Six Sigma | Never build a control chart this way! | Green Belt 2.0® Lean Six Sigma | fkiQuality HD 

Control charts have been made unnecessarily complicated and downright defective by people who don't understand their origins. For instance, some teachers say that the data has to be normally distributed - not true, as the control chart theory is independent of any statistical model, such as normality. Or that the process has to be in control before using this type chart - false, since the purpose of the tool is to detect out if control situations.

Great visual on how confusing +/- sigma with sigma can be and the effect on control limits differ between the two.

Great explanation of the difference between the Sigma lines and the standard deviation. The control chart gives you a visual representation of data over time. It shows whether a process is displaying abnormal behavior .

I left Tuesday's class thinking the UCL and LCL were the same as +/- 3 sigma. This video really helped break it down and explain the difference.

I like how this showed the side by side comparison to show that the data can be skewed by utilizing the 3s control chart and utilizing the correct control chart.

Very good explanation of the difference in meanings between the 1,2 & 3 sigma lines and the sigma used to show the standard deviation

Great visual explaining sigma and standard deviation and how to correctly use the control chart. I liked the side by side visual of showing the correct and incorrect control charts. It was easy to understand.

This helps us focus on control charts and UCL, x-bar, and LCL. It seems to me the words and labels of standard deviation, deviation, s, and sigma have taken on additional meanings and applications as the concepts behind them get (sometimes incorrectly) applied to processes. The "sigma" of a process is not the standard deviation (represented by the greek letter sigma), and as we talk about deviation and variation, it gets tempting to apply a normal distribution and it's properties to everything - we're looking for connections. Here, they don't exist. Stick with the way control charts are correctly built and use them and their terminology for what they are meant to represent.

I never realized how using the 's' in your calculator could be that detrimental to your UCL and LCL. This could throw everything else off. Great tips in this video!

Very informative video showing how using three-standard-deviation limits rather than three-sigma limits in control charts is incorrect. Using three-standard-deviation limits widens the control limits and may not show abnormal process behavior.

I like how the video shows a visual chart. Control charts shows variation of a process when built correctly.

Also really like the side by side demonstrations of the correct and incorrect control charts. The biggest and most obvious takeaway for me here is, and not to copy Francisco's summary word for word, but to NEVER use S (sigma) when calculating control limits in a control chart.

Very helpful video showing the correct way to prepare a control chart. I've seen the chart prepared using standard deviation to determine the UCL/LCL before and didn't realize this was wrong until it was discussed in the Lean Six Sigma course.

Its good to know we shouldn't use calculators for this. This video helped me understand 3+- sigma in relation to control limits and how not to create a control chart.

This video helped to understand how variables inside the control limits could still mean there is an abnormal behavior. Before this video I was under the impression only if variables were outside the control limits would they be abnormal.

Great visual on the difference of using the standard deviation on excel.

It was interesting to realize that i shouldnt use the S on my calculator. Im glad it went through the example on the right way to do it.

I will be sure to refer to this later. I took a couple screenshots to aid me during exam prep. It is a useful reminder!

Thanks for clarifying the difference between std devs and 3 sigma.

Good graphic explanation of UCL/LCL using 1/2//3s lines to study control chart value behaviors.

It's great seeing these models of correct versus incorrect running side by side for easy comparison. Thank you.

This is golden information as data accuracy is one of the biggest challenges when working with data especially when looking to derive insights. It is important to know of every little bit of tweak involved when using constants in calculations.

This video was very informative on what not to do when building a control and what to do to create a valid control chart. Thank you. =)

Great video that really shows that the LCL and UCL are different than the +/- 3 sigma or standard deviations.

Again, I like how these concepts are explained in a visual way. Makes it far easier to grasp and 'see' how data is used.

The control chart is a great tool. In class several control charts were reviewed. Understanding the control limits and identifying your targets of the natural occurrence. The process may show that the operator is not incorrect but the process experiences out of control measures.

Very helpful summary of control charts

Great demonstration

A lot of good points here, especially the misrepresentation of sigma and LCL/UCL

Good to know that there are other formulas that create a different picture. I prefer using the most conservative methods to avoid inflating results.

The explaination of sigma and standard deviation in this video was very clear and I enjoyed the visual of the effects of using sigma incorrectly on thr control chart showing the UCL and LCL with the range between the two spread out.

Thank you for the explanation-making a mental note.

This video helped clear up a major question I had: UCL/LCL is based on proven observations of organizational behavior. Also, the use of 1s/2s/3s lines is a reference, not an actual standard deviation (easy to get confused).

I definitely see how using this chart the wrong way could result a huge negative impact on a business. I look forward to using this in future projects.

i like how control chart gives you a visual representation of the data over time.

Great review

I have been using 2 Standard Deviations to establish the UCL and LCL . Now I know the correct way to do it

great explanation

Simply put, use the correct formula when creating a control chart. Nice reminder of an easy mistake that can be made when making these

Video shows proper use and representation of control charts

Very True. Many times, the easiness of having at hand a quick button to get anything done immediately, it fails the true purpose of the aimed goal. In this case, using pre-set formulas, like the stdev in excel, won't help in detecting the abnormal behavior of the analyzed individual variable....

Remembering the formulas are important to distinguishing the difference between the two and making sure that the control charts are drawn using the +/-2.66mr formula and not the standard deviation.

Interesting with the use of sigma in several different meanings how it can affect your control chart interpretation.

With the same word such as sigma reflecting multiple meanings, you can see how you could use it incorrectly if you are not careful. You point out the distinction about the control chart and measuring the variability, using the moving range and calculating the LCL and UCL (not to confuse this with the standard deviation). Important distinction about using the correct formula.

This video outlines the difference between the different sigmas and the -/+ or LCL and UCL. It's important to understand the difference, otherwise your calculation will be off. Now you know how to correctly prep your control chart.

I see the value of having the +/- sigma lines. It can offer us a tool to investigate a process that is 'in control' but also could be trending/having problems that we are unaware of. I think that I need a bit more clarification on this concept.

I also feel that the visual display helps solidify the necessity of building control charts using upper and lower control limits versus 3 times the standard deviation. It shows how “off” it really is!

Just because a specific data point is within the control limits, does not mean there is no concern. If data gets close to these limits, more analysis should take place

I can follow the control chart and think I have the understanding of how it works except for how the UCL and LCL moved as more was added. Hoping that can be gone over a little in the class tomorrow.

We've talked about not managing data correctly and the consequences of it, and now we've seen how misunderstanding the calculation portion of the UCL and LCL can also lead to complications.

Adding the +,- 3 sigma to the Upper and Lower control limits would have made me think the values are the same, good differentiation.

What I learned that just because data falls within the limits it doesn't mean there isn't something abnormal occurring

preparing by next class..... from this it looks like no one table gives you the exact reasons for those out of control points in the process

Control charts when built properly will show the variations of a process. Variations outside the upper or lower control limits show a need to investigate cause.

Bottom line, use the right formula! I'm not sure if I'm more confused now or less :)

When should you calculate new control limits to test against and when should you carry the current control limits into a new test? The demonstration in th video shows the limits being developed as the test is run, should the limits always be redeveloped when testing or should there be a baseline that is consistently referenced?

I agree with David,s comment below. I thought I had a good understanding, but am now confused after this video. Why was sigma chosen for the lines in between the UCL and LCL? Sigma was already in use for standard deviation. Or was it visa versa?

I would agree that the UCL and LCL should not also be referred to as the +/- sigma control limits. Why apply a dual meaning to a term. i.e. Sigma. This just blurs the lines of understanding between sigma and standard deviation and doesn't advance the logic. USL and LCL works for me.

By watching this video I have done a lot of charts the wrong way.

I have used a control chart to help me maintain a particular process for many years. I recall asking why would you question data that shows a number of consecutive points in line with the average. I didn't realize that the average in your example may not have been the target. Also, should you question an outlier if it is only one point that occurred some time ago?

The great statistician Donald Wheeler, lists a few myths about how to construct and use control charts. The following is from his 7 Sep 2011 Quality Digest article "Myths About Process Behavior Charts." 'It has been said that the process must be operating in control before you can place the data on a process behavior chart ... there are correct and incorrect ways to compute the limits ... The most common of the incorrect ways of computing limits consists of using three-standard-deviation limits rather than three-sigma limits. While this approach was identified as incorrect on page 302 of Shewhart’s 1931 book, it is found in virtually every piece of software available today. While three-standard-deviation limits will mimic three-sigma limits whenever the process is operated predictably, they will be severely inflated when the process is being operated unpredictably. The express purpose of the process behavior chart is to detect when a process is changing, and to do this we have to be able to get good limits from bad data.' Three-sigma limits achieve this. When this is not understood, then we start to take wrong measures to fix the data without any justification like 'censoring of the data prior to placing them on the chart (removing the outliers), and the use of two-standard-deviation limits.' Recall the method to compute three-sigma limits for the XmR chart: calculate the x average, calculate the moving ranges and their average, then multiply the moving range average by 2.66 and add/subtract from the x average to obtain the limits, known as upper and lower (control) limits.

Hi, I guess the simple question is why do we “sometimes” call the UCL/LCL 3-sigma? However I am assuming this is why the control limits are calculated using the 2.66 not 3. I can clearly see it will be our responsibility to verify how the control limits are calculated before we start any data observations.

Interesting. I can see how inaccurate information can change the end results.

Important concept that I could see could be confused

the control limits are not the standard deviation rather they are the variation within the specific data set. using just the standard deviation ignores the data variation and does not accurately reflect the actual control limits of the specific data set.

I know this is beyond the scope of this video, but if using 3s "is a totally different concept" what value/purpose does the 3s method serve? What does it reveal? In being a "totally different concept" it doesn't mean that it is useless/bad but rather not pertinent/useful in regards to an accurate control chart and evaluation of a process. When would using 3s be appropriate?

Using STDDEV to calculate movement will create insensitive and too wide control limits.

Don't we want to limit the standard deviation of a process? First get the process in control and than limit the deviation. I am a little confused.

Thanks. Simply, don't ever use 'S' in your control charts.

great explanation