iSixSigma

Shift Happens

Six Sigma – iSixSigma Forums Old Forums General Shift Happens

Viewing 42 posts - 1 through 42 (of 42 total)
  • Author
    Posts
  • #30701

    Adam Pearson
    Participant

    I just finished reading the thread on ‘where 3.4 DPM comes from’ and I am wondering if anyone has conducted studies to verify the 1.5 sigma value commonly reported as “the shift”. There was an article earlier this year in Quality Engineering that tried to decsribe it, but I have not seen any actual data to support 1.5 as a universal figure. It seems to me that each process would have a relatively unique amount of degradation – if STATISTICAL PROCESS CONTROL  is NOT used.
    Thoughts?

    0
    #80351

    Aravena
    Participant

    Adam,
    Here’s my opinion – I agree with you but I think we should, as community, stand up and yell “kill the shift”.  All it does is confuse Green Belts and force us to bring up Mikel Harry in order to explain the history.  Here’s my theory – the whole thing is made up, Mikel Harry just invented the shift in order to collect a $0.01 royalty any time “1.5 shift” shows up in litterature or knuckleheads like us are forced to try and explain to a room full of skeptics.  The bottom line is that it’s not material when explaining process capability.  Do people understand that process behave differently in the short term than over the long term? Yep – they get it.  But the artificial 1.5 is a big bag of crap that we all have to constantly explain.  Here’s my answer – shift depends on the process…and Mikel Harry – if you’re out there – thanks a lot!

    0
    #80365

    Mikel
    Member

    Each process has a relatively unique amount of degradation – even if STATISTICAL PROCESS CONTROL  is  used. Why do you think SPC prevents it?

    0
    #80366

    Dave Strouse
    Participant

    Is the shift 1.5 for all processes? Probably not, probably unique to each process. Are there empirical studies? Hear alusions to them being done at Motorola, but they appear to be “factoids” which I define as something “everyone” knows, but no one can produce the technical report. M. Harry referances a study done in the 1950’s as theoretical proof. However, the study was on tolerance stack up ,not on process drift. The “crossover” to process drift is considered fragile by most statisticians, I believe. However, I like best a statement by G.E.P. Box, in J. of Qual. Tech., I think. He said, to best of my memory “I do not know if the shift is 1.5 or some other number, but am delighted that it is finally being recognized that processes do change over time” Perhaps someone has the exact quote. But bottom line, George is saying “Shift Happens!” Ladies and gentlemen, let us consider the use of this shift. It simply allows for the very real shift of process to be incorporated into a relative measure, the process sigmna, for tracking of process improvement and comparison of unlike processes to each other. Used as such, and agreed on as the value to be used across all the practitioners of the Six Sigma methodology,what is the harm? Where is the foul? it is simply a defined metric. If you attempt to use it as an absolute measure, then there is probably reason for concern.

    0
    #80367

    Mike Carnell
    Participant

    Pablo,
    The mean does move – period. Look at a control chart. What do you think that represents? There was data and if you research the threads you will find a reference. All that is irrelevant really. If you do not like 1.5 pick another one.
    It wasn’t Mike who established the 1.5 sigma shift so you can talk about it without mentioning Mike.
    There was some point where he did something with Chaos Theory to substantiate it. You want to screw up GB’s try explaining that one.

    0
    #80368

    Mike Carnell
    Participant

    Adam,
    If you believe you are bringing something new to the the way overdone discussion of shift you need to do a search and you will find your epiphany is deja vu again and again and again……
    Of course every process has some indigenous shift. Since people typically have short term data for capability they use it and the process capability is overstated (there are those who feel that if a process is stable there won’t be any difference – quite the revelation but I haven’t seen that process yet since not only does it need to be completely sable so does every other process that feeds it). The 1.5 number is intended to represent an average shift so that the long term can be estimated. Don’t like 1.5 or you just can’t live with the inaccuracy of the stimate collect the data and get the actual number.
    Now if that really bothers you you probably need to find another field of work. As long as you are dealing with inferential statistics guess what? They are all inaccurate to some degree. That is why we have things called confidence intervals or we estimate the standard deviation, etc. The idea of an estimate of the long term is to tough to live with? I can see where it would be much better t allow the short term data to stand alone as the only process capability until we can find the one true number (at that point we have the population data – the project is over and nobody cares anymore).
    If I were running a production line I would certainly be much happier if all my suppliers were doing that. Of course since they were all overstated that would make any projections I were doing about my process, impossible because I couldn’t live with all that error and inaccuracy in my number. So we would just have to sit there until we had perfect data and no customers left. That makes more sense than using some estimate.
    Good luck.

    0
    #80387

    Aravena
    Participant

    Mike,
    Most articles I’ve read (yes, I have done some homework) attributed the 1.5 shift research to Motorola and in some cases to M Harry (guilt by association, perhaps).  Apologies to Mikel Harry for getting that wrong, thanks for pointing that out to me Mike- bad joke on my part.  I agree with a point you make in a different post about the discussion being overdone, just trying to make light of it.
    I will take issue with one of your comments, I don’t recall saying process means don’t shift and drift…so I’m not sure what to say about your “what do you think that represents?” question.  Perhaps you have some special talent for extracting meta-messages from the written word.  Hmmm…let’s see…what message might you extract from this paragraph?………………………….that’s it – you got it.
    pablo
     
     

    0
    #80392

    Withheld
    Member

    I agree with Pablo. The 1.5 figure is a fabrication.

    0
    #80399

    Adam Pearson
    Participant

    Mike,
    I sense that I’ve hit a nerve here. I never intended this to seem like any sort of epiphany – I’m simply tired of singing the 1.5 sigma shift song to GBs – without any data to support its existence. If we as six sigma practicioners stress the need for data, I’m just trying to ‘walk the walk’ so to speak. The1.5 value is intended to be an average, but surely there have been others who have tested the concept. With either sample statistics or any predictive model, they’re all wrong, to varying degrees – that’s part of what makes this so much fun, so, I’ll keep my job, thanks.
    This is a communal bulletin board for exchanging information and as such, is a proper forum for asking these questions – both for getting a sound answer, and for helping to educate the GBs and BBs who participate in this forum. As others have noted recently, there are more MBB level questions than BB, or GB here. It seems then that we, as MBBs have a certain obligation to be helpful. I enjoy reading your responses and feel that you provide sound answers, so keep it up.
    Regards.

    0
    #80404

    Kim Niles
    Participant

    Dear Adam:
    There are a couple articles already on the shift here at isixsigma. See the first one at https://www.isixsigma.com/library/content/c010701a.asp and then https://www.isixsigma.com/library/content/c010311a.asp on revisiting the shift. The second article also points out several good posts on the subject.
    While I was shocked and disillusioned about Six Sigma after first hearing of the shift, over time I began to understand how important it is to Six Sigma for the following reasons:

    It first opened my eyes to the possibility that processes can and likely do regularly shift. I had taken several statistics classes and never picked up on this.
    It allows an unreasonable goal of six sigma being 0.002 ppm out of spec to an apparently reasonable goal of Six Sigma being 3.4 dpmo.
    It is key to providing Six Sigma with novelty needed to distinguish itself from other programs such as TQM.
    Since in the real world special cause variation is the norm, data is often serial correlated and or non-normal, the shift tends to negate some of these real world data integrity problems.
    It provides insight needed to help me understand how common SPC out-of-control rules can be erroneous. See Controversies and Contradictions in Statistical Process Control at http://www.asq.org/pub/jqt/past/vol32_issue4/qtec-341.pdf
    Regarding statistically supporting the generally stated shift, it can’t be done with high confidence. I’m working on a third shift article that may or may not be published in the future with hopes of explaining the shift in more detail. I plan to explain how Mikel Harry sited Bender (1975) and Gilson (1951) to support this position in two different ways (compounding error due to tolerance stacking and normality differences with sample size), how Tadikamalla pointed out how Harry took those articles out of context, and the details of those arguments.
    Gregory Watson, ASQ president in 2000 stated in regards to the controversy over the statistics behind Six Sigma that there are three groups of thought, those in favor, those against, and now a new group of those taking the middle ground. He says that those in the middle see the controversy but recognize that management is paying more attention to Quality. He says, “It is acceptable to have a non-purist approach to statistics that gets results quickly, if that is management’s decision.”.
    I hope this helps.
    Sincerely,
    Kim Niles
    https://www.isixsigma.com/library/bio/kniles.asp
    http://www.asqsandiego.org/contacts.htm#kn

    0
    #80408

    Robert Butler
    Participant

    Adam,
      If you are referring to the Davis Bothe’s article in Quality Engineering – Statistical Reason for the 1.5 Sigma Shift, I think that he has provided the answer to your question.  At the beginning of the article he states that “Six Sigma advocates…offer only personal experiences and three dated empirical studies (which he cites) as justification.”  He also points out that these three studies are 25 and 50 years old.  In short, if you are looking for documented and published cases in support of 1.5 his article strongly suggests that you will look in vain.
      The real value of the Bothe article is that he defines the dynamics of a process that could give rise to such a change.  His examination of the dynamics of control chart sensitivity and use is reasonable and does make a case for process drift.  He identifies a range of drift from 1.3 to 1.7 and he also identifies the assumptions that have been made for drift to occur.  In his concluding paragraph he identifies the limitations of the initial assumptions.
      His concluding comments go to the heart of a number of the posts to this thread.  Specifically, if the dynamics of your process do not match the dynamics of the process he defined then your drift may or may not be greater or less than the 1.5 drift that is advocated by the Six Sigma community.

    0
    #80418

    John J. Flaig
    Participant

    Robert,
    Good  response, Dr. Bothe’s article does provide good insight into this issue.
    I’m not sure about the thinking behind the 1.5 sigma shift, but I suspect it was to provide a conservative rule-of-thumb. I think a much better approach is to model the process behavior and to draw the appropriate infernece from the model. Proper modeling would take into consideration non-normality, autocorrelation, etc. Now given the degree of statisitcal knowledge required to perform such work correctly I suspect Dr. Harry opted for a quick rule-of-thumb.
    I think this was a poor choice because it does not lead to accurate estimates of process performance and it does not improve the practitioners understanding of the process. However, from a implementation point of view it might have been a good choice. 
    John J. Flaig, Ph.D.
    Applied Technology (e-AT-USA.com)

    0
    #80421

    jerry coover
    Participant

    Of course shift happens. The only constant in life is change, so shift is change and it happens. What amazes me we deal with fact driven decisions, but if we add the shift we put in a guesstimate, why do we not just deal with facts. I personally believe the shift was a marketing ploy, since Six sigma is much more sexy than 4.5 sigma. Besides that has been the real standard for a long time,  (CpK of 2) can not have that if we are going to sell something new. Thus we have something new to sell and so we have the shift and 6 sigma. What a world we live in.

    0
    #80430

    Mike Carnell
    Participant

    Adam,
    For the most part I don’t care about the shift as long as people are up front and tell me if I am looking at long term or short term data. The part that bothers me more than anything else is that people discuss it endlessly. If the discussions were going to make any difference in our day to day lives it would be worth the effort, but it won’t. That is where the exposed nerve is.
    In many cases I choose not to participate. When I see the discussion and the responses it separates for me the people who work for a living (as defined by who delivers results) and who pontificates for a living (as defined by those pursue perfect information as an excuse not to do anything). That isn’t meant to take a shot at you because you have evidently done some thing that told you this was an issue.
    You made the comment about the people testing the theory. If 1.5 is an average and the results are normally distributed around that average (which they may not be – but suppose they are). How often will you get exactly 1.5? The results make sense if it is an average.
    Kim makes some good points in her post. The quote from the ASQ president is pretty good reflection of where it is actually sitting today. When we began this thing at Allied and GE we were focused on results only. Over the last few years you have seen the gravitation of people into the business who are turning this into a training program. They are quietly killing it with esoteric BS the same way they neutered TQM and killed the respect from a company leadership level. People need to separate the two – the tools work and always will. It is the people who either deliver results or don’t.  
    Watch the cycle. Practioners do something. Theorists hammer it. Practioners gain respectability on results. Theorists jump on bandwagon and lend their uncommitted support. Practioners do something else. Read the book by Watts Wacker “The Deviants Advantage.” It is the deviants that drive change. For all the things Mike (Harry) does that drives people crazy he did get the attention at the “C” level. It opened the doors for a lot of people to make a difference. Unfortunately we have the “me too” crowd slipping in behind them. I digress.
    Have a great day. Good luck in your search for truth.
     

    0
    #80431

    Mike Carnell
    Participant

    John,
    You obviously do not know Mike. Mike did some work and I am fairly sure he can defend his belief. Not the point.
    We live in a world where time is money. We can do the modeling and get the perfect information before we take the next step or we can make an estimate and move on. The bottom line is that the number has absolutelu zero effect on the problem resolution. It gets reported in a capability study – which is done for the sake of management. It has no effect on anyother tool used in the process with the exception of possibly the FMEA.
    The goal is results not a prfect number in relation to the long term capability.

    0
    #80446

    Mikel
    Member

    Adam,
    Instead of getting emotional like my friend Mikey (obviously not in touch with my female side like him), I think you need to ask yourself what you are doing teaching GreenBelt when you have not taken the time to quantify the difference between long term and short term performance for yourself.
    If you look you will find it, there is no way long term = short term. That is the point, the number is not relevent.
    Go get actual knowledge thru experience before you try to pass on knowledge. You will not be uncomfortable.

    0
    #80447

    Mikel
    Member

    And what exactly is your experience? Show us your data.

    0
    #80448

    RT
    Member

    The only value I’ve seen by talking about shift is that GB’s and some BB’s will complete their project with a Zst value that equates to some dollar value via a transfer function.  In this case you must remind them of the realities of process shift and random occurances that will drag their process beyond the SL.  Accounting for some shift, whether it’s actual or estimated, prevents a business from falsely reporting VCP savings etc.
    enuff said on this subject.

    0
    #80451

    Withheld
    Member

    Stans (all of you),
    No two things are exactly the same let alone two (or more) processes. Each is unique. Assuming the process varies randomly and is distributed normally, 68% of its outputs will be within +/- one standard deviation. Take a quick look at any SPC text and tell me how many rules of statistical control would be violated by a 1.5 standard deviation shift.
    If the process is not varying randomly or normally distributed, of course, anything is possible.

    0
    #80454

    Mikel
    Member

    Withhelds (any of you)
    Please tell me how many of your critical processes actually have SPC correctly applied with all rules respected. When you can tell me all of them, your arguement is a good one. Until then, your thoughts are just theories.
    Again, please tell me what your real experience is with shift. Please share your data. My data says that all who make so much noise have no data.

    0
    #80456

    Withheld
    Member

    Stan,
    In your first post to me you said show “us” the data. It struck me as funny in that it gave me a visual of a guy typing while his other personality was over his shoulder telling him what to say. I was not trying to be nasty. It was a poor atempt at humor.
    To answer your question I manage many processes and monitor them using varying methods. I draw on that experience to thumb my nose at “the shift.” I am unwilling to type out the rules of statistical control for you. Try Ishikowa’s “Guide to Quality Control,” Ott’s “Process Quality Control,” or Leavenworth’s “Statistical Quality Control.” There are others, as well.
    By way of observation, hinging acceptance of my argument on my ability to list the rules of statistical control can be described many ways. “Informed” is not one of them.
    Finally, data is plural. My data SAY that you should look elsewhere for your next argument.
    PS: If you were not so arrogant I would have left your possible typo alone. I make many typos.

    0
    #80457

    Withheld
    Member

    Unless “degredation” is intended to mean “variation” the premise is untrue and needs to be defined.
    Are we talking about a manufacturing process monitored with a variables control chart? If so, we can expect degredation over time and SPC is the tool to prevent it.
    By understanding the feedback from the variables control chart the manager of the process is able to make adjustments to drastically reduce (if not completely eliminate) degredation of the process.
    I don’t “think” SPC prevents it. I know it.

    0
    #80458

    Mikel
    Member

    Withheld,
    First of all, I am shattered that you pointed out my possible typo (although I am not clear what you pointed out).
    Second, us is the people reading this forum trying to learn. I do try to keep my other ten personalities in check when posting.
    Third, you have a reading comprehension problem. I did not ask you to list rules for me. I asked how many places you really have SPC in place. I suspect I had read and understood at least the Western Electric SPC book while you were still wetting your diapers.
    The arrogance here is that you claim to have knowledge that you don’t.
    Show us (that is the readers of the forum, not my female alter ego, Jackie) your data that gives you greater wisdom than all of us who know that process shift is real.

    0
    #80459

    Withheld
    Member

    Stan,
    I answered your questions already. Regarding your closing comment about “all of us who know the shift is real,” perhaps you should review this thread. I’m pretty sure you’re the only one standing behind the 1.5s shift.
    Others have acknowledged a shift, but I’ve yet to see anyone willing to stand behind the 1.5 constant you endorse so strongly.
    Let’s go this way: I defer to your apparent seniority and welcome the opportunity to learn other than what actual practice has shown. Here are actual outputs from actual processes. One is subgroups from an Xbar & R chart where 1 = .0001  and the target is 0. The readings are: -.4, .8, .4, -.4, .2, -.4, -.4, -1.0, -1.0, .2, -.6, .6, -.8, .6, .4, 1.2, .2, -.4, -.6, -.6, .4, -.4, 0, -.6, .2.
    The second process is volume by the week (I chose not to elaborate – what the volume pertains to need not be known) where 1 = 1000. The readings are: 98.2, 74.7, 104.1, 117.2, 93.8, 116.5, 124.1, 106, 100, 96.1, 99.7, 91.6, 112.1, 114.5, 115.8, 114.3, 106.3, 104.7, 107.7, 112.7, 104, 96.4, 104.2, 107.8, 99.2, 149.6, 102.1, 110.3, 113.8, 126.4, 112.1, 107.9, 162.7, 107.2, 91, 106.8, 100.1, 113.8, 97.8, 98.4, 176.1, 108.4, 92.4, 120.6, 100.7, 105.8, 116, 119.8, 106, 152.3Please demonstrate the 1.5s shift for me.

    0
    #80460

    Mikel
    Member

    I endorsed the 1.5 shift? Strongly?
    I refer you back to the part about your reading comprehension problem.
    Show me where I said anything about 1.5.
    By the way, I agree that properly implemented SPC will not allow a shift of more than about .5. I just don’t believe you have SPC on all (or even 10%) of your critical processes.

    0
    #80461

    Withheld
    Member

    Stan,
    I commented that the 1.5s shift was a fabrication. You challenged me without qualification of the 1.5 figure, thereby 1) endorsing it, or 2) being a troll – you choose.
    Now, after answering your bleating need for data by providing some, along with a request to demonstrate your point, you ignore the data in favor of attacking my use (or lack thereof) of SPC?
    I could let you off the hook, Stan, but I won’t. Since it simply does not matter, let’s assume that I have never used SPC and have no understanding of the topic. That done, you are left to swing in the wind until you put up or shut up.
    Either quantify the shift or go away.
     

    0
    #80462

    Mikel
    Member

    Dear Mr. Withheld,
    I am shaken that you have not let me off of the hook.
    Quantify the shift. Every properly done Black Belt project has a measure of short term capability and long term capability. Has anyone ever found the difference to be 0?
    As far as the data you sent, try putting in a shift of .4 sigma and simulate it a couple of dozen times. How many samples, on average does it take to find the shift?
    Your, or anyone elses use of SPC is a relevent question, because no one uses SPC for all critical processes and very few use it at all if you include proper interpretation and appropriate actions as part of SPC. Using SPC as eveidence that the shift is not real is just smoke.

    0
    #80467

    Mike Carnell
    Participant

    Rt,
    Financial data is full of projections so I don’t have any trouble with using them here. If you actually do them and track your results to them (and make the accuracy of them a performance metric) it will get better just like any other metric. The projections are the outcome of a process. If you work the process that creates projections you should be able to make it more accurate.
    I agree – enough said about shift. Thanks.

    0
    #80468

    John J. Flaig
    Participant

    Mike,
    Your right, I don’t know Dr. Harry — Is this a relevant issue? You indicate that you think some supporting evidence exist but you do not provide any references. Do you have them or are you accepting assertions without data.
    Further, you seem to thing that the numbers produced are only for management consumption and are hence of no consequence. Just who do you think makes decisions about funding projects?
    Final, you indicate the numbers are not important, it is results that count. However, if the numbers are wrong, then those managers (that you consider irrelevant) will not make the best project funding decisions. You conclude the since good results have be achieved, then the current program must be correct. This is a classic case of “post hoc, ergo propter hoc” thinking — a well known logical fallacy.

    0
    #80469

    Mikel
    Member

    John,
    Do you have actual experience either with working or managing this type of work? Or is your arguement strictly academic?

    0
    #80472

    Mike Carnell
    Participant

    John,
    Yes, you made knowing Mike relevant when you asserted that he used 1.5 as a rule of thumb. Since you don’t know mike it would be difficult for you to determine Mike’s motivation.
    I really hope your reading skills improve. I did not say that the numbers were for management. Capability studies are a reporting tool. They add nothing in the way of information to a person trying to solve a problem. You report your standard deviations in groups of three and refer to it as Cpk or Ppk. Really a non-value added operation. Just a ritual we insist on. As far as who makes decisions – I think I have done this long enough to realize who makes decisions. Does the capability study in and of itself let them determine funding – not the ones I consult with – maybe the ones you do.
    Nobody said the numbers were irrelevant. Sitting around waiting for perfect information will make a person and eventually a business irrelevant.
    John, you blew onto the site like everyone was waiting for your statistical enlightenment. People do appreciate the stats but the baggage that you are “the light and the way” won’t wash. Most of these people are practitioners and don’t deserve the condescending disrespect and arrogance that has been your hallmark since you posted your first response
    You have repeatedly demonstrated you are an accomplished academician who doesn’t have a clue what it is like to have to produce results for a living. I watched “Finding Forrester” last night. I thought of you repeatedly.

    0
    #80489

    Withheld
    Member

    Stan,
    In the end the only thing that matters is the 1.5s shift. I call it a fabrication. In response to my claim you challenge my experience and knowledge and make a general claim that the shift is real but completely ignore the challenge to prove your point.
    Once again – let’s stipulate that I know nothing. Walk me through it as you would anyone else who knows nothing – step by step without ambiguity.
    Quantify the shift.

    0
    #80493

    Cannizzo
    Participant

    Mike,
    Please consider ending your discussions on this thread. I’m not sure what issue you have with the other poster, but it’s becoming annoying to read it (and I keep reading this thread to learn more about the topic, not about your personal issues with the other poster). This is not the place to air your dirty laundry or settle personal vendettas — that is how it’s coming across to everyone reading this thread. Maybe you could just email him directly?
    I usually appreciate your comments, suggestions and thoughts. I usually find them enlightening. But this side of you leaves something to be desired. Please consider not revisiting this discussion thread so that we can keep things moving forward on this discussion forum.
    Thanks.
    –Carol

    0
    #80495

    Aravena
    Participant

    Nothing better that coming to work on Monday and reading a classic Carnell…  a guilty pleasure – somewhat like Doritos and Beer.
    John, or should I said Dr. John Flaig, Mike’s right.  Drop the PhD (your only impressing your wife with that one) and the latin (you hurt yourself with that one) and try to help out by applying your obvious talent and knowledge to the underlying question.  In this whole string, I have not read a single explanation of the shift that would help a practitioner explain it to a Green Belt.  I am guilty as well for making light of the shift’s origins.
    I won’t pretend to have the depth of statistical knowledge that many of the folks in this string possess (Stan, J Flaig).  I’ve been an MBB for 2 1/2 years and most of what I’ve learned is self-taught.  But, as an ex-Finance leader who will return to his function someday soon,  I do know a few things.  (1) There are many more practioners than purists (2) practioners will “rotate” back to operational roles, purists tend not to (3) practioners and ex-practioners control the $ spent on consultants (4) practioners are responsible for the spread of Six Sigma.   
    I marvel at the passion you all display.  What some of you fail to realize (but I think Mike gets it) is that what you are arguing about is immaterial.  But then again, there are no ASQ classes in “materiality” and no PhDs in common sense.  Your elegant and tangental answers serve only to demonstrate your innability to grasp the obvious.  But it’s fun to read anyway.
    pablo

    0
    #80519

    Mikel
    Member

    Dear Mr. Withheld,
    Try reading this slowly, it may help.
    Go get real data. Analyze it. See the difference between long term and short term.
    Run Spot run.

    0
    #80522

    Carl H
    Participant

    Carol,
    I agree – too many personal attacks and not enough discussion about shift!
    My customer asks for Cpk and Ppk information on certain property of lots of material we send them through the year.  The Ppk tends to be lower than Cpk by 0.2-0.6 depending on the quarter I send him the data.  Sounds like a 0.6-1.8 Zshift to me.
    Its pretty obvious where the shift is coming from.  We produce about 50 lots of this material 5-6 times per year and the SD within a run is similar each run.  What tends to change “a lot” is the mean/median between each run (due to setup differences and poor on aim control….).  For us, the SDrun-run drops our Ppk.
    Yeah, I know I should not calculate capabilities metrics on process which is out of control (we set the center and CLs based on our first “good” run), but our customer just wants us to keep the Ppk up around 1.3.  They know that we have short and long term variation and they have been good enough to give us some time to work on keeping the process centered and reducing all sources of variation.
    I dont usually dwell too much on the metrics internally.  We do use Zbench as a standard QC  metric to estimate long term DPMO our customer sees.
    Regards,
    Carl
     

    0
    #80535

    Ron
    Member

    YES!  These studies have been driven from data accumulated at Motorola.
    Why fight it. Use it and understand it origin. Te thought is that short term process variaiton differs from long term variation and we utilize the 1.5 sigma to highlight that variaition.
     
    It it sort of like understanding a new language, the words literal meaning is often different than how it is used in the country of origin.
     

    0
    #80536

    Withheld
    Member

    As I’ve stated countless times the issue is with the 1.5s shift. You can twist words any way you want if it feeds your ego, but I have not argued that there is no difference bewteen short and long term capability. I have argued that the 1.5s constant is a fabrication.
    Each process is unique. Any shift between short and long term capability will be between “0” and some unknown value that varies from process to process.
    That said, argue with yourself from here on, Stan. You have wasted my time.

    0
    #80537

    Kicab
    Participant

    Whew! A lot of words, personal attacks, ad hominem arguments, little support and data. Why not just define what each term means? As either a customer or producer, I am interested first in what my process actually produced. We call this process capability long-term in our company. It is actual performance for a specified period over many units. We try to estimate what the process potential is. We call this process capability short-term. Yes, we “bought” the 1.5 shift to make this estimate. However, we also provide an operational definition: the process centering potential is the optimal target and the process variation potential is the “smallest” variation possible. One way of estimating what the “smallest” variation is, is by using rational subgroups to separate within (“error”) variation from between variation, and using the within as the estimate of smallest variation. Then, the difference in process capability between actual and potential is the shift. At this point, the 1.5 is irrelevant. Is this the best approach? I don’t know but it does have its weaknesses. For example, how you select the rational subgroups will influence your estimate of the process shift; how you select the optimal center when there is only one specification, will influence your estimate of process shift; how you calculate potential process capability when your distribution is continuous but not normal will influence your estimate of process shift. Yes, we teach the 1.5, but personally I see no value in calculating process potential nor in calculating process shift. My only interest is in what my process is actually doing and how can I make it better. I, at least, can answer both questions without referring to the shift or the potential.

    0
    #80538

    Mikel
    Member

    Take responsibility for yourself – you wasted your own time.
    I personally consider play time and nap time as valuable time, so my time wasn’t wasted.

    0
    #80542

    Charlie Pfaff
    Participant

    I strongly agree. It is absurd.

    0
    #80556

    j
    Participant

    This discussion is absurd.
     
     

    0
Viewing 42 posts - 1 through 42 (of 42 total)

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