Be sure to check out the course overview video.

Statistical Reasoning in Public Health II (140.612.81) provides an introduction to selected important topics in biostatistical concepts and reasoning through lectures, exercises, and bulletin board discussions. The course builds on the material in Statistical Reasoning in Public Health I (140.611.81), extending the statistical procedures discussed in the first quarter to the multivariate realm, via multiple regression methods. New topics, such as methods for clinical diagnostic testing, and univariate, bivariate, and multivariate techniques for survival analysis will also be covered. These topics will be reinforced with many "real-life" examples drawn from recent biomedical literature. While there are some formulae and computational elements to the course, the emphasis is again on interpretation and concepts.

• We will define different types of study designs, including the randomized trial, observations studies and case-control studies.



• We will define confounding and its potential in non-randomized studies, and learn how to assess.



• We will define effect modification (statistical interaction) and methods for investigating it.



• We will learn how to compute necessary sample size for a study to have a desired power-level when comparing two groups on some sort of continuous or binary outcome.



• We will also study linear regression to explore multiple factors affecting a continuous outcome of interest, and we'll discuss in detail the interpretation of the model results and inference on regression estimates.



• We'll use logistic regression to explore the relationship between (potentially) multiple and a dichotomous outcome, such as presence of a disease or death.



• We'll use Cox proportional hazards regression to assess the relationship between (potentially) multiple factors and the time to event occurrence (such as death, or heart attack) in the presence of censored (loss to follow up) data.

John McGready, Course Instructor