David W.

Approachable Statistics

Approachable Statistics

My favorite part of my career has been explaining statistical concepts to students, colleagues, and the general public. I've found that many people are nervous about statistics in their coursework and career. My job is to counter their apprehension and explain things in an accessible but fully correct way. My experience includes teaching graduate-level biostatistics to public health, nursing, pharmacy, and others students on a health sciences campus. I also mentored many students through their masters and Ph.D. in both statistical and non-statistical fields. Throughout this, my focus is on connecting real-world problems to the underlying statistical technique.

Across my career, I've worked in research groups as part of a small specialty hospital, a large non-profit research group, a medium-sized clinical trials consulting group, and a major research university. I've worked with SAS, SPSS, R, and Stata in my own research as well as teaching others. Statistically, I'm an "expert" in complex longitudinal and mixed-effects models, but think that the simplest approach (a t-test, for example) is often best because people actually understand that.

When I'm helping someone understand any complicated subject, math or science, I aim to find common ground between what a student already knows and what they need to know. I teach from a simple common explanation that works for the basic case and, with only a few changes, also fits more advanced cases. It's true, statistics is not always that easy, but I find a straightforward approach provides a solid foundation for understanding the tougher material.

I apply these same keep-it-simple approaches to my non-teaching work. For example, when evaluating the effect of an anti-smoking campaign across multiple states, we developed both traditional statistical models and small-scale simulations (known as agent-based models, coded in Java) in order to understand the actual decisions people make. The models were complicated, but the simulations helped scientific readers better u