Introduction to regression modeling of longitudinal and clustered data from epidemiology and health sciences. Interpretation and familiarity with software gained by analysis of data and critiques of published analyses. Prerequisite: Either BIOST 513, BIOST 515, BIOST 518, BIOST 536, or permission of instructor. Offered: Sp.
Many epidemiological studies produce correlated data: longitudinal studies, cluster-sampled cross-sectional studies, and many ecological designs comparing risks across time or space. This course will examine the sources of correlation, its impact on prediction and estimation, and modifications of linear, logistic and Poisson regression techniques that account for these effects. Particular attention will be given to choosing appropriate regression methods and explaining them to a non-statistical audience, and to identifying problems that cannot be addressed by these techniques and require expert assistance.
Student learning goals
General method of instruction
Lectures, student presentations of readings and projects, class discussion of assigned readings and homework problems.
Students entering this class should understand linear, logistic and Poisson regression models and be able to fit these models to data and explain them to a general epidemiological audience. They should be familiar with standard epidemiologic study designs including cohort studies and matched and unmatched case-control studies and their analysis. Some familiarity with the Stata statistical package will be assumed. BIOST 536 would typically be suitable preparation for thiscourse; BIOST 512 & 513 may be sufficient.
Class assignments and grading
Assigned reading of selected chapters from texts, review articles on statistical methods and articles from medical literature that involve application of those methods. Homework problems will include data analyses, including detailed description and interpretation of results, and critique of the literature. The term project will involve analysis of correlated data of the student's own choice (in consultation with the instructor) and preparation of a short article suitable for submission to a scientific journal.
Accurate completion of homework assignments, performance on midterm and final examinations (in class, closed book and notes), quality of term paper (statement of problem, suitability of methods for problem, appropriateness of interpretation of results, clarity of exposition).