A Career in Statistics

Beyond the Numbers

Gerald J. Hahn, Necip Doganaksoy

• Publisher: Wiley
• Year: 2012
• Print ISBN: 9780470404416
• eBook ISBN: 9781118490136
• Edition: 1
• Topic: Mathematics
• Subtopic: Probability & Statistics
• Index: Mathematics

A valuable guide to a successful career as a statistician

A Career in Statistics: Beyond the Numbers prepares readers for careers in statistics by emphasizing essential concepts and practices beyond the technical tools provided in standard courses and texts. This insider's guide from internationally recognized applied statisticians helps readers decide whether a career in statistics is right for them, provides hands-on guidance on how to prepare for such a career, and shows how to succeed on the job.

The book provides non-technical guidance for a successful career. The authors' extensive industrial experience is supplemented by insights from contributing authors from government and academia, Carol Joyce Blumberg, Leonard M. Gaines, Lynne B. Hare, William Q. Meeker, and Josef Schmee. Following an introductory chapter that provides an overview of the field, the authors discuss the various dimensions of a career in applied statistics in three succinct parts:

• The Work of a Statistician describes the day-to-day activities of applied statisticians in business and industry, official government, and various other application areas, highlighting the work environment and major on-the-job challenges
• Preparing for a Successful Career in Statistics describes the personal traits that characterize successful statisticians, the education that they need to acquire, and approaches for securing the right job
• Building a Successful Career as a Statistician offers practical guidance for addressing key challenges that statisticians face on the job, such as project initiation and execution, effective communication, publicizing successes, ethical considerations, and gathering good data; alternative career paths are also described

The book concludes with an in-depth examination of careers for statisticians in academia as well as tips to help them stay on top of their field throughout their careers. Each chapter includes thought-provoking discussion questions and a Major Takeaways section that outlines key concepts. Real-world examples illustrate key points, and an FTP site provides additional information on selected topics.

A Career in Statistics is an invaluable guide for individuals who are considering or have decided on a career in statistics as well as for statisticians already on the job who want to accelerate their path to success. It also serves as a suitable book for courses on statistical consulting, statistical practice, and statistics in the workplace at the undergraduate and graduate levels.

CHAPTER 1

PROLOGUE: A CAREER IN STATISTICS

1.1 ABOUT THIS CHAPTER

We begin this introductory chapter with a brief examination of what is statistics, who is a statistician, and who employs statisticians. We then comment on the statistical thought process and what makes it special, the many skills required to be a successful statistician, and the role of statistics beyond the workplace. We provide “equal time” to presenting some downsides of a career in statistics and counter this with a brief summary of the excitement of such a career. We then indicate some alternative paths for embarking on a career in statistics, comment on ongoing efforts for accreditation, and review professional societies for statisticians. We conclude the chapter with a preview of what is to follow.

1.2 WHAT IS STATISTICS?

The rational basis for change is data. Data means statistical methods.

—W. Hunter

Some informal definitions of statistics, provided by various well-known statisticians, are

• The science of learning from (or making sense out of) data (J. Kettenring).
• The theory and methods of extracting information from observational data for solving real-world problems (C.R. Rao).
• The science of uncertainty (D.J. Hand).
• The quintessential interdisciplinary science (S. McNulty).
• The art of telling a story with data (L. Gaines).

We prefer the preceding over the more formal definitions found online,¹ such as statistics is “the science that deals with the collection, classification, analysis, and interpretation of numerical facts or data, and that, by use of mathematical theories of probability, imposes order and regularity on aggregates of more or less disparate elements.” And we note that Brown and Kass (2009) devote a 19-page article (including discussion) to an in-depth examination of “What is Statistics?”

Statistics has applicability in almost all areas of human endeavor. To take just a few examples:

• Economists need to understand statistical concepts to make predictions.
• Psychologists need to be able to interpret empirical relationships between variables.
• Business executives need to appreciate the role of uncertainty in decision making.
• Biologists need to understand that the reactions of organisms to stimuli are not deterministic and that there will be variation among individuals.
• Design engineers need to know how to conduct statistically valid experiments to develop the best possible products.

We also note with interest that in 2010 Britain joined other countries in asking their statisticians to develop a “happiness index” to be added to their existing national household survey.

The diversity of applications is further illustrated by the 24 sections of the American Statistical Association (ASA) as of the end of 2010. These include those that deal with biopharmaceutical applications, business and economics, defense and national security, education, environmental applications, epidemiology, government, health policy, marketing, physical and engineering sciences, quality and productivity, social statistics, and sports.²

We will defer discussing specifics to the next three chapters, but note that Tanur et al. (1972) and Peck et al. (2006) provide 15 and 25 (different) articles, respectively, that demonstrate the use of statistics in a wide variety of application areas.

Some areas of application have taken on lives of their own. Thus, biostatistics is the application of statistics to the analysis of biological and medical data. Going even further, actuarial science—the application of mathematical and statistical methods to assess insurance risk—has become a separate profession.

Statisticians at a particular point in their careers are commonly engaged in one or a few of these application areas. It is, however, not unusual for individuals to become involved in an appreciable number of such areas during the course of their careers.

1.3 WHO IS A STATISTICIAN?

Professional statisticians are trained in statistics and actively use statistics and statistical concepts and thinking in much of their work.

The ASA now has a program of accreditation for its members (Section 1.11) and such programs exist in various other countries. The ASA accreditation program has high requirements (e.g., 5 years of documented experience in practicing statistics) and is completely voluntary. We anticipate that there will be many statisticians in the United States, especially until accreditation becomes popular, who have not been formally accredited.

Moreover, what constitutes as adequate education in statistics to qualify an individual to be a statistician is highly job dependent (and somewhat controversial), although there are some communalities (Section 7.4). Thus, to simplify matters, we will just assume statisticians to be those who regard themselves as such.

One does not have to be a statistician to use statistics. The use of statistical methods by nonstatisticians—whom we will refer to as “practitioners”—is at an all-time high and will likely continue to increase. This places additional responsibilities on statisticians. We shall return to these topics at various junctures throughout this book.

1.4 WHO EMPLOYS STATISTICIANS?

Data are widely available; what is scarce is the ability to extract wisdom from them.

—Hal Varian (2010)

A simple, and perhaps somewhat exaggerated, answer to the question “who employs statisticians?” is “essentially all large and some medium-sized, and even small, organizations.” Somewhat arbitrarily, we categorize employers of statisticians as follows:

• Business and industry that manufacture products and/or provide services (Chapter 2).
• (Mostly government) agencies engaged in gathering, analyzing, and reporting official statistics or in related statistical activities (Chapter 3).
• Organizations involved in various other application areas, including those engaged in regulatory activities, health, national defense, other scientific research, and the social and behavioral sciences. Employers include government, research institutes, and universities (Chapter 4).
• Self-employed (typically private statistical consultants). Discussion of these is postponed to Chapter 12 since this is a role unlikely to be taken by statisticians early in their careers.
• Academia (Chapter 13).

The ASA estimates that its membership is broken down approximately as follows:

• 47%: Business, industry, nonprofit (other than government and academia), self-employed, or other.
• 42%: Academia.
• 11%: Government (national, state, provincial, or local).³

In any case, like other professionals, statisticians work for customers who benefit from their work. These customers might be direct, such as a client who has commissioned specific work, or indirect, such as users of the Consumer Price Index or students in a professor’s class.

1.5 THE STATISTICAL THOUGHT PROCESS AND WHAT MAKES IT SPECIAL⁴

1.5.1 The Scientific Method

Almost 500 years ago, during the reign of Queen Elizabeth I, Sir Francis Bacon, an English philosopher, lawyer, and statesman, addressing the Royal Society, proposed a new way to gain increased understanding of nature. Instead of drawing conclusions from their own preconceived notions, religion, or other traditional sources stemming from Aristotelian thinking, scientists should engage themselves in observation and experience. He didn’ t take the idea much further than that, but others built on it, and, a few centuries later, it led to the formal development of the scientific method.

Basically, the scientific method calls for starting with a conjecture about the state of nature, cause and effect relationships, or differences among phenomena (e.g., animal typing by physical characteristics, medical treatment differences, or rates of differently induced chemical reactions). We then make observations, that is, gather data, in order to confirm or deny that conjecture. This must be done in such a way that the results are reproducible by others. Thus, we build knowledge about the state of the universe by confirming or denying conjectures.

All of this may not sound like rocket science today, but at the time it was truly revolutionary, and the results have been spectacular. The scientific method has, for example, guided the increase of crop yields to help feed starving populations. It has led medical researchers to understand the causes of diseases and find cures. And it has resulted in electrical engineers learning how to produce microchips efficiently in mass quantities, allowing access to such modern technologies as laptop computers, the Internet, computer-based controls, and safety mechanisms in planes and automobiles.

The applicability of the scientific method is, moreover, not limited to the hard sciences, such as chemistry and physics. Our understanding of sociology and psychology and other social or human sciences has relied on its use as well. The list of beneficial applications is endless. It seems safe to say that the overwhelming majority of advances in human civilization have taken place as a consequence of the application of the scientific method.

1.5.2 Where Statistics Fits In

But what does this have to do with statistics and the way statisticians think?

Put simply, we assert that statisticians, in many ways, might be considered gatekeepers of the scientific method. There are good reasons for this lofty claim. After all, a key factor in any scientific endeavor is a concern for obtaining unbiased results with a wide range of applicability.

In many, or even most, situations, it is impractical or even impossible to enumerate completely an entire population. Thus, you need to develop precise and accurate estimates from a well-selected sample. Say, for example, that we want to characterize the mean weight of salmon in a lake. You can’t get all the salmon in the lake and weigh them—but you can take an appropriately selected sample and use its mean weight as an estimate of the mean weight of all salmon in the lake. But to draw correct conclusions, you must concern yourself with the representativeness (as well as the size) of the sample. Does it, for instance, include only farmed or only Northern Pacific salmon? If so, the estimate is clearly biased with regard to determining the lake’s entire salmon population and applies only to the limited portion of the population under examination. So, when statisticians are called upon to propose a sampling study, they ask all sorts of (both) broad and specific questions, which may perhaps initially seem impertinent, and then use the answers to help develop a plan that is maximally informative under the circumstances. When the results of the study become available, statisticians then quantify the uncertainty in the findings and provide warnings about the generality of the results. And this is where the gatekeeper status comes in.

Of course, statistics and statistical thinking are more complicated than that. Many studies are observational, and others involve designed experiments (Section 11.2). But because statistics—in light of its reliance on scientific sampling—is, in a large part, the science of making decisions under uncertainty introduced by the sampling process, and because sampling applies to almost all intellectual scientific pursuits, one could assert that statistics is relevant to just about every discipline.

1.5.3 A Peek into How Statisticians Work

It All Starts with the Theory. Statistics, just as other scientific disciplines, has theoretical components, many of which have roots in mathematics. Theoretical statisticians conduct important work in developing new methods, improving on existing ones, and coming up with novel ways to address applied problems. Expanding the theory is essential to the health and well-being of the field and requires strong mathematical skills. Most statisticians receive training in basic statistical theory and rely on such theory in applications—many of which are not straightforward textbook situations—and in understanding the basic assumptions underlying the methods that they are using (even though they are generally not conducting research in theoretical statistics).

A Traditional View. Applied statisticians work primarily on using statistics to address issues in other disciplines, generally at the behest of what we will refer to as “problem owners.” These may be administrators, economists, engineers, scientists, social scientists, or others, who are typically leaders or representatives of a larger project team.

In the past, problem owners often came to statisticians with a fait accompli; that is, they had already conducted a study and gathered data, and wanted to know what it all meant. In response, statisticians typically asked questions about the study objectives, the data gathering process, the measurement methods, and so on. This was often followed by the unenviable task of sorting through the data structure, and then the data themselves, attempting to extract meaningful findings. Sometimes, this was not possible, resulting in much wasted effort. This unfortunate situation was a consequence of a frequently held misunderstanding (often furthered by what was taught in school) of statisticians as merely appliers of a series of tools to evaluate already collected data—as opposed to being purveyors of the scientific method who concern themselves with the entire problem and the gathering of the appropriate information to address it.

A Collaborative Step-by-Step Approach. A much preferred alternative route for statisticians’ participation—and one that, fortunately, is becoming increasingly prevalent—is that of collaboration. Eff...

Table of contents

• Cover
• Half Title page
• Title page
• Copyright page
• Dedication
• Preface
• Acknowledgments
• Chapter 1: Prologue: A Career in Statistics
• Part I: The Work of A Statistician
• Part II: Preparing for A Successful Career in Statistics
• Part III: Building A Successful Career as A Statistician
• Part IV: Further Perspectives: Careers in Academia and Professional Development
• References
• Index