Graphical methods, simple and multiple linear regression; simple, partial and multiple correlation; estimation; hypothesis testing, model building and diagnosis; introduction to nonparametric regression; introduction to smoothing methods (e.g., lowess) The course will include applications to real data.
Prerequisites: Intro Epidemiology and Biostatistics and Perm. Instr
This course discusses the applications of linear regression models to medical research and public health data. We will focus on the two major goals of linear models: (1) explanation, the estimation of associations using linear regression models, and (2) prediction, the use linear regression models to predict subject outcomes, as with diagnostic tests and nomograms. Specific topics include graphical exploratory data analysis, assumptions behind simple and multivariate linear models, the use of categorical explanatory variables, identifying when transformations of explanatory and/or outcome variables are needed, assessing the presence of predictor/outcome associations through hypothesis testing, identifying when confounding and effect modification are present, assessing model fit, and model selection techniques. Prerequisite: Targeted audience members include researchers and health professionals with some basic knowledge of statistics and epidemiology who desire some in-depth exposure to the concepts and principles of linear regression models. This course will include a computer lab where students will gain experience with regression analysis using statistical software.
Site Menu
back to top