Roger W. Hoerl, GE Global Research

Do statisticians in business and industry have a sexy job? Steve Lohr—in an August 5, 2009, New York Times article —made some remarkable statements about career opportunities for statisticians in business and industry. He quoted Hal Varian, chief economist at Google, as saying, “I keep saying that the sexy job in the next 10 years will be statisticians. And I’m not kidding.”

The statistics profession has had a profound impact on such diverse areas as health care, the environment, conducting the census, and education. Perhaps less visible has been the significant effect statistics has had on business and industry—the private sector. Nonetheless, this effect has been real and noteworthy, and business and industry affords statistics students another viable career option.

The Rich Legacy of Statistics in the Private Sector

One noteworthy early development in the private sector is the work of Walter Shewhart at Western Electric, a manufacturer for Bell Telephone, in the 1920s. Shewhart realized the importance of reducing variation to improve business processes and developed the concept of “special” and “common” causes of variation, as well as the control chart as a method to distinguish between the two. While his original methods may seem basic—or perhaps even simplistic—today, Shewhart was a deep thinker who was significantly influenced by the philosopher C. I. Lewis. He also had a strong technical background in physics and mathematics, which led him to become a founding member and president of the Institute for Mathematical Statistics (IMS) and a president of the ASA.

Among those affected by Shewhart’s work, especially his unique philosophy of how to apply statistical methods, was W. Edwards Deming, another statistician with a physics background. Deming helped apply Shewhart’s methods (and inspection sampling) to U.S. munitions production during World War II. After the war, Deming promoted the application of Shewhart’s methods and the philosophy underlying them to peacetime industry—both in the United States and abroad. (He had significant influence in Japan, where he received the Order of the Sacred Treasure, Second Class, from the emperor.)

Throughout the next few decades, the application of statistics within industry grew exponentially. This growth included—but went well beyond—statistical quality to include applications in product and process design, marketing, and process improvement. Early pioneers include Dick Freund at Kodak; Mavis Carroll at General Foods; Harry Smith at Procter and Gamble; Truman Koehler at American Cyanamid; Art Hoerl (yes, my father), Bob Kennard, and Don Marquardt at DuPont; Otto Karl at 3M; Mike Free at Smith, Kline, and French; Joe Ciminera at Merck; and Horace Andrews at Swift and Company.

At some point, the distinction between industry and business applications became blurred. The traditional stereotype of “smokestack industry” faded when such high-tech industries as computer chip manufacturing, biotech startups, and software (Microsoft, Google, etc.) became critical components of the economy and began hiring statisticians. For example, my employer—General Electric—makes about a third of its profits from financial services. However, it also makes aircraft engines, power generators, and locomotives, not to mention GE Healthcare. The statisticians in our group at GE Global Research might work on molecular pathology one day, aircraft engine reliability the next, and risk analysis of financial portfolios a week later. Should we be categorized as business or industry?

In the past couple of decades, noteworthy business and industry trends have included the emergence of data mining to analyze massive data sets, growth in financial applications of statistics—particularly in risk management—and Six Sigma (more recently, Lean Six Sigma), which is a statistically based improvement methodology. We could add the more recent emergence of statistical challenges in developing Internet search engines (e.g., Google), but our time is limited.

The reason for the development of data mining is obvious: Nobody had terabyte data sets in the 1960s! With digitization of many business processes, such massive data sets are becoming more common, as everything from credit card bills to gasoline purchases tend to be electronically recorded and stored. The opportunities for mining this data, often for marketing purposes, are numerous. One obvious example is Amazon analyzing previous purchases by customers to suggest what books or music they might want to purchase in the future.