A highly noted statistician visited BYU earlier this month, giving students valuable knowledge for their future careers. Bradley Efron, of Stanford University’s Department of Statistics, spoke at a statistics seminar on April 7. He also toured BYU’s campus, met with several statistics professors, and learned about research going on at BYU.
Evan Johnson, a BYU statistics professor who attended the lecture, said he believes a visit from a famous statistician like Efron helps students get ahead in their education and careers.
“We brought somebody out who is a very high quality statistician to see what BYU is all about,” said Johnson. “[He] got to see the quality of our statistics department.”
Johnson said that BYU’s Department of Statistics believes that a good education includes propelling students forward in their careers. Exposure to important statisticians from around the world helps to accomplish that goal.
“Our major goals [are] to train our students, get them jobs in statistics, and get them into good Ph.D. programs,” said Johnson. “If famous statisticians come, they’ll see how great BYU is and how great our students are.”
Doug VanDerwerken, a graduate student in the statistics department, said he came to the lecture to see how Efron would explain complicated concepts to students who are relatively new to the field.
“I think that an important part of statistics, or any discipline, is being able to explain things,” VanDerwerken said. “I thought he did a very good job of explaining something that’s very difficult in a way that we could understand it.”
During his lecture, Efron described one way statisticians calculate accurate estimates from large amounts of data. The method uses a combination of both the Bayesian and frequentist approaches to statistics, called an empirical Bayes approach.
Efron explained that in many studies, extreme data gets the most attention. As a statistician, he said his goal is to figure out which extremes occur by chance, and then to find an accurate way to estimate what the average is.
“If I only had one or two [pieces of data], I could probably make a correction [using purely frequentist methods], but if I have 1,000 values, that’s impossible,” Efron said. “An empirical Bayes approach actually gets somewhere.”
VanDerwerken said he enjoyed learning about a concept that was new to him. In the future, he said he will know where to go for help when he has problems in statistics like those discussed by Efron.
“I’m not an expert in what [Efron] shared today, but if I ever come across a similar problem I know where I can go to find more information on that,” VanDerwerken said. “If somebody’s already done the work there’s no need for you to do it again.”