ADVANCED BIOSTATISTICAL METHODS

Topics:

  1. Modeling Binary Data
    • The logistic regression model
    • Odds ratios and logistic regression
    • Odds ratios in case-control studies
    • The Mantel-Haenszel test for stratified data
    • Adjusting odds ratios for covariates
    • The propensity score (if time permits)

      Click here for the presentation on analyzing rates and proportions.

  2. Analysis of Count Data
    • The Poisson distribution and Poisson process for count data
    • Estimating standardized mortality and morbidity ratios
    • The Poisson regression model

      Click here for a quick review of Poisson regression and of how to fit these models in SPSS.

      Click here for a summary of the main formulas used in standardizing rates.

  3. Hierarchical Models
    • Research designs that lead to hierarchical models
    • Repeated measures analysis of variance
      • Within Subject and Between Subject effects
      • Estimating and interpreting variance terms
    • Longitudinal data and the mixed model
      • Within Subject and Between Subject model components
      • Building models to answer research questions
      • Estimating and interpreting variance terms

      Click here for a brief summary on how to fit mixed models in SPSS.

  4. Survival Analysis
    • Features of survival data: outcomes and censoring
    • Basic notions in modeling survival data
    • The Kaplan-Meier estimator of survival
    • Testing for differences in survival: the log rank test
    • The Cox model for the effect of continuous covariates

      Click here for the presentation on survival analysis.

      Click here for an EXCEL file illustrating the calculations for computing the Kaplan-Meier estimator and for doing the log rank test. The data are from a clinical trial comparing two methods of treatment for acute myeloid leukemia (AML). The file has two worksheets. One worksheet has the data on group, observation time (in weeks) and event status. The second worksheet has the size of the risk sets at each event time, the number of deaths at each event time (total and by group), the expected number of deaths in group 1 (under the null hypothesis of no group differences) and the observed minus expected for group 1.

      Click here for a summary of the Kaplan-Meier estimator and how to compute a standard error for it.

      Click here for slides showing hazard functions for some common distributions.

Return to Course Home Page