Course Linear regression models for continuous and binary data

The course P127/20 Linear regression models for continuous and binary data is being offered by the Graduate School of Health, Aarhus University, 2027.

Criteria for participation: University degree in medicine, dentistry, nursing, or Master’s degree in other fields and/or postgraduate research fellows (PhD students and research-year medical students).

Requirements for participation: Basic Biostatistics (part 1 and 2) or equivalent. Familiarity with the statistical software R is recommended.

Aim: To provide participants with a solid understanding of linear and logistic regression methods for the analysis of continuous and binary outcomes in health science research.

Learning outcomes: 

Upon completion of the course, participants will be able to:

• Apply linear and logistic regression models to analyze continuous and binary outcomes.
• Interpret and critically evaluate the output from regression analyses.
• Assess model assumptions, understand the consequences of assumption violations, and apply appropriate methods to address potential problems.
• Use transformations of predictors, including restricted cubic splines for flexible modeling of continuous variables.
• Evaluate collinearity among predictors and identify influential observations.
• Develop regression models using a model-building strategy based on pre-specification, appropriate model complexity, and subject-matter knowledge.
• Communicate the main results, assumptions, and limitations of regression analyses clearly in scientific papers and academic reporting.

Workload: The full workload of the course is expected to be 60 hours

Content: The course covers interpretation and application of regression models, with a focus on linear and logistic regression for continuous and binary outcomes. Key topics include adjustment for covariates (confounders); assessment of interaction (effect modification); transformation of predictors and outcomes; flexible modeling of continuous variables using restricted cubic splines; identification of influential observations; evaluation of model misspecification; and the use of robust standard errors. Finally, the course presents a model-building strategy based on pre-specification, appropriate model complexity, and careful allocation of degrees of freedom informed by subject-matter knowledge. The statistical software R will be used throughout the course.

Instructors: Morten Overgaard (Department of Public Health)

Venue: Aarhus University, Aarhus

Participation in the course is without cost for:

• PhD students, Health Research Year students from Aarhus University
• PhD students enrolled at partner universities of the Nordoc collaboration
• PhD students from other institutions in the open market agreement for PhD courses

Course dates:

• 26 October 2027 09:00 - 16:00
• 29 October 2027 09:00 - 16:00
• 02 November 2027 09:00 - 16:00
• 05 November 2027 09:00 - 16:00