Christopher Tong has been a nonclinical and clinical biostatistician for 20 years, both in the pharmaceutical industry and in government. He has a master's degree in applied statistics, and a Ph.D. in physics, from Purdue University. He has co-authored work published in journals in fields such as fluid dynamics, atmospheric science, physiological acoustics, chemometrics, medical imaging, microbiology, and human and veterinary medicine.
Episode Description
The Gilded Age in the U.S. is perhaps best known for the great men who rose to prominence at the time. Men such as John D. Rockefeller and JP Morgan, who made money hand over fist. One of those men Andrew Carnegie was not only a shrewd businessman, he was also a shrewd statistician of sorts. His legacy is the focus of this episode of Stats and Stories with guest Christopher Tong.
+Timestamps
How you got interested in Andrew Carnegie as a statistician? (1:33), Carnegie’s background (3:44), How much influence did his work have? (8:58), The Census (10:11), the ASA today (15:48), How does his work inform today? (20:02), Statistics in business today (21:38)
+Full Transcript
Rosemary Pennington
The Gilded Age in the U.S. is perhaps best known for the great men who rose to prominence at the time. Men such as John D. Rockefeller and JP Morgan, who made money hand over fist. One of those men Andrew Carnegie was not only a shrewd businessman, he was also a shrewd statistician of sorts. His legacy is the focus of this episode of Stats and Stories, where we explore the statistics behind the stories and the stories behind the statistics. I'm Rosemary Pennington. Stats and stories is a production of Miami University's Department of Statistics and media journalism and film, as well as the American Statistical Association. Joining me as regular panelist John Bailer Chair of Miami statistics department. Our guest today is Christopher Tong Tong has been a non clinical and clinical biostatistician for 20 years, both in the pharmaceutical industry and in government. He has a master's degree in Applied Statistics and a PhD in physics from Purdue University. Tong's co authored work has published in journals and fields such as fluid dynamics, atmospheric science, physiological acoustics, medical imaging by microbiology, and human and veterinary veterinary medicine. He's also the author of an article for chance, chronicling Andrew Carnegie's adventures in statistics. Chris, thank you so much for joining us today.
Christopher Tong
Thank you, thanks for the invitation.
Rosemary Pennington
I guess I'm just kind of wondering how you got interested in Andrew Carnegie as a statistician?
Christopher Tong
Yes. So I actually when I was a graduate student in physics, this is before I transferred careers into statistics. I was at a little university bookstore, and ran into the clearance section of the store and found Harold Livesay a little book about Andrew Carnegie, Andrew Carnegie and the rise of big business. And it was like $2, or something like that. So I picked it up. It was a short book. When I read it, I thought it was a really interesting perspective on American history and the history of American business. Years later, I decided to switch careers and become a statistician. And as I started to get acquainted with the field, I remembered reading about Carnegie and his work with data. And then at some point, I joined the American Statistical Association. And back then they had some marketing material saying, Okay, here's some famous members of the ASA from an earlier age, and Andrew Carnegie's name was mentioned among them. And so that kind of got stored in my head for a while. And as I got interested in the whole discussion, back then it was Data Mining and Machine Learning, versus statistics. And Leo Brian's paper, the two cultures, where he talks about how statisticians have this probability model approach to thinking about data. And computer scientists kind of went through this algorithmic approach with machine learning, and trying to reconcile those two ways of thinking about data. And it occurred to me well, you know, Andrew Carnegie had his way of looking at data back in the 19th century, and how does that fit into all that conversation? So that all got to swirling around in my head. And I found myself living in Ames, Iowa at one point, and they have the archive of the American Statistical Association. So I was actually able to look up some primary sources about Carnegie and the ASA.
John Bailer
So when you think about your interests here, and you know that that fortuitous sale book that you encounter, you know, I think that serendipity is something in terms of my reading has has often I've benefited from the sale racks as well. So can you just just talk a little bit about some of the aspects of Carnegie's life that that kind of intrigued you, you know, what are for for people that are that that don't know much about this, this individual and the times. Can you just say just a little bit about back his background that that you found interesting and worthy of exploration?
Christopher Tong
Yeah, so he's his story is an immigrant story. He was brought to the United States at age 12. By his family, his father was a handloom Weaver in Scotland. And the industrial revolution occurred and steam powered machinery kind of made his job obsolete. So the family came to the United States in search of a future and they settled north of Pittsburgh, and the age of 13. He started working as a child laborer. Yeah, he worked 12 hour days as a bobbin boy, at a textile factory, which means running around, replacing bobbins with thread and then he moved to a factory that manufactured bobbins and started doing some bookkeeping on the side when he was there. And then he moved to become a messenger boy for a Telegraph Company, learned how to be a telegraph operator became really good at it offered his services to the Pennsylvania Railroad. Now, this is kind of a key moment, because at the time, the Pennsylvania Railroad was America's largest corporation. And so back then most companies had fewer than 250 employees. They're very local, very, what Harold Livesay, the author I mentioned before, their management methods were very casual by tradition and intuition, not very data driven. And the railroads couldn't be that way, because they had large, geographically spanning organizations, with many, many employees. And you couldn't just run a railroad with like one Foreman telling everyone what to do. You had to have some organization, and you had to be able to make money. And so the railroad executives at the time, they all had backgrounds in civil engineering, people like Daniel McCallum, for example, they started to understand how to organize how to run a railroad company. And one of the critical pieces of that was, well, when you lower the ticket prices, you'll get more ticket sales. But if you lower them too much, you're gonna lose money. So how do you solve that problem? Well, you have to understand your costs. Once you understand your costs, you know how to set your prices so that you'll stay profitable. And Carnegie picked up on all that he learned. All of these methods have cost accounting that were developed by railroad managers. And he also learned from his mentor at the Pennsylvania Railroad, how to be an investor, and he did investing on the side. And by the end of the Civil War, he became independently wealthy and was able to leave the railroad. He became a basically a financial speculator for a number of years and did some rather unsavory things, which were totally legal back then, basically, insider trading, false advertising, that kind of stuff. And he finally settled on concentrating his efforts on steel manufacturing, what he discovered in steel manufacturing was that other manufacturers really did not understand their costs. And Carnegie says in his autobiography, some managers thought they were going to make a profit at the end of the year, ended up making a loss and vice versa. And they really didn't know until they closed all the accounts at the end of the year. And Carnegie thought this was unacceptable. And he decided to adapt all the cost accounting methods from the railroad industry that he had learned. And he has in his managers took those ideas, introduced systems of measurement, which means weighing raw materials, and then weighing the final product at various stages of production, and getting an understanding of costs and production, and having a better understanding of how the mills were running. And this allowed him to run circles around his competitors, because he could set those prices in such a way where he knew he was going to make a profit, but still low enough to compete with his competitors. So and that was very much an introduction of data collection throughout his operation, and then using those data to make decisions. So that's and guess what he did join the American Statistical Association in 1892.
Rosemary Pennington
I wonder how how much influence did Carnegie have sort of after he adopted these methods? Did he did his choice to sort of take this approach influence the way other people in this industry? Were also approaching their business?
Christopher Tong
At first? The answer's no, because as I said, he was able to run circles around his competitors. But eventually, these ideas did become did begin to diffuse. And when you look at histories of American Management, Carnegie is often credited with being a pivotal figure in bringing managerial concepts set started in the railroad industry. And were for a long time confined there into these large companies, American companies, like I said, most company at the time, Carnegie started his career were pretty small, single factory, one manager and so on. And Carnegie Steel Company was one of the first big businesses and that's the title of the Herald I've seen a book I mentioned before Andrew Carnegie and the rise of big business. And turns out, you can't really manage a big business without all that methodology.
Rosemary Pennington
You're listening to stats and stories. And today we're talking to Christopher Tong about the statistical life of Andrew Carnegie. You talk at one point in the article about Carnegie and the census, do you want to share a little bit about sort of what you discussed in the chance article in relation to that?
Christopher Tong
Yeah. So Andrew Carnegie was kind of known for three things being a business tycoon, which we just talked about, he was known as a major philanthropist. And he's, he kind of thought of himself as kind of a public intellectual, even though he wasn't an educated man. He wrote books. And he was a strong supporter of the American style of government and the American type of economy, as opposed to what he called Old Europe, and the British Isles in particular. And back when he was a teenager in Pittsburgh, he carried on a debate with his cousin in Scotland. His cousin was very upset about slavery in the United States, which at the time was still practiced. And Carnegie wanted to defend his new country and, and his way of defending the United States was by going to economics statistics, and what we now call national statistics, trying to show that the United States as a productive country, growing country, and many years later, when he was in his 50s, he kind of revisited these arguments by writing a book called triumphant democracy. And that's the book where he tried to show that, yes, the United States is, in some sense, a superior country to the countries of Old Europe. And he compared all these national statistics, and economic statistics. And writing that book, he did a lot of research, including looking at the census data, the US census data, and other statistical sources at of the time, such as Scribner, statistical atlas, he went further than that I mentioned the third aspect of his life was philanthropy. He consulted census data when towns would submit requests for a donation of a church, Oregon or a public library to their town. And his assistant would ask the town to provide certain information that would then be cross checked with US census data. And the reason for that was Carnegie's rule for making the donations was $2 per person. So it's very dependent on population, as well as other factors like site availability, and so on.
Rosemary Pennington
That's really interesting. I actually, I grew up near Portsmouth, Ohio, which has a Carnegie Library, and I had no idea that's how that decision was made about whether that would be would be built there. So that's really interesting to hear.
John Bailer
You know, as part of the the paper that you wrote, In chance, you were talking about kind of the decision that was made on expanding use of a gas furnace is one of those examples of how, you know, Carney was using data to support an investment and his company, would you could you talk a little bit about that, that example that that you shared in your paper? Yeah. So,
Christopher Tong
Carnegie's steel company invested in the Siemens gas furnace, which was an open hearts, furnace technology. And at the time, this was something that was used in Europe. And then he was the first one in the United States to adopt that technology. At the time, many other steel manufacturers in the United States, were not interested in that technology, because it was very expensive. And Carnegie says in his autobiography, well, actually, it turns out that if you had doubled the price of that furnace, it still would have paid off as an investment. And that's because, again, he had a very good understanding of his costs. One of his sayings was, as long as you keep an eye on the cost, the profits will take care of themselves. Now, what he doesn't disclose in his idle autobiography, and this is very interesting is that it he had actually purchased the first one of these furnaces before it was cost effective before some of the advances in metallurgy had made those open hearth furnaces cost effective. And the reason he had bought one of these earlier was for custom orders of very high grade steel. But what this enabled him to do was to use that furnace as a laboratory for studying steel manufacturing. In fact, Carnegie was one that was the first American steel manufacturer to hire a research chemist. Now he didn't pay the chemist very well. But he was ahead of his time. Now, he he certainly tried to conceal this, one of his other sayings is pioneering don't pay. So he wanted to give people impression that he preferred to be what we would now call a fast follower, someone who piggybacks on, on all the the expense that the innovator spends on proving the new technology works. But actually, he himself was an innovator. And I guess he didn't want people to know that. But this all goes back to data, because he, he wanted to measure he wanted to measure the cost and wanted to measure product and input and output. And so because he understood his cost very well, and his competitors didn't, he had the data to confidently make those kinds of investment decisions.
John Bailer
So you have in your, in your, in your paper, a couple of lovely, contentious statements. That, you know, I you would expect me to respond to so so I will I, you know, I'll jump in and so use you, you said there that, by contemporary standards, Carnegie and his modern counterparts in business are not statisticians would not be at home in the US, as members of the NSA despite being voracious collectors and consumers of data. And then the next sentence is, this seems to be a consequence of the evolution of statistics into a highly specialized and exclusionary discipline Chris, I got a I got a call foul. You know, I, you know, because as I, as I think about this, you know, you know, my my take of statistics, when you look at the breadth of which you we even within ASA with the Deming award and, you know, the lecture that's associated with, with Demings contribution, which revolutionized manufacturing, in the last century, with the the representation within ASA, where, you know, there's rotating leadership that involves government and academia as well as business industry. And so tell me, what, what, what kind of is, what prompted you to kind of push in that direction.
Christopher Tong
Yeah, before we do that it's good. As you mentioned, W Edwards Deming, I think Deming would have had a field day, tearing Carnegie apart, because one thing that Carnegie did not have was a modern understanding of statistical variation. And so for example, when a certain furnace was doing very well, he would demand from his managers why the other furnaces weren't doing well. And when the furnace performed below average, he actually Anis asked one of his managers, why was it below average? And the manager said, well, the average is the mean of the highs and the lows. So I think Carnegie could have used a modern understanding of statistical variation and all the ideas that you heard, and Deming introduced in statistical quality control. But if you look at the the data driven methods at Carnegie was using to run his steel mills, those methods are still taught today. They're just not taught in statistics departments. They're taught in schools of Industrial Engineering, and schools of business under the title of managerial accounting, or production and operations management. We do have in statistics textbooks, whose titles call themselves things like statistics, the art and art and science of learning from data or statistics, how to use data to make decisions. And this is exactly what Carnegie was doing, using data to make decisions to learn about his factory and understand it better to learn about costs. But these methods are now taught in schools of business schools, industrial engineering. And my point is, there are a lot of data professions. Now, not just statistics, and statisticians often define their subject very broadly. And I don't disagree that statistics is a science of learning from data. It's just not the science of learning from data.
John Bailer
i Yeah, I'm not sure you'll you're gonna, even in my most prickly days, I disagree. So I think I, I see what you're saying. And I see where you're coming from. But I suspect that, you know, there's, there are lots of other traditions, particularly in some of the data that you've described, where you'd look at national statistical offices. And in particular, there's a strong tradition of people coming from an economics, political economics and social science perspective and context, where they're really engaged in these types of production of very specific types of statistics for very specific types of roles. I just one of the things that I've loved about our discipline is is the ability that had to have this, in my view, a very inclusive perspective on on this so it's a so that's why I had to just ask you about that Chris. So yes
Rosemary Pennington
I do wonder, given sort of your look back at how Carnegie use using data, and he was a member of ASA, what you think it is that we can take from Carnegie's approach to data? Well, you know, maybe he wasn't doing, you know, probabilistic sampling, but are there things that we can take to inform our own either work of data or or reading of data even?
Christopher Tong
Right, that's a very key point, rosemary, because many statisticians and I quote, a paper by Brown and Cass, who define what a statistician is, as one who recognized the centrality of probability as a way of thinking about data. And that certainly wasn't the case in the 19th century. And that's why I think people like Andrew Carnegie in the 19th century, would join organizations like the ASA statistics, Rachlin hadn't adopted this probability centered focus. The most important lesson those of us today can take from Carnegie is that often, the data you need is not the data you have. And it's very much because we have data science now. And a lot of data science is based on data that we happen to have data that we because we live in a very modern society, highly instrumented society, ecommerce, we're getting all this data copiously. But sometimes the data you need to make a good decision isn't the data that you have, it's data you need to go out and make an effort to get.
John Bailer
Now that's I think that's, that's a that's a great point. And it's, you know, I, you look at how statistics has evolved. And even like, the idea of exploratory data analysis was pretty revolutionary, when when it was first coming out that it changed are the you know, what is an aspect of practice, that became part of what we considered. But as you look, as you think about kind of the current generation that's coming up that that might be involved in impacting industrial practice. I mean, ASA is a really old organization. I mean, it's like, there's only one, one professional society older in the US than the ASA. That'll be on a quiz later. But so, so in terms of the the a, it's a very old organization, and things like industrial engineering didn't exist, you know, sort of, there are lots of things. I mean, heck, a bachelor of science and statistics are in data science and statistics didn't exist until, you know, within the last 30 years, some of this seems natural to me that there are spin offs, you know, computer science didn't exist as a separate entity at one time, you know, and there's so a lot of this type of, of evolution and separation, I don't think is, is bad, or an indictment of kind of where things started. But But if people want to work now, in this area that you were investigating at Carnegie was playing, playing and investing in and understanding this idea of the production and large mid scale manufacturing setting. What are some of the things that someone should learn to be able to do this effectively?
Christopher Tong
So yeah, the, the book that kind of got me thinking this way, is a book called factory physics by hoppin Spearman, and they start their book with a little history of Industrial Engineering, and they talk about Carnegie's role. And you don't have to necessarily study that particular book. But there's a saying in the business world, accounting is the Language of Business. And that's, I think, quite a key to understanding that it's the first step in learning how to use data to think about business decisions. And there's kind of two pieces of that financial accounting, which is actually a lot older than statistics, and managerial accounting, which is kind of the thing, the area that Carnegie and others were developing in the 19th century, that's become what we call managerial accounting. Now, most business schools do require students to take a statistics course, as we were talking about quality control, yes, you will be a better decision maker, if you do understand probability and statistics and some of the ideas from quality control. But there are all kinds of other ways of using data that you need some of these other courses to learn from. And I think those are two pillars that you can stand on to begin your study of how to use data to manage your business.
Rosemary Pennington
Thank you. That's all the time we have for this episode of stats and stories. Chris, thank you so much for joining us today.
Christopher Tong
Thank you so much been fun.
Rosemary Pennington
Stats and Stories is a partnership between Miami University’s Departments of Statistics, and Media, Journalism and Film, and the American Statistical Association. You can follow us on Twitter, Apple podcasts, or other places you can find podcasts. If you’d like to share your thoughts on the program send your email to statsandstories@miamioh.edu or check us out at statsandstories.net, and be sure to listen for future editions of Stats and Stories, where we discuss the statistics behind the stories and the stories behind the statistics.