Description

Important note: this course has been postponed until the second semester. See below for the new dates. 

‘Correlation does not imply causation’, a phrase you’ve likely heard repeatedly during your statistical training. Although a legitimate scientific goal, researchers are often discouraged from asking causal questions because, as editors and reviewers often emphasize, these questions cannot be answered reliably without experimentation or randomization. As a result, researchers commonly use statistical approaches that estimate ‘adjusted associations’, tend to avoid explicit causal language when interpreting these measures, yet often conclude their manuscripts with policy recommendations that are implicitly grounded in causal interpretations.

Advances in the interdisciplinary field of causal inference show that we can do better and that being explicit about causal study goals helps to avoid common errors and reduce bias. For example, the usual—but often unstated—goal of multivariate regression is to produce estimates adjusted for confounders to make causal interpretations more justifiable. Many researchers were, however, never taught that standard regression adjustment practices are usually inappropriate to answer the (usually implicit) causal question(s) of interest and often increase, rather than reduce, bias. In addition, not only confouning bias, but also other forms of bias—less familiar to many researchers—lurk around the corner, even though they can usually be prevented.

This course will focus on common pitfalls and how to avoid them by aligning causal goals with causal reasoning and appropriate causal methods. You’ll be introduced to key theoretical concepts—such as counterfactual outcomes—and practical tools, including causal DAGs, to strengthen causal reasoning. The course also covers the basics of implementing causal methods, such as g-computation, in R.

TARGET AUDIENCE

This course offers the opportunity to researchers in life sciences and medicine to get more familiar with statistical causal methods with expert guidance. The sessions will be interactive and practical.

FORMAT

The course will be offered through five interactive sessions.  A maximum number of 20 participants can subscribe to allow highly interactive sessions.

CONTENTS

- The three tasks of data science: description, prediction and causal inference

- Causal language, counterfactual outcomes and causal estimands

- Causal assumptions and causal DAGs

- Identification versus estimation, the causal gap

- Confounding bias, selection bias, information bias

- Regression adjustment and the Table 2 fallacy

- Marginal versus conditional effects, effect measure scales and collapsibility

- Estimation approaches: g-computation and inverse probability weighting

- A glimpse of more advanced topics (time-dependent confounding, target trial emulation, effect heterogeneity, intercurrent events)

DATES    
The course will take place on the following Mondays (2027): February 8, 15, 22 and March 1, 8 from 09:00 till 12:00. 

Deadline for doctoral school recognition: January 7, 2027

VENUE    
UZ campus - B3 (entrance 46), fifth floor – classroom 5.3 - Faculty of Medicine and Health Sciences of Ghent University, Corneel Heymanslaan 10, 9000 Ghent, Belgium

LANGUAGE    
The course will be taught in English.

PREREQUISITES    
Participants should be familiar with R and understand the basics of multivariate regression. Participants should have access to R and RStudio (installed on their laptop or via Athena).

TEACHER    
Johan Steen

FEES    
The fee for this course is 350,00 EURO. There are three payment options available. Internal through Ugent via invoice and transfers and external through invoicing. More information on the payment options will be available in the confirmation mail. 

For PhD Students - Information on reimbursement by the Doctoral Schools    
First you complete the payment to the Biostatistics Unit. After the course has ended, Doctoral Schools will reimburse this amount, provided that your reimbursement request has been approved and you can submit proof of payment.         
For questions about the application procedure, deadlines, and reimbursement, please contact the Doctoral Schools directly: doctoralschools@ugent.be

CONTACT       
- For questions regarding the content of the course, please contact Johan Steen (johan.steen@ugent.be)

- For administrative questions, you can send an e-mail to secr.biostatistiek@ugent.be

CANCELLATION POLICY