Graduate Teaching
- STAT 890 Analysis of Longitudinal Data
-
Course Outline
- Introduction:
- Examples of longitudinal and life history data; review of stochastic models; finite state Markov processes; renewal and Markov renewal processes; point processes and counting processes; regression models.
- Multistate models and prospective cohort data:
- Markov chains. Estimation, tests and model checks, regression models. Multivariate counting processes.
- Other types of longitudinal studies:
- Point process models and extensions: event and count data. Time series with discrete response. Categorical or correlated discrete response models. Model-based and estimating equation approaches.
- STAT 890 Biometrics
-
Course Outline
- Design of medical studies:
- Prospective and retrospective studies, clinical trials. Cross-over clinical trials. Sequential methods. Measures of disease occurrence and association. Confounding. Models used in the analysis of prospective and retrospective studies with binary endpoint: logistic regression models, conditional logistic regression models.
- Seminars:
- There will be four in-class seminars by statisticians working at St. Paul's Hospital, Vancouver General Hospital and the Canadian HIV Network. These statisticians will present and discuss projects in progress in Vancouver in which they are involved. Students will be expected to prepare for these seminars by reading descriptions of these projects, and literature relevant to the projects. The seminars will be followed by a discussion period where students' participation is required.
- References:
- Breslow and Day, Statistical methods in Cancer Research, Vol. 1 - The Analysis of Case-Control Studies IARC 1980. A reading list will be distributed during the first week of lectures.
- STAT 806 Lifetime Data Analysis
-
Course Outline
Statistical methodology used in analyzing failure time data. Basic features of lifetime data. Likelihoods under various censoring patters. Non-parametric and graphical procedures: lifetable methods; non-parametric estimation of a survivor function; the empirical cumulative hazard function; distributional properties of these estimators. Inference using parametric regression models including the exponential, Weibull, lognormal, generalized gamma distributions. Rank tests for equality of distributions.
The proportional hazards family, and inference under the proportional hazards model. Concepts of marginal and partial likelihood. Justification of the proportional hazards likelihood as a marginal or partial likelihood. Stratification and blocking in proportional hazards models. Time-dependent covariates. Regression methods for grouped data. Goodness-of-fit tests. Longitudinal and life history data analysis. Finite state Markov processes. Point processes and counting processes. Continuous and discrete regression models. Multivariate and multi-state models. Time series with discrete response: correlated discrete response models and inference for these models using estimating functions.