Dr. Ron Lasky
Dr. Ron Lasky is a professor of engineering and director of the Cook Engineering Design Center at Dartmouth College, where he teaches students and professionals about data-driven quality improvement. In addition, Lasky is a senior technologist at Indium Corporation, and writes regularly at the
Indium.com blog. During a 30-year career that has included stints at IBM, Universal Instruments, and Cookson Electronics, he also authored six books and numerous technical papers. We spoke with Lasky about his efforts to help people use data to improve processes, as well as his experiences using Minitab Statistical Software.
When did you first encounter Minitab?
About 15 years ago I was an engineer at a company that wanted to make a Lean-Six-Sigma-type certification program. Our statistician used Minitab, so I started using it, too. I continued with it when I started working with the Indium Corporation, and I’ve kept using Minitab almost every day. I find it extremely user-friendly.
As time goes on, I have used more and more of the features in Minitab. Indium had a technical paper on Weibull analysis, which they wanted me to rewrite. This involved redoing all of the analyses and graphs, which had originally been done in a specialized piece of software. So I learned how to do Weibull analysis using Minitab, and I started incorporating those methods in my courses. Recently I was asked to help an organization understand sampling better. To prepare for that, I went on the Internet and found out that there is software specifically for sampling—but Minitab has sampling tools, too.
How did you start teaching Lean Six Sigma at Dartmouth?
I began teaching a course at Dartmouth on optimizing manufacturing: using designed experiments to develop the processes and find the right parameters, and then applying statistical process control to manage the processes. Then Dartmouth asked me to teach statistics, too, and today I teach four courses, two of which have high statistics content. One is a required statistics course for seniors and graduate students, and the other is the manufacturing processes course. In the beginning of these courses, students do calculations by hand so they learn the fundamentals, but at the end of the course we use Minitab so students also get experience with a tool they can use in the workplace.
Around 2005, I decided to develop a Lean Six Sigma program for the engineering school. I intended it to be for industry, not for Dartmouth students. But I realized engineering students already learn most of the math and statistics required, so why not give them the option to earn a Green Belt? We started offering that to them in 2006. The engineering department is right next door to Dartmouth’s Tuck Business School, so when they heard about it, they asked if their students could take the Green Belt, too. Then students studying hospital administration asked to take it. So it really expanded, and we added a Black Belt option in 2007.
We also offer Lean Six Sigma training to industry. The typical student in industry is 40 to 50 years old, and they probably have a degree, but in business or something other than engineering. Probably more than half do not have a technical degree, and many of them struggle with the math. One of the things I’m most proud of is that in our program, we’re willing to work with people one-on-one to overcome the challenges of the math. We’ve helped some people who were really shaky in math to at least get their Green Belt.
Our program is not simply preparing you for a multiple-choice certification exam. When your manufacturing line is down, God does not send a tablet on high to you and say, “Your answer is either A, B, C, or D.” You’ve got to solve the complete problem. So we essentially give the students a stated problem with data. They have to solve it, and they use Minitab to do it.
What do you do when you're not teaching?
About one day a week I consult with the Indium Corporation, where I also write a blog about statistical issues in electronics manufacturing. Most of my blog posts feature a character named Patty, who is a professor at “Ivy University” with her mentor, the Professor. We just published a second edition of The Adventures of Patty and the Professor, a book that collects nearly 50 of my favorite posts.
Why did you choose to use characters and stories in your blog?
I think it’s just more engaging. I’ve been blogging for 10 years, and it finally occurred to me if I created some characters I could add a little humor. Most Patty and the Professor stories include a little side story that’s kind of fun.
I use these characters to talk about things I see happening in the industry. For example, I recently read a very good article about the importance of statistical process control, and what capability and being in control mean…but then they showed a chart that had Six Sigma defined as a Cp of 2.0 with 3.4 defects per million. Well, mathematically, that’s not really Six Sigma—it’s 4.5 sigma. What’s more, Cp doesn’t tell us anything about the defect level; for that, we need to know Cpk. So a true 6-sigma process, which would require a Cpk of 2.0, is about 2 defects per billion. A lot of people don’t know this, but Six Sigma as designed by Motorola should really have quotes around it, because it has a Cp of 2.0, but a Cpk of 1.5, which is how they reach 3.4 defects per million. Basically, their definition of Six Sigma includes a 1.5-sigma shift of the mean, and that is still creating confusion today.
So I wrote
a blog post in which Patty’s former boss at the company she used to work for calls to complain that they ordered some parts with a Cpk of 1.0, but the parts arrived with 5% defects. Now, a Cpk of 1.0 would be 3-Sigma, which should be 99.7% good parts—defects should be about .3%. The boss doesn’t understand how the supplier can claim they have a Cpk of 1.0. So this blog explains Cp and Cpk, and the origin of this 1.5-sigma shift which began with Motorola.
In a recent post, Lasky used humor and a Minitab graph to illustrate the impact of a 1.5-sigma shift on the true number of defects a process will produce.
Have you seen similar misunderstandings as a consultant?
I’ve seen cases where companies were averaging Cpk’s, which you can’t do. One company had an overall Cpk of 1.5. They wanted it to look better, so they split the data up in such a way that they got a Cpk of 3 and a Cpk of 1, and they averaged them to get a Cpk of 2.0. The bosses asked me, “Is it fair to do that?” But you can’t split data up and average Cpk’s, because they’re non-linear. So I took the datasets and used Minitab to show them why it’s not fair, and why you have to analyze all the data together.
Quality statistics can seem daunting to beginners. How would you recommend someone get started?
I tell my students the most useful feature in Minitab is under Basic Statistics—“Graphical Summary.” You put in a string of data and press a button and you get just about everything that’s important. You get a histogram, you get an Anderson-Darling calculation on normality, and you get confidence intervals on the mean and the median. For most engineers in manufacturing, or quality process people at a hospital, that graphical summary alone can be very, very helpful.
The great thing about Minitab is that tools like the graphical summary can help people who are just getting started. When I teach people who have never even heard of Minitab but have used Excel, they feel like they’re in a friendly place. But you could work with Minitab all of your life, and on your deathbed there would still be some features you haven’t used yet. It’s very powerful, and that makes it all the better for somebody who wants a lifelong product that they can use.
Sometimes I’ll get a consulting request, and they’ll say, “We want you to come and teach us, but we want you to use a different software package.” I say, “I can do it, but can I convince you to use Minitab instead?”
Minitab’s graphical summary makes it easy to gain a great deal of useful information.