Forum Replies Created
- January 22, 2010 at 3:29 am #188571
Not a dumb question for some. Some people like to know the history of things – about 25% of the population wants to know the history. It is like asking for the history of Six Sigma. Where did it come from? Why did it work in the past? Does what happened in the past imply it will work in the future? He asked a simple question. If you don’t know the answer, then don’t answer. So simple. But, I see from this forum that does not stop you.0January 21, 2010 at 2:16 am #188535
You know not what you just did. Let the fireworks begin.0January 7, 2010 at 9:01 pm #188099
Let’s try that code again:Function d2cal(n)Dim StartNum As Long, EndNum As Long, NumPoints As Long, Dx As DoubleDim Int1 As Double, Int2 As Double, i As Long StartNum = -200 EndNum = 200 NumPoints = 1000 Dx = (EndNum – StartNum) / NumPoints Int1 = 1 – (1 – Application.NormSDist(StartNum)) ^ n – (Application.NormSDist(StartNum)) ^ n For i = 1 To (NumPoints – 1) Int2 = 1 – (1 – Application.NormSDist(StartNum + (Dx * i))) ^ n – (Application.NormSDist(StartNum + (Dx * i))) ^ n Int1 = Int1 + 2 * Int2
Next i Int2 = 1 – (1 – Application.NormSDist(EndNum)) ^ n – (Application.NormSDist(EndNum)) ^ n Int1 = Int1 + Int2 d2cal = (Dx / 2) * Int1End Function0October 26, 2009 at 1:39 pm #186372
I read your article in your continuing debate with Dr. Wheeler. In your real data example, you say:”The example data used in this discussion is a true enterprise view of a business process. This reports the time to change from one product to another on a process line. It includes six months of data from 14 process lines that involved three different types of changeouts, all from a single factory. The factory is consistently managed to rotate through the complete product line as needed to replenish stock as it is purchased by the customers. The corporate leadership considers this process to be predictable enough, as it is run today, to manage a relatively small finished goods inventory. With this process knowledge, what is the optimal method to report the process behavior with a chart?Figure 1 is an individuals chart of changeover time. From this control chart, which has no transformation as Wheeler suggests, nine incidents are noted that should have been investigated in real time. In addition, one would conclude that this process is not stable or is out of control.”Do I understand you correctly that you used data from 14 lines and three different change outs on the same chart? Isn’t that kind of mixing apples and oranges? I don’t think I would put three different types of changeouts on the same individuals chart.It might be interesting if you provide the actual data and show the context in which it was taken including process, type of changeover, etc. Are 9 out of control points (with over 500 data points)due one type of changeout? I can’t tell based on our article.Regardless, interesting reading.Bill0August 21, 2009 at 6:56 pm #184962
Not for me. My pages still take 10-15 sec to load. Very slow response. Not that anyone cares, but I think I’ll just go play somewhere else.0August 21, 2009 at 9:51 am #184951
Andy: Sometime back you pointed out that most things were topsy-turvy – you are right. Thanks for trying to wake us up. Regretably, the sleeping giant is still asleep.Bill0March 16, 2009 at 7:46 pm #182421
Can you send me the simulation? Thanks.Bill
firstname.lastname@example.orgMarch 9, 2009 at 5:26 pm #182178
No, not a book question. I’m trying to apply this as a way to compare destructive testing methods at different manufacturing sites. I have multiple moisture analyzers that I need to compare.0March 6, 2009 at 3:05 pm #182086
Thanks for the reply. WHat I’m specifically trying to do is use a simple one-way ANOVA to show equivalence. It’s relatively easy to choose the “zone of indeifference” in a 2 sample equivalence test, since this is a pretty intuitive thing. For example, if I want to prove equivalence between the strength of two different materials, I would say “if the sample averages are within 10% of each other, this is equivalent for my purposes”. Then I would do two one-sided tests to confirm that the lower and upper confidence limits are contained within the +/- 10% zone.
THe problem with doing this with ANOVA is that the “zone of indifference” is no longer expressed in the original units of measurement, but in the ratio of the variances (F-score). I just can’t seem to make the logical leap that makes this intuitve. For example, I could say “the zone of indifference (equivalence of multiple samples) is when the F-score is less than 2”. This would be a hard limit that you could then select appropriate sample sizes to achieve, then test against. But expressing equivalence in terms of an F-score is just not intuitive – to me at least.
Clear as mud?0March 2, 2009 at 10:42 pm #181872
You are wrong.
3 sigma is only used by that crazy Shewhart – Deming – Wheeler crowd.
In the highly acclaimed and incredibly brilliant Six Sigma, we only use control charts with 6 sigma limits. The psychologist Mikel Harry has proven it to be true.0December 4, 2008 at 5:39 pm #178327
I disagree on using the p control chart. The probability of attrition is not the same for all employees. For example, younger employees tend to switch jobs more often. Thus, one basic condition for using the binomial distribution for the control limits in the p chart is not met.
I would use an individuals control chart.0November 30, 2008 at 3:17 pm #178175
Here it is free down load for cqe exam
Good luck0November 18, 2008 at 3:38 pm #177784
Can you help? Where is this open source & QCI material?
I am new to the site.
Thanks!0June 23, 2008 at 7:38 pm #173120
Why do you want to use a subgroup size greater than 25?
Bill0May 22, 2008 at 9:56 pm #61999
Michael, I am interested in knowing more about those benchmarks from GE. Could you email your contact information? My email address is email@example.com.
Thanks, Bill0May 12, 2008 at 2:00 pm #171906
Kris, good question, more subtle than some realize, Robert and BC have provided the basic math. Remember that if the slope is not statistically signficantly different from zero, then there is not a significant correlation between the x’s and y’s. The algebra provided by BC and Robert make that clear, as b –> 0, r^2 –> 0 too.0April 20, 2008 at 11:57 pm #171337
If z = 1.85:
The area under the curve to the left of z = 1.85 is found using the following in Excel:
normsdist(1.85) = 0.968
You can also use: normdist(1.85, 0, 1, true) which is for the normal curve wiht average of 0 and standar devation of 1.
For the area under to the curve to the right of z = 1.85 simply use:
1-normsdist(1.85) = 0.032
I don’t know where you get your 0.4678 result.
Bill0April 10, 2008 at 8:00 pm #170973
Just a quick add to RhinoNeil’s response. Several recent articles have been published regarding the iSSSM six sigma maturity model. Easiest place I know to find it is at the Instantis home page or perhaps ISSSP website. Our team has been wrestling with a similar question and has recently begun work to integrate the iSSSM maturity model, Michael Hammer’s Process Audit (Harvard Business Review) and a set of lean maturity criteria (Shingo). Our objective is to create a self assessment tool that; (1) helps support our longer term vision of excellence and (2) provides a gap analysis that will then become an input to the planning process.0January 24, 2008 at 10:20 pm #167694
As a motivational speaker and someone with a lot of credibility in the manufacturing field, Mario Perez Wilson is probably the best.
We used another famous individual first, and it was a disaster. Then, an ex-Motorolan recommended we contact Mario and it was unbelievable.
Our Six Sigma program runs flawless. We hired his consulting firm to proceed with the whole training and deployment.
My company is in Massachusetts we are in manufacturing. Our program got delayed by 6 months for picking the wrong person the first time.
He has two websites http://www.perez-wilson.com and http://www.mpcps.com
Good luck.0January 22, 2008 at 11:04 pm #167581
What about Six Sigma in the blow-molding industry? Have you learned anything interesting about that? You’re not still giving away extra plastic for free are you?0January 8, 2008 at 7:26 pm #166976
Why pay for training if you already understand the methodology? You could become GB certified through ASQ – though you might need to find a good Lean Six Sigma text and brush up. Good luck!0December 27, 2007 at 1:33 pm #166573
What is the best way to contact you?
Bill0December 10, 2007 at 3:10 pm #165947
I guess my first post didn’t really pertain to the original title of this thread, but then a lot of the posts are likewise guilty.
I have to agree with Brandon. No quality system can overcome people issues, only management can shoulder that load when people are not capable or willing to give their best effort.
I also have to agree with Six Sigma Shooter, mangement is mostly interested in short term performance. Many CEOs, etc. only care about their bonus or golden parachute that their cronies (since they all sit on each others’ boards) grant them. Providing the resources (ISO requirements be damned!!!) to provide excellent customer service or any other long term improvement just isn’t on their personal agenda.
Many companies hire fresh faces out of college to mid level management positions that just don’t have the experience to be effective.
Finally, even many posters to this thread, well intentioned as they may be, can’t even agree among themselves about what quality should be. It doesn’t seem to matter to me what “method” one selects as long as you get where you are going. Instead of debating flavor of the week, we should figure out how to convince and educate the masses. Many of the “problems” that are posted would disappear if more people up and down the chain of command were quality practitioners.0December 7, 2007 at 9:37 pm #165884
I am new to SS, went through TQM, ISO, etc.
Question is, with all this quality going around, why are most companies so poorly managed and why does most customer service stink?0November 15, 2007 at 10:11 pm #59207
What are you refering to when you say “contact rates”?0November 2, 2007 at 8:36 pm #61769
I would be curious too. I am a new black belt and would like to be more involved here at this hospital.
Bill0November 2, 2007 at 8:34 pm #61768
How does this book relate to healthcare? I am curious. How helpful is it?
0October 4, 2007 at 2:45 pm #61715
What do you mean by Takt? Our transcription is done by an outside vendor. I am doing a labor production analysis now. When you say TAT, do you mean from the time the exam is started to signed out?
Finally, what do I google search on, radiology, CT, MRI?
0September 12, 2007 at 7:02 pm #61684
I was not able to find anything other than the packet that is available to purchase for $9.95. Is there anything else out there, that anyone knows of?0August 29, 2007 at 9:23 pm #61668
I understand, but could some one give me some direction in getting started. Where to find the data? What data would be needed, etc.?
0August 28, 2007 at 4:45 pm #61660
That would be a great help. From what I understand, there has never been one done here before and the director would like to see one done, however, he is not sure of what he wants. But, any information to start one and to develop from there would be a good start. Do you have an email where I can contact you?
0August 6, 2007 at 5:52 pm #159618
If I have 10 million lines of software code, and each line of code needs to conform to 300 software itemized code standards, what are my opportunities for success?0July 25, 2007 at 1:09 pm #159050
There are products which perform one of the functions listed, but I’m not aware of a product which does both.
For enterprise-level metrics and SPC, you could check out Hertzler Systems or PQ Systems (and there are others).
For project tracking, you could check out Power Steering, i-Solutions, Instantis, or Traction (and there are others).
0July 10, 2007 at 10:06 pm #158431
If the annual recertification entails a fee, which is most certainly does, then it also has another name: ANNUITY. Your instincts to be cynical are correct, in my opinion. Sounds like a razor blade business model.0July 10, 2007 at 10:00 pm #158430
Anybody who has ever done real work with a control chart knows that processes shift over time. That’s not the point. The idiocy of the “1.5 Shift” argument is the claim that it is anything other than a generalized fudge factor to use when long term instability is not understood. Your prior post admitted seeing shifts ranging from 1.0 to 1.8. So why use 1.5? The whole concept is noise – a distraction best ignored since it has no bearing on the actions that must be taken to actually improve a process, in my opinion.0June 30, 2007 at 7:29 pm #158090
Yes, the courses are offered for credit, within the business school curriculum. The “Masters Certificate” would be something else entirely, probably on a non-credit basis without enrolling in a degree basis.0June 29, 2007 at 5:24 pm #158067
I am aware of a couple of schools which offer Lean Six Sigma courses within their curriculum:
The University of Notre Dame, Mendoza College of Business, offers a Green Belt course within its MBA school.
The Ohio State University, Fisher School of Business, offers a Black Belt course within its MBA program, and a Green Belt course for undergraduate business students. I understand that the Engineering school will be adding a Black Belt course at the undergraduate level this fall.
Cal Poly plans to launch a BB course in the Graduate School of Business this fall.
Cornell has offered a BB course in the MBA program. I believe the instructors may be provided by Johnson & Johnson.
I hope this helps.
0June 17, 2007 at 7:38 pm #157587
Thanks so much for sharing your knowledge.0June 17, 2007 at 5:05 pm #157583
Thanks Robert. Two more questions: It appears to be the case as long as you look at the coded factors (-1 high, 1 low). If I use actual values, I get strong correlations. For example:
The R squared for AB in this case is about 0.96 which leads to a high VIF. So, you have to stay with the coded levels?
Second question: if you just do the classical two level designs (full or fractional), do you really need to even check the VIF? Won’t it always be 1?
Thanks a lot.0June 17, 2007 at 11:46 am #157575
Thanks for the reply Robert. Then I don’t understand how VIF is calculated. From an earlier post of yours (I did try to find the answer), you said:
“When you are looking at the R2 in the VIF you are looking at the relationship between the X’s. In the VIF case R2 is the square of the multiple correlation coefficient from the regression of the ith X on all of the other X’s in the equation. An R2 of 1, in this case, would indicate a perfect linear relationship between two X’s (perfect confounding).”
Supose I have two factors, A and B. They are independent. But AB is the product of the two. Is it not impossible for VIF for AB to be 1 when it is determined by the product of A and B. Is not the VIF for AB the R2 value obtained by treating AB as if it was Y and regressing with A and B?
Than ks0June 7, 2007 at 1:16 am #157094
Lots of replies already. I don’t see the most obvious thing I would do. If you have a high Cpk and your process is really in statistical control – consistent and predictable – reduce your frequency of sampling. Whether you use a control chart to continue monitoring is your choice. I would. It was important to chart at one time. I would keep it going, just sample less frequency. I believe it always best to look at data as a time series chart – with limits wherever possible. It is the way process communicates with you. The correct interpretation will tell you if something has changed or if everything remains the same.0May 11, 2007 at 10:37 pm #155997
If you analyze the data without lag, there are a number of out of control situations. Point 8 (223.2) is just below the LCL (also out on the range chart). The last 8 points in a row are above the average with the last one being above the UCL. It appears something has caused the average to move up for those last 8 points – a special cause of variation.0May 10, 2007 at 5:49 pm #155942
ISSSP is launching an MBB assessment model. I’m not sure what credibility it will have in the market, but at least it does not force you to pay for training to develop skills that you have already mastered.
Good luck.0May 4, 2007 at 10:44 pm #155734
I stand corrected. You are for real. There is one approach you might consider. You want to create systems to manage your growth. You can also look at that in terms of creating or improving processes for growth. While every company is unique, there are many processes that most organizations perform. The American Producitivty and Quality Center (http://www.apqc.org) has developed the Process Classificaiton Framerwork to aide in benchmarking of organizations. This framework actually provides a great method for process management and determining what your company needs.
There are twelve primary processes. Five of these are operating processes. The rest are management and support processes. The primary processes are:
Developing Vision and Strategy
Designing and Developing Products and Service
Marketing and Selling Products and Services
Delivering Products and Services
Managing Customer Service
Developing and Managing Human Capital
Managing Information Technology
Managing Financial Resources
Acquiring, Constructing, and Managing Property
Managing Environmental Health and Safety
Managing External Relationships
Managing Knowledge, Improvement, and Change
If you download the Process Classification Framework from their website, it will give the secondary and work processes under each of these.
The place to start, if you haven’t already, is with the vision and strategy piece.
With each primary process, I would use a five step model to ensure the process is developed as best it can be for your company and growth. The model is Define, Document, Measure, Manage and Improve. This five-step model helps you implement process management. But enough on this for now. I think you might find the process classification framework interesting.
0May 2, 2007 at 4:08 am #155573
There are a couple of people on this site who are very good. Mike C. is one of those. But I have some concerns about you. You are now a $5 million company, up from $500,000. Hmm. You believe you will be at $20 million in 4 years. Won’t happen, but OK, lets assume it does and good for you. I always root for success. You have hired consultants and accountants but they haven’t help you. Hmm. Yet, you are visionary. So, basic question, what are you doing on this discussion forum? There are some good people here (aka, Mike C.), but most just attack each other for their views. 80% ( closer to 90 probably – a sad state for this group) of the posts here are not helpful to the person looking for assistance. Why does the owner of a $5 million a year company come here looking for advice? I have no doubt there a few here that can help you, but very few. You say you have a clear vision. Then search for someone to help you. Lots of people in the world have the skills to help you. I just dont think many of them are here. I wonder if you are real. Good luck to you and I hope you are at $30 million in 3 years.
Bill0April 27, 2007 at 2:37 pm #155375
If you believe what Bisk Education (University Alliance) is telling you about the ASQ pass rate, then you are really naive. Bisk has no access to ASQ pass rate data.0April 25, 2007 at 4:29 pm #155250
First off, zero outages is an unrealistic goal. It is not going to happen. I would track it based on time: time down/total time per week and use a X-mR chart. The number of outages really doesn’t tell you anything. I would want to know the % of time the system is not available.0April 17, 2007 at 3:56 pm #154891
Where are you located and in what industry?0April 17, 2007 at 3:52 am #154862
Thanks for answering. I have gone beyond the comic books. I have read Box, Hunter and Hunter in detail. I have Montgomery’s book. All I wanted from my questions was some answers from people who have done quite a few experimental designs.0March 29, 2007 at 12:23 pm #154156
I am glad you fellows came up with the simple definition of “opportunity”.
Now I am really confused?
For us (I may be the only one reading that is not a
Six Sigma engineer)just starting out, I take what I am reading as opportunity is an event in which either conformance or non-conformance can be added to the realization of a product. The more opportunities it takes to reach conformance to customer satisfaction, the more waste. The goal is the fewest possible.
Now you can go back and talk “baselining”.0March 9, 2007 at 2:38 am #152962
Sorry, not a stats guy here (don’t really deal in minitab speak), but if your customer told you they think your spec is too wide and you have observed out of control measurements that are within some spec that was set by engineering 4 years ago, your spec is probably wrong and you need to go back to the DM part of the process to determine proper VOC. Call me crazy, but if you work for a company that considers profits important (most do), your company really should value your customer’s input with regard to their requirements.
That being said, if you observe an out of control measurement that is over the favorable line of an upper or lower spec limit, you would be foolish not to try to determine the cause of that particular variation.
FYI, I’m not GB, BB or any other B. I just find this forum strangely interesting and informative….sometimes. So take what I say with a large grain of salt.
0March 1, 2007 at 7:25 pm #152629
There is a reason I requested a simple explanation.
I am a “grasshopper” to the Quality paradigm. I read the explanations and it intimidates me because I do not have your back ground. The experts need to consider that “grass hoppers” are lurking trying to improve.
For years I rode a motorcycle, a Yamaha. All this time Harley owners disdained foreign bike owners. They also were upset with car drivers that did not “see” them and thus there were injuries to the bike riders. Now, if car owners also rode motorcycles, any motorcycle, they would be more respectful of bike riders. Harley riders knowing that HD could never manufacture all the bikes needed to service the entire market should have been happy to see anyone riding a motorcycle because it made their life safer. The more riders the better.
The trained Six Sigma belts should consider that little “grass hoppers” are looking for tidbits of understanding and if we learn something the entire Quality paradigm strengthens thus supporting your
efforts. Maybe the real jargon explanation with an interpretation for “junior”.
I love the Quality Management concept but I am not an engineer.0March 1, 2007 at 2:04 pm #152597
Please fellows, explain this in an easy to understand
What is the difference?0January 23, 2007 at 2:18 am #150937
A good Gage R&R study will use control charts to check for bias and consistency between and within the operators. I assume this is what you mean by out of control.
For example, suppose you have done a gage R&R with the three operators, ten parts and three trials. You can and should do a range chart on each operator. You take operator’s A results and plot them on the range chart, determine Rbar and the UCLr for a subgroup size of 3. If this chart is out of control, it means that the range of results opeartor A gets is not consistent from part to part. You should investigate why this occurs. Since the results are not in control, you don’t know what would happen if you run the Gage R&R again. So, find out why the operator is out of control. You can take the same approach for checking the bias and consistency between operators.0January 20, 2007 at 6:00 am #150846
Yet another student of Harry’s Handbook of Stupid Superlatives.
Leave out the SS BS and try some facts Shoddy Roddy boy0January 19, 2007 at 3:45 pm #150800
If you have Excel 2003, there is a Weibull function built-in.0January 18, 2007 at 2:44 am #150705
As stated by the previous poster, p charts do have subgroups. For example, suppose you are monitoring the fraction of invoices that are priced correctly each day. You randomly select 100 invoices a day. This is your subgroup and n is the subgroup size – in this case 100. Suppose 98 of these are priced correctly. This is np. You can then calculate p for this subgroup as follows:
p = np/n = 98/100 = 0.98
So, there are subgroups with p charts.
Best of luck to you.0January 17, 2007 at 10:44 pm #150702
It may be that your subgroup size is too large. This makes the control limits very tight and generate out of control points. You can move to weekly data and keep the p control chart. There is also a good possibility that, if you are monitoring more than one defect, that the conditions for using the p control chart are not met.
Consider using an individuals control chart for the %.0January 16, 2007 at 12:00 pm #150618
I think the original book by Chambers and Wheeler, Understanding Statistical Process Control, is one of the best. Wheeler has some very good books on SPC at http://www.spcpress.com. But as suggested before, start with Dr. Deming. Out of the Crisis is good as suggested. Not an easy read. But understanding Dr. Deming’s system of profound knowledge is very important for continuous improvement. SPC is affected by all four bodies.0December 10, 2006 at 2:31 pm #148755
The professor is generally wrong. Use the sigma table using DPMO, but recognize the corresponding “sigma” is just an estimate of the short term capability. If your DPMO is about 6,800 (long-term capability) then your estimate of short-term capability will likely be about 4 sigma. When using defect data, you can compute your long-term capability fairly well, but without switching to continunuous data, you can not estimate short-term capability. In this case you just add “1.5 sigma” to create an approximation of the short-term capability. If you have access to continuous data, then use it and compute the actual short-term capability. The 1.5 sigma shift is just a best-guess model you can use if no other information is available. The 1.5 shift is just a way to remove assignable causes from the long-term defect data so you can “guess” what the short-term capability could be.0December 1, 2006 at 3:14 am #148255
Well said Mike. We might also wonder, since there have been far more than 58 large companies that have deployed Six Sigma, how the sample of 58 was selected.0November 30, 2006 at 1:56 pm #148194
Here at Kodak, “end to end” refers to the value chain from research all the way to manufacturing and post-sale support. It also refers to the imaging flow of capture->storage->manipulation->printing and sharing. The latter applies equally well to the old film products and the current digital products.0November 13, 2006 at 6:52 pm #147090
Do you know why Dr. Harry parted ways with ASU for six sigma training?
Please be factual in your comments.0November 11, 2006 at 12:07 am #146896
A touch of sanity. As Philip says normality can never be tested in practise beyond +/-3 sigma. Even at this level, 3,200 data points are needed !
However, there is no need for tests of normality in quality improvement. Use Shewhart Charts – Normality is not required.0November 9, 2006 at 10:45 pm #146810
ASQ Certification may be recognized, but so is the Yugo, so recognition does not equal quality. Maybe in the land of the blind, the one-eyed man is king, but in my opinion the ASQ process suffers from several serious weaknesses:
1) The project requirement is easily “gamed”. Nobody reviews the actual project, only an affidavit, which can be signed by anyone at your sponsoring organization, and there is no assessment of that person’s qualifications to determine the quality of a project. Many people “re-purpose” old projects that were not even DMAIC projects, like IT upgrades, for the purpose of satisfying the project requirement in order to sit for the exam.
2) If you accept that the purpose of a Six Sigma exam should be to validate that the person being examined can perform Six Sigma work effectively in the real world, then the ASQ exam has several severe limitations: no graphical interpretation, no large datasets, and a reliance on hand calculations. Who, in this day and age, uses paper-based statistical tables to look up values? People don’t do the work that way, so why structure and exam that doesn’t reflect the way people do the work? People in the real world use statistical software, so the exam should include a component that tests the ability to run and interpret analysis from a statistical software package. And this doesn’t even touch on the assessment of soft skills, which are probably much more important in the final analysis.
So, ASQ certification may be nice to have, and in the absence of anything else, it might represent minimum evidence of competence…but it doesn’t carry any weight with me when I hire people unless there is much more to the story and a lot of solid experience.
Just my opinion.0November 9, 2006 at 7:32 pm #146797
Try looking at the iSixSigma directory here: https://www.isixsigma.com/co/six_sigma_online_training/six_sigma_online_training.asp
Note that there is no quality assessment of the listed products.
Or search the discussion forum for previous posts. You’ll find a lot of comments.0October 25, 2006 at 9:40 pm #145721
Yes, you can add upper and lower control limits to efficiencies and yields. I would use the X-mR chart.
Bill0October 23, 2006 at 5:09 pm #145510
Bechtel has been using Six Sigma for a number of years on it’s EPC activities. I my self while at Bechtel attempted to improve ES&H preformance using Lean Six Sigma methodologies. Do you have a specific problem or question?
Bill0October 20, 2006 at 5:09 pm #145287
You are not missing anything. There is no direct correlation between the two. Cpk is calculated using the estimated standard deviation from a range chart. Ppk is calculated using the standard deviation calculated from all the data. There is no correlation between those two standard devaitions, although they should be close to one another if the process is in control. The more out of control the process is, the more difference there will be.
Bill0October 20, 2006 at 2:08 pm #145263
Please post your data. Your approach seems OK, but something is not right that 32 of 35 data points are out of control.0October 19, 2006 at 7:52 am #145142
Sorry there SS Shooter … I didn’t realise that you were one of the crowd who had woken up to how much rubbish is in SS !!!
Harry and Reigle probably don’t care much now … I can imagine they spend their days laughing at the foolishness and gullibility of the masses, while counting their millions …0October 19, 2006 at 7:37 am #145140
Bravo !! Economic limits.
Perhaps you can explain it to the other SS ignoramuses here … if they understood the basics, there would be no 1.5 sigma shift, no 3.4 DPMO, no hypothesis tests and all the other SS rubbish.
0October 19, 2006 at 4:45 am #145125
If Shewart is a genius as you say, why don’t all these SS idiots read his work ?
When will people realise that control limits are not probability limits ?
0October 13, 2006 at 3:58 pm #144716
Please, would you send me the link also?
Bill0October 11, 2006 at 6:59 pm #144560
Please re-read Darth’s reply and you will get your answer. But I will repeat it. The distribution of range samples DO NOT follow a normal distribution. The zones test should not be applied to the range chart. And please do follow his advise and read Dr. Wheeler. He will explain why it is not the central limit theorem that makes the Xbar chart work.
Bill0October 11, 2006 at 5:32 pm #144551
I agree with you. I still need to answer their question and I don’t think that the math given before works out.0September 23, 2006 at 10:10 pm #143754
Of course, Darth is correct in his analysis since this is a homework problem. But it is too bad that they don’t teach that an average has no meaning without knowing the state of the process that generated it. Is the process in control or there are special causes present? For example, the data below generates an average of 6.2. Taking the standard deviation to be the square root of 6.2 would give misleading information. Always check the state of control.
00September 5, 2006 at 11:55 am #142770
Same identical question asked by Maureen on Aug 24. ??0August 24, 2006 at 4:48 pm #142280
Three sigma limits are appropriate at all times. But don’t confuse the issue of statistical control with capability. You want to control the process but also have it capable of meeting whatever spec you have for the active ingredient. You probably want a very capable process there, at least Cpk = 2.0August 21, 2006 at 7:09 pm #142072
The type of chart you are using is an individuals control chart. The control limits for this type of chart are:
UCLx = Xbar + 2.66Rbar
LCLx = Xbar – 2.66Rbar
UCLr = 3.27Rbar
LCLr = None
Rbar is the average range from the moving range chart. The range is the difference between consecutive data points. For example, the range between the first and second data points is 0.09.
Rbar for your data is 0.551, so
UCLx = 3.742 + 2.66(0.551) = 5.208
The estimate standard deviation from the range chart is Rbar/1.128 = 0.49. This is much less than the calculated standard deviation of 0.83. This is an indication that the process is out of control, which it definitely is by looking at the X chart.
On the comment that you can’t go further because the range chart is out of control. Not true. What it does mean is that the you may not have a good Rbar, but you have only one point out of control on the range chart. Not a big deal with 56 points. Regardless, all it does is inflate Rbar which means that the points beyond the limit on the X chart are still out of control. Also, the runs you have above and below average as well as the trends on the X chart still apply.0August 21, 2006 at 5:21 pm #142062
Are you sure the data you posted was correct. I get a different average (3.66). No sense going any further.0August 21, 2006 at 3:24 pm #142048
That is not true about Excel The STDEV function of excel calculates the standard deviation using n-1 regardless of the size of the sample.
If you want the population standard deviation, you use STDEVP.
Excel does not give you the wrong answer when you have more than 30 data points when you are using STDEV. Simply not true.
0August 18, 2006 at 5:02 pm #141912
I used X-mR chart to monitor sales quite often. For a chart that is trending up, you can use the best fit line for the average line on the X chart (x = bo + b1t where t is the week number). Then the control limits become:
UCLx = bo + b1t + 2.66Rbar
LCLx = bo+b1t – 2.66Rbar
where Rbar is the average range on the moving range chart of 2. No different that then normal moving range chart.
Start with five weeks of data; recalculate limits each week with new regression line until you get about 20 weeks of data. Then you can probably use those limits into the future.
The problem I have seen with using weekly data for established products is that the last week of the month is almost always the largest.
Bill0August 8, 2006 at 10:54 am #141533
Here is a link that explains the transformation.
0August 8, 2006 at 10:47 am #141532
The Box-Cox power transformation is given by (L = lambda):
x(L)=(x^L-1)/L for L0
x(L)=ln(x) for L = 0
Then you find the value of L by maximizing the logarithm of the likelihood function (which I won’t give here).
I just wondered if there were any bounds on lambda. I don’t think so.
Bill0August 5, 2006 at 7:07 pm #141445
TQM was not “all about defect reduction.”
As Dr. Deming said, “Improve constantly and forever. Constantly strive to reduce variation.”
Bill0July 30, 2006 at 12:05 am #141195
From what I can see from previous posts there has been no critique of Dr Burns paper. All I can find is childish personal attacks and gutter sniping. Personal attacks are not appropriate with the large number of authors at the http://users.bigpond.net.au/SixSigmaFallacies/ site. Let’s see some constructive discussion of the CONTENT of these papers.
These papers do present clear and superior alternatives to six sigma. Dr Wheeler in particular spells this out clearly in his books. Perhaps you feel that Dr Wheeler is flawed ?0July 29, 2006 at 11:32 pm #141194
How was Dr Burns “torn up” as you suggest ? What are the specific errors in his paper … it sounded very convincing to me.0July 29, 2006 at 12:18 am #141166
Any one of us could die any minute … heart attack, traffic accident, whatever. I’ve had 30,485,666 such opportunities for death … which puts me well in excess of six sigma.
It makes about as much sense as attaching a sigma level to any process. For some reality, read the articles here :
http://users.bigpond.net.au/SixSigmaFallacies/0July 24, 2006 at 10:32 pm #140925
It is NOT “okay” to use the calculated standard deviation for the individuals chart. This greatly inflates the control limits for processes that are not in control. Use the moving range of two.0July 17, 2006 at 10:09 pm #140511
Thanks for all your effort Robert. Your results confirm what I got. I am not clear on what you mean about not centering and scaling the X variables. The coding (using the equation you gave as I did) does that, doesn’t it?0July 17, 2006 at 11:05 am #140464
Yes, I multiplied A and B. Also, I ran the regression on the coded factors and got the same results as for the DOE ANOVA. Below is the data:
Standard Run Number
A is temperature with range from 30 to 90
B is residence time with range from 15 to 45
Thanks.0July 12, 2006 at 12:37 pm #140212
The sample means will be normal as long as your subgroup size is large enough. Subgroup sizes of 4 to 5 are usually enough but for data that is extremely skewed.0July 9, 2006 at 4:34 pm #140095
I am not sure of your process. But your Cpk is less than 1. Without 100% inspection, you will send pieces out that are not in spec.0July 7, 2006 at 6:05 pm #140069
Great!!! Email me please at firstname.lastname@example.org
Bill0July 5, 2006 at 3:02 pm #139958
Two others to look at: SPC for MS Excel and SPC IV Excel.0July 5, 2006 at 2:59 pm #139957
There are a number of things you can do. If you want an accuracy of +/- 5 %, you will need to send out the survey to about 357 people. That is too many to call, but you could send out the survey via e-mail and follow-up with those who do not respond. If leadership does it role in making sure people know this important to respond to, you hopefully won’t have too many to follow up with.0July 5, 2006 at 2:52 pm #139955
There are a number of Excel based SPC programs out there that make it easy to do control charts and other stats in an Excel environment. Just search “spc excel” to see what is available.0June 16, 2006 at 7:33 pm #139252
Below is a link for teaching some of the basic SPC tools using colored candies (like peanut M&M’s). The info is in the Annex A7.7
http://www.ca.com/de/blast/k7/hdi_chapter_7.pdf0June 15, 2006 at 7:49 pm #139203