Examination arrangement

Course content

The course gives an introduction to statistical methods for studying associations between a continuous or categorical outcome variable and one or more explanatory variables. The course covers correlation analysis and simple and multiple linear and logistic regression analysis. A linear regression model is applicable for continuous outcome variables, and a logistic regression model for binary categorical outcome variables. General principles for estimation and hypothesis testing for unknown parameters in the statistical models will be presented, but the main focus will be on application of the models. This will include model specification, interpretation and presentation of results from the analysis, evaluation of model assumptions, and how to deal with deviation from these assumptions. Important topics to be discussed as part of the model specification are how to handle confounding, and how to allow for sub-group specific effects by using interaction terms in the model. The course also covers methods for evaluating model fit and predictive ability of a regression model (measures of goodness-of-fit, ROC curve analysis). In addition, general methods for variable selection (model selection) will be discussed briefly.

Learning outcome

After successful completion of this course the student should

• have achieved theoretical knowledge on the regression models covered by the course, including principles for estimation and hypothesis testing
• be able to choose an appropriate statistical method and model for evaluation of simple and more complex scientific questions based on analyses of empirical data
• be able to perform the data analyses by means of a statistical program package and be able to interpret the results from the analyses
• be able to choose the most appropriate statistical model in view of the model fit and inherent assumptions on the model or method
• be able to report the results from the statistical data analyses in a scientific medical journal

Learning methods and activities

Lectures and exercise sessions in the first part of the spring semester. Data analyses by means of a statistical program package (Stata and/or SPSS).

Recommended previous knowledge

Introductory course in medical statistics (KLMED8004, MH3003 or equivalent). The students need to be familiar with basic concepts of statistics (continuous and categorical variables, probability and probability distributions), as well as general principles of statistical analysis of empirical data (estimation and hypothesis testing). Some prior training in the use of statistical software is also a requirement for successful completion of this course.

Course materials

Textbook by Rosner, B: "Fundamentals of Biostatistics", 8th ed. 2016.

Textbook by Hosmer and Lemeshow: Applied logistic regression analyses

Applied Logistic Regression | Wiley Series in Probability and Statistics

Learning materials handed out during the course.

Learning materials/text book may be changed.

Credit reductions