Description: Many courses in Biostatistics focus on how to analyze data, with little attention being paid to where the data came from and how it was collected. This course focuses on the design of health investigations, with particular attention to the role of randomization in the selection of units and the allocation of treatments. The first part will focus on probability sampling designs and alternatives for the selection of units from a population. The second part concerns study designs for comparing treatments or assessing potential risk factors for health outcomes. These designs include randomized clinical trials, prospective and retrospective observational studies, and clinical data bases. Key concepts include accuracy and precision of estimates, the definition of causal effects, internal validity and the role of measured and unmeasured confounders, and external validity and the role of effect modification on the generalizability of study findings. Examples of randomized and nonrandomized studies will be included to illustrate concepts.
Students will be assigned readings and asked to assess design strengths and weaknesses. Quizzes will be assigned to assess knowledge of the key concepts.
Learning Objectives: (a) Learn key features of probability sample designs -- random sampling, stratification, clustering, multistage sampling. Understand potential limitations of purposive sampling designs, and techniques to reduce the potential bias from such designs
(b) Review the main study designs for the comparison of treatments and potential risk factors for a health outcome, including randomized clinical trials, prospective and retrospective longitudinal studies, case-control studies, analyses of clinical data bases. Understand the strengths and weaknesses of these alternative designs.
(c) Understand how the interpretation of statistical inferences is affected by the choice of study designs.
BIOSTAT842: Seminal Ideas and Controversies in Statistics
Prerequisites: Ph.D. students in Biostatistics, Statistics or related field (e.g. Survey Methodology)
Advisory Prerequisites: None
Description: Statistics has developed as a field through seminal papers and fascinating controversies. Seminal ideas and controversies in statistics will be reviewed and discussed. Students will be assigned to present and discuss key papers, with the aid of later commentaries in the literature that help elucidate the issues. The goal is to expand student's knowledge of the statistics literature and encourage a historical perspective. A draft list of papers, arranged below by topic, is provided; in additional to original papers there are some more recent commentaries that provides a modern perspective. Topics are arranged in three groupings: (a) philosophy of statistics; (b) seminal problems in statistical analysis (c) design topics, focusing on the role of randomization. The instructor will also present summaries of the topics covered.
Students will be assigned homework with a few basic discussion questions about the assigned paper or papers. Also, one "lead presenter" student or students will prepare and deliver a presentation summarizing each topic and paper(s). For the class to work it is essential that students read the assigned material, participate in class discussions, and express their own opinions on the homework questions - often there is not a "right" answer.
Learning Objectives: After completing this class, students are expected to be able to attain the following competencies:
(a) Demonstrate effective written, oral and thinking skills
(1) To learn some key ideas and concepts in statistics concerning philosophy of inference, statistical methods and statistical design, through seminal articles
(b) To learn how to read a research paper and understand the key concepts
(c) To learn how to develop clear and logical written and oral presentations based on reading seminal articles in statistics
(c) To start to develop a personal philosophy for statistical practice
Prerequisites: courses in basic statistics and standard regression
Description: This course provides an introduction to Bayesian methods in epidemiology. Topics include: contrasting the Bayesian and classical approaches to hypothesis testing and interval estimation; strengths and weaknesses of the two paradigms, and when they give similar and dissimilar answers; objective and subjective Bayes; calibrated Bayes, a conceptual approach that combines Bayesian and frequentist ideas; computational tools, including Markov Chain Monte Carlo. the Bayesian approach to some important problems in epidemiology: contingency tables, diagnostic testing, comparison of means, regression, hierarchical models, measurement error, and analysis of data from common study designs.