QBS 122: Biostatistics III: Modeling Complex Data

Course Director: Dr. James O'Malley

  • Course description: This course forms the third part of a sequence following the revised QBS 120 (Biostatistics I: Theoretical Foundations) and QBS 121 (Biostatistics II: Modeling). The first component of the course follows QBS 121 by extending regression methods for analyzing data when the data are statistically dependent. This component encompasses clustered, multi-level, longitudinal and other forms of structured data and will focus on hierarchical (mixed-effect) modeling approaches under both a frequentist perspective. However, the consideration of random effects and their conditional distribution given that data will set the scene for the following modules. Bayesian statistical methods are a feature of this course due to their affinity for solving challenging problems and their ubiquity across modern statistical applications. In an extension of QBS 120, Bayesian methodology is carefully developed and compared to the classical (frequentist) approach. A variety of applications in which the Bayesian approach is naturally suited are considered. Bayesian computation via Markov-chain Monte-Carlo (MCMC) is also developed and illustrated. The course concludes network analysis section that includes representation, visualization, and summarization of networks; models of networks; and models of peer effects and social influence processes. Graph partitioning methods will be included if time permits. For more information, click here to view a copy of the past syllabus.
  •  Prerequisites:
    • Coursework: QBS 120 and QBS 121.
    • Programming: Intermediate proficiency in R.