduce this shift to make the product or process error free.the same is the reason why 3.4 DPMO is still considered as a Black box…..if u see 3.4DPMO happens at 4.5sigma and if we want to take it up to 6sigma then 3.4 should practically reduce, which never happens in theory……i wish all can put in the research in this and modify and simplify this using statistics and mathematical analysis…..regards,
Anup

0
#172743

Taylor
Participant

1.5 sigma shift was developed because the long term model of a particular process did not meet the actual. The actual sigma shift for any given process will be different statistically, however the actual difference from 1.5 is insignificant. The 1.5 shift is for modeling of long term mature process that exhibit no assignable cause for sigma shift.
All that said it is a pointless waste of time to try to prove in most applications due to random causes that are exhibited in almost all processes, therefore cant be proven.

0
#172745

Brandon
Participant

In other words we made up the 1.5 number and since it can be neither proven nor disproven we are sticking with it. Plus it takes half a day to teach it and since we get about \$3000 per day for training we want more stuff to fill the days.
PS: There is more stuff like this in SS – good luck finding it.

0
#172747

Mikel
Member

My cynical friend, I’ve been teaching this longer than anyone and I
don’t waste more than 10 minutes on it.

0
#172748

Brandon
Participant

Good. Then I won’t either.

0
#172749

Stevo
Member

Azmat

Do not believe them; the 1.5 sigma shift is real.  We believers just call it the shift.  In the late 1970s there was even a 35 mm film proving it.  I believe it is still hiding up in the pacific NW or maybe Roswell.

Think about, too many people have taught it, not to be real.

Stevo

0
#172750

Culp
Participant

Yet another reason why Stan is Bad Ass.

0
#172754

GB
Participant

Azmat,
Don’t listen to these jokers…
Don’t forget to apply the proper shift methodology.   The Honduran shift is only 1.4162…, while the Mediteranean shift is 1.61211… The north american shift is 1.5 and the asian shift is 1.6511…
And don’t forget about African shifts.  Due to it’s location, there are 2 shifts for the african continent.   The north african shift is 1.61211 (identical to the med shift), but the sub-continent shift is 1.61511…  So, anything below the Congo is 1.61511…
The shift for the Indian subcontinent is still in development, but is expected to land somewhere in:
01.55112…
If you don’t take into account your relative geographic position to the origin of the sigma calc and apply the appropriate shift, you may miscalculate the “grid sigma” because of the influence of “magnetic sigma”,
Don’t worry, it’s a common pitfall for noob’s
Good luck.

0
#172755

Brandon
Participant

Does this have anything to do with the Calender Year? I’m getting confused.

0
#172761

GB
Participant

It’s all about the circle of life, Brandon…
The circle of life…

0
#172763

Brandon
Participant

It really all is connected isn’t it?

0
#172764

Taylor
Participant

Calendar years, the circle of life, and the 1.5 sigma shift………I need a drink
Cheers

0
#172769

RickL
Member

Seems like it become a astrology topic?? …….i am confused…

0
#172781

BC
Participant

Now that’s got to be 10 minutes worth flying across the continent for!!!

0
#172786

GB
Participant

Rick,
Once you are intiated into the 32nd degree SBAMBB level, all that was cloudy,  shall become clear as crystal…Yea, verily.

0
#172793

Mikel
Member

I don’t know about flying across the country, but all the hype about
computing sigma levels and counting opportunities never helped
anyone fix anything.

0
#172881

Eugene Lanning
Participant

As most of the replies do not seem to have a cohesive answer, try this for a response:
When calculating a process sigma, one needs to consider two things into the calculation:
#1: Since the calculation is based on a finite number of data points, we have sampled the “true” sigma, thus we have obtained an estimate of the sigma, not the sigma itself.  Thus, there is some uncertainty introduced in the sampling.  That is why, even though the process sigma can be calculated to n decimal places, I always round it to the nearest tenth.  I my more pessimistic times I round it down to the nearest integer.
#2:  Usually the data to calculate the process sigma is gathered over a relatively short time period, sometimes a day, sometimes a week or month.  That classes the data as “short term” process sigma information.  Long “term data” would be representatives of a year or more – in the time-frame when a person might look at the process again.  Now, in the process that has common cause variation due to variables a, b, c, d, e, … r, s and t, over a short term not all of the common cause variables are observed – such as machine wear.  That means that the short-term data did not include all of the variation to a significant level, when all of the common causes are included – as over a year or more, additional causes become apparent and cause the process sigma to drop.
How much is the drop?  Is it 1.5 Sigma for your process?  1.5 is a traditional correction value, based on work at Motorola (I believe), there is little reason to believe that that value determined by the electronic giant would even begin to be replicated in my industry (meat processing).  If a person wants to be exact, then you could compare short-term values with long term data for each of your processes and find the “Azmat correction of n.n”.  Personally, I usually consider that the process sigma I calculate is a sample, and that uncertainty is imbedded in the 1.5 correction factor (That is just my way of living, it does not make that true).  In reality almost none of the processes here approach a short term sigma of 6, so my most efficient option is to accept and use the traditional 1.5 correction and spend my effort on improving the processes that have lower sigma values.
So take this post as being from a idiot if you wish.

0
#172926

Jonathon Andell
Participant

I agree with Brandon, Stan, and all the others afflicted with rational though processes. I’m even more cynical than most: I think the “shift” was a load of cobbled-together baloney that allowed Motorola managers the rationalization they sought. I think they wanted to keep the sexy-sounding alliteration of “six Sigma,” without actually being forced to maintain 2 parts per billion. I cannot prove it, but as W seems to say: “That’s my story and I’m stickin’ to it.”

0
#172957

Bower Chiel
Participant

Hi All
One of the Six Sigma books on my shelf is Mario-Perez Wilson’s “Six Sigma, understanding the Concept, Implications and Challenges”. In it he states: – “The plus or minus 1.5 sigma shift surfaced when Motorola in their explanation of ‘Why Six Sigma?’, used it as a worst case scenario of a significant shift in process average. They stated that a plus or minus 1.5 sigma shift would not show a detriment in the out-of-tolerance percentage to their customers if their processes were designed to have their specification limits be at twice the process width, or at Six Sigma levels. It does NOT imply a process mean shifts about plus or minus sigma over time or as an average.” Earlier he refers to a series of technical papers on statistical tolerancing and claims that Dr Mikel Harry attempted to find a mathematical justification for a plus or minus 1.5 sigma shift in average by erroneously quoting one of them.Best WishesBower Chiel

0
#172960

Learner
Participant

Actually,  I was just thinking whether it is really a worthwhile topic to consider.  What I feel is that if we do not improve the process for a period of 12 months and continue to only measure the data for calculating DPMO then measuring the DPMO and thereby the sigma values somewhere in the 3rd month and at the end of the 12 months may show a difference of 1.5 sigma values.
But usually we do look at the ST sigma value and not the LT sigma value coz we need to take a look at the present data collected and not worry about the shift that is not present at that moment!

0
Viewing 21 posts - 1 through 21 (of 21 total)

The forum ‘General’ is closed to new topics and replies.