Today we are getting to know one of our Department's adjunct professors, Dr. Rafe Donahue. Get the inside scoop about his research and interests!
What is the focus of your research and what have been the major findings so far?
As an adjunct faculty member working in industry, my research doesn't quite line up with the research of my strictly-academic colleagues. The work I do, however, helps patients in a very direct sense, by getting products to market for the benefit of patients and health care providers.
The work that has kept me busy for the last decade or so revolves around a growth factor used in joint fusions, surgeries that eliminate pain by making two bones grow together across a joint that, typically, is damaged by arthritis and thus is painful. This growth factor, we have shown, when properly applied, can be used in place of an autograft, a piece of the patient's own bone, harvested from elsewhere in the body. These autografts can themselves result in problems, hence the value of a product that can be used in their place.
Tell us us about any recent publications you've had.
Our research typically gets published in major medical journals. I was fortunate to get recent publications in Journal of Bone and Joint Surgery, Foot and Ankle Surgery, and Foot and Ankle International.
What was your draw to statistics at Vanderbilt and what have you learned as a statistician?
The Vanderbilt biostats group is a vibrant team of fiercely-bright, top-notch professionals who are not only good statisticians, they are also good people. My relationship with Vanderbilt provides a strong connection to the academic world and provides valuable support to keep the work I do grounded.
Being a statistician has helped me understand how to look at data and see more than just means; it's about understanding variability and distributions.
What is your best advice for aspiring statisticians?
My advice to aspiring statisticians is to commit to getting details right, whether they are statistical or scientific. It's not good enough to say, "Well, that's good enough." Credibility comes with getting the details right. Learn the science; become conversant with your collaborators; don't accept being just a "Would you like fries with that analysis?" kind of statistician. If your collaborators only want someone to compute p-values and sample sizes, get different collaborators.
Good statistics is about good design. Fancy analyses will not save poor design.
Oh, and if your program takes days and days to run, either your analysis is too complex or your coding is not optimal. Or both. (That'll fire up somebody.)
What are your thoughts on controversial statistical topics such as adjusting the p-value for multiple comparison, the choice between Bayesian, likelihood, or frequentists, ethical issues, etc.?
General statistical rant:
We can be our own worst enemies sometimes because of how we teach statistics. Look at the typical class: roll some dice, deal some cards, flip some coins, probability, conditional probability, here's how to compute a mean and a standard deviation and a median, here's how to do a t-test, f-test, chi-square test, regression, Bayes rule, use this R code to compute this, use this R package to compute that, use this SAS program to get data into a format for submission to FDA: everything we teach them is _technical_.
But if you go to seminar, the discussions aren't technical; they are strategic. "How does this impact the bias?" "What does this do to the Type I error rate?" "How does one present these data in the context of treating patients?"
So, people take our courses and they think that we are technical people doing technical work and they think, "I'll just save a ton of time and money and just download a copy of the super-EZ-stat app for Excel and I'll do this all myself!" Or they think, "All I really need is a sample size and I'm too busy to do it myself; I'll just have the statistician compute that sample size for me when I'm in Boca next week."
But, as we statisticians know, it's much more complicated than that. THIS STUFF IS REALLY HARD. And it takes a lifetime to REALLY understand it. There's a great article out there somewhere by Dick DeVeaux and a buddy called something like "Math is Music, Statistics is Literature" It says that you can be a child prodigy in math and music because they are fields that are completely contained within themselves, so you can get folks like Mozart, who composed as a small child, and Guass, who showed, also as a small child, that the sum of the first n integers is (n)(n+1)/2.
But to write really good literature and to do really good statistics, you need to have life experience. You need to understand the human condition and have seen the world in action.
Statistics isn't technical, even though we teach it that way.
Feel free to do multiplicity adjustments but, better yet, show me the whole distribution of p-values (and not just the mean). How do we deal with correlation? How do we deal with the fact that in my lifetime I have look at thousands and thousands of p-values? What's the impact of that?
P-values and inferential statistics only work in a narrow, forward-looking framework: "If I run this experiment and if the null is true and if the data follow the distribution laid out in my plan, then the chance of a false positive outcome is 5%...". Once the trial is run, everything is subject to all sorts of philosophical nuance. We need to be very careful.
There's a big difference between inference from planned experiments and modeling data from EMRs and the like. Bayesian stuff is great for things like that but not so much for planned experiments.
And let's not think that Bayesian methods, as slick and sexy as the math is, are a panacea for all the ills in our modern stats world.
Ethical issues thoughts:
This should go without saying but let's not lie or cheat or stack the deck or mislead. We are nothing as statisticians if we aren't trustworthy. We shouldn't have to teach that but if we do, so be it.
Tell us about your life outside of Vanderbilt. Do you have a significant other? Children?What about your hobbies and future goals?
I have a wife of 27 plus years; we have a little doggie and live south of Nashville in warm and cheerful Nolensville. We are active members of Holy Family Catholic Church in Brentwood.
We have three children. Harry (b 1993) is a lieutenant in the Army stationed (currently!) in Tacoma, WA. Zach (b 1994) is a second-year law student at the University of St. Thomas in Minnesota. Olivia (b 1996) is an artist and care-giver living in Overland Park, KS. They are wonderful sources of Joy and Light.
I have a couple of things that keep me busy outside of work. We have a very large and well-organized Lego collection and we build things and go to conventions and all that. We are part of a local Lego club that hosts an event every November at the public library in Nashville. I typically build replicas of buildings I like or mechanical computing machines.
I also like to make sawdust in my workshop and the occasional furniture that might result from that.
I would love someday to own and run a working bison ranch but I'm thinking that's not going to happen.
Finally, what is something about you that most people at Vanderbilt still don't know about you?Here's something about me that most people don't know: I applied for the astronaut program at NASA several years ago. I actually successfully navigated the application process and my credentials were reviewed. I was, of course, rejected; but I successfully made it through the process. It was a very interesting experience!
The Vanderbilt Department of Biostatistics proudly welcomes its newest students into the Master's and PhD programs for the 2018-2019 academic year!
In our Master's program, we have Ryan Moore joining us from the College of Wooster, Yan Yan from Middle Tennessee State University, and Yue Gao who recently received her master's degree from Peabody at Vanderbilt.
In our PhD program, Rebecca Irlmeier joins us from the University of Missouri, Julia Thome from Cornell College, Chiara Di Gravio has arrived from the University of Southampton where she was working as a Statistician, and Yi Zuo from University of Boston School of Medicine where she was working as a Lead Statistician.
We are so grateful to have these new faces here on campus. Here's a big congrats to an exciting new chapter in your life!