Effective Innovation

A Reward for "Effective Innovation" Readers

noreply@blogger.com (Objective Design of Experiments) — Fri, 27 Jan 2012 19:45:00 +0000

To thank you for reading the "Effective Innovation" blog you can have access to all ObDOE eCourses for one full year for just $450 -- half the normal price! This offer is only good through January 31, 2012, so don't wait.

You can claim your reward here.

What's the Catch?

There is no catch, but you must pay by Jan 31.

Is this exactly the same as the full price eCourses?

Yes

Is coaching included?

Yes

Why are you doing this?

I put a lot of thought into the "Effective Innovation" blog. I hope this offer tells you how much I appreciate your time spent reading it.

Can I tell my friends about this?

Yes -- then they can become readers as well!

Thank You! Bill Kappele

Parity Plots

noreply@blogger.com (Objective Design of Experiments) — Tue, 10 Jan 2012 00:56:00 +0000

Chemical Engineers have a method for looking at how well theoretical values match up with measured values -- the parity plot. This plot is useful to anyone who wants to compare theoretical values with actual measured values.

Plot Courtesy of JMP, www.JMP.com

Here's the idea -- you plot the predictions from a simulation, model, mass balance, etc. on the y-axis and the corresponding actual measurements on the x-axis. A line with a slope of 1 and a high correlation indicates good agreement between theory and reality. Anything else indicates that something is missing from your theory.

For example, if you create a Response Surface model to fit you data and the plot of actual values vs. predicted values has a slope of 1 and an R^2 of 0.9, you can feel pretty comfortable that your model is describing the data well. (Warning: this does not mean that your model will predict well in regions where you haven't collected data.)

JMP automatically creates this plot for you when you use the "fit model" package. You can create this plot for yourself easily in other packages.

You can get a good idea of the quality of your theory (or model) using the simple parity plot.

Next time let's look at what R^2 really tells us.

Happy New Year!

noreply@blogger.com (Objective Design of Experiments) — Fri, 30 Dec 2011 18:25:00 +0000

We often forget that we can start fresh each day. The beginning of a new year reminds us, though, and so we set resolutions to help us re-invent ourselves.

Image: vichie81 / FreeDigitalPhotos.net

Here are a few suggestions for changes you can make in your life to make yourself a more successful innovator:

Get up early and read for one hour every day. This one habit will help you read up to 50 books a year! Focus on books that encourage creativity and the developing of good habits. Try to use what you learn each day, making new innovation skills a part of your life. Here are some recommendations to get you started:
1. The Leader's Guide to Lateral Thinking Skills: Unlocking the Creativity and Innovation in You and Your Team, by Paul Sloane. This book will open your eyes to a more complete way of thinking that will inspire your creative, innovative spirit.
2. Lateral Thinking, by Edward DeBono. Lateral Thinking is the necessary companion of logical thinking. The two together are the engine of innovation.
3. Bill & Dave: How Hewlett and Packard Built the World's Greatest Company, by Michael S. Malone. Bill Hewlett and Dave Packard are two of the world's greatest innovators. You can learn how they innovated virtually everything that is good in the workplace today and gain inspiration for your own innovation.
Listen to educational CDs and books on tape in your car. You can use your commute time to learn about anything you like. Your attention is focused on driving, not everything that is said, so you can listen over and over. Each time through the CD you will pick up different pieces of important information. Here are some CDs to get you started:
1. Smart Thinking, by Edward DeBono. Learn how to improve your thinking skills from the world's expert on thinking.
2. Eat That Frog!: 21 Great Ways to Stop Procrastinating and Get More Done in Less Time, by Brian Tracey. Procrastination is an innovation killer! Learn to stop procrastinating and innovate.
3. Getting Things Done: The Art Of Stress-Free Productivity, by David Allen. Being overwhelmed is counterproductive to innovation. Get your life under control so you can free your mind for creative thought.
Attend workshops and short courses on innovation skills. Workshops will teach you new skills and give you a chance to practice them on exercises before applying them in your work. In-person and online courses are available. Here are some suggestions:
1. Guided Tour of ObDOE eCourses For the modest sum of $75 per month you have access to a variety of on-line courses. You are assigned a Coach, a real human being to answer your questions and help you make progress.
2. Practical Measurement System Analysis This is an in-house, one-day workshop to help you learn to evaluate your measurement systems to see if you can trust the data they provide.
3. Performing Objective Experiments This in-house, 3-day workshop will teach you the practical fundamentals of the most powerful experimental strategy for innovation -- Design of Experiments. If you attend this workshop and apply the skills you learn in your work, you will become a top performer.

Happy New Year! Good luck re-inventing yourself. May you find happiness.

Next time, let's look at Parity Plots and how to make them.

How Probability Can Help us Out in Tricky Situations

noreply@blogger.com (Objective Design of Experiments) — Fri, 23 Dec 2011 01:51:00 +0000

Have you ever been confident you are doing the right thing, but you keep getting poor results? How can you explain this?

Everything we do has some probability of succeeding and some probability of failing. For example, if you play Blackjack perfectly, you will win just slightly more than half the hands and lose just slightly less than half the hands. Yes, you will win in the long run, but you will lose a lot along the way.

If you are very careful and thorough in all of your experimental work you will often draw correct conclusions -- but not always. Everything that comes from data is uncertain, so you will be misled from time to time. How often will you be misled? If you use 95% confidence limits to draw your conclusions, you can expect to be right about 95% of the time. While this is a lot better than playing Blackjack, it isn't perfect. And you can't tell when you will be misled. Sometimes you can be misled several times in a row. (This is called "rotten luck!") However, if you consistently use 95% confidence limits to draw conclusions, you will be right far more often than you are wrong over the long run.

It is very easy to get frustrated when you draw a conclusion that turns out to be wrong, but don't give up. Make sure you know the right way to draw conclusions from your data, then stay the course. You will get through the rough spots -- and there won't be that many of them. If you give up and start "winging it" instead of using good strategies, you will be frustrated far more often.

Next time, let's look at a few ideas for New Year's resolutions.

Probability: The Good News, and the Bad News

noreply@blogger.com (Objective Design of Experiments) — Fri, 09 Dec 2011 22:10:00 +0000

The world is uncertain. Data are uncertain, and everything that comes from data is uncertain.

Last week's blog discussed how Statistics was invented to help us deal with this uncertainty by making decisions with a high probability of being correct. But probability is a tricky subject.

The mathematical definition of probability is very precise, and very abstract: "Probability is a measure of sets in an abstract space of events" (The Lady Tasting Tea, p.301). What does this mean in real life?

When the weather man says, "There is a 70% chance of rain," what does this mean? Does it mean that 70% of people will get wet? Does it mean that 70% of the time it will be raining? Does it mean that 70% of the area outside will receive rain? Clearly the answer to all of these questions is, "no." but what does it really mean?

The most practical explanation of the meaning is that, based on past data, 70% of days with conditions like we have today have received rain. The take away message is: it may rain, so bring your umbrella.

In Statistics, 95% probability generally means that if we repeat an experiment 100 times, on average, 95 of those times will turn out as we expect. So, in Statistics, a 95% probability of rain means that of the next 100 days with conditions like we see today, 95 of those days will receive rain.

So the good news is we can use probability to make better decisions.

Here's the bad news -- People really don't understand probability. Most people believe that if a fair coin is tossed 10 times in a row and comes up heads every time, then the next toss has a very high probability of being tails. It does not -- it is still 50%. The probability of an independent event is not affected by past history. Most people think that the probability of a 100 year flood occurring in their state is 1 in 100 each year immediately after a flood, then it increases with time from the last flood. It doesn't change year to year. And, the probability of a 100 year flood occurring is generally much higher -- there are many 100 year flood planes in a state and they could all possibly flood any year. (Cliff Mass discusses this for Washington State in his blog.)

Here's the bottom line: probability is your friend when used correctly. Just be careful to use it correctly. Consulting a Statistician is always a good idea.

Next time, let's look at how probability can help us out of tricky situations.

Why Statistics was Invented

noreply@blogger.com (Objective Design of Experiments) — Sun, 04 Dec 2011 18:32:00 +0000

Science has always faced a problem. While the scientific method requires that experiments must be independently confirmed, repeating experiments never leads to exactly the same result.

This will come as no surprise to regular readers of this Blog -- everything that comes from data is uncertain. Measurements always vary, no matter how carefully we try to repeat everything exactly. It's a fact of Nature.

Image: Master isolated images / FreeDigitalPhotos.net

Early scientists knew they could not repeat their own work exactly, much less the work of others. They thought this was because their measurement systems were too crude. They thought that if they could improve their ability to measure, they would be able to repeat experimental results exactly.

Unfortunately, as the ability to make more and more precise measurements developed, the ability to repeat results exactly became more elusive. Scientists realized that repeating results exactly was not going to happen. So how could they objectively conclude that that their experimental results were "close enough?"

Since it was clear they would not be able to conclude with certainty that results agreed, the concept of probability became an important ally. If they could conclude that the probability of their results agreeing was very high, they could be confident that, over the long haul, they were drawing mostly correct conclusions.

So Statistics was invented to help scientists draw conclusions based on data that were very likely to be correct. While this may seem a little unsatisfying, it is the best anyone is currently able to do.

Of course scientists aren't the only people faced with the difficulty of uncertainty in data. Engineers, marketing professionals, and, in fact, anyone who uses data, face the same problem. Statistics has evolved to help people in all fields requiring decisions based on data.

If you're interested in the history of Statistics, David Salsburg has written a very entertaining book, accessible to the layman, called "The Lady Tasting Tea." This book addresses the human side of Statistics as well, including some petty squabbles among its inventors.

Next time let's look at probability a little more closely, including its weaknesses.

How Much Work will My Designed Experiment Require?

noreply@blogger.com (Objective Design of Experiments) — Sat, 26 Nov 2011 00:04:00 +0000

Before you launch into a new designed experiment it's nice to know how much work is likely to be involved. You could write out the model, count the b coefficients, and add eight to eleven for extra trials and repeated runs.

There is an easier way! You can use the free "ObDOE Minimal Trials Calculator" to do the work for you.

Suppose you are interested in studying 6 factors with a full quadratic model. You can use the calculator as seen below to discover that you should count on 36 runs.

You can bookmark this link for future use. It can save you a lot of tedious work.

Next time let's look at why Statistics was invented -- (hint: it was for your benefit).

The Gage Performance Curve

noreply@blogger.com (Objective Design of Experiments) — Sat, 19 Nov 2011 01:30:00 +0000

If you know me -- or if you read this blog regularly -- then you know how strongly I believe data are necessary for good decisions. Of course this means "good" data are necessary -- but how can you know if your data are good enough?

If your decision is whether to accept or reject a part based on data, a wonderful tool is available to help you make this judgment: "Is my measurement system capable of determining whether a part is in or out of specification?" This tool is the "Gage Performance Curve."

A Gage Performance Curve plots measured values vs the probability of accepting a part as in specification. Here's an example:

The curve above indicates that any measurement between 0.9 and 1.2 has nearly a 100% chance of being correctly identified as in specification. Any part less than 0.75 or greater than 1.35 will have essentially no chance of being accepted as in specification.

Between 0.75 and 0.9 there is a "gray area." In this region there is some chance of making a mistake. In fact, at 0.8 there is about a 50% chance of making a mistake. A similar situation holds for measurements between 1.2 and 1.35.

If your process is well in control and your control limits are 0.95 to 1.15, your data are truly trustworthy. If you ever measure a part at 0.85, you will have some doubt as to whether the data are good enough here.

Want to make your own Gage Performance Curve? You can make one on the ObDOE Web Site (obdoe.com/student/MSAResources/GPCCurve.php). You will need: lower and upper specification limits, the noise in your measurement system as a standard deviation, and any accuracy correction required.

If you prefer to do this on your own computer, macros are available for JMP, STATISTICA, and Microsoft Excel at ObDOE.com (obdoe.com/student/MSAResources/GPC.html).

You may also enjoy reading this article by John Raffaldi and me.

Next time let's look at how you can quickly estimate the amount of work a designed experiment will require.

Lateral Thinking, Data, and Innovation

noreply@blogger.com (Objective Design of Experiments) — Fri, 11 Nov 2011 23:24:00 +0000

"With lateral thinking one is allowed to be wrong on the way even though one must be right in the end. ... One may have to move to an untenable position in order to be able to find a tenable position." Edward deBono, "Lateral Thinking," p. 107

Lateral thinking is more interested in creating new ideas and opening new ways of thinking than in being right. Of course we need to be right in the end or there is no point to our work. We only need to be right at the end.

Data tell us if we are right or wrong. We can make predictions from the ideas we create through lateral thinking and then collect data for comparison. If the data agree with the predictions, we are "right." If not, we are "wrong."

Here's a wonderful example from history:

Neils Bohr had the wonderfully lateral idea that the sub-microscopic universe acts like the macroscopic universe. He suggested that electrons orbit the nucleus in an atom like planets orbit the sun. This was a brilliant new way of looking at atoms that held great promise. Unfortunately, the predictions made using the Bohr model of the atom did not agree with data -- the idea was "wrong."

However, this idea prompted new ways of thinking about atoms (a major goal of lateral thinking is to promote new ways of thinking). Erwin Schrodinger used some lateral thinking of his own and modified the model by suggesting that electrons could only have specific, quantized energy levels. This was certainly a radical idea, but it also opened up new ways of thinking about atoms. As with Dr. Bohr, Dr. Schrodinger's model could not predict what was seen in the data. He also was "wrong."

Now P.A.M. Dirac looked at what Bohr and Schrodinger had done. He also looked at what Einstein had done. In a brilliant stroke of innovation he combined the ideas into his own model of the atom, embodied in the Dirac Equation. Dirac was certain this was correct because, "God used beautiful mathematics in creating the world." Of course other scientists wanted to see if Dirac's predictions would agree with data. This turned out to be very difficult, though, because the math was so complicated that no one could really make the necessary predictions.

Then comes Richard Feynman. Feynman figured out how to solve the Dirac equation to make predictions. So far, these predictions agree with data -- Dirac was "right," as far as we can tell today.

So, while Bohr and Schrodinger were both wrong, their ideas opened the way to the "right" answer.

In your work, there is a strong temptation to avoid being "wrong." Resist that temptation. Sometimes being "wrong" is the most productive thing you can do!

Next time let's look at the Gage Performance Curve and what it can tell you.

Lateral Thinking

noreply@blogger.com (Objective Design of Experiments) — Fri, 04 Nov 2011 00:32:00 +0000

Innovation requires creativity. Creativity comes from lateral thinking.

Edward de Bono defined two important types of thinking -- vertical thinking and lateral thinking.

Vertical thinking is logical, linear thinking in which one step follows from another. If any step is wrong, the result will generally be wrong. It is the type of thinking all scientists and engineers learn in school.

Lateral thinking is humorous, non-linear thinking. No step in the process needs to be correct as long as the final result is correct. Creative people are lateral thinkers.

Both types of thinking are essential. Lateral thinking helps us create new ideas and vertical thinking helps us to establish whether they are correct.

My Friend Rob Takemura shared a great story about lateral thinking with me. He and several colleagues were brainstorming ideas for a slogan. They had been at the task for quite awhile and were all tired and slap-happy. Little progress had been made. One member of the group said, "Water- the other white meat." This was so silly, so out of touch with what they were working on that they all laughed. But it broke the pattern they were thinking in, allowing them to find a couple of really good slogans.

Here's an analogy: You drive to work by a certain route every day. One day you are forced to take a detour due to road construction. You never even noticed the small country road of the detour in your normal commute -- it had been invisible to you because you knew the route you wanted to take. Now you are forced to take this route, and you find it saves you 10 minutes on your morning commute. From now on you take this new route to work on a routine basis.

Our brains follow patterns. Once a pattern is established, our brains tend to keep following that pattern. It can become very difficult to see anything that doesn't fit the pattern. Lateral thinking breaks these patterns, using humor or other disruptive techniques.

Yes, creative thinking can be learned. An excellent book to get you started is The Leader's Guide to Lateral Thinking Skills: Unlocking the Creativity and Innovation in You and Your Team by Paul Sloane.

Hope you enjoy the book.

Next time let'a look at how lateral thinking and data work together to yield innovation.

Dealing with Rare Events

noreply@blogger.com (Objective Design of Experiments) — Tue, 01 Nov 2011 18:42:00 +0000

Donald Wheeler has an article in Quality Digest today that may be very useful to you. I will continue the promised blog flow in a couple of days.

His article is about dealing with rare events.

While this article is focused on control charting, the techniques are perfect for DOE as well.

Enjoy!

Theory versus Practice

noreply@blogger.com (Objective Design of Experiments) — Fri, 28 Oct 2011 17:51:00 +0000

Theory is great. Theory helps us make valuable predictions about Nature that saves us countless hours and dramatically improve our lives. It helped take us to the moon. Long live theory!

Unfortunately, theory is limited. No theory can take everything into account. All theories rely on simplifying assumptions.

"Essentially, all models are wrong, but some are useful. However, the approximate nature of the model must always be borne in mind…" G.E.P. Box

Nature may (or may not) be simple in principle, but it is extremely complex in its operation. Using classical physics, if we knew the exact position and velocity of every particle in existence we could calculate everything that would ever happen. However, even the most incredibly small uncertainty in the position of even one particle would make this type of prediction impossible.

"If the accuracy is taken to be one part in billions and billions and billons -- no matter how many billions we wish, provided we do stop somewhere -- then we can find a time less than the time it took to state the accuracy -- after which we can no longer predict what is going to happen!" Richard Feynman, Lectures on Physics, Vol III, p. 2-10

It is impossible to take everything into account when applying a theory.

Practice, on the other hand, has no choice but to take everything into account. We cannot stop Nature from working in order to understand it. We cannot force Nature to make simplifying assumptions. This quote sums up the situation beautifuly:

"The only difference between theory and practice is that practice takes into account all of the theory." Barney Oliver

An excellent example of theory vs. practice is in a recent news story about neutrinos traveling faster than the speed of light.

According to Einstein's Theory of Relativity it is impossible for anything to travel faster than the speed of light.

However, data collected by physicists at CERN seemed to indicate that neutrinos were travelling faster than light. The distance covered in the time measured was too long.

Was Einstein wrong? (He could have been.) It doesn't look like it. As a matter of fact, the missing theory is his! The Theory of Relativity explains the misinterpretation of the data. Because satellites moving very fast compared to the Earth were used to measure time and position, the data have to be interpreted relativistically.

"In other words, the GPS clock is bang on the nose, but since the clock is in a different reference frame, you have to compensate for relativity if you're going to use it to make highly accurate measurements."" Evan Ackerman

The data took into account all of the theory. They always do.

Use theory to save time -- but rely on the data to tell you the whole story.

Next time let's look at the value of Lateral Thinking to innovators.

The Prisoner's Dilemma and Data Driven Innovation

noreply@blogger.com (Objective Design of Experiments) — Sat, 22 Oct 2011 02:24:00 +0000

The "Prisoner's Dilemma" is puzzle that nearly everyone gets wrong. Not only do they get it wrong, but they can get really upset and refuse to believe they got it wrong.

So what is the Prisoner's Dilemma? I'll tell you -- but be forewarned -- you'll probably get it wrong, you won't believe you're wrong, and you might even get upset. If you think you can take it, here it is:

Three prisoners, Joe, Fred, and Frank, are in prison (a good place for them ;-) ). Word is out that one of them will be paroled, but none of them knows which one. Joe asks the jailer, "Which one of us is going to be paroled?" The jailer says, "I can't tell you." Joe says, "Come on -- I won't tell anyone you told me." The jailer says, "Well, I guess I can tell you someone who won't be paroled." He flips a coin, then he says, "Fred will not be paroled."

Then the jailer says, "The parole was not decided by person -- it was a random drawing of the cell. Would you like to change cells with Frank?" Joe readily agrees to change cells and sleeps much better. Does he really have any reason to be happy about this?

Don't say your answer out loud. It will probably be wrong.

By changing cells, Joe's chance of being paroled increased from 1 in 3 to 2 in 3.

Odds are good you are thinking at this point, "His odds are 50:50. It doesn't matter which cell he's in." If you are thinking this, I'm afraid you are wrong. You may be getting upset now -- sorry.

So what's the point of all this? The only way most people become convinced that his chances actually improve by switching cells is to collect data. Get a friend to help you. Your friend will pick a cell at random for parole. He will then flip a coin and tell you a cell that will not be paroled. If there is only one cell, the choice is obvious. If two cells are available for him, he will use the coin toss to tell you one of them at random. Since he flips the coin every time, you won't be tipped off. Watch how often you would be paroled if you stayed in your cell vs. changing cells. The data will convince you.

This is a problem in conditional probability. Initially your odds are 1:3 that you will be paroled and 2:3 that someone else will be paroled. After you receive more information, these odds don't change. You do know, however, which of the other two cells has the 2:3 odds of being paroled and can take advantage of this. Unfortunately, this explanation is counter-intuitive to most of us -- I got it wrong initially and had to collect data to become convinced.

Barney Oliver, the head of HP Labs in its heyday, said, "The only difference between theory and practice is that practice takes into account all of the theory." Data include all of the information available. They don't miss anything due to lack of knowledge, incorrect assumptions, etc. Data tell you the whole story -- something no theory can. Data are the best source of information for innovators.

Next time let's explore further the information content of data.

How Losing can Help You Win

noreply@blogger.com (Objective Design of Experiments) — Thu, 13 Oct 2011 23:17:00 +0000

I recently had the good fortune to hear Jeffrey Ma speak. Jeffrey is the man that won millions playing Blackjack with his MIT team -- he was the inspiration for the movie, "21."

His talk focused on what Blackjack can teach us about business. It also turns out it can teach us a lot about innovating.

Blackjack is a game that is absolutely mathematical. Played right, you will beat the house if you play long enough. (If you want to learn more about this, Jeffrey's book, "The House Advantage" is very good.)

One of the important lessons we can learn is this: even when you play Blackjack perfectly, you will lose many hands. The key is in winning more hands than you lose.

Innovators aren't always successful. Some innovations are flops. Some attempts at innovating don't work. This is only "losing a hand." Don't let it discourage you. If you are using good innovation skills (DOE being a key skill) you will win more hands than you lose.

Unlike Blackjack, innovators can learn from their losses. When an innovation "flops," you can analyze why and correct for the causes in the future. Whenever you can learn from a loss, you have the opportunity to increase your odds of winning in the future.

Next time let's look at what a Prisoner's Dilemma can tell us about the value of data when making decisions.

How Data Driven Innovation Helps You Win More Often

noreply@blogger.com (Objective Design of Experiments) — Fri, 07 Oct 2011 22:08:00 +0000

You like to win. Everyone does. Some people absolutely have to win. Everybody prefers to win.

What does it mean to win as an innovator? It means you innovate more effectively and more frequently than your competition.

You have a choice: you can use data or not use data when you innovate. Which will help you to win more often?

Here's an example of data-less innovation: you dream the solution to a big problem at work. Because you are a long term employee held in high regard, people will simply believe you when you tell them you know how to solve the problem. If your dreamed solution works, great! You would be an extraordinary individual, though, if your dreamed solution really worked. There are very few examples of dreamed solutions working in real life. To count on this method for your entire career would be foolhardy.

Here's an example of data driven innovation: You have performed several experiments and have noticed an apparent trend in your data. You investigate the trend and find that it is real. You use this information to solve the problem you are working on. This method has a long history of success. It isn't perfect -- it does require a certain amount of luck to find the trend, run the right experiments, etc. But your chances of winning are much higher.

Here's an example of better data driven innovation: You carefully plan your work to solve your problem as quickly and thoroughly as possible. You use Designed Experiments to maximize the information gathered for the work expended and Response Surface Methodology to analyze the data collected. You predict your Sweet Spot and develop it to solve your problem. Your need for luck has been significantly reduced. You will win even more often using this approach.

What's the moral to this story? We only know what our data tell us. Making decisions without data is pure gambling.

Next time, let's look at how losing can help you to win.

Understanding vs. Predicting

noreply@blogger.com (Objective Design of Experiments) — Fri, 30 Sep 2011 16:57:00 +0000

Designed experiments have been used for two primary purposes since the 1950's: understanding factor effects and making predictions.

Understanding how much of an effect a factor has on a response can help product and process developers understand how their innovations work. This is especially true when the effects of interactions between factors are understood. Interactions often are the key to interpreting otherwise confusing results.

D-Optimal designs offer a great way to create designs customized to the experimental conditions that will sort out the effects of factors and their interactions.

Predicting the results of future experiments is also extremely useful to product and process developers. This makes it possible to predict Sweet Spots, the combinations of factors that will best satisfy your customers. The ability to predict can save enormous amounts of time and money (literally millions of dollars based on the experience of our customers).

I-Optimal designs offer you the ability to create customized experiment designs that predict well.

Unfortunately, a design that predicts well may not sort out the effects of factors and their interactions. A design that does sort effects well, may not predict well. Life is not perfect!

If you have the money, you can create designs that have both I and D efficiencies that are high -- they predict and sort effects well. These designs tend to need a lot of runs.

Most of the time you will likely be budget limited. Your best bet is to decide which function is more important to your immediate needs -- the ability to predict or the ability to sort out the effects -- and go with an appropriate design.

Next time let's look at why data driven product and process development helps you win more often.

The Differences Between Theoretical and Empirical Models

noreply@blogger.com (Objective Design of Experiments) — Wed, 21 Sep 2011 21:48:00 +0000

Misunderstandings are insidious. This entry will address a common misunderstanding about theoretical and empirical models.

Theoretical models, like Newton's Laws or Maxwell's equations, are extremely useful. They cover a wide range of situations and predict extremely well over a very wide range of factor levels. When you have a theoretical model that works well, you are in an ideal position.

Unfortunately, every theoretical model is based on certain assumptions. If these assumptions are not met, the model may fail to predict correctly. It may not even be close.

Even worse, it is not uncommon to have no theoretical model that covers the work you are doing. With no model, you can make no predictions.

When you can't use a theoretical model, you can create your own empirical model. The best way to do this is using Design of Experiments with Response Surface Methodology.

So here's the misunderstanding: empirical models are not inherently better than theoretical models, or vice versa. People sometimes make their choice and claim they are superior because of their approach. Models are only useful if they help you solve problems. The best model, whether it is theoretical or empirical, is the model that predicts best for your situation.

As an example, Newton's Laws are among the best theoretical models ever devised. They can be difficult or impossible to use in some situations. If you are unable to model the frictional forces, the air drag, etc, you won't be able to use them. You will, though, be able to use an empirical model.

Theoretical models are very expensive in both time and data. Empirical models are relatively cheap in both time and data. If you can afford it, the best approach is to work on developing theoretical models for the long term while using empirical models to make products to sell along the way.

Next time lets look at the difference between predicting well and understanding the effects of your factors and their interactions.

Simplifying Innovation Tasks

noreply@blogger.com (Objective Design of Experiments) — Fri, 29 Jul 2011 17:56:00 +0000

When you examine an innovation project in its entirety it is amazing that anyone ever innovates. The complexity is staggering.

Thomas M. Sterner, in his book, The Practicing Mind: Bringing Discipline and Focus Into Your Life provides an excellent key to tackling complex projects in his four S's:

∘ Simplify -- break it down into steps.
∘ Small -- smaller steps are best.
∘ Short -- keep the time for each step short.
∘ Slow -- work at a pace that allows you to pay attention to what you are doing.

These S's are the the reason the ObDOE 7 Step Method is so effective.

∘ Simplify: The task of using Design of Experiments and analyzing using Response Surface Methodology is broken down into steps. Focusing on a step simplifies the extent of the project, while progressing through the steps insures that everything gets done.

∘ Small: Some steps are further broken down into smaller steps, such as Step 1: Check the Question. A checklist and a series of questions is used to simplify the task of identifying the real question to answer.

∘ Short: You should only work on a step for a period of time that is comfortable for you. For example, if you are holding meetings to check the question, keep them short -- 30 to 45 minutes -- so you and everyone else can remain focused on the task at hand.

∘ Slow: This is essential -- work at a pace that allows you to pay full attention to what you are doing. This is especially important to remember when collecting data. Haste definitely makes waste.

Next time you need to tackle a complex project, remember these four S's -- your life will be easier.

When we meet again let's talk about the difference between theoretical and empirical models.

Stumbling on Innovation

noreply@blogger.com (Objective Design of Experiments) — Tue, 26 Jul 2011 18:36:00 +0000

Dan Gilbert, a professor of psychology at Harvard, wrote a fascinating book, "Stumbling on Happiness"

Here are several points from the book that have a bearing on innovation:

Prospection is the act of looking forward in time or considering the future. Prospection is used to predict possible futures so we can attempt to achieve the best future and avoid the worst future. Prospection is remarkably unreliable.
Realism is the belief that things are in reality as they appear to be in the mind. People tend to consider and remember information about what did happen. They do not consider or remember what did not happen. (They remember how many times they were correct at guessing who was calling, but not how many times they were incorrect.) It is difficult for us to consider what we may not be considering.
Rationalization is the act of causing something to be or seem reasonable. We seek facts to confirm our beliefs and ignore facts that contradict them. We fool ourselves into believing that what we want to believe is true.

Each of these is real and very important to our survival, as discussed in the book. Unfortunately, each of these has disadvantages as well, some of them affecting our ability to innovate.

Prospection can cause us to believe we know the solution to a problem when in fact we do not.
Realism can cause us to fail to consider important aspects of a problem, specifically things that do not happen. One way this shows up is in our tendency to count only successful experimental trials and ignore failed experimental trials. Thus if 5 out of 105 experiments provide successful results, we tend to think we solved the problem in 5 trails, when really we needed 105.
Rationalization can cause us to cling to our belief of what nature is doing, rather than accepting data that contradict our belief.

Fortunately there are good ways to work around these disadvantages, although they do require effort.

Here are the ways:

Always trust data. If you can find something that was actually done wrong in an experiment, then of course the data are not to be trusted. Otherwise, data are much more reliable than beliefs.
Use every trick you can to collect and evaluate your data objectively, including double-blind studies, Statistical analysis, and peer review.
Remember these words of Richard Feynman: "Everything could possibly be wrong." No matter how clear and obvious your belief seems to you, it may be wrong.

Data driven innovation is the most reliable way to make progress.

Next time, let's look at some advice for tackling complicated tasks.

How You Can Excel at Innovation

noreply@blogger.com (Objective Design of Experiments) — Wed, 13 Jul 2011 19:00:00 +0000

Innovation is defined differently by different people, so let me provide my definition to avoid misunderstanding:

Innovation starts with existing capital. Capital can be a technology, a product, a service, etc.
a. Innovation improves (optimizes) existing capital,
b. Innovation finds new uses for existing capital,
c. Or innovation finds new markets for existing capital.

Samson Rope is a highly innovative company in Ferndale, WA. It has been in business for over 130 years. While its innovations could fill a book, here are three that illustrate the three modes of innovation:

Improving existing capital. No matter how good a rope is, it can break when over-taxed. US Patent 7,127,878 describes improving existing rope technology to make it safer, providing both a mechanism for early warning of rope failure and reducing the ropes potential for causing damage on breaking.
New use for existing capital. Double braided rope was invented by Samson Rope in the 1950's. It dramatically increased the strength of synthetic rope. US Patent 7,134,267 discusses the use of double braided rope technology to control the coefficient of friction for the rope. The double braided rope technology capital can, thus, be used to control rope friction rather than strength.
New market for existing capital. Metal cable, or "fire wire," is typically used as rope when fire resistance is critical. Unfortunately, metal cables are difficult to work with because they are relatively heavy and inflexible. US Patent 7,168,231 describes modifying existing rope technology to create a fire resistant synthetic rope that is lighter and more flexible than "fire wire," introducing synthetic rope to the "fire wire" market.

You can take three steps to help you excel at innovation:

Understand the three modes of innovation listed above and focus on accomplishing one or more of them in a project. Having a target will guide you to a successful conclusion
Improve your creative thinking skills. A fantastic tool for creativity is "Lateral'Thinking." Lateral Thinking, discovered by Edward DeBono, is a thinking technique not known by most engineers and scientists. It focuses on generating radical new ideas. Used in conjuction with "Vertical or Logical Thinking," it is a major tool for innovation. A good place to start learning about Lateral Thinking is the book, The Leader's Guide to Lateral Thinking Skills: Unlocking the Creativity and Innovation in You and Your Team
Learn the tools of Effective Innovation®, including Design of Experiments and Gage R&R. These tools substantially reduce the need for luck in your innovation.

Next time let's look at how our brains work and how this effects experimentation.

Common Statistical Mistakes (and How to Avoid Them!)

noreply@blogger.com (Objective Design of Experiments) — Thu, 09 Jun 2011 23:34:00 +0000

Making mistakes is one of the most common, most frustrating, ways that we learn.

You can take advantage of the mistakes of others, instead of making these mistakes yourself:

1. Failing to repeat measurements.

2. Making measurements with untested tools.

3. Not planning.

4. Relying on historical data.

5. Taking your best guess.

6. Misuse of three times the Standard Deviation.

7. Misinterpreting confidence limits.

8. Ignoring confidence limits on Cpk.

9. Drawing wrong conclusions from a t-Test

You can learn to avoid these mistakes for FREE! That's right, for free.

You can take the eCourse, "Common Statistical Mistakes and How to Avoid Them" online, any time, at your own pace.

you never need to get caught making these common mistakes.

Enjoy!

Next time, let's talk about what innovation is and how you can excel at it.

Three Types of Confidence Limits and How They Are Useful

noreply@blogger.com (Objective Design of Experiments) — Thu, 02 Jun 2011 23:33:00 +0000

Confidence limits help you judge the quality of your conclusions. They help you estimate the uncertainty in your conclusions and the probability that your estimate is correct.

Here's a simple example: suppose you have collected data and calculated 95% confidence limits on the average of your data. Suppose also that your average is 5, your lower limit is 4, and your upper limit is 6. This would mean that you think the average is 5, but because of the uncertainty in your data, you think it could be as low as 4 or as high as 6. You are 95% confident that this is a correct conclusion, but you have a 5% chance that this is not the correct range.

Confidence limits can be calculated for a wide range of properties, but three of these are particularly useful in industry: confidence limits on the average, confidence limits on the next measurement made, and confidence limits on 99% of everything we will ever make. Let's look at each of these.

Confidence Limits on the Average (commonly called "Confidence Limits"): These limits help you to determine what the average, or typical value of a property is. The range from upper to lower limit tells you how uncertain you are about your conclusion. Collecting more data will tighten the range, reducing the uncertainty. These limits help you to know if you are meeting a target on average, even though noise may be making it difficult to judge this for individual measurements.

Confidence Limits for the Next Measurement (commonly called "prediction limits"): These limits help you to test a model. Your model will predict the average behavior for a property, but any individual measurement is very likely to deviate from the average. If your measured value falls between your prediction limits, you can attribute the difference to noise. If not, your model needs improvement.

Confidence Limits on 99% of Everything We Will Ever Make (commonly called "Tolerance Limits"): These limits estimate the range of values for a property that your customer will see. At least 99% of everything you ever make in production should fall between these limits.

You can learn more about these limits and many other basic Statistical concepts in an online eCourse. You can learn more about this at obdoe.com/ecourses.html

Next time, let's talk about some very common Statistical mistakes and how to avoid them.

ANOVA

noreply@blogger.com (Objective Design of Experiments) — Tue, 10 May 2011 21:47:00 +0000

ANOVA, or ANalysis Of VAriance, is a method for determining if any of several means is (are) different from the others. This may sound a little odd, but you can actually use the variance to look at the means. If the variance between the means is larger than the variance for the means, there is likely to be a real difference in the means. This assumes, of course, that the variance is the same for each mean.

Most software packages will perform an ANOVA for you.

The downside to ANOVA is this: it tells you if there is likely to be a difference, but it gives no indication of how to tell which mean(s) is (are) different.

Next time let's talk about different types confidence limits and how they are useful.

The Role of A Measurement System

noreply@blogger.com (Objective Design of Experiments) — Fri, 15 Apr 2011 21:32:00 +0000

A measurement system is your connection to nature. It provides you with quantitative data for your observations. It is an extension of your five senses.

If you have good eyes, you have less uncertainty about what you see than does someone who is near-sighted. If your hearing is good, you are less uncertain about sounds than someone who is hearing impaired. The better your senses, the less uncertain you are about your observations, but you can never be 100% certain. We all know of optical illusions and sounds that can deceive us (like that unbalanced dryer upstairs that sounded like a helicopter passing over the house!).

Just like any of your five senses, a measurement system provides uncertain information about the world. A good measurement system provides less uncertain data than does a poor measurement system.

Just like your five senses, you can test how good your measurement system is. Just like your senses, you can correct for inadequacies. If your vision is poor, you can wear glasses. If your measurement system is poor, you can improve it.

The test for the quality of a measurement system is called a "Measurement System Analysis," or "MSA." This test is like a "checkup" at the doctor -- it tests several aspects of your measurement system's health. A key portion of this test is called "Gage Repeatability and Reproducibility," or "GR&R." This test will help you judge the quality of your measurement system and help you identify which aspect of an impaired system is most in need of correction.

In my experience measurement systems are almost never as good as people assume they are. They are, in fact, generally much worse than people think they are. When a measurement system is tested, you discover that it "needs glasses."

Unfortunately, many people take the approach that they don't want to test the measurement system because it may be bad. This is like saying, "I don't want to go to the Dr. because I might have cancer." The longer you wait, the more poor quality data you will be using to make critical decisions.

You can learn the basics of MSA in one day.

Learn about our one day workshop.

When you take control of your measurement systems, you take control of your data.

Next time, let's talk about ANOVA.

Good experimenting!

Bill Kappele.

Two Ways to Create Custom Designs

noreply@blogger.com (Objective Design of Experiments) — Fri, 08 Apr 2011 00:06:00 +0000

Custom designs make your work more realistic. They free you from unrealistic constraints in existing designs, and allow you to constrain your experiments in ways necessary to your work.

For over 20 years now Gosset has been available to you for creating custom designs. Gosset runs on Unix or Linux systems. It can now be run on Windows systems using a Linux emulator, Cygwin. Gosset is free, even for commercial use. Gosset will only create designs -- it cannot analyze them. However, you can analyze Gosset designs with most commercial Desgn of Experiments software. You can learn more about Gosset at obdoe.com/consulting/doit.html.

JMP now provides a custom design generator. JMP produces experiment designs of as high a quality as Gosset, but it is more convenient. JMP does analyze your data. The model is automatically attached to your design, making analysis much easier. You also don't need to learn any special commands, as you do with Gosset. JMP is not free, but you will find it well worth the money. You can learn more about JMP at www.jmp.com/ .

Next time, let's consider the role of the measurement system in your experimentation.