A Six Sigma Master Black Belt and long-time JMP user, Tom hosts the Central Ohio JMP Users Group on site at INEOS Composites. He holds a Master’s in Statistics from the University of Wisconsin at Madison and Bachelor’s in Electrical Engineering from Penn State University.
Tom spoke with JMP User Reference Manager Meg Eberle.
Meg: What do you think are some of the biggest challenges you’ve faced in promoting statistics within the chemical industry?
Tom: Resistance to change is a big one! A lot of people have been doing things in the same way for a long time, so you have to chip away at old habits in bits and pieces. Chemists may have a tendency to say, "We put all of our lab data into Excel – that's our electronic notebook of sorts, so we might as well just make a graph here."
To evolve away from that mindset, you have to help people to start thinking about statistics differently. Go from "I know I want to make a bar chart," to "I ran this study, now what are the questions I want to answer? I can just drag and drop in Graph Builder and answer my questions.” It becomes more of an exploratory learning experience than just, "I need to make a bar graph for my manager."
The next challenge is that sometimes management isn't as bought-in as you are on the idea of statistical approaches, so you have to find ways to demonstrate the value through case studies.
Lastly, I’d say the most common challenge today is simply a lack of resources. People have very little time to do anything extra, so when you come to them and say, "Hey, I want an hour of your time to talk about statistics," they may say “I don’t have time.”
Meg: What do you see as the role of a statistician in driving a culture change?
Tom: I see my role as acting as mentor, coach, trainer, teacher – all of that at once. I’m always looking at the organization’s growth potential.
When I first started at INEOS Composites, if I used the term DOE, no one knew what it meant. But today, it's our common language. Everybody knows what DOE is! Everybody knows how to run a DOE.
Like any organization, INEOS Composites has a huge range in statistical skillset. As a statistician, I both analyze data for people and teach them to do the analysis themselves. I generally get the problems that no one else knows how to handle; when a particular data challenge becomes very complicated, [scientists and engineers] will often give it to me to analyze.
That said, the people in our organization who have JMP generally try to do as much as they can on their own before coming to me. And then if they do need help, they usually approach another JMP user who sits near them first.
Getting people to become more independent in doing things like DOE on their own and getting them hooked into all things JMP discovery – where they can see the things that I see – is a major goal.
You need to find people who have the potential to be change agents and help them along. As they learn, they inevitably bring others along with them, and if [those people with change agent potential] are more senior in the organization, even better! You can really drive change at scale by helping them.
Meg: Given that we're having this conversation during Discovery Summit – and that cross-industry learning is in many ways its raison d'être of this conference – I’m wondering: what lessons have you brought back to INEOS Composites from previous summits?
Tom: From when I first started coming to Discovery Summit through now getting more involved on the Steering Committee, I have so enjoyed getting to learn from different people – and now being able to actually help facilitate [this exchange] myself.
Every year, there’s at least one or two things I see where I think, "I'm going to try that." For example, when Text Explorer first came out, I kept thinking how I could use it. Ultimately, I ended up using it to look at our safety near-hits data. By identifying the key themes that keep showing up, we can proactively address them. INEOS Composites is very big on safety. We’re one of the leaders in safety in our industry, and I’ve used Text Explorer to continue to [deliver on that mission].
One year at Discovery Summit, I saw a presentation on structural equation modeling (SEM) by Laura Casto-Schilo and spoke with her a couple times after that. I thought, “There's got to be a way we can use this” and I did in fact find a case study. That was just the beginning – now I’m working to implement SEM in our process data.
Meg: What advice do you typically share with new JMP users?
Tom: First, watch the 1-hour webinar Getting Started With JMP. Find a problem, learn how to import data and make some graphs. Don’t wait to get started.
Second, find a mentor who is one or two steps ahead of you [in learning statistics and/or JMP]. That's all you need – somebody who's been there in the last six months who can help move you along. Because you only need to know the next thing you need to know, not what you’ll need way down the line. Find somebody who's just a little bit more advanced than you and create a relationship with them.
Third, if you want to know how to do something in JMP, there are plenty of on-demand videos! Just go to the JMP website. Find the one you want. Watch it. They go through step-by-step, a lot of times with a sample data set that you can use to learn.
I’d also recommend attending the Mastering JMP series on Fridays. I send reminders out to all our users and try to encourage them to watch. It's free learning. If you don't do it, that's on you.
[Before the pandemic disrupted in-person gatherings,] I used to schedule a conference room and invite people to come in and watch Mastering together. Then we’d have a discussion: what did you see that you knew or didn't know? And how could we apply it to solve one of our problems? We’ve just started that in-person session up again.
Meg: Speaking of continuing education, what would you say to people in chemistry graduate programs when it comes to preparing for a career in industry?
Tom: Find a way to get some background in statistics! One of the things that I loved at the University of Wisconsin [Madison] was that in a class on nonlinear modeling, half were graduate students in statistics and half were PhD students from other disciplines – science, engineering, the business school. We got paired up – they brought in a research problem, and we worked together to model and analyze it.
Maybe your school doesn't have that kind of course yet, but you can still find ways to get connected into a community. Get some sort of statistical background analyzing data. It's important, especially if you're going to be in research of any kind. You need to have a basic understanding of statistics.
Part of the reason I chose to go to Wisconsin was because their master's program was designed to equip graduates with the capability to be their own consultant at the end. The final exam was that we had to work with two different clients who brought in their real-world problems. In one week, we had to do the analyses, write a report for each client and then defend our analyses before the exam committee. You worked day and night on those two problems, but it was a great experience.
And that's what I wanted. I came from an industrial background. I didn't want to come out being in academia. I wanted to be able to do this for a living in industry.
Meg: Before I let you go, what’s one thing you're doing with JMP right now that you're particularly excited about?
Tom: We have a data historian that collects all our process data, and it’s taken me a while to learn how to get JMP connected to it. We're sitting on a gold mine of information!
Now I'm extracting both continuous and batch data so people can ask questions like, "Why is the plant having an issue with this?" And I can give them an answer in just minutes. I've been able to help improve the quality of our products through the analysis of our process data.
Being able to have access to such valuable information directly in JMP is a complete game changer! My strategy over the next three to five years is to implement more of that. The good news is that I've found a few process engineers who are very interested in helping with the feedback loop. We’ve put together a plan to connect chemists and process engineers closer to the process so that we know what's really happening as opposed to what we assume is happening.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.