This column is written for statisticians with master’s degrees and highlights areas of employment that will benefit statisticians at the master’s level. Comments and suggestions should be sent to Megan Murphy, Amstat News managing editor.

Wayne G. Fischer is a statistician at the University of Texas Medical Branch. He provides direct support for the analysis of operations, clinical outcomes, and research data to meet the health system objectives/improvement priorities and develops predictive models.

I don’t have a degree in statistics, and I don’t play a statistician on TV. But, my accumulated knowledge and experience have qualified me to hold a position titled “statistician” at the University of Texas Medical Branch. (My degrees are in chemical engineering.)

It took me three courses over three years at the University of Cincinnati to realize applied statistics could have great potential in my career.

In graduate school at Purdue, one of the chemical engineering professors taught “Design of Experiments and Regression Modeling.” The lectures, homework, and tests were all oriented toward chemical engineering problems. Bingo! I got it. I resolved to never stop learning all I could about applied industrial statistics.

My first job out of graduate school was with Rohm and Haas Company, in Philadelphia, in its R&D Research Computing Group. The first statistical issue I tackled: How much better was the second generation of a key catalyst than the one currently used commercially? I did a two-sample t-test comparing the average conversion—and average selectivity—to the key product. Oops, can’t reject the null! I wrote the memo explaining the analysis and its conclusions. The project leader was in my office the next morning.

Fortunately, he calmly listened to my recommendation. I showed him plots of the individual measurements of conversion and selectivity of the two catalysts and pointed out there was so much variation in each set of data (the “noise”) that we couldn’t say with any confidence that the differences were not due to sampling, even though the averages were numerically different (the “signal”). I got with the chemist running the experiments, and then we went through each step of the process, made changes that reduced common cause variation (the noise), and reran both catalysts. Now the t-test conclusively showed the second-generation catalyst was better.

It was clear to me that this “ailment” (too much noise-hiding signals) probably ran through much of our experiments—bench scale and pilot plant. Plus, all experiments were conducted using OFAT: One Factor at a Time. R&D had a huge opportunity to dramatically improve both the effectiveness and efficiency of its experimentation.

That opportunity presented itself when I came across Stu Hunter’s video lecture series, Design and Analysis of Experiments—with work texts! (This was 1975, after all.) I prevailed upon our project leader to rent the series and buy the work texts; five of us took the course. I agitated to retain Stu Hunter as our consultant. Management agreed, and Stu started making monthly one-day visits. His approach was effective. He consulted with each of us about the projects we were working on, taught us what we needed to know right then, and left us with references to what we should learn next. By 1977, we had developed an interactive computer program, called “Analyzer,” for the design and analysis of full and fractional factorial experiments—the first ever, I believe.

We integrated Analyzer with a three-day, in-house short course we designed and delivered (with Stu) to chemists and engineers. (I continued to make a nuisance of myself by preaching that all R&D experiments should be conducted this way—be the norm rather than the exception—and any exceptions should be justified. Six months after I left Rohm and Haas, the vice president of R&D promulgated a requirement for promotion on the technical ladder was demonstrating at least one use of a designed experiment.)

I continued learning about, applying, and teaching applied industrial statistics through my seven years at ARCO Chemical R&D, 5½ years at Mobil Chemical R&D (where I gave 50-minute biweekly seminars), and six years in Mobil’s Olefins & Aromatics Business Group. I inculcated the use of statistical process control (SPC) in R&D’s labs and pilot plant. In the Business Group, I got to apply SPC in marketing and sales, supply and distribution, manufacturing, and even in human resources!

I made my career change in July 2000 by joining the University of Texas MD Anderson Cancer Center’s (UTMDACC) Performance Improvement Department. As a chemical engineer with 24 years’ experience in industry—and none in health care—I wasn’t sure why they hired me (even though the job description read like a summary of my résumé). But their leadership knew why, and after about six months, I said to myself, “OMG, do they need what I have!”

Back then, my kind of experience in the concepts, principles, and methods of continual improvement, data analysis, modeling, simulation, and optimization was a rarity in health care. At UTMDACC, I taught and applied SPC throughout the organization and was able to apply Monte Carlo simulation, linear programming, and multivariate linear regression.

In May 2011, I transferred to the University of Texas Medical Branch in Galveston as a “statistician.” (By now, I thought of myself as a “statistical engineer.”) I still work across the organization: faculty, residents, fellows, nurses, medical students, support staff, and administrators.

What are my life lessons after 45 years of applying, consulting, and teaching applied statistics—first in industry and then in health care?

• Never stop learning. Not only just-in-time (as project needs dictate), but “ahead-of-time”—what I call “anticipatory learning.” As you become familiar with your organization’s processes and needs, try to discern what will be needed next that you don’t yet know anything about.
• Speak the local language. Because I was an engineer trying to learn statistics, I knew firsthand the barriers of “speaking statistics,” instead of speaking about solving the problem in terms of the needs of the client. I became good at this because, starting out as an engineer, I could present the statistical methods in the “language” of the other engineers and chemists in R&D. I carried that skill into health care.
• Become known for one thing. At least one thing. While at Mobil Chemical R&D, I resolved to implement SPC everywhere I saw it was needed. When I was transferred to the Business Group, the head of R&D lamented, “Who will support our use of SPC?” Having a strong and well-known expertise in a specific area gives you credibility and serves as an “entry point” for demonstrating your other skills and knowledge.
• Volunteer with professional societies. Start with the local chapters, giving presentations. Then, when you’ve assessed the lay of the land, take on the responsibilities of an officer. Volunteer to facilitate sessions at national conferences, to chair and organize sessions, and to present your work. I found this approach motivated me to really know my subject matter, to build my professional network quickly (and widely), and—later in my career—to identify mentoring opportunities. (I use LinkedIn to learn, share my knowledge, and mentor others.) And it develops your leadership skills.
• Balance your areas of expertise. As I said, I continued to build up my knowledge and expertise in applied statistics—the “hard” side as I call it—after graduate school. It wasn’t until Mobil Chemical (16 years later) that I literally “stumbled” onto the realization I knew nothing about the “soft” side—working in groups on a shared objective. I learned about what I didn’t know from a review of The Team Handbook. I was on a “team” that had just disintegrated and I didn’t know why. The Team Handbook taught me why and opened my eyes to a critical lack in my knowledge base. I went on to learn everything I could about quickly building, leading, and facilitating high-performance teams and was able to apply it all when I was promoted to lead the implementation of team-based total quality management in that business group. I became known as the “team expert.”
• Look outside your organization/industry for the “next thing.” Cast a wide net to anticipate and identify what may lead to an important breakthrough for your department or organization. At Rohm and Haas, besides pushing designed experiments, I pushed for computer-based graphics (remember, this was 1972–1978 … multi-pen, flat-bed plotters). A coworker objected, saying, “Graphics? No one is asking for graphics!” Well, because I got HP’s public domain software for free, our project leader sprung for an HP plotter and, voila, demand surged!