Personal HistoryPhD Johns Hopkins University, 1994
MPH University of Michigan, 1983
I am interested in developing relatively simple, easy-to-use, statistical methods that can be useful in biomedical research. Recent examples of such methods can be found in the following papers:
Magder LS and Hughes J. Logistic Regression when the outcome is measured with uncertainty. American Journal of Epidemiology 1997; 146(2):195-203.
Magder LS. Simple approaches to assess the possible impact of missing outcome information on estimates of risk ratios, odds ratios, and risk differences. Controlled Clinical Trials, 2003 24: 411-421.
Magder LS, Fix A. Optimal choice of a cutpoint for a quantitative diagnostic test performed for research purposes, Journal of Clinical Epidemiolgy 2003 Oct;56(10):956-62.
A SAS Macro for implementing the method described in the first paper above can be found at the following link: http://medschool.umaryland.edu/epidemiology/software.asp.
I am also interested in promoting a shift in the view of the role of statistics in biomedical research. Statistical methods are often described in courses and in practice as methods for using data to decide whether to accept or reject hypotheses. In contrast, I view statistical methods as ways to quantify the evidence in a set of data with respect to hypotheses. Given this information, scientists can then weigh all relevant considerations in making a scientific judgement about hypotheses. This shift in thinking about the role of statistics renders many traditional statistical topics (such the use of one-sided versus two-sided tests, or the adjustment for multiple comparisons) irrelevant.
Magder LS and Brookmeyer R. Analysis of infectious disease data from partner studies with unknown source of infection. Biometrics 1993; 49:1110-1116.
Magder, LS and Zeger, SL. A smooth nonparametric estimate of a mixing distribution using mixtures of Gaussians. Journal of the American Statistical Association September 1996; 91:1141-1151
Pinsky PF and Magder LS. Evaluating the tradeoff between bias and variance through the use of prior probabilities. Communications in Statistics: Simulation and Computation, 1997, 26:399-421
Magder LS, Sloan MA, Duh SH, et al. Utilization of multiple sources of outcome information in logistic regression analysis, Statistics in Medicine, 2000, 19: 99-111