2008 Woodie Awards
More math needed on campus and in medicine
Lauren Foley
Issue date: 9/21/06 Section: News
- Page 1 of 2 next >
College students and physicians share an unwelcome bond: both may lack the mathematics training needed to understand journal articles. Two researchers at Smith, Nicholas Horton, assistant professor of mathematics and statistics, and alumna Suzanne S. Switzer '06, found that a typical introductory statistics course may not prepare college students with the skills required to understand the statistical methods used in articles in the New England Journal of Medicine (NEJM).
Horton and Switzer collaborated last summer on a science research project concerning Bayesian methods, a branch of statistics. Their research led them to a set of review articles in the NEJM by Emerson and Colditz on the use of statistics in that journal. In a rapidly changing field, these articles had not been updated in 15 years. Horton and Switzer set about mitigating this deficit.
"The NEJM is a journal widely-read by doctors-not just researchers-that helps to disseminate new medical findings," said Horton, who added, "If [doctors] only read the title and abstract, they may be misled. Since these are our doctors, we should be concerned."
Using a set of categories created by other researchers, Horton and Switzer independently coded 311 articles published in the NEJM between Jan. 1, 2004, and July 1, 2005. They found that since 1979 the number of methods in use has doubled, and that a student in a typical introductory statistics course could easily comprehend only 21 percent of the articles.
Some of the new methods are prevalent in research articles, but are only taught at the intermediate level. One common method, multiple regression, is particularly important because it is used to determine causality. Multiple regression can be used to find the effect, if any, a drug treatment may have on disease, and what the driving force is behind this relationship. "This is a key and subtle concept that requires considerable effort and practice to comprehend," said Horton.
Horton and Switzer collaborated last summer on a science research project concerning Bayesian methods, a branch of statistics. Their research led them to a set of review articles in the NEJM by Emerson and Colditz on the use of statistics in that journal. In a rapidly changing field, these articles had not been updated in 15 years. Horton and Switzer set about mitigating this deficit.
"The NEJM is a journal widely-read by doctors-not just researchers-that helps to disseminate new medical findings," said Horton, who added, "If [doctors] only read the title and abstract, they may be misled. Since these are our doctors, we should be concerned."
Using a set of categories created by other researchers, Horton and Switzer independently coded 311 articles published in the NEJM between Jan. 1, 2004, and July 1, 2005. They found that since 1979 the number of methods in use has doubled, and that a student in a typical introductory statistics course could easily comprehend only 21 percent of the articles.
Some of the new methods are prevalent in research articles, but are only taught at the intermediate level. One common method, multiple regression, is particularly important because it is used to determine causality. Multiple regression can be used to find the effect, if any, a drug treatment may have on disease, and what the driving force is behind this relationship. "This is a key and subtle concept that requires considerable effort and practice to comprehend," said Horton.
Be the first to comment on this story