Editor’s Letter – Vol. 25, No. 3
Welcome to the special issue of CHANCE, guest-edited by Jo Hardin. The main features in this issue are accompanied by our regular columns. In Here’s to Your Health, Sowmya Rao and colleagues showcase an impressive application of GPS-enabled mobile devices in conducting a health survey in Sierra Leone. Andrew Gelman writes about ethical dilemmas bound to arise in statistical consulting and journalistic work when dealing with the cigarette industry. Nicole Lazar, editor of The Big Picture, provides an articulated statistical view on Big Data. Abraham Wyner, the guest of this issue’s A Statistician Reads the Sports Pages, addresses a long-standing problem of measuring the skill of players in a game of chance. Howard Wainer reveals the imperfections of some of the methods designed for detecting cheating in examinations. Christian Robert reviews four books, one of which he finds quite “bizarre.” We close the issue with a superbly engaging puzzle by Jonathan Berkowitz.
~Sam Behseta, Executive Editor
The special issue of CHANCE on the culture of statistics in medicine came about after a handful of interesting conversations in my freshman seminar class, “9 out of 10 Freshmen Recommend This Freshman Seminar: Statistics in the Real World.” We talked about the role of statistics in medicine and topics such as common mistakes (e.g., confusion of the inverse), modern challenges (e.g., multiple comparisons with high-throughput data), and the state of statistical education in medical school—pertinent for a large number of students hoping to become doctors.
The classroom conversations eventually made their way to the article I wrote—with the course teaching assistant, Kate Brieger—in this issue. We wanted to convey a sense of importance for the second course in statistics, beyond Stat 101, for medical students. After writing the article, I spoke often about the need for more statistics in the medical community and the idea for the special issue of CHANCE was born.
One of the most substantial issues facing statisticians is reproducibility, and its importance in medicine cannot be overstated. Darrel Ince gives a perspective on reproducibility with respect to -omics data; his examples make clear why and when reproducibility is most needed. Ince provides an excellent list of suggestions for making any data analysis reproducible. We should all take his advice!
Looking at reproducibility through a slightly different lens, Christopher Barr and Jukka-Pekka Onnela discuss openness as an important step in research accountability. They value open-source journals and textbooks for disseminating information and communicating as widely as possible. Additionally, they discuss the trend toward increased data sharing and the resulting effect on reproducible science.
Dalene Stangl and her undergraduate students at Duke University discuss a transformed introductory biostatistics course that seems incredibly effective. Through active learning, decision-theoretic perspectives, and team-based work, the students are able to learn more and deeper topics than the standard lecture-based introduction to statistics. Stangl gives some fantastic examples of activities and readings many of us will be able to use in our classes.
From the other side of the classroom, Mary Kwasny and Constantine Daskalakis discuss challenges teaching introductory biostatistics in a medical center. In particular, they focus on programs in which there is no statistics or biostatistics program associated with the medical school. In such settings, adjunct professors typically teach introductory biostatistics, and those individuals face additional difficulties. Kwasny and Daskalakis provide numerous helpful suggestions to facilitate not only better working environments, but also conversations that might better the situation for all parties involved.
We are fortunate to have the perspective of Xiaohui Zhao, Keith Dowd, and Cynthia Searcy from the Association of American Medical Colleges, who were part of an extensive revision of the MCAT (Medical Colleges Admissions Test); the new version of the MCAT will be given in 2015. Through their article, we see that the changes to the MCAT were based almost exclusively on creating a means to predict success in the medical school curriculum. Unfortunately, they did not endeavor to appraise skills necessary for success after medical school, so the changes to the MCAT include very little of the standard introductory statistics material or beyond.
On the other hand, senior medical researchers were interviewed for this issue, and they universally agree that statistics (introductory and advanced) is vitally important for both medical research and medical practice. In top medical schools such as Harvard, not only students, but also residents and junior faculty, receive additional statistical training to ensure they are able to stay current with the medical field—a skill that takes increasingly more statistical training.
As statisticians, I think it is our responsibility to communicate the importance of good statistical content in the medical literature and within the medical community. With increasing data complexity, our role becomes ever more valued. But there is more work to do, and I hope the articles in this special issue help direct the conversation about the culture of statistics in medicine. I have enjoyed putting together this issue, and I hope you enjoy reading it!