October 7th

Welcome and Plenary Session – Gerald J. Hahn Achievement Award (8:30-9:15AM)

The Ubiquity of Variation: Implications for Statisticians

Willis Jensen, Brigham Young University

Abstract:

As Hahn and Doganaksoy observed in 2012, “The ever-increasing use of statistical tools by nonstatisticians has, indeed, made this a golden age for statistics. But it is not necessarily a golden age for statisticians.” Since that time, data analysis has exploded in every industry. Easy-to-use software, open-source libraries and AI have turned data analysis into something anyone can do, even if the quality of the analysis is sometimes up for debate.

The rise of data science and predictive analytics as disciplines means that even the analysis of more complex data with sophisticated methods is no longer the sole domain of statisticians. And the word “statistics” itself, fairly or not, often carries more baggage than authority with the nonstatisticians we work with. So what is a statistician to do in this environment?

The distinct value of a statistician is no longer solely in performing data analysis. Rather, a path to increasing our value stems from our recognition of and deep appreciation of the ubiquity of variation. We are experts in recognizing variation’s impact at every step of the data analysis process. Drawing on examples from my career in business and industry, I’ll show what this looks like in practice — and make the case that in the age of AI, understanding variation is the statisticians’ calling card.

Youden Address (3:30-4:30PM)

Large Computer Experiments: When Interpolation Becomes Extrapolation

Derek Bingham, Simon Fraser University

Abstract:

The design and analysis of computer experiments have evolved dramatically over the last four decades. Gaussian process emulators have become a standard tool for analyzing complex simulations, providing flexible regression together with uncertainty quantification. Yet many modern applications, ranging from nuclear engineering to climate science, now involve hundreds or thousands of input variables and simulation campaigns that appear large in absolute terms but are sparse relative to the dimensionality of the problem.

The main challenge in such settings is no longer just computation, but also internal extrapolation. In high dimensions, prediction often occurs in regions that are far from all observed simulator runs, even when those regions lie within the convex hull of the experimental design. This blurs the classical distinction between interpolation and extrapolation and raises important questions about prediction and uncertainty quantification.

Using recent work on local Gaussian process emulation with structured mean functions to aid illustration, I will discuss why conventional surrogate models can fail in sparse regions of the  input space and what this reveals about the relationship between design, modelling, and uncertainty quantification. The talk concludes with some challenges that high-dimensional computer experiments pose for statistical methodology.

October 8th

Luncheon (12:30AM-1:45PM)

Quality by Design: Strategy, AI, and Communities in Action

Jeri Mulrow, 2026 President of the American Statistical Association

Abstract: Quality is designed, built, and sustained through deliberate choices about strategy, priorities, and people. In this talk, I will explore how the American Statistical Association is putting that principle into practice: through the ASA Strategic Plan’s roadmap for the profession, an AI Strategic Plan positioning statisticians and data scientists as leaders in the rapidly evolving technical landscape, and the Communities in Action initiative’s work strengthening connections among members and elevating the value of statistical thinking. Drawing on my own path through the discipline, I will reflect on quality, not as a fixed standard, but as an ongoing commitment shared across our community.