This pilot course will focus on selected theoretical and methodologic issues related to the analysis of epidemiologic data with the purpose of drawing causal inference. The topics covered will include long-standing fundamental issues as well as new techniques or novel epidemiologic applications of methods used in other disciplines. The course will consist of 14 three hour sessions. Each session will include a brief didactic presentation of the key issues for the session by the instructor followed by a structured small group and class discussion of a selected reading or readings.
EPID820
Multilevel studies and multilevel analysis in public health research
Winter
term(s)
3 Credit Hour(s)
Instructor(s):
Diez-Roux, Ana
Last offered Winter 2007
Prerequisites: EPID600, EPID601, BIOS560
Multilevel studies and multilevel analysis are increasingly used in the public health field. This course will discuss the rationale for multilevel studies and multilevel analysis in public health as well as differences with other study designs and other analytical approaches. Although the course will not be heavily mathematical, we will review the basics of fitting multilevel models for different types of outcomes as well as the interpretation of estimates obtained from multilevel models. We will also review and critique empirical applications in the health field. The course will conclude with a discussion of causal inference in the context of multilevel reseach questions, including the utility of directed acyclic graphs, propensity scores, and instrumental variables Special emphasis will be placed on the strengths and limitations of multilevel analysis in investigating social and group-level determinants of health. The course assumes no prior knowledge of multilevel analysis, and the focus will be on fundamentals and applications rather than on statistical detail, although knowledge of linear and logistic regression is a prerequisite.
Site Menu
back to top