BMTRY701

BMTRY 701 : Biostatistical Methods II (Regression methods)

DESCRIPTION

The course is intended to focus on biostatistical applications by providing a broad coverage of critical biostatistical applications topics. The primary audience for the sequence is M.S. students (or first-year PhD students that do not have a prior M.S. degree) in biostatistics, but the course will be delivered at a level that it may be taken by graduate students in related scientific fields such as bioinformatics and epidemiology.

TOPICS:

• Simple linear regression (least squares estimation, partitioning sums of squares, hypothesis testing of slope and intercept, model fit, confidence intervals of mean response, prediction intervals)
• Correlation (the correlation coefficient, test of hypothesis and confidence intervals, testes of equality of correlation coefficients, multiple correlation coefficient, partial correlation)
• Analysis of variance (inference based on F-statistic, R-notation, sums of squares, one-way ANOVA)
• Multiple linear regression (general linear model, linear contrasts, testing of general linear hypotheses, confidence intervals, prediction intervals)
• Model Specification (biologic plausibility, interaction, confounding, indicator variables, iterative predictor selection routines)
• Model Diagnostics (diagnostic plots, multicollinearity, residual analysis, influence diagnostics)
• Nonstandard conditions (transformations, heterogeneous variance, weighted least squares)
• Maximum likelihood (principle of maximum likelihood, statistical inference via maximum likelihood, likelihood ratio tests)
• Logistic regression (ungrouped versus grouped data, interpretation, maximum likelihood estimation)
• Poisson regression (maximum likelihood estimation, interpretation, goodness of fit)
• Survival analysis (life tables, Kaplan Meier estimator, log-rank test, proportional-hazards (Cox) model, graphical examination of model assumptions)

PREREQUISITES AND OFFERINGS

BMTRY 700 or instructor consent. Typically offered each spring semester.