Here is a copy of a recent presentation I gave at a Department of Energy conference on leading indicators, SPC, and Deming's System of Profound Knowledge:

http://www.efcog.org/wg/ca/events/spring09mtg/docs/2-Steve_Prevette_Leading_Indicators.pdf

Here is a copy of a recent presentation I gave at a Department of Energy conference on leading indicators, SPC, and Deming's System of Profound Knowledge:

http://www.efcog.org/wg/ca/events/spring09mtg/docs/2-Steve_Prevette_Leading_Indicators.pdf

Thanks for sharing. I have a few questions:


You say, "Any attempt to dissect a system will destroy it." Can you elaborate? Are you saying that a "system" is such that removal of a component renders the system inoperable?
On page 11 you say "We cannot predict the future." Then on page 13, dealing with Deming's SoPK, you say, "Theory provides ability to predict." Do these statements contradict one another?

Thanks for sharing. I have a few questions:


You say, "Any attempt to dissect a system will destroy it." Can you elaborate? Are you saying that a "system" is such that removal of a component renders the system inoperable?
On page 11 you say "We cannot predict the future." Then on page 13, dealing with Deming's SoPK, you say, "Theory provides ability to predict." Do these statements contradict one another?


Good questions. For more information on dissection of a system destroying it, see Dr. Russ Ackoff's work. One of his examples I like to use is "you write, your hand does not write. To prove it, cut off your hand and lay it on the table. You will find it no longer is capable of writing". Or I use the example of the dissection of the frog in high school - how many people were able to put the frog back together again and get it to hop?

Deming has a quote to the effect that engineers are very good at predicting that failures will happen, but they cannot state the time, place, and who it will happen to for the next event. There are some that hope that some magic set of "leading indicators" will predict that a bad event is coming . .. soon.

Here is a copy of a recent presentation I gave at a Department of Energy conference on leading indicators, SPC, and Deming's System of Profound Knowledge:

http://www.efcog.org/wg/ca/events/spring09mtg/docs/2-Steve_Prevette_Leading_Indicators.pdf

That was an interesting presentation. I am a big supporter of leading indicators signalled by SPC charts. I have found some good leading indicators in range charts. Unless I skipped over the data too quickly and missed it, was there any evaluation of range chart information and leading indicators?

Nicely done as always!

I think slide 15: Common Trending Errors is a million dollar slide.

Our local ASQ section had a presentation by 3M, and EVERY company business metric was in the form of an IMR chart.

I asked the presenter and he stated that every person from the plant manager to the floor employees were well schooled in variation and understood common vs. special cause.

Now you have shown another example of best practice for management metrics.

Sadly I expect most people still "know" that 3 points in a row declining is cause for heads to roll while 3 in a row increasing is cause for bonuses and celebration.

The add trend line feature in Excel has a lot to answer for.

That was an interesting presentation. I am a big supporter of leading indicators signalled by SPC charts. I have found some good leading indicators in range charts. Unless I skipped over the data too quickly and missed it, was there any evaluation of range chart information and leading indicators?

We've been strictly using x-individuals, or more commonly, p, c, and u charts. I did originally try to use xbar-R 15 years ago, but the whole concept just was completely lost by management. They wanted their charts in monthly increments. And most of what I plot are counts of bad things, best handled with the c-chart and Poisson assumptions for the control limits.

Here is a copy of a recent presentation I gave at a Department of Energy conference on leading indicators, SPC, and Deming's System of Profound Knowledge:

http://www.efcog.org/wg/ca/events/spring09mtg/docs/2-Steve_Prevette_Leading_Indicators.pdf

Thank you Steve! Excellent work on this presentation. :applause:

Stijloor.

Great presentation -thanks for sharing :applause:

Good questions. For more information on dissection of a system destroying it, see Dr. Russ Ackoff's work. One of his examples I like to use is "you write, your hand does not write. To prove it, cut off your hand and lay it on the table. You will find it no longer is capable of writing". Or I use the example of the dissection of the frog in high school - how many people were able to put the frog back together again and get it to hop?

I was hoping for a more direct example. There are not many of us cutting off our hands and expecting the same function and killing frogs and expecting them to come back to life. Is there a more practical example?

Deming has a quote to the effect that engineers are very good at predicting that failures will happen, but they cannot state the time, place, and who it will happen to for the next event. There are some that hope that some magic set of "leading indicators" will predict that a bad event is coming . .. soon. It's all about understanding the nature of random events, I guess, and it's a good point. Statistics can help us to understand when conditions are such that we shouldn't be surprised if something bad happens, but we can't know when it'll happen if the event is really random in nature.

I was hoping for a more direct example. There are not many of us cutting off our hands and expecting the same function and killing frogs and expecting them to come back to life. Is there a more practical example?

It's all about understanding the nature of random events, I guess, and it's a good point. Statistics can help us to understand when conditions are such that we shouldn't be surprised if something bad happens, but we can't know when it'll happen if the event is really random in nature.Could it be as simple as knowing [predicting] a tire or windshield wiper on a car will fail, but we can only give a close approximation of MTBF (mean time between failures?) So perhaps the real question refers to the prediction of the time frame or number of cycles before failure. Is it wrong to say we can predict MTBF even if the date or number of cycles is kind of fuzzy?

The whole thing may be as difficult to understand as the idea broached in quantum mechanics that the real location of an electron orbiting the nucleus of an atom cannot be pinpointed except "statistically."

I was hoping for a more direct example. There are not many of us cutting off our hands and expecting the same function and killing frogs and expecting them to come back to life. Is there a more practical example?

This is the whole crux of the "systems thinking" folks. There is a school of thought that Americans as a culture are taught to dissect things as part of problem solving. For example - how do you eat an elephant? One bite at a time. Or the dissection of school curricula into independent "schools". Or the "stovepipe" organizations we see in American industry. Our focus (and this is what the standard MBA curricula teaches) is to create a "work breakdown structure" and let each piece achieve its own goas, with the hope that the addition of all of the pieces adds up to something positive.

The systems thinkers say we must integrate "upwards" to the higher level of the problem, rather than breaking it down. I really can't do as much justice as Senge and Ackoff do to this topic, and a lot of their writings are available.

An interesting point is that Systems Thinking isn't about examples (pieces parts) it is about integration and theory. Not saying that everyone buys into it, but it is out there, and I personally think there is much value to it.

Steve, Excelent presentation.:cool:

I wonder about the charts used, ¿did you used manhatan charts and moving averages to detect changes? ¿are they the tools to detect the changes, or wich other tools do you recommend?

Thanks in advance.

Steve, Excelent presentation.:cool:

I wonder about the charts used, ¿did you used manhatan charts and moving averages to detect changes? ¿are they the tools to detect the changes, or wich other tools do you recommend?

Thanks in advance.

Definitely NOT moving average charts (not sure what a manhatan chart is). I do use SPC - statistical process control and a set of rules for what is a change in the data. For more info on the details of SPC, please see http://www.hanford.gov/rl/uploadfiles/VPP_4_SPC.ppt

Definitely NOT moving average charts (not sure what a manhatan chart is). I do use SPC - statistical process control and a set of rules for what is a change in the data. For more info on the details of SPC, please see http://www.hanford.gov/rl/uploadfiles/VPP_4_SPC.ppt
Manhattan charts appear to be 3D Bar charts plotted in 2-axes. See page 12 (http://analytics.ncsu.edu/sesug/2001/P-209.pdf) of this PDF on SAS

Manhattan charts appear to be 3D Bar charts plotted in 2-axes. See page 12 (http://analytics.ncsu.edu/sesug/2001/P-209.pdf) of this PDF on SAS

Thanks for pointing out the reference. I assume they are called Manhattan charts since they look like skyscrapers on a street grid. Nope, I don't use those for detection of trends either.

The Manhattan chart may be able to serve the same purpose as a Pareto chart, just with two variables instead of one. But, my contention is that you need to know if the process is stable or has a trend prior to making such a chart. If the process has been stable, display the Manhattan (or Pareto) chart over a long period of time (duration of stability) in order to understand sources of common cause variation. If instead there is a trend, you need to look at the duration of the trend itself, and perhaps compare that to the previous stable data, to see what has changed.

I do believe that Dr. Shewhart was right - the key to managing process data is to determine if it is statistically stable (common cause variation) or if there is a changing condition/ trend. Moving averages don't provide a cutoff as to what is a trend or not, nor do Manhattan charts by themselves. The user is left to guess for themselves if there is a trend. SPC offers a specific set of rules as to what is a trend and what is not. That, I believe, is the underlying strength of SPC.

Sorry about the confusion, part was for not reading carefully (the moving averages confusion) where 2 variables on the same chart.

The other with maybe different control limits looks prety much like a manhattan chart, really it's use is to detect changes. Donald Wheeler explain their use on Advanced topics on SPC, looks more like a CUSUM chart that behaves like skycrapers.

My apologies.

I tried the new link but without any success (the ppt presentation).

nice presentation on a very key organizational topic...thanks for sharing ! :agree:

I am new to this forum, but thanks for sharing this interesting presentation.:agree1:

Slide 22, "Central committees to determin metrics are not effective".
Amen to that!
I am seriously considering quoting you and using this sentence in my signature at work. I am working in the mobilephone industry and I in these days I am bombarded with suggestions for new indicators from mgmt. Most of them are picked out of the blue without any serious thought to what and how these are actually measuring any relevant process at all. The one thing common about these indicator suggestions is that the targets are always set before the metrics is even defined :)

This is the whole crux of the "systems thinking" folks. There is a school of thought that Americans as a culture are taught to dissect things as part of problem solving. For example - how do you eat an elephant? One bite at a time. Or the dissection of school curricula into independent "schools". Or the "stovepipe" organizations we see in American industry. Our focus (and this is what the standard MBA curricula teaches) is to create a "work breakdown structure" and let each piece achieve its own goas, with the hope that the addition of all of the pieces adds up to something positive.

The systems thinkers say we must integrate "upwards" to the higher level of the problem, rather than breaking it down. I really can't do as much justice as Senge and Ackoff do to this topic, and a lot of their writings are available.

An interesting point is that Systems Thinking isn't about examples (pieces parts) it is about integration and theory. Not saying that everyone buys into it, but it is out there, and I personally think there is much value to it.

Ahhh, but even your hero Deming said that all work is a process. A process is set of steps. It is easier to break down complex systems into its components. That does not undermine systems thinking, it just means we think systematically but act locally.

The area of systems thinking that I prefer to use most is not based so much on whether the most valuable direction for discovery is upward or downward, but more the area of interrelationships - that everything has an effect on other things, and other things have have effects on it. Adjusting one factor will not solely have one effect, but can affect many things to benefit or detriment. To boil it down, the easiest way to put it is "When you fix one thing, you are probably breaking another."

A few thoughts on recent posts.

Yes, many work processes are a series of steps. BUT - what if they can be done in any order? Or is order important? What if there are things to be done "if applicable"? Or decision points? Yes, we can often build those concepts into flow charts, but sometimes a random factor hits us. And as shown by "theory of constraints", if you work to improve a step that is not in the critical path, you are not improving the process / system. And as was well pointed out - an attempted "improvement" in one step may adversely affect another step that may shift to the critical path.

The best phrase I heard was one Steve Byers used (and I believe he was quoting another author) - Systems thinking is the simplicity on the other side of complexity.

Sometimes we must integrate upwards and understand how the process fits into the system rather than dissecting the process we are looking at into component steps.

Ahhh, but even your hero Deming said that all work is a process. A process is set of steps. It is easier to break down complex systems into its components. That does not undermine systems thinking, it just means we think systematically but act locally.As with most of Deming's pronouncements, he really wraps everything up with his theory on the SYSTEM of Profound Knowledge (SoPK.)

Certainly, the system (or process) is eyed in steps or pieces, but only insofar as each piece continues to mesh with all the other steps. Deming never espoused considering any step or phase in complete isolation without considering its effect and interaction with the remaining steps.

A few thoughts on recent posts.

Yes, many work processes are a series of steps. BUT - what if they can be done in any order? Or is order important? What if there are things to be done "if applicable"? Or decision points? Yes, we can often build those concepts into flow charts, but sometimes a random factor hits us. And as shown by "theory of constraints", if you work to improve a step that is not in the critical path, you are not improving the process / system. And as was well pointed out - an attempted "improvement" in one step may adversely affect another step that may shift to the critical path.

The best phrase I heard was one Steve Byers used (and I believe he was quoting another author) - Systems thinking is the simplicity on the other side of complexity.

Sometimes we must integrate upwards and understand how the process fits into the system rather than dissecting the process we are looking at into component steps.


I would really like to see a process where steps can be done in any order.

Edit: a real process. not a made up example.

Here is a copy of a recent presentation I gave at a Department of Energy conference on leading indicators, SPC, and Deming's System of Profound Knowledge:


Hi

Thank you Mr Steve! Excellent work on this presentation.

rgs
Ngo

One of the examples I use when training in systems thinking, and the concept of the effect of interrelationships is this simple exercise:

Here is the scenario:

You are standing in front of the sink in your bathroom at home. You have a tissue that you wish to discard. You have a decision: You can either throw it into the wastebasket or flush it down the toilet.

What can you write down about this decision? It seems like a very benign choice, and most people at this point do not have much to say.

The first step is to describe how the decision can affect you. Those points are included in the upper portion of the attached cart.

The next step is to describe how the decision affects the rest of the world around you - not only those you directly affect, but those beyond you. To do this the flows from you to the endpoint are mapped out. Along the flows all of the people and materials (essentially the fishbone of each action) that make the action happen are added. That includes all of the "overhead" - management, unions, accounting, purchasing, taxes, government, etc. Once that is developed, you look at the decision and see who will benefit from the choice, and who will not benefit from not being chosen.

For this particular choice, it is also noted that the wastebasket choice heavily supports the fuel industry, where the flush choice supports the electric industry. You can see how an electric company may try to influence one to flush their tissue instead of throwing it away.

After going through the exercise (it is given blank, and the participants help fill it out), I ask how much one can write about the decision - and it becomes much longer with the additional thought processes, based on systems thinking.