October 9th

Welcome and Plenary Session (8:00-9:00AM)

Experimental Mathematics—Friend or Foe?

Fred Faltin, Virginia Tech

ABSTRACT: In the years since World War II, the practice of statistics has been transformed in many ways. Arguably the most substantive of these has been the field’s often rocky interplay with “experimental mathematics”, most of whose practitioners consider themselves computer scientists, not statisticians. The rise of experimental math has undeniably contributed a great deal to current modeling methodology (consider, for example, many of the tools of Machine Learning). But it has also led to some conspicuous failures in applications. With decisions of increasingly greater importance being based on these methods, the stakes for getting it right are becoming ever higher.

From my perspectives as a student, consultant, businessperson, and academic, I’d like to share some observations and experiences regarding the evolution of the relationship between statistics and experimental mathematics. The root causes of modeling errors are often not difficult to identify, and suggest that in some quarters a basic understanding of statistical fundamentals has been lost. Identifying where naïve modelers can go wrong provides us a better understanding of what we need to address if we are to be successful in exerting a positive influence on the practice of data science, and the societally important outcomes that will ensue.

Luncheon (12:15-1:45PM)

Reflections on a Career at Eastman in Statistics

Kevin White, Eastman

ABSTRACT: Historically, the FTC has featured a local speaker who provides an engaging non-technical presentation. In this spirit and being somewhat local (we’ll see about the engaging part), I will begin by offering insights into Eastman, one of the state’s largest employers. Eastman boasts a long-standing history of quality and statistics, and I will discuss how the applied statistics group has operated to contribute to the company’s success while addressing some of the current challenges they face. Additionally, I will explore Eastman’s newer AI and Machine Learning team, sharing my experiences and observations from leading both the Applied Statistics and AI/ML groups during my career. I will conclude with some experiences and advice that was particularly helpful during my career involving professional societies and the Fall Technical Conference.

Youden Address (4:00-5:00PM)

Youden’s Enduring Legacy at NIST

Adam Pintar, NIST

ABSTRACT: In this address, I highlight some of Dr. Youden’s vast body of work at the National Bureau of Standards (NBS) while discussing how these works continue to impact the culture, policy, and statisticians at today’s National Institute of Standards and Technology (NIST). As an example, I describe Dr. Youden’s contributions to the resolution of the 1953 AD-X2 controversy at NBS that severely tested NIST’s core value of integrity. The resolution of this controversy is still held high as a shining illustration of NIST’s integrity.

The title of my address is derived from a later article of Dr. Youden’s called “Enduring Values” where he argues that systematic errors in the measurements of physical constants are best explored and quantified through designed experiments. While this ideal may not always be fully embraced, there is recognition that multiple “independent” measurement methods are crucial for rigorous uncertainty quantification, and these ideas are found in documents governing the creation of NIST standard reference materials (SRMs).

Beyond the influence that Dr. Youden had on the policy and culture of NIST, the tools that he added to the applied statistician’s toolbox are timeless. Techniques for exploratory data analysis like the Youden plot and for experiment design like the Youden square remain relevant. I highlight the power and continued relevance of these tools through their use in several recent NIST projects.

October 10th

Luncheon (11:45AM-1:15PM)

Statistics Is a Core Competency for Effective Collaboration and Sound Science

Madhumita (Bonnie) Ghosh-Dastidar, RAND

ABSTRACT: The American Statistical Association vision imagines a world that relies on data and statistical thinking to drive discovery and inform decisions. We know the challenges to attaining this vision are significant, so collaboration is key. As a statistician working to inform policy and decision making, I know it will take collaboration across disciplines to address society’s biggest challenges—e.g., pandemic recovery, climate change, precision medicine, education reform, or criminal justice. In an era of data ubiquity and rapid analysis, statisticians and data scientists are positioned to play a central role across application areas. The gold standard for public policy is evidence-based decision making—deliberate and strategic application of real facts and research-supported principles that yields objective evidence.

Statistical science is the foundation for evidence-based decision making. As an interdisciplinary science, it has applications to every field imaginable, making statisticians uniquely qualified to lend their expertise in multiple policy domains. Effectively informing policy requires becoming involved early in the design phase; understanding the nature of the issue; and knowing how to communicate, educate, and explain. In this talk, I will provide multiple examples from health policy to highlight both valuable contributions made by statistical scientists and lessons learned – and how this model of collaboration is relevant across other fields and application areas. I will suggest areas of improvement based on lessons learned. And extrapolating from these successes, I will suggest areas for future contributions in which the stakes are very high and involving statistics will be essential.

Reception with SPES Special Panel Session (3:15-5:15PM)

How to Attract and Prepare Students for Careers in Industrial Statistics

Panel discussion: Maria Weese, Miami University, Yeng Saanchi, JMP, Peter Parker, NASA, and Kade Young, Eli Lilly & Co.

ABSTRACT: In a world of big data, cloud analytics, and artificial intelligence, graduate students in statistics and data science programs might not readily think of careers in industrial statistics. This panel will explore strategies for recruiting students into careers in industrial statistics and preparing them for those careers. On the recruiting front, we will discuss recruitment locations, effective advertising strategies, and the types of problems to highlight. On the preparation front, we will explore the essential technical and non-technical skills that are needed for success and whether current programs are adequately preparing new hires for industry roles. We hope to generate a lively discussion on these issues and more.